Armin K.

Thron

Vascular Anatomy
of the Spinal Cord

Radioanatomy as the Key to

Diagnosis and Treatment
Second Edition

123
Vascular Anatomy of the Spinal Cord
Armin K. Thron

Vascular Anatomy
of the Spinal Cord
Radioanatomy as the Key to Diagnosis
and Treatment

Second Edition

With collaboration of
Ch. Rossberg, A. Mironov, M. Mull,
T. Krings, J. Otto, and J.M. Schröder
Armin K. Thron
Aachen
Germany

ISBN 978-3-319-27438-6 ISBN 978-3-319-27440-9 (eBook)

DOI 10.1007/978-3-319-27440-9

Library of Congress Control Number: 2016933135

© Springer International Publishing 2016

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed
to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been
made.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland
Foreword

Between 1977 and 1979, Armin Thron, at that time a young general radiologist, taught himself
neuroradiology. He and the upcoming crème de la crème in neurology (H. J. v. Büdingen,
P. Clarenbach, J. Dichgans, Ch. Diener, V. Dietz, H.-J. Freund, V. Koenig, G. Leopold,
K.-H. Mauritz, J. Noth, G. Oepen, G.-M. von Reutern, U. Thoden, U. Weitbrecht) belonged to
the famous Department of Neurology and Neurophysiology at the University of Freiburg,
Germany. Being a neurosurgical consultant, I had the chance to follow Armin’s metamorphosis
from a common radiologist to a neuroradiologist in this steam boiler of alpha-types led by
Professor Richard Jung.
And yet again, from 1979 to 1987, Armin Thron provided a large number of neurologists
and neurosurgeons (E.H. Grote, W. Hassler, H. Steinmetz, J. Zentner) with his neuroradiologi-
cal knowledge as assistant professor under the directorship of Karsten Voigt in the
Neuroadiological Department at the University of Tübingen. It was there that he started his
experimental and clinical investigations about the vascular blood supply and vascular diseases
of the spinal cord, thus creating the basis for the first edition of his book.
After fulfilling necessary criteria to obtain a leading position, Professor Armin Thron spent
the rest of his professional life, from 1987 to 2010, at the University Hospital of the RWTH
Aachen University. From 1989 until our respective retirements, we cooperated well: he as the
leading neuroradiologist, me as his neurosurgical counterpart.
Every day from 7:30 to 8:30 a.m., we held our neurosurgical-neuroradiological conference
where Armin and his compatriots taught us neuroradiology and we got the chance to defend
our surgical misadventures and the sometimes controversial indications for surgical interven-
tions. To my personal disappointment, many of my neurosurgical co-workers identified these
conferences as the most defining part of their training.
Just as at Freiburg and Tübingen, at the University Hospital in Aachen, Armin again trained
many upcoming leading neurosurgeons (H. Bertalanffy, V. Coenen, F.H. Hans, A. Harders,
V. Rohde, K. Schmieder, U. Sure, E. Uhl, J. Warnke).
Armin and I were both born in 1945 and belong to the 68-generation. Therefore, it was not
surprising that we first had to define our professional and team positions in Aachen before we
could become friends and also enjoy a good and successful professional time together.
For this new version of his book, Professor Thron spent a lot of his sparse private time as a
pensioner in order to help his readers. But as is typical for Armin, he is fully dedicated to the
cause of his work without regard for the achievement of personal recognition.
Thanks to Armin’s favourite topic, spinal vascular disease, we had the chance to come
across numerous interesting cases and most fortunately could help many of the concerned
patients owing to our fruitful neuroradiological-neurosurgical cooperation. Therefore, I highly
recommend that one consider this book not only as a basic textbook but also as a key to micro-
surgical treatments of spinal vascular disorders.

Aachen, Germany Joachim Gilsbach

September 2015

v
Preface

The excitement which the subject represents to us lies in the fact that buildings with what is in principle
an identical function exist in such a variety of diverse forms
Bernd and Hilla Bechers

The idea for a treatise on the radiological anatomy of superficial and deep spinal cord blood
vessels evolved from daily routine neuroradiological work in the early 1980s at the University
of Tübingen. The topic was not induced but promoted by guest stays at hospitals in Paris,
especially at the Neuroradiology Department of Lariboisière Hospital. Jean-Jacques Merland,
the head of the department at that time, and his predecessor René Djindjian were highly
renowned for practising selective spinal angiography at the highest level. This met very well
with my long-lasting interest in diseases of the spinal cord.
Progress in selective angiography of the spinal cord in suspected disease of vascular origin
demanded an advanced understanding of spinal cord blood vessel anatomy and its variations.
Parallel to this development, progress in neurosurgery and especially the rapidly developing
endovascular treatments in interventional neuroradiology improved our knowledge in this field
and broadened the spectrum of treatment options. Significantly ameliorated disease classifica-
tions could be established, reflecting an improved understanding of pathogenetic aspects of
vascular diseases like in cases of arteriovenous malformations, dural arteriovenous fistulas or
vascular tumours of spine and spinal cord.
The first version of this book, Vascular Anatomy of the Spinal Cord, published in 1988, was
a monograph, bearing the subtitle Neuroradiological Investigations and Clinical Syndromes.
Thus, it was subdivided into an anatomical part with post-mortem examinations of arteries,
capillaries and veins of the spinal cord in normal and pathological conditions and in a clinical
section. The latter reflected the progress in diagnostic imaging mentioned above, mainly in
selective digital subtraction angiography (DSA). Computed tomography (CT) was not very
helpful for intraspinal details at that time and magnetic resonance imaging (MRI) was in its
modest beginnings, at least concerning imaging of the spinal cord. Only two figures of this
book from 1988 illustrate a clinical case example using MRI. Myelography using water-
soluble contrast media was still the established standard technique to outline the structures
within the dural sac.
The rapid technical progress in non-invasive imaging techniques in the fields of MRI, MR
angiography or modern CT technologies, but also upgraded DSA units with higher spatial/
contrast resolution, offered new detailed insights into the intraspinal compartments.
Consequently the diagnostic methods illustrating the clinical part of the treatise published
in 1988 are only of historical value today with the exception of selective angiography. Not only
for this reason, we have abstained from including this type of clinical section in this second
edition. We concentrate on what we consider fundamental in this context: vascular anatomy
and its correlation with spinal angiography in normal and pathological conditions.
This field of basic classical anatomy has maintained its essential importance. It has been
established for a long time and has not been the subject of fast and dramatic modifications. The
basic principles are generally well known, but the knowledge of details depends on whether
physicians are able to make use of them for patient care or scientific purposes. But, to give an

vii
viii Preface

example, details about the blood vessels and the blood circulation within and around the spinal
cord could be neglected to a certain degree as long as nobody was in need of this knowledge;
and as long as nobody was able to perform successful diagnostic and therapeutic interventions
based on this knowledge. But during the last 25 years this situation has changed substantially
both for diagnostic and therapeutic modalities.
Nevertheless, a clear conception of the vascular radioanatomy of the spinal cord and its
variations remains an obvious challenge for many physicians, even for those working in disci-
plines which are involved in patients suffering from spinal cord diseases. This is the reason
why the anatomical knowledge presented in the previous edition has maintained if not aug-
mented its significance.
We have tried to simplify the basic principles by using schemes and graphics to facilitate
learning and training by an improved didactic presentation. Anatomical evaluations of angio-
graphic findings in vascular malformations and special notes on dangerous pitfalls or examina-
tion requirements are included in order to address some important problems in the clinical
application of blood vessel anatomy.
The many details presented in the microangiogram section mainly address those concerned
with either scientific questions or with invasive therapeutic techniques and who are familiar
with the interpretation of radioanatomic findings.
A comprehensive description of medullary vascular syndromes would be beyond the scope
of this treatise. It would require a different approach with more interdisciplinary contributions
from physiologists, neurologists, neuroradiologists, neurosurgeons and neuropathologists.

Aachen, Germany Armin K. Thron

January 2015
Acknowledgement

The first group of radioanatomical investigations on the spinal cord were performed at the
Department of Neuroradiology, University of Tübingen, in the 1980s. It is to the department’s
head at that time, Prof. Dr. K. Voigt, that I express my sincere gratitude for stimulating and
supervising my initial scientific projects, and for constant support.
Encouraged by Prof. Dr. W. Dauber from the Department of Anatomy of the University of
Tübingen, we made the first steps to inspect the anatomy of spinal cord blood vessels and to
test techniques of radiographic work-up of the contrast-injected specimens.
Prof. Dr. J. Peiffer, a neuropathologist, and his collaborators from the Institute for Brain
Research at the University of Tübingen also gave valuable support and kindly provided their
laboratory facilities for us.
After I had changed to the University Hospital of Aachen in 1987 to become Head of the
Neuroradiological Department, additional anatomical, functional and clinical studies were
planned and could be carried out during the 1990s thanks to excellent support from the Institute
of Neuropathology. Therefore I owe special thanks to Prof. Dr. J. Schröder, the former Head of
this Institute, for not only providing laboratory facilities for us but also for participating in
cooperative investigations initiated by our department.
Special thanks are further expressed to Mrs. Virginia Müller and Prof. Dr. H. Steinmetz for
the translation of the first edition of the book and Springer International Publishing AG for
copy-editing the final manuscript of the second edition.
Furthermore, I would like to thank Prof. Dr. J. J. Merland (Hopital Lariboisière, Paris) and
Prof. Dr. B. KendaIl (National Hospital, Queens Square, London) for the inspiration gained
through their work and for the fruitful exchange of ideas and friendly personal contact.
To accomplish a scientific project resulting in a monograph and a textbook like the one
presented, you need help from many people.
This is why I want to list those who practically assisted in one way or another in the realiza-
tion of this project with my grateful thanks.

Dr. Christine Rossberg

At the time of the initial neuroanatomical study she worked at the Neuropathological
Department of the University of Marburg (Head Prof. Dr. H. D. Mennel). I owe special thanks
to her for removing and preparing most of the post-mortem specimens which we were able to
evaluate macro- and microangiographically. Her contribution included the first radiographic
documentation of the prepared specimen. She was a member of the Department of
Neuroradiology in Tübingen in the early 1980s before she changed to the Department of
Neuropathology in Marburg. She died in October 2001. Without her practical and professional
assistance, the post-mortem anatomical study could not have been realized, at least not with
this quality of specimen preparations.

Prof. Dr. Angel Mironow

Comprehensive post-mortem and clinical investigations are impossible without the assistance
of colleagues and the exchange of views and experiences on an advanced level. I owe special
thanks to my former colleague Prof. Dr. A. Mironow for his initial help in conceptual consid-
erations and post-mortem preparations.
ix
x Acknowledgement

Walter Korr
He is a former graphic artist of the University Clinic of the RWTH Aachen University.
The concept for the second edition of the book followed the ambitious plan to make the
book useful not only as an instructive atlas for scientific purposes but at the same time as a
textbook or a teaching file for the neuro-specialities. Therefore I was very happy to meet him
and to benefit from his special abilities. His professional help is evident in the computer graphic
design of the diagrams in this book, which I had conceptualized and outlined for didactic pur-
poses over several years.

In the context of our continuous research on anatomy, clinical imaging and endovascular treat-
ment of vascular diseases of the spine and spinal cord at the University Clinic of Aachen, I owe
sincere thanks to the following colleagues and collaborators:

Assistant professor Dr. Michael Mull, and Prof. Dr. Timo Krings
For their highly qualified everyday support and especially for their keen interest in the basic
scientific and advanced clinical imaging projects of the department. Guided by our primary
principles of intellectual honesty in research and patient safety in interventional procedures
they helped to maintain a high level of competence and patient confidence in diagnostic and
interventional challenges over many years.

Dr. Juergen Otto, and Dr. Tibor Jacobs

Were medical students working on a thesis for a doctorate in medicine under my guidance.
They elaborated topics on morphological and functional anatomy of spinal cord veins.
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Historical Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Previous Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Anatomical Methods and Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.3 Injection Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Radiological Examination Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Post-mortem Macro- and Microangiography. . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Clinical Spinal Angiography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.3 CT and Magnetic Resonance Angiography . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Own Investigations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 Contrast Medium and Injection Technique . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 X-Ray Techniques for Standard Images . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.3 Preservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.4 Microangiography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Arterial Blood Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1 Embryology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Segmental Supply to the Spine and Spinal Cord . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Extra- and Intraspinal Extradural Anastomoses. . . . . . . . . . . . . . . . . . . . 10
2.2.2 The Segmental Spinal Branch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Radicular and Radiculomedullary Arteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Superficial Arteries of the Spinal Cord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5 Intrinsic Arteries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.1 Central Arteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5.2 Peripheral System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5.3 Capillaries of the Spinal Cord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.6 Arterial Territories of Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.7 Arterioarterial Anastomoses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.7.1 Superficial Anastomoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.7.2 Transparenchymal Anastomoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.7.3 Intrinsic Anastomoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.8 Specific Patterns of Arterial Supply in the Cervical Spinal Cord . . . . . . . . . . . . 49
2.9 Specific Patterns of Arterial Supply in the Upper- and Mid-Thoracic
Spinal Cord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.10 Specific Patterns of Arterial Supply in the Thoracolumbar
Spinal Cord and Cauda Equina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.11 Anatomical and Haemodynamical Aspects of Arterial Supply . . . . . . . . . . . . . . 79
2.11.1 Anastomotic Capacity of Longitudinal Trunks and Distribution
of Supplying Arteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
2.11.2 Spinal Cord Ischemia and Infarct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

xi
xii Contents

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3 Venous Drainage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.1 Intrinsic Veins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2 Superficial Veins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.2.1 Longitudinal Venous Trunks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.2.2 Pial Venous Networks of Lower Degree . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.3 Transmedullary Venous Anastomoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.4 Specific Patterns of Venous Drainage in the Cervical Spinal Cord . . . . . . . . . . 108
3.5 Specific Patterns of Venous Drainage in the Thoracic Spinal Cord . . . . . . . . . . 110
3.6 Specific Patterns of Venous Drainage in the Thoracolumbar
and Sacral Spinal Cord. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.7 Radiculomedullary Veins and the Transdural Course . . . . . . . . . . . . . . . . . . . . 122
3.8 Extradural Intraspinal and Extraspinal Venous Systems . . . . . . . . . . . . . . . . . . 131
3.9 Anatomical and Haemodynamical Aspects of Venous Drainage . . . . . . . . . . . . 134
3.9.1 Intrinsic Veins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
3.9.2 Intrinsic Venous Anastomoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
3.9.3 Superficial and Radiculomedullary Veins . . . . . . . . . . . . . . . . . . . . . . . 135
3.9.4 Circulatory Disturbances of Venous Origin . . . . . . . . . . . . . . . . . . . . . . 135
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
4 Clinical Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.1 Applications in Normal Spinal Digital Subtraction Angiography
and Interpretation of Uncommon Blood Vessels
on Tomographic Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.2 Applications in SAVMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.2.1 Definition and Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.2.2 Angiographical Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
4.2.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.3 Applications in Spinal Dural AV Fistulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
4.3.1 History and Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
4.3.2 Localization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
4.3.3 Aetiological and Pathophysiological Aspects . . . . . . . . . . . . . . . . . . . . 161
4.3.4 Pathogenesis of Spinal Cord Damage . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.3.5 Angiographical Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.3.6 Diagnostic Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4.3.7 The Question of Chronic Myelopathies in the Absence
of AV Shunt Disorders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4.3.8 Open Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4.4 Applications in Vertebrospinal and Medullary Vascular Tumours. . . . . . . . . . . 181
4.5 Potential Pitfalls in Spinal Angiographic Examinations and Evaluations . . . . . 186
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
About the Author

Armin K. Thron is Professor of Neuroradiology. Until March

2010 he was the Head of the Department of Neuroradiology at
the University Clinic of the University of Technology Aachen,
Germany. From 1996 to 1999 he was the President of the
German Society of Neuroradiology and from 1996 to 2002 he
was the President of the Professional Association of German
Neuroradiologists. He was member of the Board of the
European Society of Neuroradiology from 1993 to 1996 and
President of the XXIX Congress of the European Society of
Neuroradiology 2004 in Aachen, Germany. Prof. Thron has a
special interest in spinal diseases and vascular diseases of the
spinal cord. He is author of various textbook contributions con-
cerning mainly diagnostic and interventional neuroradiology.

xiii
 Introduction
Because of progress in microneurosurgery and interven-
tional neuroradiology, intramedullary lesions have become
more and more accessible and treatable. Unfortunately, a
lack of knowledge about spinal vascular anatomy is evident
in many conferences with neurologists and sometimes even
with neurosurgeons and neuroradiologists. This lack of
knowledge might be a reason for unsatisfactory clinical
results in the treatment of spinal vascular diseases by inva-
sive therapeutic techniques. Furthermore, magnetic reso-
nance imaging (MRI) and magnetic resonance angiography
(MRA) of blood vessels in and around the spinal cord have
substantially improved. To provide a correct anatomical
interpretation of the demonstrated blood vessels, knowledge
of the anatomy of spinal cord blood vessels is the first
prerequisite.
1.1 Historical Background
1.1.1 Previous Studies
Our limited knowledge about the physiology and pathology
of the spinal blood supply contrasts with anatomical
descriptions of the spinal vascular system from as early as
the end of the nineteenth century by Adamkiewicz (1881,
1882) and, with a precision that has hardly been achieved
since, by Kadyi (1889). Earlier studies, such as those by
Duret (1873) and Ross (1880), were less defined, incom-
plete and unsystematic. Until about the middle of the twen-
tieth century, the spinal vascular system was seldom the
object of thorough anatomical research. However, one must
mention the comprehensive works of Tureen (1938) and
Suh and Alexander (1939), who called back to mind the
results of Kadyi (1889).
According to these previous investigations, one ventral
and two dorsal arterial trunks supply the spinal cord. They
© Springer International Publishing 2016
A.K. Thron, Vascular Anatomy of the Spinal Cord: Radioanatomy as the Key to Diagnosis and Treatment,
DOI 10.1007/978-3-319-27440-9_1
1
originate at the cranial end by feeders from the distal seg-
ment of the vertebral arteries, as well as by a varying number
of lateral feeders of different calibre entering with the nerve
roots at different levels. These radicular arteries originate
from segmental arteries or homologous vessels.
On the basis of this anatomical knowledge, Bolton (1939)
carried out injection studies to examine blood flow direction
and vascularization territories of the medulla. As of 1950, the
interest in problems of spinal vascular supply was stimulated
by progress and new questions raised by clinical work. Some
examples of these are neurological complications following
surgery on the aorta and spinal column (Adams and van
Geertruyden 1956; Hogan and Romanul 1966; Adams 1984),
neurological syndromes of assumed vascular origin (Kalm
1953; Zülch 1954, 1976; Gruner and Lapresle 1962) and the
spinal vascular malformations that had become diagnosable
with the introduction of clinical spinal angiography (Djindjian
et al. 1963, 1970; Di Chiro et al. 1967; Doppman et al. 1969).
These clinical challenges have led to an expansion and deep-
ening of our knowledge. Noteworthy are the studies of Gillilan
(1958, 1970), of Lazorthes (1972) and co-workers (1958,
1962, 1966, 1971, 1973), Noeske (1958), Roll (1958), Perese
and Fracasso (1959), Clemens and v. Quast (1960), Corbin
(1961), Houdart et al. (1965), Romanes (1965), Hassler
(1966), Jellinger and Neumayer (1972), Mannen (1966),
Turnbull et al. (1966), Fazio and Agnoli (1970), Manelfe et al.
(1972), Piscol (1972), Dommisse (1975), Tveten (1976a, b, c),
Crock and Yoshizawa (1977), Moes and Maillot (1981), Crock
et al. (1986). Except for a few studies by authors such as Kadyi
(1889), Suh and Alexander (1939), Clemens and v. Quast
(1960) and Crock and Yoshizawa (1977), the arterial system
was more extensively, if not exclusively, examined. Venous
anatomy was focussed upon in the studies of v. Quast (1961)
as well as Gillilan (1970) and in the detailed treatise of Moes
and Maillot (1981) about the superficial veins of the human
spinal cord. Breschet (1828–1832) and later on Batson (1957)
1
2 1
and Clemens (1961) demonstrated quite early the vertebral
venous system and its connection to the veins of the medulla.
Mainly Woollam and Millem (1955), Jellinger (1966a, b) and
Tveten (1976a) conducted comparative animal experiments.
1.1.2 Anatomical Methods and Materials
Anatomical investigative techniques for the spinal vascular
system are based upon preparations using macroscopic-
microscopic inspection, histological serial sections and
numerous methods of vascular injection. The last of these are
especially valuable for demonstrating continuous vessel
courses. In addition to non-hardening staining fluids such as
ink, usually mixed with gelatine (Bolton 1939; Herren and
Alexander 1939; Suh and Alexander 1939; Gillilan 1958;
Corbin 1961; Vuia and Alexaniu 1969; Moes and Maillot
1981), many hardening substances were used as well.
Adamkiewicz (1881) and Kadyi (1889) injected stained gluten
and a Teichman cement of chalk and linseed oil. Other notable
substances are PANSCH-mass (Noeske 1958; Roll 1958), cel-
loidin (Bolton 1939), latex solution (Gillilan 1958; Clemens
1961; Jellinger 1966a; Dommisse 1975) or neoprene latex
solution (Perese and Fracasso 1959), synthetic resin (Lazorthes
et al. 1958; Corbin 1961) or polymerized plastics such as plas-
toid (Gillilan 1958; Piscol 1972). These and other substances
which harden out well facilitate corrosive preparations after
maceration in potassium hydroxide solution.
Not all substances are equally good for filling both large
and small vessels and may easily cause extravasation.
Therefore, different methods were often applied separately,
especially if visualization of the outer and inner medullary
vessels was intended. For the study of vessels within the
parenchyma, Adamkiewicz (1881) and Kadyi (1889) used a
clearing method with clove oil. This technique, further devel-
oped by Spalteholz (1914), renders tissue sections transpar-
ent, so that the stained vessels can be pursued over a certain
length under the microscope by turning the knurled screw.
Even without vascular injection, the superficial medullary
vessels of the exposed spinal cord can be studied both macro-
scopically and with a microscope. This is easy with veins of large
calibre (v. Quast 1961), but can also be applied to the determina-
tion of radicular feeders and the course and calibre of the ventral
and dorsal spinal arteries (Jellinger 1966a; Mannen 1966).
1.1.3 Injection Techniques
As feeding and draining vessels of the medulla originate
from various vascular territories, particular problems are
encountered in injection techniques. No method exists as of
yet which is qualitatively suitable for a complete demonstra-
tion of both the extra- and intraspinal vessels.
Introduction
In principle, the most appropriate technique is considered
to be a global filling of all feeders via the aorta, imitating
natural conditions. This was employed by Gillilan (1958),
Dommisse (1975) and Tveten (1976) for all of the examined
specimens, and by Corbin (1961), Hassler (1966) and
Jellinger (1966a) only for a part. Apart from the unfeasibility
for routine examinations, the large injection volume of sev-
eral litres necessary even after ligation of the peripheral
major arterial trunks poses an additional problem. Gillilan
(1958), Dommisse (1975) and Tveten (1976b, c) reduced
this problem by studying perinatal autopsy cases. The injec-
tion technique is well suited for determining the extraspinal
sources of arterial supply and their collaterals. Corbin (1961)
and Hassler (1966) additionally carried out selective filling
of radicular arteries on the removed spinal cord to demon-
strate the inner spinal vessels.
Adamkiewicz (1881) and Kadyi (1889) had already
applied this selective injection technique. Despite numerous
collaterals on the medullary surface, complete filling is usu-
ally impossible, unless several anterior radicular feeders are
injected simultaneously or successively. This deficiency, on
the other hand, is advantageous for studying vascular territo-
ries on transverse and longitudinal sections if enough frac-
tioned injections can be applied. Nevertheless, only very
restricted conclusions concerning in vivo haemodynamics
can be deduced from post-mortem injection experiments
(Turnbull et al. 1966; Piscol 1972).
Noeske (1958), Roll (1958), and Corbin (1961) discussed
at length the difficulties of an adequate filling technique, yet
the authors rarely quote exact data on the rate of success.
The quality and completeness of filling depend on many
factors:
• The type and quantity of the injected medium
• The specimen’s state of preservation (autolysis, thrombi)
• The occurrence of vascular perforation in the vicinity of
the injection site or intraparenchymal extravasations
• Leakage of contrast medium from small disrupted vessels
in removed spinal cords
• The impossibility of filling very small radicular arteries
For that reason the material for evaluating the outer and inner
microvascular architecture must be assembled like a mosaic
from a large number of cases in order to obtain a complete pic-
ture of all medullary segments. One of the main reasons for
divergent results can certainly be seen in these methodological
problems, which can only be solved at great expense.
The injection techniques applicable for the spinal drain-
age system differ because satisfactory filling of the subarach-
noid veins from outside of the dural sac was supposed to be
impossible. The disappointing results from post-mortem
investigations (Breschet 1828– 1832; Batson 1957; Clemens
1961; Crock et al. 1986; Crock 1996) were supported by
1.2 Radiological Examination Techniques
results of spinal in vivo phlebography with only occasional
simultaneous filling of intradural spinal cord veins (Théron
and Moret 1978; Vogelsang 1980). Consequently, the pres-
ence of valves in the radiculomedullary veins was assumed
and the few existing studies on intradural and intramedullary
veins were based on macroscopic-microscopic inspection (v.
Quast 1961; Jellinger 1966a; Dommisse 1975) or selective
intradural filling of radicular veins (Adamkiewicz 1881,
1882; Kadyi 1889; Suh and Alexander 1939; Lazorthes et al.
1962; Gillilan 1970). Satisfactory post-mortem filling
through arteries and capillaries into the veins has seldom
been achieved (Dommisse 1975).
1.2 Radiological Examination Techniques
1.2.1 Post-mortem
Macro- and Microangiography
Lazorthes et al. (1958) and later Corbin (1961) combined
classical methods and injection techniques with post-mortem
X-ray arteriograms after injecting barium sulphate, thereby
contributing essentially to the radioanatomy of the spinal
vascular system. Several further studies, e.g. by Houdart
et al. (1965), Doppman (1975), Tveten (1976a, b, c),
Doppman and Girton (1977), and Doppman et al. (1979),
made use of the excellent contrast characteristics of barium
sulphate. With its average grain size of 0.5 μm (Treichel
et al. 1977) it is likewise applicable to capillary preparations.
Piscol (1972) carried out flow tests with iodine contrast
media, partially using serial angiography.
The possibility of also demonstrating histological struc-
tures by the use of extremely fine-grained film and soft
X-rays led to the development of microangiography, thus
named by Bellman (1953). This technique of radiological
microscopy and microradiography is based on the physical
investigations of Cosslet et al. (1957) and Engström et al.
(1960). Microangiographical techniques were conducted on
the central nervous system with a focus on the cerebral ves-
sels (Saunders 1960; Lazorthes 1961; Dor and Salomon
1970; Salomon 1973; Duvernoy 1978).
Neuroradiological studies of embryonic cerebral develop-
ment by Voigt and Stoeter (1980) and of experimental
cerebral tumours by Schumacher (1981) employed micro-
angiographical techniques. These works include extensive
surveys of the literature and methodological discussions on
microangiography.
Microangiography of spinal vascular structures has
received only sporadic attention. Saunders (1960) demon-
strated it exemplarily on a foetal spinal cord in the context of
an extensive study on the cerebral vessels. The first systematic,
strictly microradiographical investigations are the studies by
Turnbull, Breig and Hassler (1966) on the cervical spinal cord
3
and by Hassler (1966) on the thoracolumbar region. Yet these
studies only refer to the arterial component. Furthermore, one
must mention the microangiographical investigations of the
spinal dura mater by Manelfe et al. (1972).
The studies of Tveten (1976a, b, c) on the spinal vascular
system in men and rats are partially based on microradiologi-
cal methods. Extended applications of this method can be
found in animal experiments (Doppman 1975; Doppman and
Girton 1976; Doppman and Girton 1977, 1979).
1.2.2 Clinical Spinal Angiography
The present state of selective spinal angiography as a rela-
tively safe diagnostic procedure is a result of methodological
developments in the early 1960s. On the one hand, these con-
sisted of improvements in radiological technology and the
possibilities of image subtraction (photographically or
through digital image processing). On the other hand, global
injection methods were replaced by a selective catheteriza-
tion of segmental nutrifying arteries or their equivalents. In
addition, the application of less neurotoxic contrast media
was facilitated by the introduction of non-ionic and almost
iso-osmolar substances at the beginning of the 1980s.
The now largely standardized technique of selective spi-
nal angiography is based on the pioneering studies of
Djindjian (1978a, b), Djindjian and co-workers (1963, 1975)
as well as Di Chiro and Doppman (Di Chiro et al. 1967; Di
Chiro and Doppman 1969; Di Chiro 1972). Under the pres-
ent technological conditions, only radicular arteries and sec-
tions of the anterior and posterolateral spinal arteries with
calibres of more than 0.2 mm can be demonstrated angio-
graphically. For the demonstration of blood flow directions
in the longitudinal anastomoses, in vivo angiographies of
normal spinal cords are most suitable, as they reflect the
physiological conditions most realistically. According to
Adamkiewicz’s partial flow theory (1881) deduced from
anatomical findings, the bloodstreams of the radicular arter-
ies divide and reach the medullary surface in two partial
streams, one running cranially and one caudally.
Consequently, opposing partial flows approach each other in
the longitudinal anastomosis between two inflow regions.
Spinal angiography is diagnostically essential in arterio-
venous disorders and in highly vascularized tumours of the
spine and spinal cord (Djindjian et al. 1970; Merland et al.
1980; Vogelsang 1980). All surgical procedures on the spinal
column that involve a mechanical alteration of the roots
should be preceded by a preoperative visualization of the
large anterior radicular feeders (Hilal and Keim 1972). On
the other hand, spinal angiography is of limited value for
arteriosclerotic vascular disease or the so-called anterior spi-
nal artery syndrome, although such cases have occasionally
been examined (Di Chiro 1971).
4 1
The major limitations of this method are due to the time-
consuming examination technique as well as to the insuffi-
cient visualization of small vessels and of the venous part of
the spinal circulation. Even in spinal phlebography (Théron
and Moret 1978), a reflux-impeding or at least-modulating
mechanism at the dural or epidural level normally interferes
with free flow of contrast medium from the epidural venous
plexus into the intradural superficial system.
1.2.3 CT and Magnetic Resonance
Angiography
MR angiography of the spine and spinal cord has become an
attractive non-invasive tool for the detection and evaluation
of spinal cord blood vessels in normal and pathological con-
ditions (Mull et al. 2007; Nijenhuis et al. 2006; Backes and
Nijenhuis 2008).
1.3 Own Investigations
For the radioanatomical study of the previous edition
macro- and microangiographies were performed post-mor-
tem on a total of 66 human spinal cords. Following a com-
plete laminectomy, the intact dural sac was excised with
the roots severed as peripherally as possible. The only
selective factor applied to autopsy cases was an effort to
examine space-occupying lesions or spinal vascular dis-
eases. Such pathological cases represented 12 % (8/66).
Consequently, the majority of the examined spinal cords
were chosen from randomly distributed autopsy cases with
varying causes of death. Age distribution had its maximum
in the sixth and seventh decade of life and an average of
64 years (36–84 years). Male cases predominated over
female by 2:1. Classification according to spinal cord seg-
ment and arterial or venous vascular filling shows the fol-
lowing distribution (Table 1.1).
In the context of two doctoral (research) studies, per-
formed in our department of neuroradiology in Aachen, sup-
ported by the department of neuropathology, on the
reflux-impeding mechanism by J. Otto and on the specifica-
tion of radicular and radiculomedullary veins by T. Jacobs,
Table 1.1 Distribution of spinal cord regions and site of contrast injection in the examined specimens
Spinal cord region Site of contrast injection
Arterial Venous
Cervicothoracic 7 4
Thoracolumbar 24 13
Complete spinal cord 7 0
Introduction
another 71 specimens could be evaluated, using different
techniques of examination.
1.3.1 Contrast Medium and Injection
Technique
The contrast medium prepared for filling the unfixated speci-
mens consisted of 8 parts of a 90 % aqueous suspension of
barium sulphate (Micropaque °R) mixed with 2 parts of gela-
tine solution (25 g gelatine dissolved in 100 ml physiological
saline solution at 40 °C) and 4 % phenol. In this dilution, bar-
ium sulphate suspension reaches at least the precapillary arte-
rioles. Tests with a less concentrated suspension were
unsatisfactory as a result of increased contrast medium extrav-
asations and poorer contrast in the X-ray images. Adding gela-
tine (heated to 37–40 °C) hardens the contrast bolus and
prevents contrast medium from leaking when dissecting the
fixated specimens. This was occasionally unavoidable mainly
for superficial veins with wide lumen and thin walls.
After dissection of the dural sac the radicular arteries (or
veins) of largest calibre were localized, probed with a blunt
lymphographic needle and bound. Further radicular vessels of
large calibre in the same filling territory were ligated temporar-
ily along with the cranial end of the cord. Thus, rapid leakage
of contrast medium was avoided, while unligated small calibre
vessels permitted injection into the thereby partially open sys-
tem. Contrast suspension was injected manually under visual
control of filling in the superficial vascular network. Depending
on the filling achieved, additional vessels were probed and
injected in order to procure the most complete vascular visual-
ization possible. On the arterial side, this was most difficult in
the mid and upper thoracic regions. For the upper cervical
region, ligatures were applied to the proximal basilar artery
below the origin of the anterior inferior cerebellar artery
(AICA) and to the posterior inferior cerebellar artery (PICA)
followed by contrast injection in both vertebral arteries.
1.3.2 X-Ray Techniques for Standard Images
Following vascular filling, but before fixation, radiographs
were taken using a soft tissue technique on a mammography
Arterial + venous Total no.
1 12
9 46
1 8
–
66
1.3 Own Investigations 5

device with 25 kV and 0.5 s, or on a Packard Faxitron with Duo2.A.). The paraffin-embedded pieces were suitable for
30 kV and 1.5 min exposure. The images thus obtained allowed microradiography after excess wax had been removed on a
the evaluation of filling quality and were valuable in determin- hot plate. Contact radiographs were obtained with a special
ing the plane of section for subsequent microradiography. microstructure X-ray device (Philips X-ray generator PW
1720) under the following exposure conditions: 30 kV,
23 mA, 6–15 min depending on specimen thickness. The
1.3.3 Preservation film material exposed in direct contact with the specimens
consisted of Kodak spectroscopic plates, type 649–0, with
Specimens were fixed in 4 % formaldehyde for at least especially fine-grained emulsion. The use of consistent pho-
5 days. Deposition in a stretched position prevented defor- tographic enlargement allowed consecutive segments of
mations due to incorrect storage, but could not exclude the frontal or sagittal sections to be joined together without gaps,
effects of erratic shrinking processes. making a longitudinal reconstruction of extensive cord
regions possible.
Inner vessel diameters of the most important arteries
1.3.4 Microangiography and veins were determined from the contact radiographs
in ten different specimens with a PET 2001 series mini-
For microradiography the specimens were cut with micro- computer (Commodore) and a Summagraphics magnetic
tome blades into 1- to 3-mm-thick axial (horizontal), plate (Table 1.2).
frontal or sagittal sections, as seen in Fig. 1.1. The format Histology was obtained from all cases of suspected spinal
of spectroscopic blades used in microradiography allowed cord disease and from the majority of routine autopsy cases
a maximal section length of 2.5 cm. A few spinal cords (provided by Christine Rossberg M.D., Department of
were cut into contiguous sagittal or frontal sections. In Neuropathology, Institute of Pathology, University of
most cases, sectioning alternated in order to obtain differ- Marburg, FRG).
ent views of each spinal cord segment from various
specimens.
The slices were prepared for paraplast embedment by
alcohol and benzene treatment in an Autotechnicon (model
Table 1.2 Range of inner vessel diameters for the most important
arteries and veins of the spinal cord
Spinal cord region
Cervical Thoracic (upper Thoracolumbar
Vessel (mm) two thirds) (mm) (mm)
Ant. rad. med. <0.6 <0.5 <1
artery
Ant. spinal artery 0.2–0.5 0.2–0.4 0.5–0.8
“Narrow 0.1 0.1 –
segments”
Arcades of the <0.3
cone
Post. rad. med. 0.2–0.4 <0.2 0.2–0.5
artery
Posterolat. + post. 0.1–0.2 0.1–0.2 0.1–0.4
spinal artery
Central artery 0.1–0.25 0.08–0.2 0.2–0.26
Penetr. branch <0.05 0.06
(Vasocor.)
Ant. + post. rad. 0.5–1.5 0.4–1.0 <2.0
med. veins
(+median spin.
veins)
Central vein 0.1–0.25 0.08–0.2 0.1–0.26
Radial (peripheral) 0.1–0.25 0.1–0.25 0.1–0.2
vein
Fig. 1.1 Orientation of the sections cut for the microangiographic eval- Midline <0.4 <0.7 <0.2
uation (Reproduced from Thron 1988) anastomoses
6 1 Introduction

References Dommisse GF (1975) The arteries and veins of the human spinal cord
from birth. Churchill Livingstone, Edinburgh
Doppman J (1975) The mechanism of ischemia in anteroposterior com-
Adamkiewicz A (1881) Die Blutgefäße des menschlichen
pression of the spinal cord. Invest Radiol 10:543–551
Rückenmarkes. 1. Teil: Die Gefäße der Rückenmarkssubstanz. S
Doppman J, Girton M (1976) Angiographic study of the effect of
Ber Akad Wiss Wien Math-Naturw Kl (Abt III) 84:469–502
laminectomy in the presence of acute anterior epidural masses.
Adamkiewicz A (1882) Die Blutgefäße des menschlichen
J Neurosurg 45:195–202
Rückenmarkes. II. Teil: Die Gefäße der Rückenmarkoberfläche.
Doppman J, Girton M (1977) Absence of vasospastic response of the
Sitzungsberichte der Akademie der Wissenschaften Wien.
anterior spinal artery to subarachnoid blood. J Neurosurg 47:64–67
Mathematisch-Naturwissenschaftliche Klasse (Abt.III). 85:101–130
Doppman J, Di Chiro G, Ommaya AK (1969) Selective arteriography
Adams CBT (1984) Vascular catastrophe following the Dandy
of the spinal cord. H Green, St Louis
McKenzie operation for spasmodic torticollis. J Neurol Neurosurg
Doppman J, Girton M, Papovski MA (1979) Acute occlusion of the
Psychiatry 47:990–994
posterior spinal vein. J Neurosurg 51:201–205
Adams HD, van Geertruyden HH (1956) Neurologie complications of
Dor P, Salomon G (1970) The arterioles and capillaries of the brain-
aortic surgery. Ann Surg 144:574–609
stem and cerebellum: a microangiographic study. Neuroradiology
Backes WH, Nijenhuis RJ (2008) Advances in spinal cord MR angiog-
1:27–29
raphy. AJNR Am J Neuroradiol 29:619–631
Duret H (1873) Note sur les artères nourricières et sur les vaisseaux
Batson OV (1957) The vertebral vein system. Am J Roentgenol
capillaires de la moelle épiniere. Progr Med 1:284
78:195–212
Duvernoy H (1978) Human brainstem vessels. Springer, Berlin/
Bellman S (1953) Microangiography. Acta Radiol (Stockh) 102(Suppl):
Heidelberg
1–140
Engström A, Cosslett V, Pattee H (1960) X-ray microscopy and X-ray
Bolton B (1939) The blood supply of the human spinal cord. J Neurol
microanalysis. Elsevier, Amsterdam
Psychiatry 2:137–148
Fazio C, Agnoli A (1970) The vascularization of the spinal cord.
Breschet G (1828–1832) Recherches anatomiques, physiologiques
Anatomical and pathophysiological aspects. Vasc Surg 4:245–257
et pathologiques sur les systemes veineux et specialement sur les
Gillilan LA (1958) The arterial blood supply of the human spinal cord.
canaux veineux des os. Villaret, Paris
J Comp Neurol 110:75–103
Clemens HJ (1961) Die Venensysteme der menschlichen Wirbelsäule.
Gillilan LA (1970) Veins of the spinal cord. Neurology 20:860–868
De Gruyter, Berlin
Gruner JE, Lapresle J (1962) Etude anatomo-pathologique des
Clemens HJ, Quast H v (1960) Untersuchungen über die Gefäße des
médullopathies d’origine vasculaire. Rev Neurol 106:592–631
Rückenmarkes. Acta Anat (Basel) 42:277–306
Hassler O (1966) Blood supply to human spinal cord. Arch Neurol
Corbin JL (1961) Recherches anatomiques sur la vascularisation arté-
15:302–307
rielle de la moelle: leur contribution a l’étude de l’ischémie médul-
Herren RY, Alexander L (1939) Sulcal and intrinsic blood vessels of
laire d’origine arterielle. Masson, Paris
human spinal cord. Arch Neurol Psychiatry 41:678–687
Cosslett VE, Engström A, Pattee HH (1957) X-ray microscopy and
Hilal S, Keim H (1972) Selective spinal angiography in adolescent sco-
microradiography. Academic, New York
liosis. Radiology 102:349–359
Crock HV (1996) Atlas of vascular anatomy of skeleton and spinal
Hogan EL, Romanul FC (1966) Spinal cord infarction occurring during
cord. Martin Dunitz, London
insertion of aortic graft. Neurology (Minneap) 16:67–74
Crock HV, Yoshizawa H (1977) The blood supply of the vertebral col-
Houdart R, Djindjian R, Julian H, Hurth M (1965) Données nouvelles
umn and spinal cord in man. Springer, Wien/New York
sur la vascularisation de la moelle dorso-lombaire (Application
Crock HV, Yamagishi M, Crock MC (1986) The conus medullaris and
radiologique et intéret chirurgical). Rev Neurol 112:472–476
cauda equina in man. An atlas of the arteries and veins. Springer,
Jellinger K (1966a) Zur Orthologie und Pathologie der Rücken-
Wien/New York
marksdurchblutung. Springer, Wien/New York
Di Chiro G (1971) Angiography of obstructive vascular disease of the
Jellinger K (1966b) Experimentelle Untersuchungen zur Frage der arteriel-
spinal cord. Radiology 100:607–611
len Versorgungsgebiete des Rückenmarks. Acta Neuropath 6:200–207
Di Chiro G (1972) Development of spinal cord angiography. Acta
Jellinger K, Neumayer E (1972) Chapter 20: Claudication of the spinal
Radiol Diagn 13:767–770
cord and cauda equina. In: Vinken PJ, Bruyn GW (eds) Handbook of
Di Chiro G, Doppman J (1969) Differential angiographic features of
clinical neurology, vol 12. North-Holland, Amsterdam, pp 507–547
hemangioblastomas and arteriovenous malformations of the spinal
Kadyi H (1889) Über die Blutgefäße des menschlichenRückenmarkes.
cord. Radiology 93:25–30
Grubnowicz u Schmidt, Lemberg
Di Chiro G, Doppman J, Ommaya K (1967) Selective arteriography of
Kalm H (1953) Über Entstehung und Lokalisation der Quer-
arteriovenous aneurysms of spinal cord. Radiology 88:1065–1077
schnittslähmung. Dtsch Z Nervenheilk 170:261–273
Djindjian R (1978a) Angiography in angiomas of the spinal cord.
Lazorthes G (1961) Vascularisation et circulations cérébrales. Masson,
In: Pia HW, Djindjian R (eds) Spinal angiomas. Springer, Berlin/
Paris
Heidelberg/New York, pp 98–136
Lazorthes G (1972) Chapter 19: Pathology, classification and clinical
Djindjian R (1978b) Treatment of spinal angiomas by embolization.
aspects of vascular disease of the spinal cord. In: Vinken PJ, Bruyn
In: Pia HW, Djindjian R (eds) Spinal angiomas. Springer, Berlin/
GW (eds) Handbook of clinical neurology, vol 12. North-Holland,
Heidelberg/New York, pp 189–200
Amsterdam, pp 492–506
Djindjian R, Dumesnil M, Faure C, Tavernier C (1963) Angiome
Lazorthes G, Poulhes J, Bastide G, Rouleau J, Chancollé AR (1958) La
medullaire dorsal. Etude clinique et arteriographique. Rev Neurol
vascularisation artérielle de la moelle. Recherches anatomiques et
108:432–434
applications à la pathologie médullaire et à la pathologie aortique.
Djindjian R, Hurth M, Houdart R (1970) L’angiographie de la moe1le
Neurochirurgie 4:3–19
épinière. Masson, Paris
Lazorthes G, Poulhes J, Bastide G, Rouleau J, Chancollé AR, Zadek O
Djindjian R, Houdart R, Hurt M, Cophignon J, Rey A, Thurel J (1975)
(1962) La vascularisation de le moelle épinière. (Etude anatomique
Embolisation dans les angiomes de la moelle. J Neuroradiol
et physiologique). Rev Neurol 106:535–557
2:73–172
References
Lazorthes G, Gouaze A, Bastide G, Soutoul JH, Zadeh O, Sautini JJ
(1966) La vascularisation artérielle du renflement lombaire. Etude
des variants et des suppleances. Rev Neurol 114:109–122
Lazorthes G, Gouazé A, Zadeh O, Santini JJ, Lazorthes Y, Burdin P
(1971) Arterial vascularization of the spinal cord. Recent studies of
the anatomic substitution pathways. J Neurosurg 35:253–262
Lazorthes G, Gouazé A, Djindjian R (1973) Vascularization et patholo-
gie vasculaire de la moelle épinière. Masson, Paris
Manelfe C, Lazorthes G, Roulleau J (1972) Artères de la dure-mère
rachidienne chez l’homme. Acta Radiol 13:829–841
Mannen T (1966) Vascular lesions in the spinal cord in the aged.
Geriatrics 21:151–160
Moes P, Maillot C (1981) Les veines superficielles de la moelle epin-
ière chez l’homme. Essai de systematisation. Archives D’Anatomie
D’Histologie D’Embryologie normales et expérimentales. Extrait
du tome 64. Edition Alsatia, Paris, pp 5–110
Mull M, Nijenhuis RJ, Krings T, Backes WH, Wilmink JT, Thron A
(2007) Value and limitations of contrast-enhanced MR angiography
in spinal arteriovenous malformations and dural arteriovenous fistu-
las. AJNR Am J Neuroradiol 28:1249–1258
Nijenhuis RJ, Mull M, Wilmink JT, Thron AK, Backes WH (2006)
MR angiography of the great anterior radiculomedullary artery
(Adamkiewicz artery) validated by digital subtraction angiography
AJNR Am J Neuroradiol 27:1565–1572
Noeske K (1958) Über die arterielle Versorgung des menschlichen
Rückenmarkes. Gegenbauers Morph Jb 99:455–497
Perese DM, Fracasso JE (1959) Anatomical consideration in surgery of
the spinal cord. A study of vessels and measurements of the cord.
J Neurosurg 16:314–325
Piscol K (1972) Die Blutversorgung des Rückenmarks und ihre
klinische Relevanz. Springer, Berlin/Heidelberg/New York
Roll D (1958) Über die Arterien der Pars caudalis des menschlichen
Rückenmarkes und das Vorkommen arterio-venöser Anastomosen
im Stromgebiet der A. radicularis magna. Gegenbaurs Morph Jb
99:425–454
Romanes GJ (1965) The arterial blood supply of the human spinal cord.
Paraplegia (Edinb) 2:199–207
Ross J (1880) Distribution of the arteries of the spinal cord. Brain
3:80–84
Salomon G (1973) Microangiographic studies of injected brain speci-
mens in the normal and pathologic states. In: Hilal S (ed) Small
vessel angiography. Mosby, St Louis, pp 183–206
Saunders RL (1960) Microangiography of the brain and spinal cord.
In: Engström A, Cosslett V, Pattee H (eds) X-ray microscopy and
microanalysis. Elsevier, Amsterdam
7
Schumacher M (1981) Neuroradiologie experimenteller Hirntumoren.
Neuroanatomie und Tumorpathologie des Rattenhirns. Enke, Stuttgart
Spalteholz W (1914) Über das Durchsichtigmachen von menschlichen
und tierischen Präparaten und seine theoretischen Bedingungen,
2nd edn. Hirzel, Leipzig
Suh TH, Alexander L (1939) Vascular system of the human spinal cord.
Arch Neurol Psychiatry 41(4):659–677
Théron J, Moret J (1978) Spinal phlebography. Springer, Berlin/
Heidelberg/New York
Treichel J, Koeppe P, Dörflinger B, Trüber E, Löwis R v (1977) Klinisch-
radiologische Beurteilung von Bariumsulphatsuspensionen für die
Doppelkontrastuntersuchung des Magens. Fortschr Röntgenstr
127:308–314
Tureen LL (1938) Circulation of the spinal cord and the effect of vascu-
lar occlusion. Res Publ Ass Res Nerv Ment Dis 18:394–437
Turnbull JM, Breig A, Hassler O (1966) Blood supply of cervi-
cal spinal cord in man; a microangiographic cadaver study.
J Neurosurg 24:951–965
Tveten L (1976a) Spinal cord vascularity. The venous drainage of the
spinal cord in the rat. Acta Radiol Diagn 17:653–662
Tveten L (1976b) Spinal cord vascularity. The spinal cord arteries in
man. Acta Radiol Diagn 17:257–273
Tveten L (1976c) Spinal cord vascularity. I Extraspinal sources of
spinal cord arteries in man. Acta Radiol Diagn 17:1–16
Vogelsang H (1980) Angiographische Untersuchungen von Wirbelsäule,
Spinalkanal und Rückenmark. In: Diethe MK, Wende S (eds) Handb
Med Radiologie, vol XIV/I, Neuroradiologie. Springer, Berlin/
Heidelberg/New York
Voigt K, Stoeter P (1980) Neuroradiologie der embryonalen
Hirnentwicklung. Enke, Stuttgart
Quast H v (1961) Die Venen der Rückenmarksoberfläche. Gegenbaurs
Morph Jb 102:33–64
Vuia O, Alexaniu M (1969) Arteriovenous shunt in the spinal cord
circulation. Acta Neurol Scand 45:216–223
Woollam HHM, Millem JW (1955) The arterial supply of the spinal
cord and its significance. J Neurol Neurosurg Psychiatry 18:97–102
Zülch KJ (1954) Mangeldurchblutung an der Grenzzone zweier
Gefäßgebiete als Ursache bisher ungeklärter Rückenmarks-
schädigungen. Dtsch Z Nervenheilk 172:81–101
Zülch KJ (1976) Pathogenetic and clinical observations in spino-
vascular insufficiency. Zbl Neurochir 37:1–13
 Arterial Blood Supply
2.1 Embryology
Only a few investigations and observations exist on the
embryonic development of the spinal vessels in humans.
Ontogenetic studies by His (1887) and more recently by Torr
(1957a, b) and Di Chiro et al. (1973) are supplemented by
detailed phylogenetic studies, especially those by Sterzi
(1904). The examination of early vascularization patterns is
facilitated by the apparent similarity of the embryonic devel-
opment of the spinal vessels in higher vertebrates.
The segmental feeders to the developing spine are branches
of the dorsolateral somatic intersegmental arteries originating
from the aorta. In the thoracic and upper lumbar regions the
intersegmental arteries persist as intercostal and lumbar arter-
ies (Hamilton et al. 1959). In the cervical and sacral regions
the intersegmental arterial pattern is considerably modified by
the development of the vertebral and iliac arteries, respec-
tively. Longitudinal anastomoses, which are formed between
successive intersegmental arteries in the cervical region in a
postcostal position, form a large portion of the vertebral artery.
Precostal anastomoses in the cervical and upper thoracic
regions persist to form the thyrocervical trunk and the supe-
rior intercostal stem. Persistence of the post-transverse anas-
tomoses provides an origin for the deep cervical artery.
The spinal cord vascularization originates from a bilateral
capillary network on the ventrolateral surface of the medulla
connected to the segmental, dorsal branches of the aorta.
After undergoing numerous transformations, two primitive
longitudinal systems are formed at the same site. The capil-
lary networks expand to the ventral and dorsal sides. By the
end of the second embryonic month, the primitive ventrolat-
eral longitudinal systems are transformed into a solitary
anteromedian longitudinal vessel, the so-called anterior spi-
nal artery. This artery was shown by Torr (1957a, b) and later
by Di Chiro et al. (1973) using microangiograms in 10- to
14-week-old embryos to be a quite attenuated vessel.
Although at this time the differentiation of the vertical sys-
tems on the dorsal side already takes place, a plexus-like
© Springer International Publishing 2016
A.K. Thron, Vascular Anatomy of the Spinal Cord: Radioanatomy as the Key to Diagnosis and Treatment,
DOI 10.1007/978-3-319-27440-9_2
2
pattern of vessels remains visible much longer there, as well
as on the ventral side of the upper cervical region (Di Chiro
et al. 1973). According to His (1887) and Torr (1957b), the
formation of the anterior spinal artery is the result of a medial
movement and fusion of the primitive ventrolateral tracts
(analogous to the formation of the aorta), whereas Sterzi
(1904) assumes a segmental and irregularly alternating invo-
lution of the primitive longitudinal tracts.
The development of the anterior spinal artery is followed
by an individually varying regression of the originally 31
bilateral segmental feeders. This numerical reduction is most
pronounced in the lumbar region, where usually only one
artery remains as a ventral feeder. The reduction is com-
pleted around the end of the fourth embryonic month.
Consequently, the final number and distribution of the arter-
ies supplying the medulla is determined by this time.
At the craniocervical junction the modifications of segmen-
tal supply need to be explained in more detail. The terminal
intracranial part of the vertebral artery and its descending
branch to the ventral surface of the upper spinal cord are
homologous to the anterior radiculomedullary artery of the first
spinal nerve (Arteria radicalis anterioris nervi primi; Kadyi
1889; Fig. 2.10). The ascending branch corresponds to the
intracranial end-portion of the vertebral artery, running to the
ventral side of the medulla oblongata. The descending branch
is the cephalad origin of the so-called anterior spinal artery. The
basilar artery has to be regarded as the continuation of this ante-
rior arterial longitudinal tract of the spinal cord towards the
brain stem. The other branches of the vertebral artery are,
together, homologous to a posterior radiculomedullary artery
of the first spinal nerve. The so-called posterior spinal artery,
originating from the vertebral artery or from the posterior infe-
rior cerebellar artery (PICA), corresponds to the descending
branch. The PICA, with branches to brain and medulla oblon-
gata, is homologous to an ascending branch of the posterior
radiculomedullary artery of the first spinal nerve (Kadyi 1889).
To our knowledge there are no studies of the embryonic
development of the medullary veins.
9
10
2.2 Segmental Supply to the Spine
and Spinal Cord
The blood supply to the vertebral body, the paraspinal
muscles, the dura, the nerve root and the spinal cord is
derived from segmental arteries. This is modified but not
essentially changed during embryonic life.
As outlined in Sect. 2.1, the pattern of segmental supply
changes in the cervical and lumbosacral regions. Intrauterine
vascular rearrangements result in the formation of longitudi-
nal arteries in the cervical region. On each side, three vessels
become potential sources of segmental spinal blood supply
in this region, namely, the vertebral artery, the deep cervical
artery and the ascending cervical artery.
In the sacral and lower lumbar regions, sacral arteries and
the iliolumbar artery (supplying the L5 level) derived from
the internal iliac arteries are established for the most impor-
tant supply to the caudal spine. In the majority of the thora-
columbar region, segmental vessels persist as intercostal and
lumbar arteries. The supreme intercostal artery connects sev-
eral segmental arteries in the upper thoracic region. An angi-
ographic evaluation of the entire spinal canal has to
demonstrate all these potential sources of arterial supply.
But parallel to the vascular rearrangements concerning
the pattern of segmental supply and parallel to the forma-
tion of the superficial spinal cord arteries, the embryonic
transformation process also results in a reduced segmental
supply of the spinal cord (Figs. 2.2, 2.7, and 2.8). Therefore,
after the end of the fourth embryonic month most of the
spinal branches do not have a branch supplying the spinal
cord. They only divide into an anterior and a posterior
artery of the vertebral canal, which supply the spinal col-
umn, and a radicular artery, supplying the dura and the
nerve root at this level (Fig. 2.1). According to Manelfe
et al. (1972), this radicular branch can be demonstrated at
almost every segmental level and reflects the initially
strictly segmental pattern of supply. Numerous illustrations
concerning segmental supply of the spine are provided in
the atlases and textbooks of Crock (1996), Crock and
Yoshizawa (1977) and Crock et al. (1986).
2.2.1 Extra- and Intraspinal Extradural
Anastomoses
1. An extraspinal system connects the neighbouring seg-
mental arteries longitudinally. The vessels course on the
lateral aspect of the vertebra or transverse process
(Figs. 2.1, 2.3, and 2.4). This system is highly developed
in the cervical region, where the vertebral artery and the
2 Arterial Blood Supply
deep cervical and ascending cervical arteries form the
most effective longitudinal anastomoses.
2. The intraspinal extradural system is mainly a transverse
anastomosis, but also has longitudinal interconnections.
The retrocorporeal and prelaminar arteries are the rele-
vant vessels for the supply of bone and dura (Figs. 2.1 and
2.3). These anastomoses provide an excellent collateral
circulation and therefore numerous segmental arteries
can be visualized by injection into one segmental artery
(Fig. 2.3). The extra- and intraspinal anastomoses protect
the spinal cord against ischemia when pathologies, such
as arteriosclerotic disease of the aorta, cause focal vessel
occlusion.
2.2.2 The Segmental Spinal Branch
The posterior intercostal (segmental) artery divides into a
dorsal and a ventral branch. In the thoracic region the ventral
branch is running circumferentially under the rib and meets
the anterior intercostal artery. The dorsal branch, occasion-
ally named the dorsospinal artery (Chiras and Merland 1979;
Clavier et al. 1987), has a spinal branch to the vertebral canal
(Fig. 2.1). It enters the vertebral canal through the interverte-
bral foramen and regularly divides into three branches: an
anterior and posterior artery of the vertebral canal, which
supply the spinal column, and a radicular artery, which sup-
plies the dura and nerve root at every segmental level. Dorsal
muscles and bony parts of the spinal column are other struc-
tures supplied by this dorsal branch (Fig. 2.1).
The hemivertebral blush, resulting from injection into a
segmental artery, may help in identifying the artery
(Fig. 2.5c). At the thoracic level, the artery is named accord-
ing to the number of the rib under which it courses.
Occasionally, the proximal part of a segmental artery gives
rise to two (or more) ventral branches of neighbouring inter-
costal arteries, as shown in Fig. 2.5a. This means that the com-
mon stem supplies two intercostal spaces. But as long as only
one dorsal branch can be shown, demonstration of the other
one is missing and the presence of a pseudo-bimetameric
trunk has to be presumed. Consequently, a separate small pos-
terior intercostal artery, from which the spinal branch of the
second metamere arises, has to be looked for. This constella-
tion is demonstrated by Fig. 2.5b. A small seperate artery
from the aorta represents the posterior branch of the T8 seg-
mental artery and supplies the corresponding spinal struc-
tures, examplified by the intense hemivertebral blush, shown
in Fig. 2.5c. This may be crucial when searching for the site of
a dural arteriovenous fistula or the origin of a radiculomedul-
lary artery (Lasjaunias et al. 2001; Mosimann et al. 2014).
2.2 Segmental Supply to the Spine and Spinal Cord 11

Posterior
radiculomedullary Dorsal muscular
artery branch

Spinal branch
Posterior vertebral
canal artery

Ventral branch
Ventral branch

Dorsal branch Anterior

radiculomedullary
artery

Dorsal branch

Anterior vertebral
canal artery
Post. intercostal
artery Anterior
spinal artery

Anterolateral
anastomotic artery

Aorta

Fig. 2.1 Blood supply of spinal column and spinal cord. Schematic
representation in axial view. With regard to the supply of the spinal
(vertebral) canal and it’s content the division of the posterior intercostal
artery in a ventral and dorsal branch is the essential anatomical detail.
The ventral branch follows the wall of the chest anteriorly in the inter-
costal space. Only the dorsal branch further divides into arteries which
supply structures in and around the vertebral canal. Aside of branches
to the dorsal paraspinal muscles a spinal branch regularly enters the
spinal canal. But as a result of the embryonic desegmentation process
the territory of supply is restricted to nerve root , dura mater and bony
structures in the majority of segmental spinal branches. But at certain
levels spinal branches have preserved anterior and/or posterior radicu-
lomedullary branches to the spinal cord as shown on this schematic
illustration (Modified from Thron 2008)
12 2 Arterial Blood Supply

Fig. 2.2 Main sources of

arterial supply to the anterior
surface of the spinal cord.
(a) X-ray in a.p. view.
Infantile autopsy case
following global filling of
the large perispinal arteries
Anterior spinal artery
with barium sulphate
suspension. Only the two Costocervical trunk
(deep cervical artery) Vertebral artery
dominant feeders to the
cervical and thoracolumbar
enlargements of the spinal Thyreocervical trunk
(ascending Segmental arteries
cord can be delineated cervical artery)
(arrows). The artery of the
Subclavian artery
cervical enlargement,
originates from the
thyreocervical trunk. The
artery of the thoracolumbar Carotid artery
enlargement, also called the
great anterior radicular artery
of Adamkiewicz originates Aorta

from T12 on the right side in

this example. Differences in
vessel size between spinal
cord arteries compared to Intercostal
arteries
segmental and paraspinal
sources of supply can well
be appreciated.
(b) Schematic illustration
Great anterior radiculo-
of the potential sources of medullary artery
anterior arterial supply with (Adamkiewicz artery)
accentuation of the main Low intercostal
features of radiculomedullary or upper lumber artery
arteries and the anterior
superficial anastomotic tract, Anastomotic circle of the cone
called anterior spinal artery.
Connection with the Artery of the filum terminale
posterior longitudinal
channels of arteries at the
level of the cone is
delineated with dotted lines
(b Modified from Thron 2008) a b
2.2 Segmental Supply to the Spine and Spinal Cord 13

Fig. 2.3 Extra- and intraspinal anastomoses between segmental

arteries at thoracolumbar levels. Contrast medium was injected
selectively into the 12th intercostal artery. Note the hemivertebral
blush at the level and side of injection. Numerous homolateral as well
as contralateral segmental vessels are opacified through the anasto-
motic channels. The typical hexagonal configuration of the retrocorpo-
real intraspinal anastomosis (large arrows) is seen. The extraspinal
pretransverse or anterolateral anastomoses are smaller in this example
(small arrow) (Reproduced from Thron 2002)

a b

Fig. 2.4 Anastomoses between segmental arteries at upper thoracic
and lower cervical levels. (a) Supreme intercostal artery, supplying sev-
eral upper cervical segments. An anastomosis exists to a muscular
branch of the thyreocervical trunk (large arrow). The arrowhead points
to a retrocorporeal anastomosis. (b) The thyreocervical trunk and its
branches give rise to a cervical radiculomedullary artery (small arrow-
heads). The anastomosis with the supreme intercostal artery is again
marked with a large arrow, as shown in (a) (same patient) (Reproduced
from Thron 2002)
14
Fig. 2.5 Spinal angiography demonstrating pseudo-bimetameric sup-
ply of T9 and T8. (a) One posterior intercostal (segmental) artery (T9)
is the common stem for two ventral branches (large arrow). (b) The
proximal posterior intercostal artery T8 together with the posterior
branch of this segment are originating as a separate small artery from
2 Arterial Blood Supply
the aorta (small branch). (c) Note the vascular blush of the T8 hemiver-
tebra on the digital subtraction image. Concerning arterial supply of
spinal column and spinal cord and its surrounding structures the poste-
rior branch of the segmental artery is the relevant feeder which has to be
looked for (Reproduced from Thron 1988)
2.3 Radicular and Radiculomedullary Arteries
2.3 Radicular and Radiculomedullary
Arteries
Several nomenclatures and classifications have been used to
describe spinal cord arteries. This is an ongoing cause of
misunderstanding. In the past, the main problem occurred
with the term “radicular artery”. It was not clear whether it
meant (a) a blood vessel supplying only a “root” (“radix”) or
(b) an artery supplying a root and going along with the
“radix” to reach and supply the spinal cord. These two
different types of arteries both exist, because of the individu-
ally varying regression of an initially purely segmental sup-
ply of the entire myelomere during embryological life.
Therefore, the introduction of the term “radiculomedullary
artery” is a helpful and anatomically correct specification of
terminology. We therefore suggest this slight modification of
the older anatomical classifications with the following dif-
ferentiation of spinal radicular arteries:
1. Radicular arteries supplying only nerve root and dura
mater, but not the spinal cord. This type of vessel exists at
every level and both sides.
2. Anterior radiculomedullary arteries in which the persis-
tent medullary branch is running with the anterior nerve
root to join the longitudinal trunk, which has been called
the anterior spinal artery (Figs. 2.1 and 2.6).
3. Posterior radiculomedullary arteries in which the persis-
tent medullary branch accompanies the posterior nerve
root and joins the longitudinal systems on the posterior
aspect of the spinal cord. They have been called postero-
lateral and/or posterior spinal (Figs. 2.1 and 2.6).
A classification proposed by Lasjaunias and Berenstein
(1990) differentiates three types of spinal radicular arteries:
radicular, radiculopial and radiculomedullary.
The first type of spinal radicular artery is the small branch,
present at every segmental level, which is restricted to the
supply of the nerve root. The second type supplies the nerve
root and superficial pial plexus (e.g. posterior radicular
artery). The third type is characterized by it’s predominant
ramification in the central grey matter of the medulla.
This classification may offer some advantages for the
interventional neuroradiologist when compared to a classical
anatomical differentiation. It stresses the importance of the
anterior (central artery) supply for the grey matter of the spi-
nal cord parenchyma and the potentially higher risk when
embolization in this territory is performed. From an anatomi-
cal and functional point of view, however, this classification
is not convincing. Some arguments shall be mentioned.
Anterior and posterior radiculomedullary arteries share in
the blood supply not only of the outer fibre tracts (white mat-
ter), but also of the centrally located grey matter. The respec-
tive proportions vary between individuals and between
15
different spinal cord regions. In the upper cervical spinal
cord an anteromedian tract may not be present at all. Instead,
a pial network of superficial arteries may cover the ventral
surface (Figs. 2.35a, c and 2.36b). In other spinal cord
regions only the very central part of the grey matter and the
very peripheral part of the fibre tracts have exclusive supply
from the central or the peripheral system. In an intermediate
zone, branches from both systems may replace each other.
This can be appreciated in Figs. 2.28, 2.29, and 2.30 demon-
strating the areas of supply in the transverse section. Anterior
and posterior longitudinal tracts represent pial arteries
(Kadyi 1889) contributing to the pial network covering the
entire spinal cord. Both systems anastomose with each other,
not only around the cone but also through other superficial
and different types of intrinsic anastomotic interconnections
(Sects. 2.4 and 2.7).
Radiculomedullary arteries feed into the longitudinal
tracts at various segmental levels. The branching of a radicu-
lomedullary artery in a ventral and dorsal feeder at the same
level is the exception in man (Jellinger 1966a). The vessel
pierces the dura ventrocaudally of the nerve root.
The number, calibre and entrance level of anterior and
posterior radiculomedullary vessels were determined by
numerous authors. As Jellinger (1966a) and Piscol (1972)
have shown, statistical evaluation of a big number of cases
and comparisons with previous studies produced relatively
reliably results without essentially diverging from the
description of Kadyi (1889).
Out of initially 31 paired feeders from all segmental lev-
els in early embryonic life only 2–14 anterior radiculomed-
ullary feeders, on average six persist at the end of the fourth
month of embryonic life to reinforce the spinal cord’s super-
ficial arteries on the anterior and anterolateral surface
(Fig. 2.7) (Jellinger 1966a; Piscol 1972).
Kadyi (1889) and Piscol (1972) described an inverse
correlation between number and calibre of the anterior radic-
ulomedullary vessels supplying the entire spinal cord. When
they are few in number (two to five), which is called pau-
cisegmental type, these vessels are of large calibre and they
are concentrated in the cervical and thoracolumbar regions.
The presence of at least six feeders is called plurisegmental
type and results in small calibres (Fig. 2.9).
Dominant radiculomedullary arteries typically supply the
large volumes of grey matter in the enlargements of the spi-
nal cord. The arteria radicularis magna (Adamkiewicz artery)
is the dominant anterior radiculomedullary feeder at the tho-
racolumbar level. It mostly arises at the lower thoracic or
upper lumbar level (T9–L1) and more frequently on the left
than on the right side. The artery of the cervical enlargement
(Lazorthes et al. 1958) arises from the lower thoracic level
(C5–C7) more often as a segmental branch of the ascending
or deep cervical arteries than of the vertebral artery (Figs. 2.7
and 2.34).
16 2 Arterial Blood Supply

Post. and posterolat.

spinal arteries

Post. radic.-medullary
artery

Ant. radic.-medullary
artery

Ant. spinal artery

Fig. 2.6 Schematic illustration of radiculomedullary supply. These the longitudinal anastomotic systems. The tract of the anterior spinal
arteries, restricted to some segmental levels, accompany either an ante- artery is running on the anterior midline. It is supplemented by longitu-
rior or a posterior nerve root or occasionally anterior and posterior roots dinal systems, which may be located in posterior and/or posterolateral
at the same level to reach the surface of the spinal cord and constitute position with regard to the entry of the posterior nerve root
2.3 Radicular and Radiculomedullary Arteries 17

Fig. 2.7 Origin and number of anterior

radiculomedullary arteries, which
are reduced to an average of six segmental Anterior spinal artery
feeders at the end of the fourth month
of embryonic life. This numerical Costocervical trunk
reduction is most pronounced in the (deep cervical artery) Vertebral artery
lumbar region, where usually only one
artery remains as an anterior feeder.
(Modified from Thron 2008) Thyreocervical trunk
(ascending Segmental arteries
cervical artery)

Subclavian artery

Carotid artery

Aorta

Intercostal
arteries

Great anterior radiculo-

medullary artery
(Adamkiewicz artery)

Low intercostal
or upper lumber aftery

Anastomotic circle of the cone

Artery of the filum terminale

18 2 Arterial Blood Supply

Fig. 2.8 Origin and number of posterior

radiculomedullary arteries (schematic drawing).
Posterior spinal artery
The number of these arteries is less drastically
Posterolateral
reduced to 11–16 feeders during the embryonic
spain artery
transformation processes. The sources of Costocervical trunk
supply do not differ from the arteries, which (deep cervical artery) Vertebral artery
reinforce the anterior longitudinal tract. The
complex posterior network of longitudinal and
transverse channels is demonstrated in a
schematic, simplified manner Thyreocervical trunk
(ascending
cervical artery)

Subclavian artery

Carotid artery

Aorta

Intercostal
arteries

Upper lumber artery Low intercostal

or upper lumber artery
2.3 Radicular and Radiculomedullary Arteries 19

Anterior spinal artery

Costocervical trunk
(deep cervical artery) Vertebral artery

Thyreocervical trunk
(ascending Segmental arteries
cervical artery)

Subclavian artery

Carotid artery

Aorta

Intercostal
arteries

Great anterior radiculo-

medullary artery
(Adamkiewicz artery)
Low intercostal
or upper lumbar artery

Anastomotic circle of the cone

Artery of the filum terminale

a b c

Fig. 2.9 Pluri- and paucisegmental supply exemplified on X-rays of
two spinal cord specimens. (a) This specimen is an example of a
rather plurisegmental type of supply with at least six anterior radicu-
lomedullary arteries (open arrowheads). The feeder to the thoraco-
lumbar enlargement entering at the T11 level is considerably larger
than the additional five radiculomedullary contributors to the remain-
ing spinal cord. (b) Three equally large radiculomedullary feeders
(open arrowheads) accomplish blood supply to the entire anterior two
thirds of the spinal cord (paucisegmental type). Posterior radiculom-
edullary and superficial arteries of less size (small arrowhead in a and
b) are connected to the anterior spinal artery by anastomotic circles
around the cone. (c) Diagram showing an average number of radicular
feeders between the extremes of 2–14 (a Reproduced from
Thron 1988, c modified from Thron 2008)
20
2.4 Superficial Arteries of the Spinal Cord
The network of superficial arteries of the spinal pia mater
includes two systems.
1. The anterior and posterolateral/posterior spinal arteries
as the primary longitudinal arterial chains and in
addition
2. The pial arterial networks of lower degree.
The so-called anterior spinal artery and the posterior/pos-
terolateral spinal arteries are not arteries in a common sense.
Instead, they are systems of longitudinal anastomoses, con-
necting the ascending and descending branches of neigh-
bouring radiculomedullary arteries. Therefore, they are
called longitudinal (arterial) tracts or trunks in many text-
books. Kadyi (1889) suggested the term Tractus arteriosus
medulla spinalis for these longitudinal arterial chains
(Additionally, he stated that spinal artery is an inappropriate
term for an artery supplying the medulla and not the spine.)
Nevertheless, we suggest keeping the terms anterior, pos-
terolateral and posterior spinal artery and we will concur-
rently use them in this treatise, as they are commonly used in
clinical practice.
The course and branching pattern of the anterior
radiculomedullary arteries on the ventral surface of the
spinal cord show characteristic features. Before reaching the
midline, the ascending branch is issued (from the radiculom-
edullary artery) and diverges cranially in a slight curve to the
anterior median fissure. The main artery continues on an
ascending course, before it bends in a hairpin curve into the
descending branch. At low thoracolumbar levels of origin
the ascending branch of the radiculomedullary artery is
usually considerably larger than the ascending branch
(Figs. 2.12, 2.15, 2.51, and 2.52).
In cases in which branching occurs distinctly apart from
the midline, the unconcealed segment of the anterior fissure
is penetrated by multiple steeply ascending sulcal (central)
arteries, issuing from the hairpin curve (Figs. 2.14 and 2.57).
This mode of branching and the appearance of a hairpin
course are similarly found in the posterior radiculomedullary
arteries (Fig. 2.16c) and are typical for all radiculomedullary
blood vessels, including veins (Fig. 4.2), which have a steep,
cranially directed course to the spinal cord (as a result of the
ascension of the cord). Consequently, branching in the upper
2 Arterial Blood Supply
thoracic region is more often T-shaped. The difference in
calibre between the ascending and descending branches in
the lumbar region is quite variable and, according to our
observations, also independent of the entrance level, fre-
quency or calibre of the thoracic feeders.
The anterior spinal artery may be very thin or even inter-
rupted as a continuous tract in the upper thoracic and upper
cervical regions of the spinal cord (Figs. 2.12, 2.35, and
2.36). At the lower end of the spinal cord the small artery of
the filum terminale constitutes the direct continuation of the
anterior spinal artery. Additionally, lateral branches form
anastomotic circles between the anterior with the posterolat-
eral longitudinal trunks (Figs. 2.31, 2.51, and 2.52).
Nevertheless, the main source of arterial supply to the
cord is the anterior spinal artery (ventral axis), with a multi-
segmental distribution of blood and a distinct territory of
supply. It gives rise to the hemodynamically important cen-
tral (centrifugal) system, which supplies the major part of the
grey matter. Additionally, the anterior longitudinal trunk pro-
vides branches to the ventral two-thirds of the pial arterial
network of lower degree on the anterior and lateral surface
(Figs. 2.13 and 2.19).
Both systems supply a superficial network of smaller
pial arteries that covers the spinal cord and from which
the peripheral arteries of the centripetal system originate
(Figs. 2.13, 2.14, 2.15, and 2.16).
The posterior/posterolateral arteries do not have such a
distinct territory of supply as the anterior spinal artery, which
means that they predominantly reinforce the rope-ladder-like
network of posterior pial arteries.
The posterolateral tracts are located laterally and the pos-
terior tracts medially to the posterior root entry zone. These
longitudinally oriented vessels may not be continuous over
the whole length of the spinal cord and they may replace
each other. The posterior/posterolateral spinal arteries dis-
tribute blood to the dorsal one-third of this pial network, and
in this way they share with central artery branches in the sup-
ply of the posterior horn and marginal parts of the central
grey matter (Fig. 2.13).
However, there is a significant anastomotic interconnec-
tion between anterior and posterior longitudinal trunks
formed by two anastomotic semicircles around the cone. It
is called the arcade of the cone. The calibre and symmetry
of the two sides are rather variable (Figs. 2.7, 2.31, 2.51,
and 2.58).
2.4 Superficial Arteries of the Spinal Cord 21

a b

Fig. 2.11 Superficial spinal cord arteries of the lower thoracic and
lumbosacral region on the anterior (a) and posterior (b) surface.
(Reproduced from Kadyi’s monograph, 1889; Lower half of Tafel I
Figs. 1 (a), 2 (b). The unsurpassed realistic and instructive anatomical
drawings were designed by himself and realized together with the
painter Heinrich Dyrdon)

a b

Fig. 2.10 Superficial spinal cord arteries of the cervicothoracic region

on the anterior (a) and posterior (b) surface. The anatomical illustration
of the relationship between arteries and the basal cranial and spinal
nerves at the craniocervical junction is very informative. Compare
Section 2.1. (Reproduced from Kadyi’s monograph, 1889; Upper half
of Tafel I Fig. 1 (a), 2 (b). The unsurpassed realistic and instructive
anatomical drawings were designed by himself and realized together
with the painter Heinrich Dyrdon)
22 2 Arterial Blood Supply

Mid-sagittal section
S1

Coronal sections
S3

S2

S1

Mid- and parasagittal sections

S2 S1
S3

a b

Fig. 2.12 X-rays of contrast-filled spinal cord specimens of the cervi-
cothoracic (a) and thoracolumbosacral (b) region in a.p. view. It is dif-
ficult to define localization and contiguity of blood vessels on these
plain films in a.p. view because many structures superimpose each
other. Therefore sectioning of the specimens was performed in axial,
coronal (frontal) and sagittal planes, as demonstrated on the diagrams.
The results are comparable to tomographic imaging techniques and pro-
vide images from different parts of the volume without superimposi-
tion. In this way the three-dimensional course of intrinsic spinal cord
blood vessels can be followed
2.4 Superficial Arteries of the Spinal Cord
Fig. 2.13 Diagram demonstrating the position of the superficial
arteries. The longitudinal tracts are reinforced at certain levels from
radiculomedullary arteries, which accompany an anterior or
posterior nerve root to the surface of the spinal cord. The central
arteries and the pial arterial plexus (vasocorona) on the ventral and
23
lateral surface receive blood from the anterior spinal artery. The pial
arterial network on the dorsal aspect of the cord is supplied by
posterior longitudinal tracts. They differ in their position to the
posterior root entry. The posterolateral artery is running lateral to it
24
Fig. 2.14 Superficial spinal cord arteries. Details of pial arteries at the
entrance of the great anterior radiculomedullary artery. (a) X-ray film of
the injected specimen. The division of the anterior radicular feeder
(large arrow) into the ascending and descending part to the anterior
longitudinal tract may be lateral from the midline. Note the
asymmetrically developed arcade of the cone and the varying develop-
ment of posterior longitudinal tracts in either posterior (small arrow) or
posterolateral position (median sized arrow). (b) Microangiogram of
2 Arterial Blood Supply
b c
Coronal sections
S3
S2
S1
the anterior surface (section S1). The paramedian position of the
ascending and descending anterior tract results in a prolongation of
central arteries to reach the entrance of the anterior fissure. Numerous
branches derived from the anterior spinal artery form a dense superficial
pial network on the anterior surface and on the nerve roots.
(c) Microangiogram of the posterior surface (section S3). Posterior
longitudinal tracts in changing positions and calibres (arrows)
2.4 Superficial Arteries of the Spinal Cord 25

b c

a d e

Fig. 2.15 Superficial arteries of the anterior and the posterior surface
above and below the entry of the great radiculomedullary artery. (a)
X-ray film of the injected specimen in a.p. view. The vertical lines indi-
cate the position and vertical extension of the anterior and posterior sec-
tions shown in (b–e). (b, c) Low thoracic level. Almost equal size of
anterior and posterior longitudinal tracts. (d, e) Lumbar level. Clearly
dominating anterior spinal artery (d) with numerous branches in oblique
or horizontal direction to supply the anterior two-thirds of the superficial
pial network (vasocorona). Posteriorly (e), rope-ladder-like linked tracts
with a stronger posterolateral component (Reproduced from Thron 1988)
26 2 Arterial Blood Supply

Fig. 2.16 Superficial arteries at S3

Coronal sections

S2
the lower end of the spinal cord.
S1
(a) X-ray film of the injected
specimen in a.p. view. In addition
to the great anterior
radiculomedullary artery of
Adamkiewicz, entering the spinal
canal at the T10 level, a posterior
radiculomedullary artery (large
arrowhead) reinforces two
posterolateral longitudinal tracts
of the posterior surface (small
arrowheads). The anastomotic
connection to the anterior spinal
artery at both sides of the cone
occurs at different levels. (b) and
(c) are vertical reconstructions of
frontal microangiograms. (b)
Anteriorly, the superficial pial
branches of the anterior spinal
artery are outlined without
superimposition. (c) On the
posterior surface the rope ladder
like network of pial arteries is
reinforced from a posterior
radiculomedullary feeder (large
arrowhead). The longitudinal
trunks in lateral position (small
arrowheads) are the dominating
arterial structures of the posterior
pial plexus in this specimen. This
figure is a good examples of the
supreme information given by
radiographic techniques which can
be obtained following sectioning
of the injected spinal cord in
different planes (Reproduced from
Thron 1988)

a b c
2.5 Intrinsic Arteries 27

2.5 Intrinsic Arteries tal

git
sa
Ever since the fundamental studies of Adamkiewicz (1881,
1882) and Kadyi (1889), the arteries directly supplying the
spinal cord are divided into a central (centrifugal) and a
peripheral (centripetal) system.
The central (sulcal) arteries constitute the centrifugal
system. They approach from the anterior arterial tract and
pass into the depth of the anterior median fissure. They, pen-
etrate into the myelon on the left or right before the white
commissure and branch tree-like (centrifugally) mainly in
the grey matter of a hemicord (Figs. 2.17 and 2.18).
The peripheral arteries (perforating branches; rami
perforantes) are derived from the pial network covering the
surface of the spinal cord including the surface of the ante-
rior fissure. They mainly consist of numerous small arteries
and pass centripetally through the white matter in a radial axia
l
course (Figs. 2.13, 2.17, 2.18, and 2.22). They are of differ-
ent size and length and have to be regarded as end arteries
like the central arteries. This entire centrifugal system was
designated as the corona vasorum or vasocorona
(Adamkiewicz 1881). Perforating branches do not keep to
borders of different nervous tissue. They may end within the
white matter or go deeper to the peripheral parts of the grey
matter. Vice versa, centrifugal branches of the central arter-
ies may pass the borders of grey matter and additionally sup-
ply white matter. The differences between the architecture of
nervous tissue components in the spinal cord and the archi- Fig. 2.17 Schematic illustration of intrinsic arterial blood vessel
tecture of intrinsic blood vessels was repeatedly mentioned distribution in axial and sagittal plane. It should facilitate the under-
standing of blood vessel configurations shown on microangiograms of
and stressed by Kadyi (1889). They can be confirmed by a slices, cut in different planes. The central and the peripheral systems
big number of microangiograms in this book. can easily be discerned in this drawing. Be aware that the central arter-
ies go either to the right or to the left hemicord. Course of arterioarterial
anastomoses between the central artery branches and the superficial
systems is outlined in the axial plane on top
28
2.5.1 Central Arteries
The number, course, branching pattern and vascular territory
of the central arteries can only be adequately viewed three-
dimensionally. The X-ray visualization chosen here in axial
(horizontal), coronal (frontal) or sagittal sections (Fig. 1.1)
offers the additional possibility of eliminating superimposi-
tion of structures at various depths. In this manner, the deep,
inner vessels of the myelon can be visualized separately
from the superficial systems.
The number of central arteries is most easily determined
in mediosagittal sections of the spinal cord. It varies substan-
tially from individual to individual and in different spinal
cord regions. The same applies to calibre. The data deter-
mined here in contiguous sagittal sections range from 170 to
210 central arteries per spinal cord and are somewhat lower
than generally assumed until now. A false calculation from
lateral projections could be due to superimposition of arter-
ies going to left and right at the same level. However evalua-
tion of consecutive frontal sections produced identical
results. The observed discrepancy may be explained by the
inclusion of very small arteries of the cone region and by
inclusion of small arteries to the anterior white column next
to the anterior fissure, which actually are branches of the
vasocorona or of the central arteries within the anterior
fissure.
The number of central arteries is 60–80 for the lumbosa-
cral region (length 9–11 cm), 60–70 for the thoracic region
(length 20–22 cm) and 50–60 for the cervical cord (length
10–12 cm). As shown impressively by Fig. 2.26, the densest
sequence of side-alternating central arteries is found in the
region of the enlargements with 7–12 arteries per centimetre.
In the thoracic region there are only 2–3/cm. This equals an
average of 5–9 per segment.
Determining the angle between the sulcal arteries and the
anterior spinal artery is not essential, because deviations are
numerous and the course may change from cranial to caudal
direction even among central arteries issued from the same
ascending or descending branch.
2 Arterial Blood Supply
The essential features established from evaluation of our
material can be summarized by the following statements:
• Spinal cords showing a straight course of the central arter-
ies and those showing a very tortuous course in the sagit-
tal plane are found with almost equal frequency
(Figs. 2.58, 2.59, and 2.60). Whether the second type rep-
resents a primary condition or a secondary vessel elonga-
tion cannot be discerned because of the high average age
of our patients.
• In relation to the longitudinal axis of the cord (or the ante-
rior spinal artery), the central arteries can have an ascend-
ing, horizontal or descending course. Their course in the
region of the enlargements is predominantly horizontal
(more likely ascending). The mid- and upper thoracic
regions mainly exhibit clearly ascending vessels (Figs. 2.24
and 2.25) while descending vessels are found sporadically
scattered in all segments (Fig. 2.47). A high entrance level
of the great radiculomedullary artery was observed in
some specimens showing long descending vessels and
vice versa (Figs. 2.59 and 2.60).
• Steeply ascending and very long central arteries, which
parallel the course of the anterior spinal artery, can be
found at the paramedian bifurcations of large anterior
radiculomedullary feeders (Fig. 2.57). Similar features
are displayed by the central arteries in the entire thoracic
region, where they can ascend parallel to the anterior
spinal artery up to 1 cm, before they bend horizontally
into the anterior median fissure (Figs. 2.46 and 2.47).
• The vessel calibre of the central arteries varies consider-
ably intra- and interindividually (Fig. 2.48), e.g. between
0.1 and 0.2 mm in the region of the lumbar enlargement
(Table 1.2).
As mentioned above, the central arteries bend either to the
right or to the left into the medulla and do not ramify in the
depths of the fissure into bilateral branches as described at
one time. This false impression is produced in axial sections
when neighbouring central arteries overlap or become
2.5 Intrinsic Arteries
projected upon one another (Figs. 2.18 and 2.53). The true
situation can be easily assessed in oblique sections of the
central artery’s origin and in frontal views: a short, common
trunk may be found; however, a course alternating from side
to side predominates (Figs. 2.19 and 2.21). This fact, too,
reflects the originally bilateral, symmetrical arrangement.
The horizontal and vertical arborization of the central arter-
ies within the spinal cord can be readily recognized in micro-
radiographic sections. Horizontal branches of close succession
are predominant in the enlargements while thoracically the
steeply ascending and descending branches of large calibre pre-
vail (Fig. 2.25). Inspections of our early cord preparations led
to the impression that the ascending and descending branches
could join to form an inner longitudinal anastomosis, which
Adamkiewicz (1881) assumed to be the paracentral artery. In
the course of this study, however, careful analysis of vascular
branching in frontal and sagittal sections corroborated Kadyi’s
findings (1889) of a close intertwining and overlapping of the
terminal branches. As shown by the microradiographs of all
spinal cord segments, a continuous inner longitudinal anasto-
mosis does not exist, the same is true for significant anastomo-
ses between two neighbouring central arteries on the arterial or
arteriolar level (Figs. 2.20, 2.23 and 2.26).
29
The horizontal distribution of central arterial branches is
demonstrated best in axial sections (Figs. 2.18, 2.22, and
2.37). After issuing smaller branches to the anterior white
column, located next to the anterior fissure, and to the ventral
marginal zones of the anterior horn, and after entering the
grey matter, the main stem branches out fascicularly into the
anterior horn, the lateral column, the central gelatinous mat-
ter and the central components of the posterior horn.
In view of their great variability, denomination of the
individual branches or branch groups is not essential.
Isolated branches of this ramification also reach neighbour-
ing fibre tracts, yet locally and individually to a very vari-
able extent.
Whereas consecutive, approximately 2-mm-thick axial
sections of the enlargements each contain at least 1–2 central
arteries whose horizontal branching can be examined, many
thoracic sections show none or only a short vessel transec-
tion in the ventral fissure (Figs. 2.37 and 2.45). This becomes
clear when inspecting the vessel courses in sagittal sections
(Figs. 2.46 and 2.47). A picture of the spatial distribution of
these arteries, which are vital for the supply of the medullary
grey matter, is provided by microradiography in different
sectional planes (Figs. 2.23, 2.27, and 2.50).
30
Fig. 2.18 Superficial and intrinsic arteries of the spinal cord.
Microangiogram of a transverse section at the L2/3 level. Three
millimetre slice thickness. Several superimposed sulcal arteries pass to
one side or the other of the cord and branch out mainly in the grey
matter. Additional small vessels to the funiculus anterior and the cornu
anterius originate in the anterior median fissure as branches of the
central arteries or as branches from the pial plexus of lower degree
2 Arterial Blood Supply
(arrow). Note the variability of branches supplying the posterior horns.
They may be derived from the central arteries (centrifugal system) or
from the pial arterial network on the dorsal aspect of the cord (centrip-
etal system), which is supplied from the posterior longitudinal tracts.
The pial arterial plexus on the ventral and lateral surface receives blood
from the anterior spinal artery (see section 2.6) (Microangiogram
reproduced from Thron 1988)
2.5 Intrinsic Arteries 31

Coronal sections
S3
S2

S1

a b

Fig. 2.19 Superficial and intrinsic branches derived from the anterior
spinal artery. (a) Coronal microangiography of the anterior surface. The
anterior spinal artery has lost its previous content of contrast medium.
The stems of the central arteries, running into the anterior fissure
(arrows), are therefore not covered by the arterial midline tract.
Extension of the superficial network of pial branches derived from the
anterior spinal artery includes the depth of the anterior fissure. (b)
Coronal microangiography of the midline section. Ramification of the
central arteries in horizontal and vertical direction mainly within the
grey matter of the anterior horns. The small arteries in the outer fibre
tracts are penetrating branches derived from the lateral parts of the
vasocorona
32 2 Arterial Blood Supply

Mid-sagittal section
S1

Fig. 2.20 Intrinsic arteries of the spinal cord. Microangiogram of a

mid-sagittal section at the level of entry of the great anterior radiculom-
edullary artery (arrow). The central arteries take a horizontal or slightly
ascending course (arrowheads). In this plane the vertical distribution of
long central artery branches near the central canal can be demonstrated
very well and the perforating arteries within the posterior columns can
be appreciated in their typical tree-like configuration (Reproduced from
Thron 1988)
2.5 Intrinsic Arteries 33

R
Cervical
enlargement

mid-coronal
sections S 2

Coronal sections
S3
S2

S1

Fig. 2.21 Intrinsic arteries of the spinal cord. Microangiogram of a
median coronal section at the C4–C5 level. The central arteries (small
arrowhead) show an irregular side-alternating distribution. In this
median part of the cervical enlargement the number of central arteries
is relatively high, the vertical ramification is therefore only moderate.
The branches of neighbouring central arteries intertwine, but do not
form inner anastomotic interconnections. Superficial arteries of the pial
network (large arrowhead) provide blood supply through the centripe-
tal system of small perforating arteries (Reproduced from Thron 1988)
34
a b
Fig. 2.22 Superficial and intrinsic arteries in a specimen presenting
with an optimal result of contrast filling. (a) X-ray film of the injected
spinal cord specimen in a.p. view. The great anterior radiculomedullary
artery and two posterior radiculomedullary feeders are connected to
each other by the anastomotic arcades around the cone. (b, c) Axial
microangiograms of mid-lumbar (b) and lower thoracic level (c). There
is an equally good distribution of the contrast medium to all vascular
territories. Even the very small precapillary arteries are opacified.
2 Arterial Blood Supply
Differentiation of vascular territories of supply is not possible. Posterior
longitudinal tracts are situated medially, posterolateral ones laterally of
the posterior root entry zone. These tracts are supplemented by smaller
ones (arrowheads) in anterior, lateral, or posterior positions. They con-
stitute together with additional smaller branches the pial arterial net-
work of lower degree (b Reproduced from Thron 2002; c Reproduced
from Thron 2008)
2.5 Intrinsic Arteries 35

Fig. 2.23 Intrinsic arteries at a b d

thoracic level.
Microangiograms of median
coronal sections. (a) X-ray
film of the injected spinal cord
specimen in a.p. view. The
vertical lines indicate the
position of the sections shown
in (b–e). There are small gaps,
caused by axial slices, as
indicated. (b–e) The number
of central arteries is consider-
ably smaller, the vertical
extension of the central artery
branches considerably larger
compared to the blood supply
in the enlargements.
Arterioarterial anastomoses
between central arteries and
the superficial system
(arrowheads in c and d) are
frequent findings. But intrinsic
longitudinal anastomoses
between central artery
branches are not verifiable.
Supply from the posterior
superficial system is much
stronger than in other parts of
the spinal cord (Reproduced
from Thron 1988)

c e

Coronal sections
S3
S2

S1
36 2 Arterial Blood Supply

Mid-sagittal section
S1

b d

a c e

Fig. 2.24 Intrinsic arteries. Course of central arteries throughout the
length of cervicothoracic spinal cord (Figs. 2.24 and 2.25). (a) X-ray
film of the injected specimen in a.p. view. (b–e) Microangiograms of
mid-sagittal sections at cervical and upper thoracic levels. The sections
c and d are contiguous. The course of the central arteries is tortuous and
slightly ascending in the cervical region (b, c) and becomes increas-
ingly steeper in the upper thoracic levels (d, e). Solitary or intercon-
nected central arteries run parallel to the anterior spinal artery over
some distance before turning into the depth of the anterior fissure.
Compare Fig. 2.25b
2.5 Intrinsic Arteries
Mid-sagittal section
S1
b
a c
Fig. 2.25 Intrinsic arteries. Course of central arteries throughout the
length of cervicothoracic spinal cord Continuation of Fig. 2.24. (a)
X-ray film of the injected specimen in a.p. view. (b–e) Microangiograms
of mid-sagittal sections at upper to mid-thoracic levels. The sections (c
and d) are contiguous. In (b), the contrast medium has been lost in a
part of the anterior spinal artery. This reveals the primarily ascending
37
e
f
course of the small number of central arteries in the pial covering of the
anterior fissure of the upper thoracic spinal cord (b, c). In the mid and
lower thoracic cord central arteries may be more numerous but small in
size, often matched by a large number of large perforating arteries from
the vasocorona (d, e)
38 2 Arterial Blood Supply

S2 S2 S2

C5 T5 L3
Coronal sections
S3
S2

S1

Fig. 2.26 Intrinsic arteries. Microangiograms from mid-coronal sec-
tions including the grey matter of the anterior horns. The sections were
obtained from the cervical enlargement (cranial to C5), the upper tho-
racic region (cranial to T5) and the thoracolumbar enlargement (cranial
to L3). Section length 2.5 cm, slice thickness 2–3 mm. Note the differ-
ences in the spatial distribution of central arteries and the preferential
plane of their ramification. Furthermore, size course and degree of over-
lap of central arteries and their branches can be compared at these lev-
els. The number of central arteries is about 16 per 2.5 cm in the cervical,
9 per 2.5 cm in the thoracic and 26 per 2.5 cm in the lumbar region.
Anastomotic connection between the central system and the superficial
pial network at the T5 level (Reproduced from Thron 1988)
2.5 Intrinsic Arteries
2.5.2 Peripheral System
The posterior longitudinal anastomoses (posterolateral and/
or posterior spinal arteries) function as a superficial pial sup-
ply system of first degree. Numerous superficial branches of
the anterior spinal artery and the longitudinally directed
network of posterolateral and/or posterior arteries feed a
dense pial network of lower degree, from which the perforat-
ing branches of the peripheral system arise. This is very well
demonstrated on microradiographs of frontal and sagittal
sections (Figs. 2.14, 2.20, 2.55, and 2.58). The larger vessels
of this pial network branch irregularly into ascending and
descending vessels that can join the course of the roots
(Figs. 2.14, 2.15, and 2.16). They, however, end after a few
centimetres. These superficial vessels of second and third
degree should not be specially designated as anterolateral or
lateral arteries. Instead, they should be assigned to this part
of the pial network, which supplies the peripheral system of
penetrating branches. Especially above the enlargements,
these smaller vessels form an impressive secondary anasto-
motic system between the anterior and posterior longitudinal
anastomoses, although no significant anastomotic capacity
can be attributed to the individual rami.
The penetrating branches appear angiographically as a
largely uniform, radial vascular system. The almost exclu-
sively thin arteries with calibres between 10 and 50 μm ter-
minate in the external fibre tracts or in the marginal zone of
the grey matter. Longer branches of larger calibre exist only
on both sides of the posterior root’s entry zone and as
branches of the anterior system to the anterior column and
anterior horn (Figs. 2.19, 2.37, and 2.53). Even oblique
39
longitudinal sections, which follow the course of a postero-
lateral artery over several centimetres, do not display larger
constant branches going directly to the cord. For this reason
there is no need to develop such an extensive nomenclature
for the vasocorona, as several anatomists have done, neither
for its superficial nor its intramedullary components. From
an angiographic-morphological standpoint, the superficial
longitudinal and transverse connections attributed to the pial
systems of lower degree do not suggest the function of sup-
plying selective territories via larger branches. Rather, they
guarantee a constant blood distribution from the periphery.
2.5.3 Capillaries of the Spinal Cord
Since only the diameter of injected vessels can be assessed in
microradiographs, the exact transition of arterioles into cap-
illaries and venules cannot be accurately determined.
Nevertheless, evaluation of all specimens showing good,
simultaneous filling of capillary vessels (Figs. 2.27 and 2.28)
provides an impressive visualization of the difference in cap-
illary density between grey and white matter, as has been
shown by histological studies. The capillary networks are
densest within the grey matter of the enlargements. In accor-
dance with the microradiographical findings of Turnbull
et al. (1966), no unequivocal difference between anterior and
posterior horn was detectable here. In contrast, the capillar-
ization of the fibre tracts is much more loosely meshed. The
occurence of capillaries in the spinal pia mater was exam-
ined exhaustively by Kadji (1889). He denied the existence
of own pial capillaries insistently.
40 2 Arterial Blood Supply

Fig. 2.27 Capillary density

(a) X-ray of the specimen in X-ray lateral
lateral view. Low entry of the
great anterior Ant.
radiculomedullary artery
(arrow). Poor filling of the
posterior superficial pial
network. (b–d) Axial c,d
microangiograms at the levels,
indicated in a. There is a
striking difference in the
density of small pre-capillaries
and capillaries between gray
and white matter, not only in
the lumbosacral enlargement
(b), but also at the T9 level
(c, d). The artery of the T9
posterior median septum,
d asa
opacified in c and d
(arrowhead), is filled from the
anterior spinal artery. asa
Abbreviation: asa, anterior
spinal artery (as empty vessel
in b–d)

b c asa

L5

a b asa
2.5 Intrinsic Arteries 41

a b

Fig. 2.28 Capillary density and areas of supply. (a–c) Axial microan-
giograms from thoraric levels. Distinctly higher density of arterioles
and capillaries in the grey matter of the spinal cord compared to the
white matter. The contribution of the anterior spinal artery to the pial
system in the anterior two thirds is very well shown. Due to the low
proportion of gray matter, central arteries do not play a major role for
the supply of the cross sectional area in this region. The artery of the
posterior median septum (arrowhead in a and b) is filled from the
anterior circulation. Pial branches to the anterior funiculus and anterior
horn originate from the anterior fissure (arrowhead in c)
42
2.6 Arterial Territories of Supply
Demonstration of the intramedullary territories of arterial
supply can be achieved by demarcating the central from the
peripheral supply system. This can also include a separation
of the area fed by the anterior or posterior spinal arteries
(Fig. 2.30).
Figure 2.31 illustrates three axial sections of a cord at
different levels, together with their microangiographic
images. It is clearly shown that the cross-sectional areas
differ considerably in size and that the proportion of grey
matter is variable. Determining the distribution area of the
intrinsic arteries allows a basic pattern to be observed. The
ramification territory of the central arteries corresponds
rather closely to the proportion of grey matter. In the region
of the thoracolumbar enlargement the central artery branches
cover a third to a half of the sectional area. Their contribution
decreases thoracically to a fifth or a sixth and lies cervically
between these extremes. However, this relates only to the
intramedullary branching of the central arteries and not to
the region of supply served by the anterior spinal artery,
which also nutrifies more than the ventral half of the outer
tracts by its contribution to the pial arterial networks of lower
degree. Furthermore, a strict division between the supply of
the grey columns solely from the centre and that of the white
matter solely from the periphery is not justified. In fact, with
2 Arterial Blood Supply
complete filling of even very small vessels, both territories
can hardly be distinguished from one another in the marginal
zone of the grey matter (Figs. 2.22, 2.29, and 2.30). Moreover,
considerable local and interindividual variations occur here.
Yet at the apex of the posterior horn, supply via branches of
the vasocorona dominates (Figs. 2.37 and 2.45). Also at the
edge of the anterior horn, centripetal branches of the periph-
eral system almost always contribute to the supply.
In the cervical region, the amount of grey matter on the
transactional surface is proportionately less in comparison
to lumbar. This corresponds to a reduced dominance of the
main ventral artery here as well as to a smaller proportion
of the transactional surface that is supplied by central arter-
ies (Fig. 2.30). If sagittal and coronal sections are also
included in the evaluation of supply areas, it can be seen
that a lesser volume of grey matter is accompanied by both
a reduction in calibre and number of central arteries. This
consequently causes greater axial vascular ramification.
The supply region of neighbouring central arteries, “over-
lapping” longitudinally, correlates more closely to a cone-
shaped intermeshing when viewed three-dimensionally. In
the presence of vascular occlusion in a short segment, defi-
cient compensation can be most likely expected in the ter-
ritory of the central arteries, since they are end arteries like
the perforating branches but with a more extensive territory
of supply.
2.6 Arterial Territories of Supply
a b c
Fig. 2.29 Vascular territories of supply. Limited filling of the anterior
spinal artery at the lumbar enlargement. (a) X-ray of a specimen in a.p.
view. Exceptionally in this case, injection of contrast medium into the
great anterior radiculomedullary artery did not proceed to the posterior
longitudinal channels via the arcades of the cone. In other spinal cord
regions, this is an expected and common observation. (b) X-ray of a
specimen in a.p. view. Typically, injection of a single anterior or
posterior radiculomedullary artery in this lumbosacral region results in
filling of the complete network of superficial arteries. (c) Axial micro-
angiogram taken from the specimen in a at the L1 level. The arterial
territory of supply from the anterior spinal artery is restricted to the
43
anterior parts of the cross-sectional area. However, completion of filling
seems to be suboptimal in this case. (d) Axial microangiogram taken
from the specimen in (b). Seperation of territories in this lumbosacral
region, connected by the arcades of the cone is generally not possible.
Contrary, at cervical and thoracic levels the anterior spinal artery terri-
torry can easily be defined by injecting this artery (Figs. 2.30 and 2.42).
At lumbar levels the large calibre of the anastomoses at the cone inevi-
tably includes filling of the posterior longitudinal tracts and conse-
quently contrasts the entire arterial system on the cross-sectional area
(d Reproduced from Thron 2002)
44
a b
Fig. 2.30 Territories of arterial supply. This composition offers a mor-
phological and microangiographical comparison between the cervical,
the thoracic and the lumbar region with regard to cross-sectional sur-
face areas, relations between grey and white matter and the schemati-
cally defined territories of supply. (a) Paraffin-embedded sections,
prepared for axial microangiograms, demonstrate the differences of
cross-sectional surface areas of the same specimen at cervical, thoracic
and lumbar levels. The different proportions of grey and white matter
are obvious. The grey matter presents as a dark butterfly-shaped area in
a bright surrounding of white matter. (b) The middle row of images
present microangiogams of axial sections cut at the level of the cervical
enlargement, the mid-thoracic region and the lumbar enlargement.
Examples have been selected, separating the contrast-injected area of
anterior supply from the non-contrasted posterior area of supply. (c)
2 Arterial Blood Supply
cervical
c
thoracic
lumbar
Diagrams of the approximate territories of arterial supply. The grey
zone corresponds to the area specifically supplied by the anterior spinal
artery through the central arteries and the branches to the anterior pial
system of lower degree (about two-thirds of the whole cross-sectional
area). The dorsal white zone is supplied by the peripheral system, fed
from posterolateral and posterior spinal arteries. The central territory
inside the hatched area is supplied by the central arteries. It is not dis-
tinctly seperated from the peripheral system. The hatched intermediate
area is variably supplied by one system or the other. The cross-sectional
extension of the central system corresponds approximately to the vol-
ume and relative amount of grey matter, which is quite different at vari-
ous levels. The conclusion is that rough territories of supply exist but
that there is no strict attribution between vascular supply and anatomi-
cal structures (a, c Reproduced from Thron 1988)
2.7 Arterioarterial Anastomoses
2.7 Arterioarterial Anastomoses
2.7.1 Superficial Anastomoses
1. The spinal cord is covered in its entire length by a net-like
anastomosing vascular system, in which the anterior spi-
nal artery and the bilateral posterolateral and/or posterior
spinal arteries can be regarded as the most constant longi-
tudinal anastomotic trunks of greatest calibre. The ante-
rior trunk is directly connected to the posterior ones by
arterial circles at the conus medullaris (rami cruciantes)
(Fig. 2.31).
2. Even though these main trunks may occasionally exhibit
short interruptions or at least narrow parts, it should be
noted that the network is supplemented by a system of
closely meshed longitudinal and transverse anastomoses
of second and third degree.
3. In the thoracic region longitudinal connections between
central arteries can be observed that run parallel to the
anterior spinal artery at the anterior median fissure
(Figs. 2.49 and 2.50b).
2.7.2 Transparenchymal Anastomoses
Transparenchymal arterioarterial anastomoses between cen-
tral artery branches and superficial arteries exist in the tho-
racic region (Figs. 2.26, 2.32, and 2.33). They are, however,
rare and demonstration of their course requires a special
45
sectional plane. Direct proof of their existence only succeeds
if they can already be assumed on a standard X-ray survey.
The calibre of these vessels does not taper from the central
artery to the surface. The direct connection to superficial lon-
gitudinal trunks leaves no doubt as to their arterial character.
Nevertheless, these findings were additionally checked
against series of histological sections to eliminate arteriove-
nous anastomoses.
Depending on the direction of course, we can differentiate
the following types:
1. Centro-anterolateral anastomoses (Fig. 2.32b)
2. Centro-posterolateral anastomoses (Figs. 2.32c, d and 2.33)
These directions do not correlate to the site of primary
or secondary longitudinal anastomotic trunks. In no case
are arterial interconnections found in anteroposterior
directions as with the venous anastomoses. They form an
additional collateral system between central and superfi-
cial in the thoracic region, thereby demonstrating the plas-
ticity of the spinal blood supply.
2.7.3 Intrinsic Anastomoses
A longitudinal anastomosis within the parenchyma between
the ascending and descending branches of the central arteries,
as was assumed by Adamkiewicz (1881) and later on by Fazio
and Agnoli (1970), could not been verified in our specimens.
46 2 Arterial Blood Supply

Fig. 2.31 X-ray films of

specimens from cadavers that R L lateral R L lateral
were contrast-injected after
fixation through the great
anterior radiculomedullary
artery (arrowhead). There is
non-filling of small vessels,
but the anastomotic circles
around the cone (at the S3–S5
level; arrow) and the
transverse interconnections
between the posterior
longitudinal tracts are
exhibited impressively. (a)
and (b) specimen with high
entry of the artery of the
lumbar enlargement
(arrowhead) and very
symmetrically develloped
“arterial basket” around the
cone. (c) and (d) low entry of
the artery of the lumbar
enlargement (arrowhead).
Compensating larger caliber
of the ascending branch of
the anterior
radiculomedullary artery and
of the posterior system. In
such anatomically well
develloped cases of the
arcades around the cone an
important collateral function,
similar to that of the circle of
Willis at the base of the brain,
is obvious (Reproduced from
Thron 1988)

a b c d
2.7 Arterioarterial Anastomoses 47

Fig. 2.32 Arterioarterial l

anastomoses between the central gitta
artery branches and the sa
superficial peripheral system.
(a) Schematic illustration of the
two typical courses. The vertical
arrow marks an anterolateral
anastomotic connection
exemplified in (b). The axia
horizontal arrow to a l
posterolateral anastomosis,
exemplified in (c and d).
(b) Axial microangiogram
demonstrating a centro-
anterolateral connection
(Type 1). (c) X-ray film of the
whole specimen in a.p. view. a b
Undissected anastomosis from
the central system to the
posterolateral surface (Type 2)
(arrow). (d) Microangiography
of this specimen in an oblique
section from the anterior midline
to the posterolateral surface.
Only a short part of this
anastomotic channel could not
be included in the slice of about
2 mm thickness (arrow). The
majority of intrinsic
anastomoses are disrupted by
sectioning because of their
irregular and variable courses
(b–d Reproduced from
Thron 1988)

c d
48 2 Arterial Blood Supply

a c

Coronal sections
d S3

S2

S1

Fig. 2.33 (a–c) The anastomoses demonstrated on these mid-coronal network (arrows). (d) position of the mid-coronal sections of the micro-
microangiograms at thoracic levels from different specimens are angiograms related to the cross-sectional area of the spinal cord
directed from central arteries to the lateral surface of the superficial pial
2.8 Specific Patterns of Arterial Supply in the Cervical Spinal Cord
2.8 Specific Patterns of Arterial Supply
in the Cervical Spinal Cord
Variations of arterial supply are very pronounced in this
region (Figs. 2.35 and 2.36).
Furthermore, various collaterals connect the arteries,
which are extramedullary sources of supply in the cervical
region (see Fig. 2.34). This can be demonstrated using super-
selective in vivo angiography of the vertebral, deep cervical
and ascending cervical arteries (Fig. 2.4). Neglecting this
knowledge makes embolization procedures in the regions of
the neck and skull base hazardeous.
The most important arterial feeders for the anterior med-
ullary supply of the cervical region may be the descending
branches from the intracranial section of the vertebral arter-
ies like in Figs. 2.35a, b and 2.36a. But these arteries may
also have a relatively small calibre like in Figs. 2.35c and
2.36b, c. More relevant radiculomedullary vessels for the
cervical enlargement may enter between C4 and C7. These
segmental branches originating from the extracranial part of
the vertebral arteries or from the neck arteries may exhibit
calibres of 0.4–0.6 mm. The largest feeder was named the
artery of the cervical enlargement by Lazorthes (1958)
(Figs. 2.9, 2.34, and 2.36). It is more often derived from the
deep and ascending cervical arteries than from the vertebral
artery. The average number of anterior radicular feeders for
the cervical medulla is 2–3, with no significant lateralization
in this region. In the individual situation all the three men-
tioned potential sources have to be checked bilaterally.
Well-developed descending branches from the terminal
intracranial part of the vertebral arteries, uniting to form the
anterior spinal artery, are a rarely observed pattern of supply
to the upper spinal cord. As outlined in Sect. 2.1 the terminal
vertebral artery corresponds to the stem of the ascending
branch of the anterior radiculomedullary artery of the first
spinal nerve. The descending branch of this radiculomedul-
lary artery represents the cephalad origin of the anterior
spinal artery (Kadyi 1889).
Descending feeders of equal size forming this median
arterial axis were the exception in our specimens (1/14).
More often, the vessel on one side dominated (right/
left = 1:1), the contralateral being small or even rudimentary
(Figs. 2.35 and 2.36a).
Often, two branches run parallel over a number of
segments. The major branch then continues in the midline as
49
the typical median longitudinal vessel. The minor one may
also reach the midline, yet upon closer inspection, it rather
terminates as a large central artery (Fig. 2.35a).
The continuity of a median arterial trunk is less constant
in the cervical region than in the thoracic or lumbar regions.
Furthermore, dorsolateral spinal arteries of large calibre are
not unusual here.
All of the additional details described in the following,
such as vessel duplication, pseudo-islands and net-like vas-
cular patterns, have to be regarded as a state more closely
related to the embryonic (or ontogenetic) condition.
Sometimes a bilateral network extends as far as the mid-
cervical region with decreasing vessel calibre (Fig. 2.35c).
Further branching of the cervical anterior spinal artery
may also occur. True island formation is less frequent than
duplication according to our experience. Again, the major
branch continues as the anterior spinal artery, whereas the
minor one ends in a central artery (Fig. 2.35b). Duplication
over some distance was seen in 7 out of 14 filled cervical
cords (Figs. 2.35b, 2.26a, b, and 2.41). Anterior
radiculo-medullary arteries of larger calibre most often
enter from segments C4 to C8 (8 out of 14 of our cases)
(Figs. 2.36, 2.37, and 2.42). Discontinuities of the ventral
longitudinal axis (3 out of 14 of our cases are usually found
between two zones, the upper one being directly supplied by
the vertebral artery and the lower two-thirds being supplied
by segmental tributaries of the deep cervical or vertebral
arteries (Fig. 2.35c). The posterior system of cervical supply
is usually derived from the PICA or the vertebral artery. The
descending branches further divide into the dorsal and dorso-
lateral spinal cord vessels of variable calibre. Lasjaunias
et al. (1985) call this vessel the “lateral cervical artery”.
However, corresponding to conditions on the ventral aspect
it must be an equivalent of the descending branch of a dorsal
radiculomedullary artery. Kadji (1889) therefore termed it
“Ramus descendens arteriae radicalis posterioris nervi
primi”.
The average number for the cervical posterior radicu-
lar arteries is approximately 3–4. They approach the
dorsal cervical medulla with a calibre of 0.15–0.4 mm
(Jellinger 1966a; Piscol 1972).
The posterior longitudinal systems form a network of
concatenated fragments, only exhibiting more dominant
individual tracts towards the thoracic region (Figs. 2.10
and 2.38).
50 2 Arterial Blood Supply

Basilar artery

Anterior spinal artery

Posterolateral
spinal artery

Costocervital trunk Vertebral artery

(deep cervical artery)

Anterior
radiculomedullary artery

Thyreocervical trunk
(ascending cervical artery) Posterior
radiculomedullary artery

Subclavian artery

Cartoid artery

Aorta

Fig. 2.34 Sources of spinal cord blood supply to the cervical region. anterior spinal artery are presented in red, the posterior superficial
The arteries from which segmental and medullary supply is derived channels in light red. (Reproduced from Dafotakis et al. 2008 as
(vertebral, deep cervical and ascending cervical arteries) and the modification of Thron 2008)
2.8 Specific Patterns of Arterial Supply in the Cervical Spinal Cord
a b
Fig. 2.35 Embryologically defined patterns of arterial supply on the
ventral surface of the medulla oblongata and the upper cervical spinal
cord. Photographs of injected specimens. (a) The typical finding of
blood supply to this area is a large descending branch from one verte-
bral artery (arrowheads) and a less developed, often rudimentary vessel
on the other side (black arrow). They do not unite. The small one (small
black arrow) is usually ending separately as a central artery. The longi-
tudinal tracts of the posterior surface originate as a descending branch
from the vertebral artery or from the posterior inferior cerebellar artery
in a (postero-)lateral position (white arrow; cf. Fig. 2.10a). (b) In many
cases only a single descending branch from one vertebral artery is pres-
51
c
ent. A plexiform network of short superficial vessels supplies the con-
tralateral part of the medulla oblongata. Concerning the cervical
anterior spinal artery, non-fusion of the originally bilateral tracts
(“duplication”) is frequent in variable physical forms (white arrow-
heads). True “island formations” are at least rare. Reunion of the two
tracts was not observed in our cases. The smaller descending branch
generally enters the anterior fissure and terminates in a central artery
(white arrow). (c) Plexiform pattern of supply from descending
branches of the vertebral arteries and radiculomedullary feeders, but
without formation of a significant unpaired anterior spinal artery
(Reproduced from Thron 1988)
52 2 Arterial Blood Supply

C3 C3

C1

C5

C6

C7

T2

C6

a b c

Fig. 2.36 X-ray films of three injected specimens in a.p. view.
Variations of arterial supply to the anterior surface of the cervical spinal
cord. (a) The strong unilateral descending branch from the left vertebral
artery establishes the median anterior arterial tract (big arrow). A short
duplication on the opposite left side below C1 with pseudo-island for-
mation (small arrow) is small and terminates as a central artery in the
median fissure. On the right side a plexiform pattern of short branches
covers the medulla oblongata. The main radiculomedullary contribution
enters the anastomotic tract at the C6 level. (b) “Duplication” (non-
fusion) of the anterior spinal artery cranial to the C4 level with small
radiculomedullary feeders entering at both sides. The artery of the cer-
vical enlargement joins the anterior spinal axis at the C6 level. (c) The
main feeder is a radiculomedullary artery entering at the C7 level.
Additional anterior supply by several small arteries from higher cervi-
cal levels (a Reproduced from Thron 1988)
2.8 Specific Patterns of Arterial Supply in the Cervical Spinal Cord
a d
Fig. 2.37 Changing pattern of arterial supply at cervical and upper tho-
racic level. (a) X-ray of the injected specimen and localization of the
axial sections in (b–d). Axial Microangiograms at the C5 (b) and C7/8
level (c) and, for comparison, at the T6 level (d). In the cervical region
(b, c), arterial filling is equally well distributed in all territories of
supply. Note the dense capillary network within the grey matter and the
53
dominant role of the anterior spinal artery (arrow in c). The section at
the thoracic level T6 (d) demonstrates incomplete filling in the posterior
column area but the calibre of anterior and posterior longitudinal tracts
(arrow and open arrowhead) has significantly changed in favour of the
posterior system. (Reproduced from Thron 1988)
54
S1
C7
a b
Fig. 2.38 Superficial arteries of the lower cervical spinal cord. Coronal
microangiograms of the anterior and posterior superficial section taken
from the specimen shown in Fig. 2.37a. (a) Anterior section S1. The
main anterior radiculomedullary artery enters at the C7 level. The
extensive network of arteries shown on this superficial frontal section
are branches which constitute the anterior pial network of lower degree
2 Arterial Blood Supply
S3
including supply of the nerve roots. The central artery branches, run-
ning posteriorly into the anterior fissure are not well shown because of
superimposition of the anterior spinal artery. (b) Posterior section S3.
The posterolateral and posterior longitudinal systems form an irregular
pial network with fragmentary chains (Reproduced from Thron 1988)
2.8 Specific Patterns of Arterial Supply in the Cervical Spinal Cord 55

S2 S2
C2 C2 C 5-6

b,
c

C5

C7

b d

C 4-5 C 7-8

T3

a c e

Fig. 2.39 Vertical ramification of central arteries in the cervical spinal
cord. (a) X-ray of the specimen in a.p. view. The vertical lines indicate
the position and length of the median (S2) coronal slices which are
contiguous between C2 and C5. (b, c) Upper and middle cervical levels.
The irregular side-alternating distribution of the central arteries to the
right or to the left side are best shown in this coronal sections in median
position (S2). Each of them covers approximately 5 cm of length. The
vertical ramification is intertwining but does not form an inner anasto-
motic tract. (d, e) Lower cervical and upper thoracic levels. The density
of central arteries in the slim cervicothoracic spinal cord region is far
less. The vertical ramification is more pronounced. The supply by per-
forating branches of the pial network is more important. This corre-
sponds to the different proportions of grey and white matter in these
regions (Reproduced from Thron 1988)
56
a b c d
Fig. 2.40 Microangiograms of mid-sagittal sections from different lev-
els of the cervical region (b–d). Their position is marked by vertical
lines on the X-ray survey shown in (a). (b) The anterior spinal artery
appears to be very thin at the high cervical level (arrow), but this false
impression is caused by non-fusion of the two primary anterior tracts
(compare with a). (c) In the region of the cervical enlargement which
involves parts (b and c), the number of central arteries is high. The
blood vessels take an almost horizontal course. Additionally, numerous
small perforating arteries arise from the pial plexus, covering the outer
surface and the anterior fissure. (d) The calibre of the anterior arterial
2 Arterial Blood Supply
tract very often decreases at high thoracic levels. Rather typical is the
steeply ascending course of the central arteries at high thoracic levels.
Solitary or sometimes interconnected central arteries run parallel to the
anterior spinal artery over some distance before turning into the depth
of the anterior fissure. As a result of the small calibre, central arteries of
this region are difficult to distinguish from perforating vessels of the
peripheral system on mid-sagittal sections. Compare the course and
ramification of intrinsic arteries with the mid-coronal sections of this
region, shown in Fig. 2.39 (b, d Reproduced from Thron 1988)
2.8 Specific Patterns of Arterial Supply in the Cervical Spinal Cord 57

X-ray a.p.

"

C3

C6

a b c d

Fig. 2.41 Superficial and intrinsic cervical arteries. Non-fusion (small arrows), supplying the corresponding semi-cord. (c, d) Mid-
(“duplication”) of the anterior spinal artery. (a) X-ray film in a.p. view. sagittal microangiograms of mid-cervical level. Short common stems of
(b) Mid-sagittal microangiogram of upper cervical level. The calibre of two central arteries frequently occur in unpaired anterior spinal arteries
the central arteries is equal to that of the separated longitudinal tracts (arrowheads)
58 2 Arterial Blood Supply

Fig. 2.42 Anterior (and

right lateral) epidural spinal R
tumour (Hodgkin’s disease)
at the C7–C8 level. (a)
X-ray film of the injected
cord which is flattened and
compressed from the right
side (large arrow). The
anterior spinal artery is
narrowed, but there is
extensive pial
hypervascularization at the
tumour level. (b) Photo of
the macroscopic aspect of
the epidural tumour
position. (c) Mid-sagittal
and (d) axial
microangiograms show
anterior indentation and
flatting of the cord (large
arrows). There is still no
obstruction of the anterior
spinal artery and the sulcal
arteries are patent although C7
they are considerably
displaced posteriorly into
the anterior median fissure.
Histology revealed normal
findings (Reproduced from
Thron 1988)

b c

C8

C 7/8

a d
2.9 Specific Patterns of Arterial Supply in the Upper- and Mid-Thoracic Spinal Cord
2.9 Specific Patterns of Arterial Supply
in the Upper- and Mid-Thoracic
Spinal Cord
The upper- and mid-thoracic region is commonly sup-
plied by small radiculomedullary arteries 0.2–0.4 mm.
Therefore, angiographical demonstration of feeders may fail
(Figs. 2.43b, 2.47, and 4.1). The average number of ante-
rior radiculomedullary arteries in the thoracic region is 2–3
(Figs. 2.23, 2.37, and 2.43). In several of our specimens, we
found the anterior longitudinal anastomosis in this region to
be extremely narrow (Figs. 2.40, 2.43b, c, 2.46, and 2.47), or
even interrupted (1 case). Usually, however, it took the form
of a continuous trunk of small calibre (Figs. 2.43 and 2.46).
Corresponding to the smaller cross-sectional area of this
region and the relatively smaller portion of grey matter, the
number of central arteries is only 2–3 per cm for this region.
This explains the prevalence of steeply ascending and
descending central artery branches (Figs. 2.25, 2.40d, 2.46,
and 2.47). As pointed out earlier, the impression of an intrin-
59
sic longitudinal anastomosis on sagittal images was not con-
firmed on coronal images of this spinal cord region where
there is a preponderance of white matter tracts (Fig. 2.44).
On the posterior surface of the cord the longitudinal tracts
may run in posterior or posterolateral directions, thus
indicating the functional identity of these longitudinal
anastomotic tracts (Fig. 2.44).
For the same reason, the arteries of the posterior longitu-
dinal system can either uni- or bilaterally exceed the vessels
of the posterolateral system in calibre or even completely
substitute it at least for some segments (Figs. 2.37d and
2.44c). The relation of vessel diameters clearly shows that
the posterior as well as the anterior vascular supply is
adjusted to the thinner thoracic cord.
As could be demonstrated on the thoracic cord, a continu-
ous posterior longitudinal system does not exist. The frag-
mentary trunks in posterior and/or posterolateral position
mainly supply the superficial network of smaller branches
and from this second-degree pial network the perforating
rami emerge to the white matter.
60 2 Arterial Blood Supply

X-ray a.p. X-ray a.p.

T2

T4

T6

T8

T11 T8
a b c d

Fig. 2.43 Variability of number and calibre of radiculomedullary feed-
ers and longitudinal arterial tracts in the thoracic region. (a) One single
feeder reinforces the strong anterior spinal artery in the entire length of
the thoracic spinal cord. (b) Two small thoracic radiculomedullary feed-
ers are obvious, the lower one nicely demonstrates additional supply of
the corresponding spinal ganglion. The calibre of the anterior spinal
artery is almost filiform and hardly exceeds that of the posterolateral
tracts. (c) Two medium-sized segmental feeders constitute an equally
sized anterior spinal artery (arrows). Again combined supply of a spinal
ganglion from the radiculomedullary artery. (d) A large-sized radiculo-
medullary artery at high thoracic level merges with anterior spinal tracts
which are doubled (not-fused) for a certain distance (arrowheads)
2.9 Specific Patterns of Arterial Supply in the Upper- and Mid-Thoracic Spinal Cord
S1
T3
a b
Fig. 2.44 Typical relationships of anterior and posterior patterns of
supply in the upper thoracic region demonstrated on coronal microan-
giograms. (a) The anterior spinal artery is of medium size and contrib-
utes to the dense pial network with transverse and longitudinal
interconnecting branches, also supplying the nerve roots. (b) The central
arteries are less numerous compared to cervical or lumbar levels. The
few arteries mainly branch in a vertical direction but they do not form
61
S2 S3
c
inner longitudinal anastomoses. Instead, anastomoses between central
artery branches and the superficial pial plexus can be observed (arrow)
(see Sect. 2.7). (c) Posterior radicular arteries (arrowheads) reinforce a
rope-ladder-like system of tracts. The longitudinal components may be
situated in posterior or posterolateral position (small arrows). Both com-
ponents, the vertical and the horizontal, distribute blood within the pos-
terior pial network of lower degree (Reproduced from Thron 1988)
62
a b
Fig. 2.45 Superficial and intrinsic arteries of the thoracic spinal cord
on an X-ray of the injected specimen and at different levels of the
thoracic spinal cord on axial microangiograms. Extension of the
centrifugal and the centripetal supply system depending on grey/white
matter relationship. (a) X-ray film of the injected specimen in a.p. view.
(b–d) Axial sections from the thoracic levels indicated in (a). In
2 Arterial Blood Supply
between the enlargements the posterior tracts often match the anterior
spinal artery in size. Supply of the fibre tracts through the superficial
pial network gains increasing importance. The decreasing numbers and
calibres of central arteries in the upper and mid-thoracic spinal cord
correspond to the decreasing proportion of grey matter on the cross-
sectional area in this region area (a Reproduced from Thron 1988)
2.9 Specific Patterns of Arterial Supply in the Upper- and Mid-Thoracic Spinal Cord 63

a b c

Fig. 2.46 Course and vertical branching pattern of central arteries in fissure. They run parallel to the anterior spinal artery as a pial subsys-
the upper thoracic region. (a–c) Microangiograms of mid-sagittal sec- tem before they bend horizontally and give off branches to the central
tions extending from C8 to T5 in the same specimen. Narrow anterior grey matter. During this course a significant contribution to the anterior
spinal artery. Only few central arteries ascend steeply in the anterior pial network (vasocorona) is obvious, especially in this imaging plane
64 2 Arterial Blood Supply

a b c d

Fig. 2.47 Anterior spinal artery and distribution of central arteries in a
specimen at mid-thoracic levels. (a) X-ray of the specimen and local-
ization of the parts shown in (b–d). The great anterior radiculomedul-
lary artery is joining the anterior trunk at T11. Its ascending branch is
very thin. (b–d) Microangiograms of mid-sagittal sections extending
from T5 to T7/8 in the same specimen. Each specimen part measures
2.5 cm in length and exhibits only four to eight central arteries (artificial
interruptions of the anterior spinal artery in parts of (b and d). The
central arteries ascend steeply in the anterior fissure, run parallel to the
anterior spinal artery as a pial subsystem before they bend horizontally
and give off branches to the central grey matter. During this course an
additional significant contribution to the anterior pial network
(vasocorona) is obvious in this imaging plane (c, d Reproduced from
Thron 1988)
2.9 Specific Patterns of Arterial Supply in the Upper- and Mid-Thoracic Spinal Cord 65

a b c

Fig. 2.48 “Duplication” (non-fusion) of the anterior spinal artery at
T7–T8 level of a spinal cord. (a) X-ray in a.p. view. Entry of the great
anterior radiculomedullary artery at T10 (large arrow) and of a poste-
rior feeder at T11 (small arrow). Duplication is shown at T7/8 (small
arrowheads), which is unusual at this location. (b) Microangiogram of
the mid-sagittal section indicated in (a). Few but strong central arteries
originate from the ascending branch of the anterior spinal artery above
T10 and ramify vertically. (c) Bilateral anterior arterial channels each of
which gives off the central arteries to the corresponding hemi-cord
(arrowheads). They issue long branches in vertical direction (section
length 2.0–2.5 cm) (b, c Reproduced from Thron 1988)
66
b c
Fig. 2.49 Central artery courses at T6 in sagittal plane and at the entry
of a radiculomedullary artery at T4. (a) X-ray of an injected thoracic
spinal cord in a.p. view. (b) Microangiogram of an oblique section cut
from the anterior midline to the surface posterolateral right. It roughly
follows the ascending oblique course of a centro-dorsolateral anasto-
mosis (vertical arrows in a and b). From the hairpin curve of the ante-
rior spinal artery a steeply ascending central artery (short thick arrow)
arises, running parallel to the ascending branch of the anterior spinal
2 Arterial Blood Supply
artery. Not only central artery branches arise from this stem but also
strong superficial pial branches, running parallel to the anterior
longitudinal tract. (c) Other central arteries may also arise from this
secondary pial system, which can easily be seen on the mid-sagittal
section of c (arrow). This is illustrated in the schematic drawing of
Fig. 2.17 showing two or three central arteries arising from a common
stem (Reproduced from Thron 1988)
2.9 Specific Patterns of Arterial Supply in the Upper- and Mid-Thoracic Spinal Cord 67

Fig. 2.50 Anterolateral a b

extradural spinal tumour
(metastasis) at the T3–T4
level. (a) X-ray film of the
contrast-filled spinal cord.
At the upper extension of
the tumour, which
coincides with the upper
margin of section (b), there
is a narrow segment of the
anterior spinal artery
(diameter 100–200 μm).
The posterolateral spinal
artery on the side of the
tumour is considerably
dilated (300 μm) and
receives blood from the
anterior arterial axis from
above. (b) Mid-sagittal
microangiogram at the
level of the upper tumour
extension. Narrow anterior
spinal artery (arrowhead).
Interconnections between
central arteries at the
entrance into the anterior
median fissure are very
pronounced (arrows). (c)
Axial microangiogram of
the tumour and the highly
deformed spinal cord.
Histology revealed only
demyelinization of the
posterior columns,
probably as a result of
venous congestion
(Reproduced from Thron
1988)

c
68
2.10 Specific Patterns of Arterial Supply
in the Thoracolumbar Spinal Cord
and Cauda Equina
One of the ventral feeders between T9 and L1 (exceptionally
at L2 or L3) is distinctly dominant in calibre (800–1000 μm)
and was therefore called the “artery of the lumbar enlarge-
ment” (Lazorthes) or the “great radicular artery”
(Adamkiewicz). In 73 % of all cases it enters from the left
side with its origin between T9 and 12 in 62 %, lower in the
lumbar region in 26 % or higher between T6 and T8 in 12 %
(Jellinger 1966a).
Additional significant ventral feeders are unusual below
its level of entrance.
The average number of anterior radiculomedullary arter-
ies to the lumbosacral medulla is 0–1 (Jellinger 1966a).
The predominant role of this great (anterior) radiculom-
edullary artery in supplying the thoracolumbar enlargement
and the calibre and course of the anterior spinal artery, often
deviating in a zigzag from the midline, are apparent in all the
illustrations of this region. Complete filling of the low tho-
racic and lumbosacral spinal cord was easily achieved for
almost all injections into this vessel, irrespective of whether
the entrance level of this artery was located between T9 and
T12, exceptionally high between T6 and T8, or low at L1 to
L3 (Figs. 2.11, 2.14, 2.15, 2.16, 2.51, and 2.52).
Supply to the posterior system in this region often includes
two almost equally large dorsal feeders (400–500 μm), which
enter higher or lower than the great radiculomedullary artery
(Figs. 2.8, 2.11, 2.51, and 2.52).
The existence of a great posterior radiculomedullary
artery, corresponding to the anterior one and possessing a
distinctly greater calibre than the other posterior radiculom-
edullary arteries, is affirmed by a few authors (Gillilan 1958;
Lazorthes et al. 1958; Jellinger 1966a), but rejected by others
(Corbin 1961; Piscol 1972). We found an especially large
posterior radiculomedullary artery on only one side as an
unusual observation.
The number of dorsal radiculomedullary arteries is
reported with an average of 8 (Piscol 1972) to an average of
9–12 (Jellinger 1966a) for the total thoracolumbar region.
No lateral preference is observed; the vessel calibres lie
between 0.15–0.4 mm.
Posterior radiculomedullary arteries to the thoracolumbar
region show a differently configured hairpin course com-
pared to the anterior vessels and branch into ascending and
descending vessels of mostly different calibre.
The transverse network is formed by external anastomo-
ses between the anterior tract and the posterolateral spinal
arteries yet even more by the intercommunication between
them and the so-called posterior spinal arteries (Figs. 2.55
2 Arterial Blood Supply
and 2.56). According to the description given by Piscol
(1972), the tract of the posterolateral spinal arteries that runs
in the “axilla” of the posterior roots is very constant in the
lower half of the spinal cord. Its location and calibre in com-
parison to the posterior spinal arteries running medially and
dorsally of the posterior roots can be easily recognized in the
microradiographic reproductions of Figs. 2.13, 2.14, 2.15,
2.16, and 2.18. The numerous intercommunications of equal
calibre produce a rope-ladder-like system.
The posterolateral longitudinal tracts clearly dominate
above the posterior side of the lumbar enlargement, but they
are only the largest components in a longitudinally oriented
network that already becomes more irregular and fragmen-
tary in the lower thoracic region.
The ventral and dorsal systems are connected to each
other around the conus (arcade of the cone, rami anastomot-
ici arcuati) (Kadyi 1889). This pattern may constitute a sig-
nificant anastomosis, comparable to the circle of Willis
(Fig. 2.31). It was seen in the vast majority of specimens, but
with variable calibres and not necessarily at the same level
on both sides (Figs. 2.14 and 2.16). Simultaneous filling of
the posterolateral and posterior longitudinal tracts in the
lumbar region occurred with regularity if the great anterior
radiculomedullary artery was injected. Vice versa, filling of
the anterior spinal artery was easily possible by injecting one
of the neighbouring posterior radiculomedullary arteries.
The filling was achieved primarily via these rami cruciantes
(Adamkiewicz 1882), laterally surrounding the cone. These
interconnecting vessels lie concealed between the sacral
roots, but can be demonstrated well radiographically. They
are usually of considerable calibre. Anastomotic function
and supply of the cauda fibres might explain why the
descending branch of the anterior spinal artery, but also that
of the posterolateral tracts, maintains such a large calibre
until the end of the conus even at low entrance (Figs. 2.52
and 2.60a).
The straight continuation of the anterior spinal artery to
the filum terminale is an insignificant branch (Figs. 2.51b, c
and 2.52c, d). The small arteries running with the fibres of
the cauda equine (Fig. 2.54) are derived from the pial plexus
(Fig. 2.11). Both crura of the conus arcade are usually rela-
tively symmetrical, but at least one of both connecting
branches may be missing (Fig. 2.51b, c).
The effective anastomotic function of this circle might
also explain the difficulty in demonstrating different areas
of supply in our specimen in this region. Injecting the
solution of contrast medium in either anterior or posterior
radiculomedullary arteries of the lumbar enlargement always
resulted in immediate filling of the entire superficial longitu-
dinal tracts and, furthermore, of the small arteries running
with the fibres of the cauda equine.
2.10 Specific Patterns of Arterial Supply in the Thoracolumbar Spinal Cord and Cauda Equina 69

a b c d

Fig. 2.51 Frequent constellations in arterial supply of the lower third
of the spinal cord. Selected examples of contrast-injected spinal cords,
demonstrated on X-ray images in a.p. view. (a) Classical constellation:
entry of the A. radicularis magna at the T11 level on the left. Bilaterally
well-developed posterior radiculomedullary feeders at high lumbar
levels and bilaterally well-constituted anastomotic circles around the
cone. (b) Equally large-sized anterior feeders entering at the levels T10
and T12. Anterior spinal artery with a straight course. Asymmetrically
formed anastomotic circle on the left and distinct artery of the filum
terminale. (c) A radicularis magna entering at T10 level on the left side
with convoluted course of the anterior spinal artery and an asymmetri-
cal development of the anastomotic circle around the cone. (d) Similar
to (a) with a very symmetrical supply situation, stronger posterior
tracts, but only small-sized anastomotic arterial basket around the cone
(a Reproduced from Thron 1988)
70
T6
T7
T9 T9
a b
Fig. 2.52 Variability of supply conditions in individual specimens.
High or low entry of the artery of the lumbar enlargement, variable
patterns of supply between anterior and posterior superficial systems
and variable developments of anastomotic circles around the cone.
X-ray films of specimens in a.p.view. (a) High entry of the great ante-
rior radicular artery at the T9 level. Well developed posterolateral arte-
rial tracts but only sparse anastomotic circles around the cone. (b) High
entry of the artery of the thoracolumbar enlargement. The ascending
2 Arterial Blood Supply
X-rays a.p.
T7
T 10
T 11
L2
c d
branch to form a part of the anterior spinal artery is extremely narrow.
Medium-sized posterolateral tracts including symmetrical arcades
around the cone. (c) Low entry of the artery of Adamkiewicz at the L2
level. Narrow ascending branch and anterior spinal artery up to the T7
level. Compensating strong posterolateral tracts. Incomplete unilateral
“anastomotic basket” at the cone. (d) “Textbook example” with average
findings in every respect including the artery of the filum terminale
2.10 Specific Patterns of Arterial Supply in the Thoracolumbar Spinal Cord and Cauda Equina 71

a c

Fig. 2.53 Axial microangiograms at a thoracic and a lumbar level for
comparison. (a) X-ray in a.p. view. The dotted lines indicate the position
of the axial sections. (b) Thoracic region. The position of the posterolat-
eral (arrows) and the posterior longitudinal tracts (arrowheads) in rela-
tion to the posterior root is clearly shown. The two central arteries
included in this section have remarkably different calibres. Filling of
intrinsic vessels in the posterior third is poor. (c) Lumbar region.
Transverse interconnection between the posterior longitudinal tracts on
the right side (sharp arrow). In this slice of 3 mm thickness 4 central
arteries of different size are included, reflecting the high vascular density
(c Reproduced from Thron 1988)
72
a d
Fig. 2.54 Arteries of the thoracolumbar enlargement. (a) X-ray film of
the injected specimen in a.p. view. The supply pattern and the calibre
relations of anterior and posterior radiculomedullary feeders are not
uncommon. Well-formed anastomotic circles around the cone bilater-
ally. Posterior longitudinal trunks of almost equal size. (b) Lumbar
level as indicated in (a). The equal filling of the superficial and deep
2 Arterial Blood Supply
arteries is rarely achieved in post-mortem studies. (c) Lower thoracic
level as indicated in (a). Strong posterior tracts exist only on the right
side. The difference in the density of arteries within grey and white
matter is impressive. (d) Poor filling of superficial and intrinsic arteries
especially in the territory of the posterior systems (c Reproduced from
Thron 2002)
2.10 Specific Patterns of Arterial Supply in the Thoracolumbar Spinal Cord and Cauda Equina 73

a b

Fig. 2.55 Arteries of the thoracolumbar spinal cord. (a) X-ray film of
the injected specimen in a.p. view. The levels chosen for the coronal
sections in (b and c) are marked by vertical arrows. (b) Microangiograms
of four contiguous coronal sections in the middle of the lumbar enlarge-
ment. Arrangement and distribution of the superficial (S1and S4) and
intrinsic spinal cord arteries (S2 and S3) at this level demonstrate the
extremely dense vascularization of this region. (c) Microangiograms of
four contiguous coronal sections cranial to the lumbar enlargement,
above the entry of the great radiculomedullary artery. A more loosely
meshed superficial arterial network is obvious, the number of central
arteries is far smaller and their ramification mainly occurs in a vertical
direction (Reproduced from Thron 1988)
74 2 Arterial Blood Supply

Fig. 2.56 Arteries of the

a b c
lumbosacral spinal cord.
(a–c) Frontal
microangiograms (S1, S2,
S3) in vertical reconstruction
images. The arteries of the
anterior and posterior
surface as well as the
intrinsic vessels (S2) can be
seen without
superimposition of each
other. The longitudinal
arterial trunks on the
posterior surface (S3) form a
rope-ladder-like network, in
which the posterolateral
arteries dominate in size
(Reproduced from Thron
1988)
2.10 Specific Patterns of Arterial Supply in the Thoracolumbar Spinal Cord and Cauda Equina 75

Fig. 2.57 Configuration of central arteries at the hairpin curve of

radiculomedullary arteries at thoracolumbar levels. (a) X-ray film
in a.p. view with significant radiculomedullary feeders at thoracic
and lumbar levels. (b) Microangiogram of an oblique antero-
posterior section at high thoracic level. A single very elongated b
central artery ascends parallel to the ascending branch of the
anterior spinal artery. There are longitudinal interconnections
between pial arteries at the entrance into the anterior fissure, but
not between the central artery branches. A transverse anastomosis
between a central artey branch and the superficial system is
dissected (compare Fig. 2.32) (c) Microangiogram of an anterior
frontal section at the entry of the great anterior radiculomedullary
artery. Division of these large feeders into an ascending and
descending branch most often takes place lateral from the midline.
A series of steeply ascending central arteries covers the gap of
supply before the anterior anastomotic tract returns to the midline
(arrows) (Reproduced from Thron 1988)

a c
76 2 Arterial Blood Supply

S1 Mid-and parasagittal sections

S2 S1
S3

S2 S1 S3

c,d,e

c e

d
a

Fig. 2.58 Arteries of the lumbosacral spinal cord. (a) X-ray of the
injected specimen in a.p. view. Small damaged area on the right side
near the cone. Complete filling of superficial and intrinsic arteries. (b)
Microangiogram of a mid-sagittal section at the level of entrance of the
great anterior radiculomedullary artery as indicated in (a). Cranial of
the hairpin curve, the number and calibre of central arteries decreases
abruptly. (c–e) Microangiograms of the mid-sagittal section (d) and the
two lateral parts of the cone. The anastomotic tracts, also known as
“rami cruciantes”, connect the anterior spinal artery with the posterior
longitudinal system. In this specimen they are well developed at both
sides. The branch to the filum terminale is small (Reproduced from
Thron 1988)
2.10 Specific Patterns of Arterial Supply in the Thoracolumbar Spinal Cord and Cauda Equina 77

Fig. 2.59 Vascularisation pattern of

b c
central and anterior peripheral arteries in
the thoracolumbar enlargement.
Mid-sagittal microangiograms from a
sacral level (a) to a lower thoracic level
(c). The great anterior radicular artery is
entering high, cranially of these sections.
This might explain the continuously
strong calibre of the anterior spinal artery.
Horizontal or slightly descending course
of central arteries, some of them showing
short common stems (arrows in (a and b).
In some cases in the thoracic region long
superficial interconnections exist between
two central arteries (arrows in c)

a
78 2 Arterial Blood Supply

a b

Fig. 2.60 Microangiograms of mid-sagittal sections at lumbosacral lev-
els demonstrating the variability of central artery courses. (a) Exceptionally
low entrance of the great anterior radiculomedullary artery at the L3 level
(large arrow; specimen shown in Fig. 2.9b). Ascending steeply from the
hairpin curve, several central arteries have a long common stem running
parallel to the anterior spinal tract before dividing into two or more central
arteries (small arrows). This finding is more common in the thoracic cord
and is therefore included in the schematic drawings of Figs. 2.17 and 2.49.
(b) Elongated and dilated arteries with a tortuous course in a patient of
about 80 years of age (Reproduced from Thron 1988)
2.11 Anatomical and Haemodynamical Aspects of Arterial Supply
2.11 Anatomical and Haemodynamical
Aspects of Arterial Supply
2.11.1 Anastomotic Capacity of Longitudinal
Trunks and Distribution of Supplying
Arteries
In comparing the radiographic anatomy presented here with
earlier anatomical investigations and later studies, it can be
appropriately remarked that inaccuracies and dubious clini-
cal conclusions could have been avoided had the careful,
minutely conducted study by Kadyi from 1889 received
more attention, rather than just being quoted. One of the
main reasons for divergent and discrepant descriptions of
spinal vascular anatomy is certainly the gross regional and
interindividual variability of vascular conditions. Unless all
spinal cord segments are studied in as many specimens as
possible, the observed findings can differ considerably. This
is why we tried to present the large variability of supply pat-
terns in this book.
If one compares surveys and monographs, which statisti-
cally evaluate a large number of specimens (Lazorthes,
1958; Corbin 1961; Jellinger 1966a; Piscol 1972; Dommisse
1975; Tveten 1976b), no essential discrepancies in the num-
ber of anterior and posterior radicular feeders are found.
Also the distinction between a paucisegmental type with
2–5 ventral feeders per cord in approx. 45 % of specimens
and a plurisegmental type supplied by 6–14 anterior radicu-
lar arteries in approx. 55 % (Kadyi 1889; Jellinger 1966a;
Piscol 1972) shows largely conforming results. To a certain
extent, this also applies to the segmental distribution of
radicular feeders in percent of cases, which shows no great
deviations comparing the evaluations by Piscol (1972) or
Domisse (1975) with those by Jellinger (1966a), who
derived his results from several hundred spinal cords. From
this it can be deduced that both the regions C5 to C8 and T9
to L2 show an increased incidence of medullary feeders and,
at the same time, almost all larger vessels join the spinal
cord in these regions (Piscol 1972). These conclusions are
important for a basic understanding of spinal cord blood
supply, though not applicable in each individual case
(Domisse, 1975). The medullary segments with a high inci-
dence of large tributaries correspond to the cervical or tho-
racolumbar enlargement, respectively, i.e. the regions with
the greatest accumulation of nerve cells. Analogous to the
large-calibre radicular feeders, the anterior spinal artery,
which is viewed as their anastomotic connection on the sur-
face of the cord, also has its greatest diameter in these
regions. These areas contrast with the thoracic region, which
is supplied by fewer, mainly smaller arteries, and whose
anterior anastomotic tract can show an accordingly weak
development. Main functional zones of medullary supply
were deduced from these conditions.
79
Thus, Lazorthes (1958 to 1962) and Domisse (1975) dif-
ferentiate three territories, yet with differing boundaries.
Their division roughly separates the cervical region from
the thoracolumbar enlargement and from a thoracic inter-
mediate zone. Other authors distinguish between two
regions: an oral, cervicodorsal (supplied by subclavian
artery branches) and a caudal, thoracolumbar region (sup-
plied by aortic segmental arteries), which they further sub-
divide into sections with a low or high incidence of feeders
(Corbin 1961; Jellinger 1966a). Piscol (1972) distinguishes
between four territories that result from segmental distribu-
tion curves of anterior radicular arteries and various pro-
portions of very small to large arteries (vessel categories
I–IV). From this he defines zones of characteristic supply
situations in respect to the total blood supply of the spinal
cord, namely:
• Obligatory vasoafference regions: the regions C5 to C8
and T9 to L2 with the radicular arteries relevant for the
enlargements and the entire myelon.
• Facultative vasoafference regions: the regions C3 to C4
and T4 to T6 with radicular arteries in varying number
and size, and consequently a modifying effect on the total
supply.
• Vasodeficiency regions: the segments T1 to T3 and L3 to
the caudal end without relevant feeders.
• Variable vasoafference region: C1/C2, which always
receives feeders from the vertebral arteries, but has a very
inconsistent contribution to the total supply.
This short summary of the statistical extramedullary sup-
ply situation is indispensable as a basis for the very contro-
versial discussions on sufficiency or regional insufficiency of
arterial spinal vascular supply, and on the importance of the
anterior spinal artery as a ventral arterial anastomotic trunk.
The calibre of the anterior spinal artery and of its radicu-
lar tributaries is approximately proportionate to the volume
of grey matter. Very few authors have referred to this fact, yet
it is emphasized here in agreement with Gillilan (1958). She
stated, “The blood supply to the thoracic spinal cord is
entirely adequate for the volume of grey matter present and
it is relatively as good as for any other cord segment”.
The reduced density of central arteries in the thoracic
region correlates on the one hand with the greater longitudi-
nal magnitude of the thoracic segments (Kadyi 1889; Corbin
1961). On the other hand, it corresponds to the decreasing
proportion of grey matter here, down to one-sixth of the
sectional area according to our measurements (compared
with approx. 1/2 to 1/3 in the lumbar and cervical region).
Furthermore, there is a corresponding absolute reduction of
the total cross-sectional area (Fig. 2.30). If the number of
central arteries is seen in relation to the varying lengths of
myelon segments rather than to centimetre lengths of myelon,
80
not much less than the normal incidence of 6–7 central arter-
ies per segment results for the thoracic region as well
(Clemens 1966).
On the other hand, selective injections into single radicu-
lar arteries showed that in most cases the entire spinal cord
arteries cannot be filled by this procedure (Adamkiewicz
1882; Bolton 1939; Suh and Alexander 1939; Noeske 1958;
Roll 1958; Corbin 1961; Lazorthes et al. 1962), since the
anterior longitudinal anastomosis can narrow noticeably or
even be interrupted at different locations. This supports the
concept of circulatory partial systems (Adamkiewicz 1882;
Piscol 1972). Concerning the anastomotic capacity of the
anterior median trunk it means that this vessel may be an
insufficient anastomosis for highly perfused distant regions,
at least as far as acute demands are concerned.
But this situation is not restricted to the upper thoracic
cord. It also exists in the upper cervical region according to
our observations that agree with the findings of Corbin
(1961), Lazorthes et al. (1971) and Piscol (1972) and with
animal in vivo studies by Fried et al. (1970). The descending
feeders from the vertebral arteries (homologous to the
descending branches of an anterior radicular artery at C0/C1
level) may not only run as bilateral vessels over several seg-
ments but may also fail to unite with the median longitudinal
trunk. This is just one of many variations for a territory
whose vascularization often corresponds to the more primi-
tive embryonic pattern of a plexus-like anastomosis
(Figs. 2.35 and 2.36). Therefore, the potential anastomotic
capacity of the anterior arterial axis may be as insufficient
above the cervical enlargement as it may be in segments of
the thoracic cord.
According to concurring reports, the second area with
facultative narrow segments of the anterior spinal artery is
the mid to upper thoracic region, i.e. the region between the
entrance of the great radicular artery and of the largest, cervi-
cal anterior radicular artery. Here, we measured calibres as
low as 0.2–0.1 mm, corresponding to those of the central
arteries, which is in agreement with other investigators
(Romanes 1965; Jellinger 1966a). Even interruptions were
sporadically observed (Corbin 1961; Hassler 1966). But
there are no set rules for the appearance of small-calibre seg-
ments in the thoracic region. Rather, it depends solely on
individual distribution patterns.
If one accepts the existence of circulatory partial systems
in the spinal cord, as conceived by Adamkiewicz, experi-
mentally substantiated by Piscol (1972), and in agreement
with spinal angiographical experience, it follows that the
anterior spinal artery as an anastomotic trunk has segments
with very good anastomotic capacities, but others with
poorer anastomotic capacities in regard to the more distant
circulatory partial systems of the spinal cord.
According to our findings, the calibre, arrangement and
distribution of the posterolateral and posterior spinal
2 Arterial Blood Supply
arteries indicate that they merely represent the largest, lon-
gitudinally oriented components of the posterior part of the
anastomotic pial network. Gillilan (1958) reaches the same
conclusion, but does not place special emphasis on these
vessels pursuing constant courses over quite a distance.
Microradiographs show that these anastomotic trunks have
less to do with the direct supply of certain medullary struc-
tures. Instead, they serve to distribute a constant blood sup-
ply in the dorsal part of the vasocorona. Evidence for this
fact is that the dorsal longitudinal anastomoses cannot only
complement each other by varying calibre, but also com-
pletely replace one another in certain segments. This is one
reason for conflicting opinions on their location. However,
it must be noted that the dorsolateral trunk is normally
more constant and has greater calibre. A pial arterial net-
work mediates between the longitudinal trunks and the ves-
sels that directly enter the spinal cord. Visualization of the
perforating arteries was not obtained by filling the longitu-
dinal anastomoses alone, but only by additionally filling
this superficial plexus. Although the pial plexus, fed by the
anterior spinal artery and the posterior longitudinal anasto-
moses, is extremely dense over the enlargements, the parts
of the vasocorona filled from ventral or dorsal are largely
independent of each other in post-mortem injections.
The pattern of arterial supply to the spinal cord, via a sys-
tem of longitudinal anastomoses reinforced from various
segmental levels, protects the neural tissue against ischemia.
Primarily the conus arcade constitutes an anastomotic func-
tion comparable to that of the cerebral circle of Willis. The
additional pial and intramedullary anastomoses may only be
important for slower circulatory adjustments, since their cal-
ibres are inadequate for a sudden intake of larger volumes of
blood. Other intramedullary anastomoses, e.g. longitudinally
and/or transversely between the central arteries, as postu-
lated at the arteriolar level by Adamkiewicz (1881), Suh and
Alexander (1939) and Fazio and Agnoli (1970), were not
detectable in our specimens.
The number, distribution and branching pattern of the
central arteries should no longer be a matter of dispute since
Kadyi (1889) recognized them as essentially unilateral arter-
ies occasionally branching from a short common stem.
Hassler (1966) observed by microradiographs that bifurca-
tion of a short common trunk takes place in the vertical
plane. This was corroborated in our material (Figs. 2.59 and
2.60a). Noteworthy is the observation also mentioned by
Hassler (1966) that very elongated, tortuous courses of the
central arteries can be observed in some specimens, espe-
cially in the lumbosacral region (Fig. 2.60b). Their possible
pathological or age-dependent significance, however,
remains to be clarified by corresponding histological studies.
Alterations in the vessel walls or luminal narrowing, as seen
in cases of arteriosclerosis, was not observed in any of our
specimens.
2.11 Anatomical and Haemodynamical Aspects of Arterial Supply
Considering the practical experience with spinal angiog-
raphy, doubts as to the validity of the “partial flow theory”
(Adamkiewicz 1882) are no longer reasonable. Variability in
vessel calibre of the ascending and descending branches of
the anterior radicular arteries, however, demonstrates that the
haemodynamics of these circulatory partial systems cannot
be strictly classified, and that the proportions of blood run-
ning cranially or caudally vary.
Suh and Alexander (1939) and later Zuelch (1954; Zülch
1976) postulated their traditional hypothesis of a watershed
zone of increased ischemic vulnerability near T4. This model
was based on the relative hypovascularity of this region.
Cheshire et al. (1996) described 44 patients with spinal
cord infarctions. Surprisingly, the mean sensory level of the
deficits was observed at T8 and T9 in cases of global isch-
emia. A review of the clinical literature by Cheshire and col-
leagues (1996) suggested that the most commonly sensory
level found after spinal cord infarction is centred at T12.
We avoided using terms such as “vulnerable zone” or
“undersupplied zone” in respect to the “poor supply” of the
upper thoracic region. Such designations can be traced back
to Adamkiewicz (1882). Kadyi (1889) subsequently rejected
them with good reasons, but Suh and Alexander adopted
them again in 1939. These terms were particularly revaluated
and extended in the 1950s by Zülch’s introduction of a mar-
ginal zone hypothesis (Zülch 1954, 1976). Numerical dis-
putes were the result, and the participating authors declared
almost every thoracic segment and many of the cervical ones
to be marginal zones or areas particularly vulnerable to
impaired blood flow. Since spinal vascular supply has been
clarified better by following studies, this hypothesis need not
be discussed in detail. Stochdorph (1961) criticized its fun-
daments with good reasons and it never found definite, path-
oanatomical proof (Jellinger 1966a, b). This is discussed
thoroughly in Schneider’s study and review (1980).
Our investigation also shows that no vascular anatomical
evidence supports the assumption of undersupplied areas.
Differences in calibre of the anterior spinal artery as well as
varying calibre and number of anterior radicular feeders
depend on locally variable blood requirements especially of
the grey matter. The only potential limit of anastomotic
capacity is a small-calibre segment or interruption in the
anterior spinal artery. This can be regarded as the cause of
spinal complications in aortic surgery and also in dissecting
aortic aneurysms (Hogan and Romanul 1966). Operative
complications only occur—according to Adams and
Geertruyden (1956)—if the great radicular artery entering in
the lower thoracic region is ligated upon serial elimination of
the intercostal arteries and if there is simultaneous, tempo-
rary occlusion of the aorta below the subclavian artery. Then,
the possible occurrence of medullary ischemia signals insuf-
ficient anastomotic capacity of the anterior spinal artery with
its sole remaining cervical supply. Nevertheless, ischemia
81
affects only the lower cord in these cases. On the other hand,
the anterior spinal artery can even function as a collateral of
the basilar artery, as shown in cases of occlusion of both ver-
tebral arteries. Furthermore, occlusion of the anterior spinal
artery without ensuing effects in cases of arteriovenous mal-
formation (AVM) treatment is not an unusual observation
(Decker et al. 1975; Riché et al. 1983). But this is not really
comparable with an exclusively spinal cord–supplying ante-
rior spinal artery occluded over a critical distance.
As a rule, the numerous extra- and intraspinal anastomoses
guarantee collateral supply in the presence of extravertebral
occlusion. According to animal experiments on rhesus mon-
keys (Fried and Aparicio 1973) this also applies to ligature of
the great radicular artery, unless its descending branch is
directly ligated. Experimental findings by Shimomura et al.
(1968) show that only occlusion of the anterior spinal artery
extending over 1–2 segments leads to ischemia. This corre-
lates fairly well to the pial longitudinal connections, which
we could demonstrate between several central arteries at the
anterior fissure parallel to the anterior spinal artery (Figs. 2.17,
2.46, 2.49, 2.50, and 2.60a). They are frequent in the thoracic
region, but not in the cervical or lumbar enlargements.
Deficits in the rare cases with proven occlusion of the
anterior spinal artery (Garland et al. 1966) vary considerably.
Only a part of these showed the expected lesions of the ven-
tral two-thirds of the transverse sectional surface (Schneider
1980). This accords with the general principle that develop-
ment of an infarct becomes more probable the more distal a
vessel is obstructed. Collaterals may not compensate for
embolic occlusion of a central artery, but obstruction of a
single artery will probably lack functional consequences
owing to an overlapping of neighbouring regions.
Piscol (1972) stressed the fact that a paucisegmental arte-
rial supply may function well in situations demanding rapid
anastomotic compensation, but can be deficient in haemody-
namical crises. This explains the verified necroses of the ante-
rior horn in the enlargements at autopsy following circulatory
arrest, which are consequently the most vulnerable structures.
Displacement even of the central arteries by ventral-
compressing masses (Ramsay and Doppman 1973) does not
lead to early ischemia, since they have some room within the
anterior fissure (Fig. 2.42). The experimental studies by
Doppman and Girton (1976) on the effects of intraspinal
masses support the opinion voiced by a neuropathologist
(Schneider 1980) that disorders in microcirculation play the
decisive role in the genesis of medullary vascular lesions.
2.11.2 Spinal Cord Ischemia and Infarct
Compared with brain ischemia, spinal cord strokes are
caused by more diverse etiologies. Up to now there is no
satisfactory and accepted classification of spinal infarcts.
82
Etiologies include circulatory arterial and venous disorders
(Thron et al. 1987; Mull and Thron 2006). From a clinical
and pathoanatomical point of view it seems reasonable to
differentiate between acute ischemic myelomalacia and
subacute to chronic vascular myelopathy. In most cases MRI
enables the differentiation of these two main etiologies. A
deficient spinal arterial blood flow generally has various
causes, ranging from the occlusion of intercostal or lumbar
arteries to affection of the intrinsic arteries of the spinal cord.
Many causes of acute spinal cord infarction (of arterial
and venous origin) have been reported. They include dis-
eases of the aorta and aortic surgery, thromboembolic events
and cartilaginous disc embolism, vasculitis, coagulopathy,
radiation-induced vasculopathy, toxic effects of contrast
medium, epidural anaesthesia, periradicular nerve root ther-
apy with crystalline corticoids (Dafotakis et al. 2008),
decompression illness, shock or cardiac arrest, lumbar artery
compression and other etiologies (Schneider 1980).
Cartilaginous disc embolism is meanwhile a widely
accepted cause of acute spinal cord ischemia mainly on the
basis of post-mortem findings (Pfeiffer et al. 1976; Mikulis
et al. 1992; Toro et al. 1994). Most reported post-mortem
cases are cervical, but conus medullaris infarctions are also
reported. The exact mechanism of disc embolism remains
unclear and speculative. An increased intraosseous pressure
within the vertebral body as a result of acute vertical disc
herniations—often after axial trauma—may explain the ini-
tial step in a process, which ends up in spinal cord ischemia
(Tosi et al. 1996). In the clinical course of the histologically
established cases of fibrocartilaginous embolism three main
characteristics could be identified: (1) the sudden severe pain
at onset, (2) the free interval of neurological signs and symp-
toms (“spinal stroke in progress”), (3) no improvement of the
neurological deficit. In our experience, there is a surprisingly
high rate of disc herniations in close proximity to the vascu-
lar lesions of the cord (Mull et al. 1992).
Lumbar artery compression by the diaphragmatic crus
has been described as a new aetiology for spinal cord isch-
emia (Rogopoulos et al. 2000). A prolonged hyperlordotic
position may play a predisposing role in this new syndrome.
The first right and left and the second right lumbar arteries
course through an osteotendinous passage between the verte-
bral body and the diaphragmatic crus. If spinal cord–supply-
ing arteries arise at the level of the first and second lumbar
artery, compromise of the spinal cord arterial supply can
result in spinal cord ischemia, especially if lordotic positions
reduce the space within the vertebral canal and intervertebral
foramen.
We had a young patient who could provoke weakness in
both legs by performing press-ups in a hyperlordotic posi-
tion. One day, repeating this exercise he experienced definite
paraparesis. Spinal angiography revealed an origin of the
great anterior radicular artery from the L2 level.
2 Arterial Blood Supply
Additional information on acute spinal cord ischemia can
be obtained from special literature on this subject (Mull et al.
1992, Mull and Thron 2006).
References
Adamkiewicz A (1881) Die Blutgefäße des menschlichen
Rückenmarkes. I. Teil: Die Gefäße der Rückenmarkssubstanz. S
Ber Akad Wiss Wien Math-Naturw Kl (Abt III) 84:469–502
Adamkiewicz A (1882) Die Blutgefäße des menschlichen
Rückenmarkes. II. Teil: Die Gefäße der Rückenmarksoberfläche. S
Ber Akad Wiss Wien Math-Naturw KI (Abt III) 85:101–130
Adams HD, van Geertruyden HH (1956) Neurologic complications of
aortic surgery. Ann Surg 144:574–609
Bolton B (1939) The blood supply of the human spinal cord. J Neurol
Psychiatry 2:137–148
Cheshire WP, Santos CC, Massey EW, Howard JF (1996) Spinal cord
infarction: etiology, and outcome. Neurology 47:321–330
Chiras J, Merland JJ (1979) The dorsospinal artery. A little known ana-
tomical variant. Its importance in spinal angiography. J Neuroradiol
6:93–100
Clavier E, Guiamaraems L, Chiras J, Merland JJ, Vasquez J (1987)
Isolated dorsospinal artery supplying anterior spinal artery.
Neuroradiology 29:213
Clemens HJ (1966) Beitrag des Morphologen zum Problem der spina-
len Mangeldurchblutung. Verh Dtsch Ges Inn Med 72:1059–1080
Corbin JL (1961) Recherches anatomiques sur la vascularisation artéri-
elle de la moelle: leur contribution a l’étude de l’ischémie médul-
laire d’origine arterielle. Masson, Paris
Crock HV (1996) Atlas of vascular anatomy of skeleton and spinal
cord. Martin Dunitz, London
Crock HV, Yoshizawa H (1977) The blood supply of the vertebral
column and spinal cord in man. Springer, Wien/New York
Crock HV, Yamagishi M, Crock MC (1986) The conus medullaris
andcauda equina in man. An atlas of the arteries and veins. Springer,
Wien/New York
Dafotakis M, Fink GR, Schlangen C, Block F, Sparing R, Thron A
(2008) [Acute spinal cord ischemia following deep injection ther-
apy of the neck] Nervenarzt 79:927-931 [Article in German]
Decker RE, Stein HL, Epstein JA (1975) Complete embolization of
artery of Adamkiewicz to obliterate an intramedullary arteriovenous
aneurysm. J Neurosurg 43:486–489
Di Chiro G, Harrington T, Fried LC (1973) Microangiography of
human fetal spinal cord. Am J Roentgenol 118:193–199
Dommisse GF (1975) The arteries and veins of the human spinal cord
from birth. Churchill Livingstone, Edinburgh/London/New York
Doppman J, Girton M (1976) Angiographic study of the effect of lami-
nectomy in the presence of acute anterior epidural masses.
J Neurosurg 45:195–202
Fazio C, Agnoli A (1970) The vascularization of the spinal cord.
Anatomical and pathophysiological aspects. Vasc Surg 4:245–257
Fried LC, Doppman JL, Di Chiro G (1970) Direction of blood flow in
the primate cervical spinal cord. Neurosurgery 33: 325-330
Fried LC, Aparicio O (1973) Experimental ischemia of the spinal cord.
Histologic studies after anterior spinal artery occlusion. Neurology
23:289–293
Garland H, Greenberg J, Harriman DGF (1966) Infarction of the spinal
cord. Brain 89:645–662
Gillilan LA (1958) The arterial blood supply of the human spinal cord.
J Comp Neurol 110:75–103
Hamilton WJ, Boyd JD, Mossman HW (1959) Human embryology
(prenatal development of form and function), 3rd edn. Heffer,
Cambridge
References
Hassler O (1966) Blood supply to human spinal cord. Arch Neurol
15:302–307
His W (1887) Zur Geschichte des menschlichen Rückenmarks und der
Nervenwurzeln. Abh Math Phys Klasse Sächs Ges Wiss (Leipzig)
13:477
Hogan EL, Romanul FC (1966) Spinal cord infarction occurring during
insertion of aortic graft. Neurology (Minneap) 16:67–74
Jellinger K (1966a) Zur Orthologie und Pathologie der Rückenmarks-
durchblutung. Springer, Wien/New York
Jellinger K (1966b) Experimentelle Untersuchungen zur Frage der
arteriellen Versorgungsgebiete des Rückenmarks. Acta Neuropathol
6:200–220
Kadyi H (1889) Über die Blutgefäße des menschlichen Rückenmarks.
Gubrynowicz u Schmidt, Lemberg
Lasjaunias P, Berenstein A (1990) Spinal and spinal cord arteries and
veins. Surgical neuro-angiography 3: functional vascular anatomy
of the brain, spinal cord and the spine. Springer, Berlin
Lasjaunias P, Vallee B, Person H, TerBrugge K, Chin M (1985) The
lateral spinal artery of the upper cervical spinal cord. J Neurosurg
63:235–241
Lasjaunias P, Berenstein A, ter Brugge K (2001) Surgical neuroangiog-
raphy. Volume 1: clinical vascular anatomy and variations, 2nd edn.
Springer, Berlin
Lazorthes G, Poulhes J, Bastide G, Rouleau J, Chancollé AR (1958) La
vascularisation artérielle de la moelle. Recherches anatomiques et
applications à la pathologie médullaire et à la pathologie aortique.
Neuro-Chirurgie 4:3–19
Lazorthes G, Poulhes J, Bastide G, Rouleau J, Chancollé AR, Zadek O
(1962) La vascularisation de le moelle épinière. (Etude anatomique
et physiologique). Rev Neurol 106:535–557
Lazorthes G, Gouazé A, Zadeh O, Santini JJ, Lazorthes Y, Burdin
Ph (1971) Arterial vascularization of the spinal cord. Recent
studies of the anatomic substitution pathways. J Neurosurg 35:
253–262
Manelfe C, Lazorthes G, Roulleau J (1972) Artères de la dure-mère
rachidienne chez l’homme. Acta Radiol 13:829–841
Mikulis DJ, Ogilvy CS, McKee A, Davis KR, Ojeman RG (1992)
Spinal cord infarction and fibrocartilaginous emboli. AJNR Am
J Neuroradiol 13:155–160
Mosimann PJ, Stauder M, Nordmeyer H, Chapot H (2014) Isolated
emergence of a dorsospinal artery supplying the Adamkiewicz
artery. Clin Neuroradiol 34:43
Mull M, Thron A (2006) Spinal infarcts. In: von Kummer R, Back T
(eds) Magnetic resonance imaging in ischemic stroke. Springer,
Berlin, pp 251–267
Mull M, Thron A, Petersen D, Stoeter P (1992) Acute infarction of the
spinal cord: clinical and neuroradiological findings in 20 patients.
Neuroradiology 29:85–88
Noeske K (1958) Über die arterielle Versorgung des menschlichen
Rückenmarkes. Gegenbauers Morph Jb 99:455–497
Pfeiffer J, Wenig C, Mäusle E (1976) Akutes Querschnittssyndrom
durch Embolien von Nucleus-pulposus-Gewebe. Dtsch Med
Wochenschr 101:583–586
Piscol K (1972) Die Blutversorgung des Rückenmarks und ihre
klinische Relevanz. Springer, Berlin/Heidelberg/New York
Ramsay R, Doppman JL (1973) The effects of epidural masses on spi-
nal cord blood flow. Radiology 107:99–103
Riché MC, Scialfa G, Gueguen B, Merland JJ (1983) Giant extramedul-
lary arteriovenous fistula supplied by the anterior spinal artery:
83
treatment by detachable balloons. AJNR Am J Neuroradiol
4:391–394
Rogopoulos A, Benchimol D, Paquis P, Mahagne MH, Bourgeon A
(2000) Lumbar artery compression by the diaphragmatic crus: a
new etiology for spinal cord ischemia. Ann Neurol 48:261–264
Roll D (1958) Über die Arterien der Pars caudalis des menschlichen
Rückenmarkes und das Vorkommen arterio-venöser Anastomosen
im Stromgebiet der A. radicularis magna. Gegenbaurs Morph Jb
99:425–454
Romanes GJ (1965) The arterial blood supply of the human spinal cord.
Paraplegia (Edinb) 2:199–207
Schneider A (1980) Kreislaufstörungen und Gefäßprozesse des
Rückenmarks. In: Doerr W, Seifert G (eds) Spezielle patholo-
gische Anatomie, Bd 13/1. Springer, Berlin/Heidelberg/New York,
pp 511–649
Shimomura Y, Hukuda S, Mizuno S (1968) Experimental study of
ischemic damage to the cervical spinal cord. J Neurosurg
28:565–581
Sterzi G (1904) Die Blutgefäße des Rückenmarks. Untersuchungen
über ihre vergleichende Anatomie und Entwicklungsgeschichte.
Anat Hefte I Abt 24:1–364
Stochdorph O (1961) Zur Deutung histologischer Befunde bei chro-
nischen Kreislaufstörungen des Rückenmarks. Acta Neurochir
(Wien) [Suppl] 7: 386–387
Suh TH, Alexander L (1939) Vascular system of the human spinal cord.
Arch NeurPsych 41:659–677
Thron A (1988) Vascular anatomy of the spinal cord. Neuroradiological
investigations and clinical syndromes. Springer, Wien/New York
Thron A (2002) Vascular anatomy of the spine. In: Byrne J (ed)
Interventional neuroradiology. Oxford University Press, Oxford
Thron A (2008) Vascular anatomy of the spine and spinal cord. In:
Hurst RW, Rosenwasser RH, (eds) Interventional Neuroradiology.
Informa Health Care, New York/London
Thron A, Koenig E, Pfeiffer P, Rossberg C (1987) Dural vascular anom-
alies of the spine – an important cause of progressive radiculomy-
elopathy. In: Cervos-Navarro J, Ferszt R (eds) Stroke and
microcirculation. Raven, New York
Toro G, Roman GC, Navarro-Roman L, Cantillo J, Serrano B, Vergara
I (1994) Natural history of spinal cord infarction caused by nucleus
pulposus embolism. Spine 19:360–366
Torr JBD (1957a) The arterial supply of the foetal spinal cord. J Anat
(London) 91:576
Torr JBD (1957b) The embryonal development of the anterior spinal
artery in man. J Anat (London) 91:587
Tosi L, Rigoli G, Beltramello A (1996) Fibrocartilaginous embolism of
the spinal cord: a clinical and pathogenetic reconsideration. J Neurol
Neurosurg Psychiatry 60:55–60
Turnbull JM, Breig A, Hassler O (1966) Blood supply of cervical spinal
cord in man; a microangiographic cadaver study. J Neurosurg 24:
951–965
Tveten L (1976a) Spinal cord vascularity. Extraspinal sources of spinal
cord arteries in man. Acta Radiol Diagn 17:1–16
Tveten L (1976b) Spinal cord vascularity. The spinal cord arteries in
man. Acta Radiol Diagn 17:257–273
Zülch KJ (1954) Mangeldurchblutung an der Grenzzone zweier
Gefäßgebiete als Ursache bisher ungeklärter Rückenmarksschä-
digungen. Dtsch Z Nervenheilk 172:81–101
Zülch KJ (1976) Pathogenetic and clinical observations in spinovascu-
lar insufficiency. Zbl Neurochir 37:1–13
 Venous Drainage
The pattern of venous drainage substantially deviates from
that of the arteries (Figs. 3.1 and 3.2). It will be described in
the direction of venous drainage from the spinal cord paren-
chyma to the epidural plexuses.
Four compartments, representing four different levels of
intercommunicating drainage from the spinal cord and from
the spine, can be discerned:
• The intrinsic veins of the spinal cord parenchyma.
• The superficial pial system covering the surface of the
cord, comprising a network of small collecting veins. It is
complemented by large, longitudinally oriented channels
carrying the blood to the dural border of the thecal sac.
Additionally, numerous intra- and extramedullary anasto-
motic connections between all components guarantee
flexible blood flow deviation.
• The extradural but intraspinal internal venous vertebral plexus.
• The extradural and extraspinal venous vertebral plexus
communicating with the azygos and cava system.
© Springer International Publishing 2016
A.K. Thron, Vascular Anatomy of the Spinal Cord: Radioanatomy as the Key to Diagnosis and Treatment,
DOI 10.1007/978-3-319-27440-9_3
3
Technical Remarks
In comparison to the spinal arteries, the variability of the
draining venous system is even greater. Due to a restricted
number of specimens (Table 1.1 and Sect. 1.3), the pre-
sented images in this chapter should be considered rather
as illustrative examples than as statistically confimed
results. For the veins, we re-evaluated the radicular drain-
age, which is easier on the opened vertebral canal in situ
(Jacobs 1996). Concerning the intramedullary system of
veins importance was attached to the most complete venous
filling possible for large medullary segments, which
requires even better preserved specimens and much practi-
cal experience to avoid extravasations. Filling defects in
some of the specimens are probably a result of intraluminal
coagulation. On the other hand, a loss of contrast medium
from large superficial vessels upon sectioning must be
expected. In certain illustrative microangiograms the empty
lumen of a large superficial vessel was therefore marked
with “ev” (empty vessel).
85
86 3 Venous Drainage

Post. median
vein

* *Midline
anastomosis

Post. radiculo- Radial veins

medullary vein

Ant. radiculo-
medullary vein
Pial venous network
of lower degree

Ant. median vein

Fig. 3.1 Diagram of superficial veins, large anastomotic midline
connections and intrinsic veins. Radially oriented peripheral (and
central) veins within the spinal cord parenchyma are connected to a pial
venous network of lower degree. Superficial pial tracts of drainage, the
large longitudinal venous tracts, typically, but not exclusively in a
midline position, collect the blood. The inconstantly developed superfi-
cial midline veins may be supplemented or replaced over some distance
by superficial venous trunks in an antero- or posterolateral position.
They merge with anterior or posterior radiculomedullary veins (r.m.v.)
at different levels

tal
git
Sa

Axi
al

Fig. 3.2 Course and distribution of intrinsic veins in axial and sagittal
planes. The characteristic features of the system are radially symmetric,
largely horizontally oriented veins of the peripheral system with differ-
ent depths of penetration. Central veins play a minor role. Different
transparenchymal anastomotic channels in an anteroposterior (a.p.) or
centro-posterolateral direction (on top of the diagram) are outlined
3.1 Intrinsic Veins
3.1 Intrinsic Veins
Radial symmetry and dominance of intrinsic peripheral veins
is evident in many parts of the spinal cord (Figs. 3.1, 3.3, 3.6,
and 3.23). Mainly in the lower thoracolumbar cord, includ-
ing the conus medullaris, do veins of the anterior fissure play
an accentuated role and may display vessel diameters of up
to 0.3 mm (Figs. 3.4, 3.29, and 3.31).
Therefore it seems justified to maintain the differentiation of
– central veins and
– peripheral veins
similar to the classification of the arterial system.
However, compared with arterial supply, an essentially
different orientation of intrinsic drainage paths is obvious.
Generally, intrinsic veins running along with arteries (venae
comitantes) only occur along the proximal stem of the central
arteries.
Peripheral veins tend to take a separate course, starting
deep in the grey matter. In comparison with the numerous
perforating branches of the arterial pial network, the radial
veins are fewer in number but exhibit larger calibres. Their
territory of drainage covers not only the outer rim of the fibre
tracts. They also collect blood from the central grey matter
area (Figs. 3.3, 3.15, and 3.25). Therefore their calibre equals
or exceeds that of central veins (Figs. 3.4 and 3.26), at least
in the cervical and thoracic regions (Kadyi 1889).
This basic radial drainage pattern becomes evident in
axial as well as sagittal and frontal sections from most speci-
mens of these regions (Figs. 3.4, 3.5, 3.6, 3.7, and 3.8). In
the higher medullary segments, central veins may be more
numerous, but more often they are of smaller calibre than the
corresponding arteries. If veins of large calibre are observed
in the anterior fissure at greater intervals, they are usually
the anterior part of an anastomosis of large calibre to the
dorsal or dorsolateral surface (Figs. 3.16, 3.17, 3.18, and
3.25; Sect. 3.3). Tortuous central veins that drain the white
commissure and anterior central parts of the grey matter
often join to a common stem deep in the anterior fissure. A
common trunk is at least more frequent than for the arter-
ies (Figs. 3.28, 3.29, and 3.30). Longitudinal anastomoses of
central veins occur only at the entrance into the fissure, that
is, directly under the main anterior median vein (Fig. 3.30b).
Small peripheral veins (so-called marginal branches) that
originate in the white matter take a short course to the sur-
face of the spinal cord. Other peripheral veins with diam-
eters of 0.1 mm to more than 0.2 mm are larger and longer
than the perforating arteries of the vasocorona and extend
deep into the grey matter. They are morphologically equiva-
lent to the central veins. This is clearly indicated not only by
87
their almost equal calibres but also by the frequent anasto-
motic connections between both of them (Figs. 3.16, 3.20,
3.24b, and 3.26). However, this does not refer to even larger
transmedullary connections between the anterior and poste-
rior main venous trunks (Sect. 3.3).
As shown in X-rays of axial sections, a nomenclatural
classification of the veins draining the medullary transverse
sectional area is no less problematic than that of the perforat-
ing arteries, since an almost arbitrary number of venous
trunks of approximately the same calibre can be found.
Owing to the constancy of their occurrence, location in the
anterior fissure and dominant calibre in the lumbosacral
region, the central veins are of special importance that
deserves mention. Furthermore, of the peripheral veins, only
those of the posterior median septum should receive special
attention because of their length (Figs. 3.3, 3.5, 3.6, and 3.8).
Despite their inconsistent yet very remarkable calibres in
certain sections, they were nevertheless hard to fill in our
injection specimens.
Since the medullary territories drained by the central
veins and the peripheral veins do not coincide with the sup-
ply regions of the central and peripheral arterial systems, an
analogous classification should not falsely suggest an equal
hemodynamical significance. The major impact for the
drainage of the whole cross-sectional area has to be attrib-
uted to the peripheral system because the long radial vessels
drain from the central grey matter and along their passage to
the surface receive branches from the white matter, including
veins of the posterior white columns.
All larger veins extending through the white matter into the
medullary grey matter show a typical form of branching that
Kadyi correctly characterized as a “compact, profusely rami-
fied, small tree trunk” (Figs. 3.5, 3.6a, and 3.8b). By contrast,
the central veins do not branch until they reach the white com-
missure and the central grey matter (Figs. 3.3, 3.4, and 3.29).
Another difference between the intrinsic arterial and venous
system with respect to the extent of the longitudinal territo-
ries is found in the thoracic region. There, larger veins may
sometimes ascend obliquely, yet show no significant branch-
ing into long ascending or descending vessels (Fig. 3.27).
In summary, the intrinsic medullary drainage pattern can
be characterized as a radially symmetric, largely horizon-
tally oriented system. Mainly in the section of the thoraco-
lumbar enlargement it exhibits a slight dominance towards
the anterior or posterior midline veins yet drains mainly via
radial peripheral veins and numerous oblique and trans-
verse anastomoses almost uniformly to the surface. Not until
the level of the spinal pia mater is blood accumulated in
larger longitudinal trunks, of which the anterior and poste-
rior median spinal veins are the most constant vessels with
the greatest calibre.
88 3 Venous Drainage

a c

b d

Fig. 3.3 Characteristic features of intrinsic venous drainage.
Microangiograms of axial sections from various levels of the mid and
lower thoracic region. (a, b) Symmetrically oriented peripheral veins of
different length drain the deep central parts of the spinal cord as well as
the parts near to the surface on almost the complete cross-sectional area.
(c, d) Veins of striking larger calibre running within the midline struc-
tures might be veins of the posterior septum (c, arrow) or central veins
running in the anterior fissure (d, arrow). If they do not receive tributar-
ies they might also be parts of dissected anastomotic connections
between superficial (compare Fig. 3.25). The large superficial venous
trunks of this specimen are in variant positions
3.1 Intrinsic Veins 89

Fig. 3.4 Comparison between peripheral and central veins at

a
cervical (a), thoracic (b) and lumbar (c) levels. Central veins are
generally less important in the pattern of venous drainage. The main
part of the cross-sectional area is drained by the horizontally oriented
peripheral veins (Reproduced from Thron 1988)

C6

T3

L5
90 3 Venous Drainage

a b

Fig. 3.5 Microangiograms of intrinsic and superficial veins of the
cervical spinal cord (incomplete filling of small vessels in some cir-
cumscribed posterior areas of the posterior columns and septum).
(a) Mediosagittal section at midcervical level. Course and ramifica-
tion of the central veins anteriorly (small arrows) differ very much
from that of the larger veins of the posterior septum and columns
(arrowheads). Typical mediosagittal a.p. anastomosis (large arrow).
Anterior and posterior longitudinal venous trunks course in midline
position. (b) Mediosagittal section of the lower cervical cord. The
incomplete filling of the posterior column area offers a unique image
of the tree-like branching pattern of these compact, straightly cours-
ing veins in the posterior septum (arrowheads). Numerous smaller
central veins descend in the CSF-space of the anterior fissure (arrow)
(a Reproduced from Thron 1988)
3.1 Intrinsic Veins 91

a b

Fig. 3.6 Intrinsic spinal cord veins. Microangiograms at the lower cervi- compared to the large number of peripheral veins and strong veins of
cal level. (a) Mid-sagittal section at the C7–C8 level. Only contrast rem- the posterior septum penetrating deeply into white and grey matter. (b)
nants are visible in the anterior and posterior median spinal veins. Central Paramedian sagittal section. The predominant role of slightly descending
veins can be identified (arrows), but they seem to be of minor importance peripheral veins of varying size is impressively demonstrated in this image
92
S1
a b
Fig. 3.7 Microangiograms of frontal sections from the upper thoracic
spinal cord C8–T3. (a) Anterior frontal section. The large median
and anterolateral pial veins on the anterior surface contain only little
contrast medium (arrow). The intrinsic peripheral veins are symmetri-
cally arranged and take a horizontal or slightly oblique course in the
3 Venous Drainage
S2
c
craniocaudal direction. (b) Mid-frontal section. There are only scant
central veins. Midline vessels of larger calibre are supposed to be anas-
tomotic channels. (c) Posterior frontal section. Serpentine superficial
vein (arrow) (Reproduced from Thron 1988)
3.1 Intrinsic Veins 93

a b

Fig. 3.8 Central veins and posterior column veins at lumbosacral lev-
els. (a) Microangiogram of a mid-sagittal section in the region near the
cone, obtained following injection of stronger diluted contrast medium.
Central veins occur in higher number and larger calibre at lumbosacral
levels compared to other spinal cord regions. Additionally, a high
density of smaller peripheral veins is obvious in the nervous tissue of
the lumbar enlargement. (b) Microangiogram of a mid-sagittal section
of the posterior columns and the posterior septum at low thoracic levels
of the same specimen. The high density and tree-like configuration of
peripheral veins in this area is clearly demonstrated
94
3.2 Superficial Veins
As demonstrated by the dorsal aspect of a spinal cord in situ
(Fig. 3.13b, c) and the radiographs in Figs. 3.9, 3.14, 3.29,
3.30, 3.31, 3.32, and 3.33 a reliable systematization of the
superficial venous vessels is difficult. Kadyi’s anatomical
drawings (Figs. 3.10 and 3.11) are a perfect compromise to
explain the principle and the margins of spinal cord venous
architecture. Apparently, there is again no well-defined par-
allelism to the superficial arteries. This impression is gener-
ated mainly by thick, convoluted venous formations on the
posterior side of the spinal cord and by two longitudinally
directed vessels which merge with radicular veins at differ-
ent levels, and whose sections of largest calibre run medi-
ally in some segments, in some paramedially. Despite the
great variety, common characteristics can be observed, and
basic differences in comparison to the pattern of arterial
supply can be demonstrated. Moes and Maillot (1981) have
presented a very thorough study on these superficial vessels
with an essay of systematization.
Two main systems can be differentiated within the pia
mater:
1. The longitudinal venous trunks (Tractus venosus media-
nus anterior et posterior), also called the anterior and
posterior median spinal veins as the most common longi-
tudinally directed vessels of the spinal cord. Anterolateral
and posterolateral venous tracts partially replace the
median trunks at least for short distances.
2. The pial venous networks of lower degree collecting
blood from the intrinsic peripheral and less important
central veins.
3.2.1 Longitudinal Venous Trunks
The anterior median vein receives the draining vessels from
the ventral fissure and can be considered as the vena comi-
tans of the anterior spinal artery.
This mostly continuous tract is located in the anterior fis-
sure where it is running below the anterior spinal artery
(Fig. 3.12a). It starts at the craniocervical junction, has its
largest calibre lumbosacrally and proceeds as a sometimes
very large terminal vein in 60–80 % of the cases (Kadyi
1889; v. Quast 1961; Thron 1988). It continues parallel with
3 Venous Drainage
the filum terminale to the end of the dural sac and can actu-
ally be regarded as a coccygeal radicular vein. Occasionally,
a sacral radicular vein of equally large calibre assumes this
function. In this case, the terminal vein is then less well
developed (Figs. 3.31 and 3.33e).
The remainder of the superficial veins exhibit courses
independent of the arteries. The most important of these ves-
sels is the posterior median spinal vein. It is significant in the
cervical region, especially large and sometimes convoluted
and dilated above the thoracolumbar enlargement and it ends
at the level of the cone (Figs. 3.11, 3.31, 3.33, and 3.34). As
a rule, this large-calibre, posterior drainage system is formed
especially in the thoracic region by longitudinal vessels that
cannot be schematically standardized. They are sometimes
larger, sometimes smaller, now running more to the left, then
more to the right at the dorsal circumference. Varicose
convolutions are frequent and can also appear as extremely
tortuous, radicular venous convolutions, as seen in Figs. 3.28
and 3.32.
Venous circles at the conus medullaris, connecting ante-
rior and posterior channels are also found (Fig. 3.33c, d).
However, they are often supplemented or replaced by intra-
medullary anastomoses and are much less constant than the
arterial conus arcade. The contrast in calibres of the arteries
and veins is striking. Even if one takes into consideration
that veins can dilate unphysiologically under injection pres-
sure, the figures obtained here are within the dimensions pre-
viously reported (Table 1.2). The veins of the thoracolumbar
enlargement, with a diameter up to 1.5 mm, are undoubtedly
the medullary vessels of largest calibre and are only rarely
matched by superficial cervical veins (Figs. 3.19, 3.24, and
3.31).
3.2.2 Pial Venous Networks of Lower Degree
Ventrally as well as dorsally and laterally, short, secondary
systems of smaller calibre exist. Of these, anterolateral veins
show the greatest continuity, replacing the median vein in its
function if it is interrupted (Figs. 3.11a, 3.12, 3.25, and 3.26).
Superficial transverse vessels connect the longitudinal ones,
although not as densely as in the case of the arteries. Moes
and Maillot suggest to differentiate between a truncal and
plexiform type of drainage especially in the cervical region
(Fig. 3.10).
3.2 Superficial Veins
a b
Fig. 3.9 Characteristics of superficial spinal cord veins. (a, b) X-ray
films of an injected specimen in a.p. (a) and lateral view including a
schematic illustration (b). The superficial longitudinal trunks anteriorly
and posteriorly are typically located in the midline. They drain the pial
venous networks of lower degree. In the lower half of the spinal cord the
anterior median vein follows the filum terminale as terminal vein. The
posterior median vein often exhibits a plexiform convolution of large
95
veins on the posterior surface at low thoracic levels. It is called the
posterior venous plexus. In certain regions of the cord frequent anasto-
motic channels of large calibre exist. They interconnect the main longi-
tudinal trunks through the midline of the spinal cord parenchyma as
shown on the diagram and on the lateral X-ray, some of them marked
with arrowheads (b) (X-rays reproduced from Thron 1988)
96 3 Venous Drainage

a b a b

Fig. 3.11 Superficial veins of the anterior (a) and posterior (b) surface
in the lower half of the spinal cord. In this anatomical illustration, large
anterior and posterior midline veins drain the lumbar enlargement. (a)
The anterior midline vein follows the filum terminale. Two large radicu-
lomedullary veins accompany low lumber or sacral nerve roots to exit
the dural sac (compare Fig. 3.33e). (b) The posterior veins are stronger
and more convoluted in the lower thoracic area (posterior venous
plexus). The posterior midline vein terminates at the level of the cone.
Fig. 3.10 Superficial veins of the anterior (a) and posterior (b) surface
(Reduced reproduction from Kadyi’s monograph, 1889; Tafel II Figs. 5
in the upper half of the spinal cord, including the craniocervical junc-
and 6; lower half. The unsurpassed realistic and instructive anatomical
tion. Note the convoluted veins of the posterior venous plexus and con-
drawings were designed by himself and realized together with the
joint radicular veins or radicular vein complexes on the posterior
painter Heinrich Dyrdon)
surface at level T4/5 and T5/6. (b) (Reduced reproduction from Kadyi’s
monograph, 1889; Tafel II Figs. 5 and 6; upper half. The unsurpassed
realistic and instructive anatomical drawings were designed by himself
and realized together with the painter Heinrich Dyrdon)
3.2 Superficial Veins 97

a b

Fig. 3.12 Superficial veins of the spinal cord. Photographs of injected
spinal cord specimens. (a) Photograph of the anterior aspect at tho-
racic level. The hairpin configuration in the course of venous outflow,
formed by the connection between anterior median vein (thick arrow)
and radiculomedullary vein (slim arrow) is very similar to the arterial
counterpart. The dark blood vessel (small arrowhead) is the non-filled
anterior spinal artery, running beside or over the vein (large arrowhead)
(Reproduced from Thron 2008). (b) Photograph of the large venous
trunk at the posterior aspect at thoracic level. Cranial to the posterior
radiculomedullary vein the median trunk is replaced by a secondary
system of pial veins
98
a
Fig. 3.13 Superficial spinal cord veins on the posterior surface. (a)
Simplified diagram. A more or less continuous posterior median vein,
exhibiting plexiform parts especially on the posterior surface of the
lower thoracic region, extends over the whole length of the spinal cord.
It gets smaller at the cranial and caudal end of the medulla spinalis. In
between, numerous radiculomedullary veins provide outflow to the epi-
dural venous plexus (symbolized by the blue rectangular blocks). (b)
Photograph of a specimen at cervical and thoracic levels. The posterior
median spinal vein is continuous from the cervical to the lower thoracic
3 Venous Drainage
b c
region, but shows a variable size at different levels. Four large r.m.v.
and some smaller ones can be seen. c Photograph of the dorsal aspect
of a spinal cord in situ (post-mortem) at thoracic and lumbosacral level.
There are discontinuous longitudinal veins of median and parame-
dian location. The vessels show tortuosities and varicose convolutions
above the lumbar enlargement. One of the r.m.v. can be attributed to the
“great radiculomedullary vein” (arrow). (b Reproduced from Jacobs
1996; c Courtesy of Prof. Dauber, Institute of Anatomy, University of
Tuebingen, Germany)
3.2 Superficial Veins
a b
Fig. 3.14 Superficial veins of the spinal cord. X-ray film of two
injected specimens. (a) X-ray in a.p. view. The lower cervical and upper
thoracic part of this specimen demonstrates typical drainage through a
continuous longitudinal midline tract and several radiculomedullary
veins. These veins follow a nerve root and empty into the internal ver-
tebral venous plexus. The descending courses of radiculomedullary
veins result in hairpin configurations, which are very similar to the nar-
row curve of radiculomedullary arteries. (b, c) A.p. and lateral view of
99
c
the lower half of an injected spinal cord. Three anterior radiculomedul-
lary veins and a small posterior one at lumbosacral level are exhibited
in this specimen. The anterior midline tract is more pronounced than the
posterior one and includes plexiform at the low thoracic level as well as
a radicular vein complex at mid-thoracic level on the left side. Large
anastomotic tracts at different sides and levels of the cone, comparable
to the arterial arcades around the cone, are additional unusual findings
(arrows) (a Reproduced from Jacobs 1996)
100 3 Venous Drainage

3.3 Transmedullary Venous Anastomoses ascend and descend in wide arcs or peaks (Figs. 3.22 and
3.41). In axial sections, lateral bending around the central
Venous channels within the spinal cord parenchyma connect canal is typical (Fig. 3.17). Guided by these characteristic
the superficial venous tracts. Along their course they show a courses, the large, transmedullary venous anastomoses can
constant calibre, and the large ones do not receive intrinsic be distinguished from superficial transverse anastomoses
tributaries. They have preferential sites, different courses and on X-rays of injected spinal cords.
vary considerably in diameter. They are not distributed uniformly over the length of a
Two types of these venous anastomoses can be distinguished: spinal cord. Their greatest incidence is found in the high tho-
racic region, where up to two such anastomoses per centime-
Type 1: Centro-dorsolateral anastomoses Frequent oblique tre can be detected (Fig. 3.9).
anastomotic connections between peripheral or central This will be described in detail in the Sects. 3.4–3.6 about
and peripheral branches (0.1–0.2 mm) (Figs. 3.16, 3.18, specific patterns of drainage in the different spinal cord regions.
3.20, 3.21b, and 3.26). In the mid and lower thoracic regions, internal venous
Type 2: Midline anastomoses They connect the main superfi- anastomoses of large calibre occur at greater distances. As of
cial venous tracts without receiving intrinsic tributaries yet, none were found to have any significant calibre in the
(0.3–0.7 mm) (Figs. 3.1, 3.2, 3.5, 3.9, 3.17, 3.22, and 3.27). lumbar enlargement, although in one case we detected an
abnormally large anastomosis near the conus, replacing super-
Kadyi (1889) as well as Herren and Alexander (1939) already ficial anastomoses or running next to these (Fig. 3.22d).
described the variable vessels of 0.1–0.2 mm thickness connect-
ing central and peripheral branches (Fig. 3.15).
In addition, anastomoses of much larger calibre run
between the superficial longitudinal trunks through the spi-
nal cord. To our knowledge Crock and Yoshizawa were the
first to mention and illustrate this type of venous channel in
their book from 1977.
Our radiological evaluation of a large number of contrast-
injected human spinal cord specimens clearly shows that
transmedullary venous anastomoses are an essential compo-
nent for the venous return of the spinal cord and that they are
not rare, occasional findings. Instead, they constitute an
essential principle of superficial and intrinsic spinal cord
drainage: anastomotic pathways and interconnections.
The diameters of Type 2 anastomoses approach those of
the superficial main channels (0.3–0.7 mm). These veins
receive few if any veins draining directly from the spinal
cord. Their assumed function could be to provide collateral
pathways for the venous return of blood from the spinal cord.
Their sinuously oscillating, mostly anteroposterior
courses visible in thicker mediosagittal sections make it
clear why they are missed in thin, axial as well as coronary Fig. 3.15 Intrinsic venous anastomosis type 1 at high cervical level
(arrowhead). Anastomoses of 0.1–0.2 mm between different peripheral
sections (Figs. 3.17, 3.18, 3.19, 3.20, 3.21, and 3.28). They
veins or branches of central and radial veins are frequent findings. They
can show varicose enlargement and elongation similar to often course central areas to the dorsolateral surface. They receive intra-
the superficial veins (Fig. 3.19), or just run straight, or medullary tributaries (Reproduced from Thron 1988)
3.3 Transmedullary Venous Anastomoses
a b
Fig. 3.16 Intrinsic venous anastomoses type 1 demonstrated on sag-
ittal microangiograms at thoracic level. (a) Right paramedian sec-
tion. (b) Mid-sagittal section. (c) Left paramedian section. Several
dissected venous anastomoses between radial veins can be identified.
The small and large arrowheads point to the vessel segments,
belonging to two continuous anastomotic veins. These vessels take
different ascending and descending oblique courses through the
101
c
spinal cord substance and receive smaller tributaries. Peripheral
veins in anterolateral or posterolateral location (large and small
arrow in c) may exhibit very different configurations. The unique
tree-like ramification of the veins of the posterior septum and col-
umns, demonstrated on the midsagittal section in (b) is again very
different (b Reproduced from Thron 1988)
102
a d
Fig. 3.17 Large transmedullary anteroposterior venous anastomosis
type 2 demonstrated on a series of axial microangiograms (a–d) and
outlined in a diagram (e). (a–c) The anastomotic venous channels
through the midline of the cord show ascending and descending
courses. Therefore, several contiguous axial sections are needed to
reconstruct the vessel through the spinal cord substance between the
two midline veins (arrowheads in a, b and c). (d) Image obtained by
3 Venous Drainage
superimposing a and c. The short parts of the anastomosis included in
b have to be added (arrowheads). The deviation of the anastomotic
midline veins type 1 around the central canal is best demonstrated in the
axial view (c). (e) Illustrative diagram of the different courses of
anastomotic vessels type 1 (on Top) and type 2 (in the sagittal plane)
(a, c Reproduced from Thron 1988)
3.3 Transmedullary Venous Anastomoses
a b c
Fig. 3.18 Frequency of occurrence and variability of anastomotic
channels type 2 in the thoracic spinal cord. (a, b) X-ray in a.p. and lat-
eral view. A few of the many anastomoses type 2 exhibited in the lateral
view of the X-ray (b) were marked with an arrow. Number and distribu-
tion of type 2 intrinsic venous anastomoses can best be estimated on the
lateral view of X-rays. The often steeply ascending and descending
courses near the central canal make it clear that the vessels are regularly
103
dissected on axial sections of 2–4 mm thickness and that they are very
difficult to follow on coronal sections. (c, d) Two axial microangio-
grams at the mid-thoracic levels indicated in a, b. At least two contigu-
ous slices are needed to follow the anastomotic vessel course
(arrowheads in c and d). More dissected anastomotic connections can
be supposed in view of additional venous segments of large calibre
between the smaller peripheral veins
104 3 Venous Drainage

b c

Fig. 3.19 Varicosity of intrinsic venous midline anastomoses type 2 at veins (vertical arrow in b and c). A similarly large and tortuous anasto-
upper thoracic level. (a, b) X-ray in a.p. and enlarged in lateral view. (c) motic vein can be identified on the lateral view of an X-ray of the
Microangiogram of the mid-sagittal section at the T3 level. A large- contrast-injected specimen shown in b, based on the typical course and
sized tortuous midline anastomosis connects the superficial midline calibre (oblique arrow) (Reproduced from Thron 1988)
3.3 Transmedullary Venous Anastomoses
a b
Fig. 3.20 Different types of intrinsic venous anastomoses at thoracic
level. (a) X-ray of an injected specimen in a.p. view. (b, c) Mid-sagittal
microangiograms at upper thoracic level, contiguous in vertical direction,
as outlined by the vertical bar in a. Courses, calibres and drainage func-
tions of different types of anastomotic veins through the midline of the
105
c
spinal cord parenchyma are obvious on these slices. The large channels
type 2 do not receive tributaries from the parenchyma, whereas smaller
anastomoses between central and radial veins do. Posterior peripheral
veins (at the side of the included diagram) at this thoracic level deeply
penetrate into the medullary parenchyma and take a descending course (c)
106 3 Venous Drainage

a b c d

Fig. 3.21 Intrinsic venous anastomoses demonstrated in coronal sec- midline anastomotic channel type 2 can be followed, but it is cut into
tions. (a) X-ray in a.p. view. (b–d) Three microangiograms of coronal short pieces (arrows). The arrowhead in b points to an interesting verti-
sections corresponding to the section positions S2–S3 of the schema. cal connection between two large radial veins in the border zone of the
The anterior and posterior superficial sections are not shown. The large anterior horn

Fig. 3.22 Microangiograms of large midline anastomoses at different different courses but a similar deviation around the central canal
levels of the spinal cord. (a–d) Mid-sagittal sections of injected speci- (cf. Figs. 3.1 and 3.17), often combined with an up- and downward
mens. The large channels connect the superficial longitudinal midline movement at this point. Even varicose dilatation and elongations were
veins and do not receive from intramedullary tributaries. They show observed in some specimens (Fig. 3.19) (Reproduced from Thron 1988)
3.3 Transmedullary Venous Anastomoses 107

a c

b d
108 3 Venous Drainage

3.4 Specific Patterns of Venous Drainage
in the Cervical Spinal Cord
Radial symmetry and horizontal distribution of intrinsic veins
are very pronounced in the cervical spinal cord (Fig. 3.23).
Transmedullary midline anastomoses are frequent in the
upper and mid-cervical regions, but they are of smaller cali-
bre than those in the lower cervical and upper thoracic
regions (Figs. 3.5, 3.15, and 3.24).
The anterior median vein is frequently larger than the pos-
terior median vein. Radicular outflow to the epidural plexus
occurs at many levels (Figs. 3.10, 3.13, and 3.14a).
Additionally, both veins connect to the brainstem veins
(Fig. 3.10) and basal sinuses around the foramen magnum
(Fig. 3.36). On the posterior aspect of the upper spinal cord
and medulla oblongata differentiation between a truncular type
of drainage in 80 % and a plexiform venous drainage pattern
in 20 % of cases was described by Moes and Maillot (1981).

Fig. 3.23 Microangiograms of

transverse sections at various
levels of the cervical spinal cord.
The intrinsic drainage pattern is
characterized by radially and
horizontally oriented peripheral
veins draining the central as well
as the peripheral nervous tissue.
The central veins play a minor role
compared to central arterial supply
(Reproduced from Thron 1988)
3.4 Specific Patterns of Venous Drainage in the Cervical Spinal Cord 109

a b c

Fig. 3.24 Intrinsic radial veins at a high cervical level (C2–C3). (a) running in the anterior fissure with and without anastomotic connections
X-ray in lateral view. The superficial anterior midline vein has a large to posterior veins can be discriminated (arrows). (c) Microangiogram
calibre. (b) Microangiogram of a mid-sagittal section at the C2–C3 of a parasagittal section. The main drainage occurs through horizontally
level as indicated in a. Apart from peripheral veins some central veins oriented peripheral veins
110
3.5 Specific Patterns of Venous Drainage
in the Thoracic Spinal Cord
Anterior and posterior median veins are commonly of
equal size in the upper thoracic region. At lower thoracic
levels the longitudinal venous trunks on the posterior sur-
face show variable positions and courses, which deviate
from the midline and may form plexiform convolutions,
also called the posterior venous plexus (Figs. 3.9, 3.10,
3.11, and 3.13).
3 Venous Drainage
In the thoracic cord the course of peripheral and central
veins as well as of small anastomotic veins is much more
vertically oriented, compared to the mainly horizontal distri-
bution of intrinsic veins in the enlargements. The greatest
incidence of type 1 and type 2 intrinsic anastomotic veins is
found in the thoracic region.
Among the posterior radiculomedullary veins in the tho-
racic region non-functional vessels exist (Figs. 3.38 and
3.43a). Moes and Maillot (1981) mention fibrotic veins in
this location in the majority of their specimens.
3.5 Specific Patterns of Venous Drainage in the Thoracic Spinal Cord 111

a b c g

d h

e i

f
j

Fig. 3.25 Veins of the thoracic spinal cord. Variability of peripheral
veins and intrinsic venous anastomoses. (a, b) X-ray film of an injected
spinal cord in a.p. (a) and lateral view (b). The broken lines indicate
the level of the cross sections in c–f. (c) Axial microangiogram at a
mid-thoracic level. In this section, a vein of the left intermediate pos-
terior septum is the most prominent and deepest penetrating venous
blood vessel. (d–h) Series of axial microangiograms at mid-thoracic
level, demonstrating the changing pattern of peripheral veins in nearby
sections. Strong vein of the posterior septum in f. Parts of very large
vessels in the anterior fissure like in e, f and h (arrow) or around the
central canal like in g (arrow) are most likely dissected parts of intrinsic
anastomoses between central and peripheral veins. (i, j) Axial micro-
angiograms at low thoracic level. The very large vessel in the anterior
median fissure is not a central vein, but certainly part of a large midline
anastomosis type 2 (arrowhead) (Reproduced from Thron 1988)
112
a b
Fig. 3.26 Superficial and deep thoracic spinal cord veins of unsystem-
atic and irregular appearance due to a missing anterior midline trunk.
(a) X-ray film of the specimen in a.p. view. (b) Mid-sagittal section at
the T3 level as indicated by the vertical line in a. At the level of this
section posterior veins are in midline position (arrowhead). But there is
no significant venous trunk in anterior midline position. Numerous
transparenchymal venous anastomoses of moderate calibre cross this
midline section. Most of them take an oblique path with ascending and
descending loops and are cut in this median section at one point or
3 Venous Drainage
c
another (arrows). The smaller veins are irregularly distributed and
linked to one another in a pattern, appearing unsystematic. (c) Axial
section, contiguous to the lower end of b. Corresponding to b, the radial
veins follow a horizontal course only for short distances. The arrow
points to a dissected vessel of larger calibre situated at the posterior
intermediate septum on the right. It should be the posterior part of the
lowest transparenchymal anastomoses marked with an arrow in b
(Reproduced from Thron 1988)
3.5 Specific Patterns of Venous Drainage in the Thoracic Spinal Cord 113

Fig. 3.27 Intrinsic veins in midline position between anterior and

posterior midline trunks. Caudal continuation of Fig. 3.26. (a) X-ray a b
in a.p. view. (b) Microangiograms of two consecutive mid-sagittal
sections at the T4–T6 level. Numerous intrinsic anastomotic veins of
different size (0.3–0.5 mm arrows and arrowheads) connect the
median longitudinal trunks in the midline. Median or intermedian
veins of the posterior columns start in the centre of the cord and take a
long descending course. Small anastomotic connections (0.1 mm)
with branches of central veins are numerous. Compare Fig. 3.20. The
vertical component of the vessel courses is similar to Fig. 3.26b, but
the pattern of drainage through peripheral and anastomotic veins
follows a more systematic pattern (Reproduced from Thron 1988)
114
a b
Fig. 3.28 Superficial and intrinsic thoracic cord veins at the level of
the posterior venous plexus including large midline anastomoses
type 2 in coronal microangiograms. (a) The anterior frontal section
shows the anterior median vein (arrows) collecting numerous superfi-
cial branches from the parenchyma and the roots as well as some large
veins emerging out of the anterior fissure (arrowheads). (b) The mid-
frontal section demonstrates three disconnected and partly dislocated
large-sized veins (arrowheads) in the location of the arrowhead-marked
3 Venous Drainage
c
vessels in (a). Sporadic, small central veins (small arrows) and periph-
eral veins directed to the lateral surface. (c) On the posterior surface the
superficial veins exhibit varicose convolutions corresponding to the
posterior venous plexus (arrows). The numerous large-sized anasto-
motic midline channels, connecting the posterior venous plexus with
the anterior median vein are disrupted by dissection and partly dislo-
cated and are therefore not adequately demonstrated (Reproduced from
Thron 1988)
3.6 Specific Patterns of Venous Drainage in the Thoracolumbar and Sacral Spinal Cord
3.6 Specific Patterns of Venous Drainage
in the Thoracolumbar and Sacral
Spinal Cord
At the thoracolumbar enlargement including medullary
cone, intrinsic venous drainage differs from the rest of the
spinal cord and is more similar to the arterial supply pattern.
Hence, central veins may be considerably larger than periph-
eral veins (Figs. 3.29, 3.30, and 3.31).
Compact sequences of central veins can possibly be seen,
exhibiting calibres of 0.1–0.25 mm, which correspond to
those of the arterial counterparts. But they exceed them in
number, as should be expected. In 2.5-cm-long mid-sagittal
sections of the lumbar enlargement, as shown in Figs. 3.29
and 3.31, approximately 25–40 sulcal veins are found, in
contrast to 25 sulcal arteries, which would be expected in
this location. (Compare arterial supply in Figs. 2.59 and 2.60
with venous drainage in Fig. 3.31.)
No midline anastomoses of significant calibre were found
in the lumbar enlargement, although in one case, seen at the
conus, a very large anastomotic tract type 2 apparently sub-
stituted superficial anastomoses (Figs. 3.22d and 3.30d).
115
The anterior midline vein reaches its maximum calibre in
this region and it is important to note that the vein of the
filum terminale is the continuation of this anterior median
vein, which is very clearly shown in the anatomical drawing
by Kadyi (Fig. 3.11a) and in many radio-anatomical images,
e.g. Figs. 3.29, 3.30, and 3.32. Alternatively, the anterior
midline vein on the enlargement or cone follows a lumbar or
sacral nerve root to reach the sacral epidural space (Figs. 3.11,
3.31, and 3.33c–e).
The posterior median spinal vein is particularly large
above the thoracolumbar enlargement (Figs. 3.10b, 3.11b,
3.13, 3.30, and 3.31), frequently forming varicose convolu-
tions called posterior venous plexus. This trunk ends at the
level of the cone and anastomoses with the anterior median
spinal vein. The anterior and posterior venous tracts on the
surface of the thoracolumbar enlargement and their radicu-
lar venous outlet(s) are the most prominent blood vessels of
the spinal cord with diameters up to 2 mm (Figs. 3.30, 3.31,
3.32, and 3.33). When demonstrated by tomographic imag-
ing techniques without adequate time resolution they should
not be mistaken for spinal cord arteries.
116 3 Venous Drainage

a b c d

Fig. 3.29 Predominant anterior median and terminal vein. (a, b) X-ray veins in which central veins play a substantial role. (d–f) Axial slices
of the specimen in a.p. and lateral view. (c) Two mid-sagittal contiguous from the lumbar level indicated in c exhibit a radial symmetry of veins
microangiograms from the region of the lumbar enlargement and cone in the contrast-filled central and anterior parts of the cord (a–c
demonstrate a predominant terminal vein (arrowhead) and intrinsic Reproduced from Thron 1988)
3.6 Specific Patterns of Venous Drainage in the Thoracolumbar and Sacral Spinal Cord 117

a b

Fig. 3.30 Veins of the thoracic and lumbosacral spinal cord. (a, b)
X-ray films of an injected spinal cord in a.p. and lateral view. (c)
Microangiogram of a mid-sagittal sections from the levels indicated in
a. Central veins in the lower thoracic region are less numerous than the
peripheral veins of the posterior columns, but they are somewhat larger
in calibre (arrowheads). Several of them may join to form a common
stem (anterior arrowhead). (d) The large transparenchymal anasto-
mosis near the conus medullaris (arrow) has a calibre of about 1 mm
(Reproduced from Thron 1988)
118 3 Venous Drainage

Fig. 3.31 Variations of

lumbosacral vein patterns. (a, b)
X-ray of the injected specimen in
a.p. and lateral view. The thick
and convoluted vein
(arrowheads) belongs to the
posterior venous plexus of the
low thoracic region. It passes to
the anterior surface using
superficial branches around the
cord (not midline anastomoses as
might be suspected from b).
Anteriorly, it divides into two
radiculomedullary veins which
follow different lumbar and
sacral nerve filaments of the
cauda equina (large and small
arrow). Only a very small branch
accompanies the filum terminale
(small open arrow). (c) Two
contiguous midline sections
include the large
radiculomedullary veins and the
small filum terminale vein
(arrows). The midline sections
exhibit central and peripheral
intrinsic veins in a distribution
typical for this region without
significant intrinsic anastomotic
interconnections

a b c
3.6 Specific Patterns of Venous Drainage in the Thoracolumbar and Sacral Spinal Cord 119

a b c d

Fig. 3.32 X-ray films of injected spinal cord specimens in a.p. view. findings as well as a distinct terminal vein (arrowhead). The large radic-
Different patterns of superficial veins at the thoracolumbar enlargement ulomedullary vein in a is a conjoint vein from two superficial trunks. The
and cauda equina. (a–d) The posterior venous plexus with varicose con- specimen shown in d features an extreme varicosity of a sacral radicular
volutions and a “great radiculomedullary vein” (arrows) are common vein on the right side (arrow) (c, d Reproduced from Thron 1988)
120 3 Venous Drainage

c d e

a b

Fig. 3.33 X-ray films of injected spinal cord specimens. Variations of
venous drainage at the lower end of the spinal cord. (a, b) A.p. and lat-
eral view. There is a large “great anterior radicular vein” and a distinct
terminal vein (arrowheads), but only a weakly developed posterior
venous plexus. (c, d) A.p. and lateral view. In this specimen the terminal
vein is replaced by two strong anterior thoracolumbar radicular veins.
Furthermore, lateral anastomoses around the cone on both sides (but dif-
ferent levels) resemble the arterial arcades. Radiculomedullary vein
complex at mid-thoracic level. (e) A.p. view. Pronounced varicosity of
veins on the thoracolumbar enlargement. Two sacral veins on the right
side (arrowheads) and a great radicular vein on the left side substitute the
drainage function of a terminal vein (e Reproduced from Jacobs 1996)
3.6 Specific Patterns of Venous Drainage in the Thoracolumbar and Sacral Spinal Cord 121

a b

Posterior Posterior venous plexus

radiculomedullary veins
Anterior
radiculomedullary veins

Posterior median spinal vein

vein of the filum

teminale

Fig. 3.34 Basic pattern of anterior and posterior superficial veins at
thoracolumbar level. (a) Diagram of anterior veins. Typical findings in
the lower part of the lumbar enlargement are a large anterior midline
vein, proceeding as vein of the filum terminale in variable size.
Moreover, additional radiculomedullary veins of considerable size are
frequent findings, some of them originating from very deep sacral lev-
els (cf. Fig. 3.31). (b) Diagram of posterior veins. A plexiform pattern
of veins is almost regularly observed in the upper part of the enlarge-
ment (posterior venous plexus), as well as large posterior radiculomed-
ullary arteries
122
3.7 Radiculomedullary Veins
and the Transdural Course
Venous blood collected in the large longitudinal venous
trunks on the surface of the spinal cord issues into anterior or
posterior radiculomedullary veins, which cross the dura and
empty into the internal vertebral venous plexus (Fig. 3.35). A
hairpin bend in the course of radiculomedullary blood vessels
is not only typical for spinal cord supplying arteries. As a
result of the rising of the spinal cord, the radiculomedullary
veins take similar hairpin courses. Commonly, they follow a
nerve root and descend to an intervertebral foramen where
they pierce the wall of the dural sac close to the nerve root in
about 70 %, but their exit may be completely independent
from the nerve root in 30 % (Thron et al. 2015). After cross-
ing the dura mater, the so far distinct vessels empty in about
95 % into the lacunae and irregular spaces of the internal
vertebral venous plexus. Rarely, a direct continuation to the
intervertebral veins can be seen (Fig. 3.42c).
Documented by the experience of selected spinal angiog-
raphy, most of the venous return passes through radiculom-
edullary veins (Fig. 3.35). At the craniocervical junction,
connection occurs between the superficial venous trunks and
the intracranial dural sinuses at the base of the posterior cra-
nial fossa. Figure 3.36 demonstrates the case of a dural arte-
riovenous fistula located at an upper thoracic level with
ascending drainage. Apart from segmental radicular outflow
in the cervical region the dilated anterior median spinal vein
seems to join directly the basilar or marginal sinus and con-
tinue through petrosal inferior and superior sinuses bilater-
ally. Whether this occurs really directly or via passage of
parts of the internal vertebral venous plexus is not apparent
in our case. In a second example of increased descending
flow, caused by a dural arteriovenous fistula, the anterior
median spinal vein can be followed downwards to radicular
veins of the cauda equina (Fig. 3.37b; small arrow) until
these vessels empty in an extended sacral epidural plexus at
the end of the dural sac.
The true number of radicular spinal cord draining veins
faces a problem of anatomical nomenclature, similar to the
one encountered with the arteries.
In an older nomenclature the terms “radicular artery” and
“radicular vein” were both used for blood vessels running
with a nerve root, irrespective of their area of supply or
drainage. Radicular blood vessels with and without relevance
for the spinal cord blood circulation were not clearly dis-
cerned by nomenclature. Only in the French literature, the
specification “radiculo-médullaire” was commonly used to
specify spinal cord–supplying radicular arteries (Lazorthes
et al. 1973), but also spinal cord–draining veins (Tadié et al.
1979, 1985). In any case, for a well-defined attribution of
spinal cord–draining veins this specification seems indis-
pensable. The term “radiculospinal”, introduced in the
3 Venous Drainage
Anglo-American nomenclature for this purpose, is less cor-
rect from an anatomic or semantic point of view, because we
are not dealing with the spine, but with the spinal cord. That
is different.
We suggest to use a nomenclature adjusted to the one pro-
posed for the arteries, despite the fact that the pattern of
venous drainage is predominantly independent from the pat-
tern of arterial supply. But the different anatomical regions
and functional aspects make it necessary to discriminate
Radicular veins draining the metameric elements like bone,
dura mater, nerve roots of a given metamere and
Anterior and posterior radiculomedullary veins, which have
additional drainage function for the spinal cord.
The question of how many radiculomedullary veins are
preserved after embryonic life for the venous outlet of the
entire spinal cord has been a matter of dispute. The group of
Tadié et al. had insisted on an average number of only nine to
ten veins, definitely carrying the venous return.
According to the older literature data, the total number of
morphologically countable spinal cord draining veins should
add up to 30–70 (Kadyi 1889; Tureen 1938; v. Quast 1961;
Jellinger 1966), with an average of over 50 per spinal cord.
An equivocal numerical dominance of anterior or posterior
radicular veins has not been reported. Jellinger (1966) calcu-
lated an average of 23 ventral and 25 dorsal outflows,
whereas v. Quast (1961) calculated 30–26. This almost sym-
metrical anterior-posterior distribution was accompanied by
a minimal fluctuation in the segmental distribution. A signifi-
cant reduction should only occur in the uppermost cervical
and sacral segments. Likewise, no obvious lateral preference
was observed.
These data supported the widely accepted conclusion of a
much more pronounced symmetry and metameric structure,
preserved in the radicular drainage system.
The calibre of the anterior radiculomedullary veins lies
between 0.25 and 1 mm, whereas some of the posterior radicu-
lomedullary veins can reach calibres up to nearly 2.00 mm.
Compared with the great anterior radiculomedullary
artery, an especially large anterior or posterior radiculomed-
ullary vein can be demonstrated with less frequency. This
vein drains the caudal intumescence and is located ventrally
between T6 and S3. The dorsal vessel of maximal calibre is
situated between T9 and S3 with a distribution maximum at
L1 (Jellinger 1966). The radiculomedullary veins in this
region are the largest blood vessels around the cord
intradurally.
Differentiation of radicular and radiculomedullary veins
solely on the basis of macroscopic inspection may be prob-
lematic, especially for the numerous small veins running
with the nerve roots or the filaments of the cauda equina.
Morphological criteria include that radiculomedullary veins
3.7 Radiculomedullary Veins and the Transdural Course
can easily be separated from the nerve root or cauda fila-
ment and that purely radicular veins increase in size if they
are followed towards the spinal cord surface (Moes and
Maillot 1981).
Tadié et al. (1985) used radiographic and cinematographic
films to define radiculomedullary veins “through which the
contrast medium leaves the spinal veins and the intradural
sector to reach the extradural sector and the periphery”.
In the large numbers of radicular veins counted by pre-
vious investigators, purely radicular veins might have been
included. But even if all small veins less than 0.25 mm
are excluded, the number of at least medium-sized
radiculomedullary veins draining the spinal cord (and not
only the radix or small parts of the pial plexus) remains in
the range of 6–11 for the anterior and 5–10 for the posterior
systems. These numbers can be determined from the data
of Kadyi (1889) and Suh and Alexander (1939), who listed
and categorized the number, localization and calibre of the
observed radicular veins meticulously. They are in agreement
with a study performed at our department (Jacobs 1996). If
one accepts the principle that the volume for venous return
has to exceed the volume of arterial input by a factor of 2.5
because of the different flow characteristics of arteries and
veins (Kadyi 1889), the large numbers are reasonable and
necessary.
Similar to the arterial supply individual protocols of num-
ber, distribution and calibre of radiculomedullary veins in
post-mortem studies reveal highly variable patterns of drain-
age. This justifies the differentiation between a pauci- and a
plurisegmental type of drainage similar to the arterial supply.
Conjoint and non-functional radicular veins are special
observations made at the final intradural vein segment. Two
or more veins coming from the medullary surface can merge
to form a single radiculomedullary vein just before reaching
the dura mater. Then, together or independent from the root,
they pass as a common trunk through the dura and drain
into the internal vertebral venous plexus (Fig. 3.38). The
arrangement of several veins of ascending, descending or
horizontal course converging radially to form a single radic-
ulomedullary vein, as demonstrated on Figs. 3.10b, 3.35, and
3.38, has been called a radiculomedullary vein complex
(Kadyi 1889). These veins may cross the dura mater in very
atypical locations between two nerve roots.
This configuration of surface veins conjoining to a single
radicular vein can also be observed in a variation or patho-
logic finding, in which the singular terminal segment is
occluded immediately behind the communication. Only a
few authors have reported the occurrence of such blind end-
ing radicular veins. Moes and Maillot (1981) called these
more or less fibrotically occluded radiculomedullary veins
“non-functional veins”, occurring mainly among the
posterior thoracic radicular veins. Crock et al. (1986) pre-
sented an image of this arrangement, which they described as
123
“V-shaped anastomosis” or “arrow-tipped loop of the dorsal
median longitudinal trunk”.
Examples from our investigations are shown in Figs. 3.38b
and 3.43a. The aetiology and clinical relevance of such
fibrotic radicular veins are unclear (Thron et al. 2015) and in
so far open for hypothetical considerations (Gailloud 2014).
The fibrotic segment may represent a residue from the
embryonic desegmentation process. Likewise, the occur-
rence mainly among posterior thoracic veins might be related
to the high variability and sometimes poor development of
the posterior part of the internal vertebral venous plexus as
reported by Groen et al. (1997).
Figure 3.46, reproduced from a cadaver study using injec-
tion of a modern polymer (Araldite CY 221), illustrates this
high segmental and interindividual variability, which is in
contrast to the fairly constant anterior part. Poor develop-
ment of this posterior plexus could be the cause of secondary
fibrotic regression of initially patent posterior radicular
veins. Other possible aetiologies like focal inflammation or
thrombosis appear unlikely but cannot be ruled out. The clin-
ical significance of non-functional radicular veins is so far
unclear. Hypothetically, they could play a role in the notice-
able delay of hindered spinal drainage, observed in many
cases of spinal dural arteriovenous fistula (SDAVF) (Merland
et al. 1980).
Varicose convolutions, more frequent on the posterior
than on the anterior surface of the spinal cord, are not unusual
(Figs. 3.31 and 3.33). Even extreme varicosities of radiculo-
medullary veins like in Fig. 3.32d may be observed as inci-
dental findings. This is why the term “spinal varicosis” was
frequently used in the older neuropathology literature and in
the evaluation of myelograms. Even today, accentuation of
plexiform veins on tomographic images may cause confu-
sion with regard to the differentiation between a normal and
a pathological finding. In cases of doubt, especially if com-
bined with clinical suspicion, exclusion of an arteriovenous
shunt disorder is necessary.
The dural segment of radiculomedullary veins connects
the veins of the subarachnoid space with the internal verte-
bral venous plexus (Batson’s plexus).
Many investigators found that filling of intrathecal spinal
cord veins in a retrograde manner from the outside of the
dura mater could not be achieved or only with bad results
(Breschet 1828–1832; Kadyi 1889; Batson 1957; Clemens
1961; Dommisse 1975; Crock et al. 1986). This is why the
presence of valves close to the dura mater was generally
assumed. Another convincing clinical argument for the
necessity of valves was seen in the prevention of reflux to the
spinal cord veins in case of extraspinal venous hypertension.
Oswald (1961) then reported the existence of bicuspid
venous valves in the radiculomedullary veins, immediately
after passage through the dura. But subsequent investigators
(Tveten 1976; Tadié et al. 1979, 1985) did not confirm this
124
observation. Instead, Tadié et al. suggested a different
“anti-backflow control system” consisting in a high reduc-
tion of vessel calibre at the dural segment of radicular veins
and an oblique zigzag course of the vessel.
To improve our understanding of the assumed reflux-
impeding arrangement in the dural segment of spinal cord–
draining veins, a microscopic and microangiographic
investigation was performed in the late 1980s at our institu-
tion (Otto 1990; Thron et al. 1990; Schroeder et al. 1993).
Secondly, the unknown vulnerability of venous return from
the spinal cord in the absence of an arteriovenous shunt
should be studied. Again, no valves could be detected
(Fig. 3.41). Similar features as described by Tadié et al.
(1985) were identified. An oblique and sometimes bending
transdural course associated with narrowing of the vessel
lumen was observed in about 60 % of the studied specimens.
However, microscopic evaluation of a total of 102 radiculo-
medullary veins revealed a second arrangement, comprising
35 % of examined cases.
Exemplifying histological sections of these two main
types, called “slit” type and “bulge or nodular” type by Otto
(1996), are demonstrated in Fig. 3.39. A diagram of both
types is presented in Fig. 3.40.
The slit type occurs mainly in the vicinity of nerve roots.
The dural segment shows narrowing of the lumen. The vessel
wall is mostly replaced by fibrous tissue of the dura. The ves-
sel courses in an oblique direction, sometimes bending or
kinking.
The bulge or nodular type mainly occurs with vessels
crossing the dura at some distance from a nerve root. There is
considerable broadening of dural collagen fibres at the site of
dural penetration, which follows the blood vessel in the direc-
tion of the epidural compartment. Narrowing or splitting of
the lumen occurs within the fibrous tissue formation, which
eccentrically encapsulates the disintegrated blood vessel.
Why these different arrangements at the dural exit of
veins exist is yet unclear.
A reflux-impeding or better flow-modulating effect might
be attributed to both devices. To what extent both exit
arrangements with narrowing of the lumen impede reflux
from the outside of the dura mater or influence flow in either
direction is again unclear. Out of 14 attempts to fill the
3 Venous Drainage
intrathecal veins from the extradural space, two were suc-
cessful in our specimens. Figure 3.44 shows the radiculom-
edullary vein and the beginning longitudinal tract contrasted
via the lacunae of the epidural plexus.
Clinical situations in which these devices did not strictly
prevent reflux are spinal phlebography (Fig. 3.45) with
abdominal compression (Théron and Moret 1978;
Meijenhorst 1977) or SDAVF shunts to the internal vertebral
venous plexus and secondary, at another place, reflux to
superficial spinal cord veins (Krings et al. 2006). Examples
in this book are Figs. 4.34, 4.38 and 4.39. But in the mean-
time not only the presence of valves but generally the exis-
tence and necessity of anti-reflux arrangements are not
undisputed (Groen 1997; van der Kuip et al. 1999).
In two post-mortem studies from the end of the 1990s,
cadavers were thoroughly prepared including removal of
clotted blood before a modern polymer (Araldite CY 221)
was injected to demonstrate the internal vertebral venous
plexuses (Groen et al. 1997) and the human radicular veins
(van der Kuip et al. 1999). These procedures allowed retro-
grade filling of intradural radicular veins to be documented
as a rather regular and not as an exceptional observation.
Therefore the conclusion is justified that reflux into radiculo-
medullary veins might be a physiological phenomenon (van
der Kuip et al. 1999). Our reconsiderations concerning anti-
reflux arrangements have been published recently (Thron
et al. 2015).
The X-rays of the specimens injected for microangiogra-
phy (Figs. 3.42, 3.43, and 3.44) confirm oblique vessel
courses through the dural sheet and narrowing of the lumen.
Moreover, they confirm a connection between regularly
composed veins and a system, which in the majority of our
specimens comprises multitudinous communicating lacunar
spaces. Most of the epidural venous plexus compartments in
our specimens injected for microangiography have a similar
sponge-like aspect. In only a few cases the epidural plexus
was composed of defined venous blood vessels or a mixture
of both.
The functional role of this probably predominant internal
vertebral venous plexus for blood flow adjustments between
the intra- and extradural compartments of the vertebral canal
remains poorly understood.
3.7 Radiculomedullary Veins and the Transdural Course 125

a b

Fig. 3.35 Superficial spinal cord veins and radiculomedullary veins.
X-rays of contrast-injected human spinal cord specimens in a.p. view.
(a) Upper thoracic region. The anterior median venous trunk communi-
cates through radiculomedullary veins with the internal vertebral
venous plexus (Pl.ep.) after crossing the dura mater (D; arrows)
(Reproduced from Thron 2015). (b) Mid-thoracic region. Not only sin-
gular veins branch from the longitudinal midline trunks to join the epi-
dural plexus (arrow). Two or more radicular and radiculomedullary
veins from different parts of the surface can merge to a common stem
just before penetrating the dura mater (D), thus forming a “radiculom-
edullary vein complex” (Kadyi 1889)
126 3 Venous Drainage

a b

Fig. 3.36 Digital subtraction angiogram of the superficial spinal cord rather directly than via the basilar or marginal sinuses. The posterior
veins in case of a dural arteriovenous fistula at thoracic level. (a, b) median spinal vein and the cervical radicular outflow at the C2/3 level
Cervical region: lateral view (a) and a.p. view (b). The large anterior (arrowhead) are additionally opacified (Reproduced from Thron 2002)
median spinal vein (arrow) joins the petrosal sinuses (white arrows)
3.7 Radiculomedullary Veins and the Transdural Course 127

Fig. 3.37 Unsubtracted and

subtracted image of digital
subtraction angiography in case
of a spinal dural arteriovenous
shunt. Drainage is observed
along the veins of the cauda
equina (small arrow) downwards
to the sacral epidural plexus
(large arrow) (Reproduced from
Thron 2002)

a b

Fig. 3.38 Photographs of

conjoint radiculomedullary veins
presenting a fibrotic common
radicular stem in one of the two
cases. (a) Posterior surface veins
merge to a common radicular
stem immediately before
penetrating the dura mater at low
thoracic level (arrow)
(Reproduced from Jacobs 1996).
(b) The median longitudinal
venous trunk deviates from the
midline to form a radicular vein.
The point of junction to form a
single radicular vein is halfway to
the dura mater in this specimen.
The communication between
upper and lower part of the
longitudinal channel is preserved,
but the single radicular vein is a
non-functional fibrotic vein
(arrow)
128 3 Venous Drainage

a a

Fig. 3.39 Structurally different types of blood vessel exit through the
dura mater. Histological sections, H&E, original magnification ×60. (a)
“Slit type” showing a vein accompanying a nerve root to the nerve root
sheath before penetrating the dura mater. The vessel course is marked
with arrows. The narrowing and oblique transdural course are evident,
as are the alteration of the blood vessel into a fissure-like narrowed Fig. 3.40 Schematic illustrations of the “slit type” of dural vein pas-
channel. (b) “Bulge or nodular type” refers to the bulge of collagen in sage (a) and the “bulge or nodular type” (b) (Reproduced from Thron
which the vein loses its normal structural integrity (Reproduced from et al. 2015)
Thron et al. 2015)

Fig. 3.41 Histological sections of the dural segment of a radiculomed-

ullary vein at the T9 level. Even in this technique, providing semi-thin
sections, valves could not be demonstrated. Moderate narrowing of the
vessel calibre and an oblique course when crossing the dura mater seem
to be the only specifications to provide anti-reflux properties in this
example (Reproduced from Thron et al. 2015)
3.7 Radiculomedullary Veins and the Transdural Course
a b
c d
Fig. 3.42 Transition of radiculomedullary veins to the internal vertebral
venous plexus. Different configurations as demonstrated by post-mortem
X-ray microangiography following orthograde contrast injection. (a)
Short transdural passage of a radiculomedullary vein running with the
nerve root. The lacunar spaces of the epidural venous plexus surround
the nerve root sheath (slit type, level T7). (b) Tidy narrowing of the dural
vein segment and acute angle between the subarachnoid vein and the
recipient epidural vein. In this case the internal vertebral venous plexus
is composed of an irregular plexiform convolution of distinct veins (his-
tologically, the type of dural transition could not be assigned to slit or
129
bulge type, level T6). (c) Rare dural transit of a radicular vein which
continues in the epidural space as an intervertebral vein of equal size.
There is only circumscribed narrowing of the vessel lumen while cross-
ing the dura mater in an oblique course (slit-type, level T7). (d) Two
pairs of radiculomedullary veins of different size, merge directly before
they cross the dura mater. Narrowing of the lumen and kinking or bend-
ing course are similar (arrows). The larger veins empty in a lacunar part
of the epidural plexus distant from the nerve root, the smaller ones in an
extensively contrasted epidural venous space surrounding the nerve root
sleeve (bulging type, level T10) (Reproduced from Thron et al. 2015)
130 3 Venous Drainage

a b

Fig. 3.43 Microangiograms of the dural segment of radicular veins.
(a) Fibrotic, “non-functional” radicular vein. Specimen at the L1 level
following injection of an X-ray contrast medium. Two radiculomedul-
lary veins of the cauda equina conjoin to a common radicular stem very
close to the wall of the dural sac. Whereas free communication between
the two veins is preserved, only a filiform residual lumen of the com-
mon stem can be demonstrated radiographically. It can be followed to a
tiny dural or epidural compartment (arrows) (Reproduced from Thron
et al. 2015). (b) Microangiogram of a specimen containing two conjoin-
ing small radicular veins and two conjoining large radiculomedullary
veins (T9 level). Oblique passage through the dura mater and narrowing
of the lumen (arrow) are evident. Blood flow is carried forward into
irregular lacunar spaces of the epidural plexus (Reproduced from
Krings et al. 2006)
3.8 Extradural Intraspinal and Extraspinal Venous Systems
Fig. 3.44 Malfunction of the presumed anti-reflux arrangement at the
transdural exit of radiculomedullary veins (arrow). X-ray of the speci-
men in a.p. view. Following contrast injection of the internal vertebral
venous plexus surrounding the dural sac, retrograde filling of a r.m.v.
(arrowhead) at the T5 level was observed (D dura mater; r.m.v. radicu-
lomedullary vein) (Reproduced from Thron et al. 2015)
131
3.8 Extradural Intraspinal and Extraspinal
Venous Systems
Drainage of blood from the spine (including spinal cord)
occurs through the internal and external vertebral venous
plexuses of Batson (1957), which are connected to each
other through intervertebral veins. Both are valveless epi-
dural systems with an anterior and posterior part. The inter-
nal vertebral venous plexus lies within the vertebral canal
between the dural sac and the outer margins of the vertebral
canal. This system is still controversial in nature; either
plexal or venular (Lasjaunias and Berenstein 1990). It is the
first compartment, which receives blood from the radiculom-
edullary veins draining the spinal cord. Most of the compart-
ments of the internal vertebral venous plexus close to the
dura, which were injected and opacified for microangiogra-
phy in our specimens, have a similar aspect. It can best be
described as a system of smaller and larger venous lakes,
lacunae and fissures (Figs. 3.42, 3.43b, and 3.44). In only a
few cases the epidural plexus was composed of defined
venous blood vessels or a mixture of both (Fig. 3.42c). This
sponge-like system of venous pools extends over the whole
length of the vertebral column (Fig. 3.45) from the sacrum to
the skull base and connects pelvic veins with the basal cra-
nial sinuses.
An example of spinal angiography in case of an arteriove-
nous shunt disorder shown in Fig. 3.36 indicates that the
anterior spinal vein communicates with the basal sinuses
around the foramen magnum. As an important collateral
pathway, the internal vertebral venous plexus bypasses extra-
spinal longitudinal systems like the cava or azygos system
(Lasjaunias and Berenstein 1990). The external vertebral
venous plexus, lying on the outer surface of the vertebral col-
umn, is connected to the inner one (Fig. 3.45).
The anterior internal vertebral plexus is larger and more
constantly developed than the posterior one (Figs. 3.45 and
3.46). The reason for this is that it receives venous blood not
only from the spinal cord but also from the basivertebral
veins of the vertebral body.
The valveless system of the plexuses is connected to the
azygos and hemiazygos systems by intercostal or segmental
veins and in the cervical region to the vertebral and deep
cervical veins (Fig. 3.47).
132 3 Venous Drainage

Fig. 3.45 Lumbar venograms

a b
demonstrate large parts of the
internal vertebral venous
plexuses. The anterior venous
plexus as the longitudinal part
(small open arrow) together with
the retrocorporeal network
including the basivertebral vein
of the vertebral body (large open
arrow) form the oval-shaped
configurations at each level. They
are preferentially demonstrated
in these clinical venograms. The
posterior internal vertebral
venous plexus is of high
interindividual variability, better
shown in the schematic drawings
of Fig. 3.46. Intervertebral and
emissary radicular veins are
lateral connections (small
arrows) to longitudinal external
efferents, which may not be
continuously developed
(arrowheads). (a) Normal
findings. (b) Venogram in case of
lateral disk herniation. The lack
of the vein(s) accompanying the
S1 nerve root on the right side to
the foramen (small black arrow)
substantiate the suspicion of
lateral disk herniation (Courtesy
of Professor H. Becker, image
archive of Professor
H. Vogelsang, Institute for S1
Neuroradiology, Hannover
Medical School, Germany)
3.8 Extradural Intraspinal and Extraspinal Venous Systems 133

a b c

Fig. 3.47 Diagram of drainage routes and anastomotic interconnec-

tions between spinal cord draining veins and systemic veins of thorax,
abdomen and pelvis. The small blue blocks in the course of radiculom-
edullary veins indicate the exit of these vessels through the dura mater
and the compartment of the internal vertebral venous plexus, interposed
in the drainage route

Fig. 3.46 The posterior internal vertebral plexus exemplified in three

schematic drawings obtained in an intravenous postmortem injection
study. They outline and stress the prominent differences in the segmen-
tal and interindividual configurations (Sect. 3.8) (Reproduced from
Groen et al. 1997 © 1997 WILEY-LISS, INC, artist’s impression,
C. Zondergeld)
134
3.9 Anatomical and Haemodynamical
Aspects of Venous Drainage
3.9.1 Intrinsic Veins
The spinal veins, especially the intramedullary ones, have
been mostly neglected in neuro-anatomical studies (Gillilan
1970). The fundamental structure of this drainage system is
assessed very diversely. The division into central and periph-
eral systems proposed in several studies (Jellinger 1966;
Tveten 1976; Lanz and Wachsmuth 1982) is modelled after
the pattern of arterial supply. However, since the pattern
of venous drainage deviates substantially from that of the
arteries, the sense of an analogous classification is at least
questionable.
The radioanatomic images presented in this book visual-
ize the intrinsic venous drainage of the spinal cord as a radi-
ally arranged vascular system. These mostly peripheral
veins originate at different depths of the cross-sectional area
and run directly to the periphery. Anteromedian and postero-
median vessels, including central veins, dominate to a cer-
tain extent in different regions. But basically, venous
drainage more likely complements the arterial supply pat-
tern. These findings are in agreement with the findings of
other investigations (Tveten 1976e; Crock and Yoshizawa
1977). The conclusion that, in comparison to the arterial sys-
tem, a very different venous drainage pattern exists can
therefore be confirmed. As outlined above, the basic pattern
described exhibits regional variations that are conducive to
divergent descriptions. In the lumbosacral region of the
human spinal and to a certain extent in the cervical enlarge-
ment, central veins are similar in calibre and number com-
pared to the arteries. In other regions, posteromedian veins
of the septum are the largest and longest veins of the medul-
lary transverse section (Kadyi 1889).
To accomplish drainage of the richly vascularized grey
matter, the numerous, deeply extending radial veins form a
system with greater capacity. Calculating the number and
calibre of central arteries and veins, Kadyi (1889) had
already pointed out that only some of the blood entering via
the central arteries can return via the veins of the anterior
fissure as the capacity in the venous “low-pressure system”
with slow flow of blood must be higher by a factor of more
than 2 compared to that in the arterial system. This stresses
once more the opposition of a centrally dominating supply
and a peripherally dominating drainage system.
3.9.2 Intrinsic Venous Anastomoses
Radiographic imaging of vascular anatomy, combining
X-rays with tomographic sections in three planes, offers sig-
nificant advantages for new insights into the number, local-
ization and course of different types of intrinsic venous
anastomoses, which were described in detail. Although
3 Venous Drainage
venous anastomoses have been mentioned in some reports
(Kadyi 1889; Herren and Alexander 1939; Crock and
Yoshizawa 1977), other authors disputed them (Gillilan
1970; Tadié et al. 1985). As Tveten (1976) could not detect
them in rats, he considered them to be random findings.
Consequently, their physiological and pathophysiological
importance has not been appreciated until now.
Kadyi (1889) describes diverse anastomoses between the
central and the peripheral veins that allow drainage along
several routes, yet his examples merely show the connec-
tions of smaller and medium-sized calibre. Herren and
Alexander (1939) interpreted these centro-dorsolateral anas-
tomoses as an arrangement to equalize venous pressure. The
findings of our investigations support this view. Crock and
Yoshizawa (1977) first mentioned a large midline communi-
cation between the main median trunks (type I). We had the
possibility to examine a large variety of cases and, thanks to
a novel technique of specimen preparation and my personal
experience in X-ray examinations, we were able to image,
understand and demonstrate the principles of extrinsic and
intrinsic spinal cord arterial supply and venous drainage.
We are able to demonstrate that these anastomotic channels
are still underestimated. They are frequent, they serve differ-
ent needs and are therefore not randomly distributed over the
length of the spinal cord. Whereas one type receives tributar-
ies from the spinal cord, the second type does not. The con-
stant calibre, mostly equal to that of the veins to which they
connect, suggests them to be arrangements for pressure and
flow adaptation. A rapid exchange between both of the large
blood reservoirs anteriorly and posteriorly can occur through
them. Furthermore, it can be assumed that the density in which
they occur in the upper thoracic and cervicothoracic regions is
related to the particular mobility of the spinal column in these
regions. In contrast to the brain, the spinal cord must toler-
ate inner deformation under movements of the spinal axis.
The draining vessels running on the surface are submitted to
ensuing variations in volume and pressure. As can be seen
in myelographic studies with maximal flexion and extension
of the head, compression effects occur ventrally or dorsally,
depending on the curvature of the spine, which probably affect
the more easily compressed veins first, and the arteries later.
Thus, if the ventral venous longitudinal tract is compressed by
extreme movements, blood can be diverted dorsally not only
through circular, superficial anastomoses but even more quickly
and efficiently through the transmedullary anastomoses.
By means of this blood-deviating system, regional pres-
sure differences in the veins are quickly adjusted, even under
extreme hyperflexion of hyperextension. Also pathological
conditions are compensated over long periods. Only a circu-
lar, collar-like constriction seems to be appropriate for an
effective obstruction of venous drainage.
Our systematic radio-tomographic investigation of many
spinal cord specimens support the conclusion that the pre-
dominantly radial arrangement of the intrinsic veins and
the high number and variation of different anastomotic
3.9 Anatomical and Haemodynamical Aspects of Venous Drainage
connections near the mobile segments of the spinal cord
are one of the most characterizing features of spinal cord
drainage. The anastomoses are locally adjusted to different
requirements and protect venous drainage against circum-
scribed space-occupying effects as well as parenchymal
deformation through movement.
3.9.3 Superficial and Radiculomedullary
Veins
The anterior and posterior median spinal veins are only the most
common superficial venous channels. The large blood-collecting
anastomotic trunks can extend over the entire length of the spi-
nal cord or they can be replaced or accompanied by antero- and
posterolaterally located blood vessel tracts. In fact, they only
represent the most noticeable longitudinally oriented compo-
nents within a pial vascular network that is termed the “venous
plexus of the pia mater” (Tveten 1976), “coronal (pial) plexus”
(Gillilan 1970) or “venous pial plexus” (Crock and Yoshizawa
1977). Moes and Maillot (1981) presented a meticulous study
on the superficial veins of the human spinal cord and for good
reasons they added the subtitle “Essay of a systematization”.
The main reason is that deviations from the basic pattern are
almost a rule instead of an exception and the changes in calibre
and position of these veins along the longitudinal axis show a
high interindividual variability. Looking at the numerous X-rays
and photographs of contrast-filled specimens in this book, this
statement should be convincing.
The clinical significance of the frequent varicosities of the
venous pial plexus, especially on the dorsal side of the thora-
columbar enlargement, or of the extreme varicosity of a radic-
ular vein shown in Fig. 3.32 remains unclear and unexplained.
Drainage routes of the superficial system run towards the
inner extradural vertebral venous plexus via the anterior and
posterior radiculomedullary veins.
3.9.4 Circulatory Disturbances of Venous
Origin
Acute venous infarcts caused by spinal venous thromboses are
considered rare. As a result of the lack of clinical criteria for
such a diagnosis and because of the difficulties of post-mortem
documentation, the low number of venous thromboses and
phlebitides that are confirmed by autopsy must be considered
with caution (Kulenkampff and Matheis 1961; Gruner and
Lapresle 1962; Neumayer 1966; Hughes 1971; Rao et al. 1982;
Kim et al. 1984). Disturbances in the venous section can hardly
be distinguished from microcirculatory disorders.
In cases with local masses, venous drainage is only effec-
tively obstructed by a circular constriction of the spinal cord,
since the internal venous anastomoses allow for a change in
direction of drainage. This is the reason why Doppman et al.
(1979) failed in their attempt to produce spinal cord lesions
135
by ligating the dorsal veins in rhesus monkeys. It further
explains the switch of drainage towards the ventral system,
which these authors could not interpret. Since intradural
venous blood channels as protected as the cerebral sinus do
not exist in the spinal canal, intramedullary or circular extra-
medullary masses lead to a compression of the superficial
medullary veins after the subarachnoid space is consumed.
Venous stasis then becomes relevant in addition to the pri-
mary process (tumour, trauma, inflammation, ischemia).
In case of a SDAVF, arterialisation of one of these veins at
the site of dural penetration with flow backwards towards the
perimedullary veins causes massive venous hypertension
and in the long term severe damage of the spinal cord paren-
chyma (Foix–Alajouanine disease). The experience that spi-
nal cord–draining veins may play an important pathogenetic
role in vascular diseases created the expectation that chronic
impairment of spinal venous drainage alone, in the absence
of an arteriovenous shunt, could be the cause of important
myelopathic conditions (Tadié et al. 1985; Aboulker et al.
1977; Merland et al. 1980; Thron 1988).
One of the two main aetiological concepts was the idea
that contrary to the reported high numbers of radicular
veins only a small number of the anatomically counted
radicular veins definitely carry the venous return from the
spinal cord (Tadié et al. 1985). To support this idea Tadié
et al. define the restricted role of pure radicular veins and
conclude that these considerably smaller veins of generally
0.1–0.2 mm diameter “do not play an important role in spi-
nal venous drainage”. Moes and Maillot (1981) argue with
good reasons that vessel diameter should neither be the
only nor the crucial argument to discern radicular from
radiculomedullary veins. But Tadié et al. (1985) insisted on
an average number of only nine to ten radiculomedullary
arteries for the venous return of the entire spinal cord, com-
pared to an average of about 50 in the majority of other
investigations. But this low number would not only demon-
strate high vulnerability, its capacity would principally be
far beyond that needed for venous return.
Non-functional veins and narrowing of the veins in the
transdural segment have not been taken into account in this
assessment.
Therefore, the idea that possible diseases might result
from impairment of venous drainage alone, in the absence of
dural arteriovenous shunts (Sect. 3.4) has not received greater
attention during the past three decades. Arguments that a low
number of veins effectively draining the spinal cord should
be responsible for chronic vascular myelopathies (Tadié
et al. 1985) have not been substantiated.
The same is true for the second aetiological concept that
possible malfunction of an assumed reflux-impeding device
at the dural transit of radicular veins might occur. We must
admit that we cannot measure normal or abnormal venous
return from the spinal cord and imaging of parenchymal dam-
age beyond the level of “subacute necrotizing myelopathy” is
also not possible at the moment. But on the other hand we are
136
not confronted with significant clinical consequences. It may
be that there is a causal co-relationship between the high
number of varicosities in and around the spinal cords demon-
strated in cadaver studies of an older population and chronic
impairment of venous drainage.
Our assured knowledge about the impact of circulatory
disturbances in venous drainage is mostly restricted to our
knowledge about the clinical impact of disclosed or undis-
closed dural arteriovenous shunts. This impact is very high.
And in numerous cases of unexplained hyperintensity of the
spinal cord and progressive symptoms, failure to disclose
the site of the pathological arteriovenous shunt was the real
reason for a misdiagnosis. Missing experience in the neces-
sary examination procedure or technical limitations were the
concrete obstacles.
References
Aboulker J, Bar D, Mursault L, Khoudja F, Redondo A, Gard L, Nahum
H (1977) L’hypertension veineuse intra-rachidienne par anomalies
multiples du systeme cave: une cause majeure de souffrance medul-
laire. Chirurgie 103:1003–1015
Batson OV (1957) The vertebral vein system. Am J Roentgenol 78:195–212
Breschet G (1828–1832) Recherches anatomiques, physiologiques et
pathologiques sur les systemes veineux et specialement sur les
canaux veineux des os. Villaret, Paris
Clemens HJ (1961) Die Venensysteme der menschlichen Wirbelsäule.
De Gruyter, Berlin
Crock HV, Yoshizawa H (1977) The blood supply of the vertebral col-
umn and spinal cord in man. Springer, Wien/New York
Crock HV, Yamagishi M, Crock MC (1986) The conus medullaris and
cauda equina in man. An atlas of the arteries and veins. Springer,
Wien/New York
Dommisse GF (1975) The arteries and veins of the human spinal cord
from birth. Churchill Livingstone, Edinburgh/London/New York
Doppman J, Girton M, Papovski MA (1979) Acute occlusion of the
posterior spinal vein. J Neurosurg 51:201–205
Gailloud P (2014) The arrow-tipped loop is a marker of radiculomedul-
lary vein thrombosis linked to the anti-reflux mechanism–angio-
graphic anatomy and clinical implications. Neuroradiology
56:859–864
Gillilan LA (1970) Veins of the spinal cord. Neurology 20:860–868
Groen RJM, Groenewegen HJ, van Alphen HAM, Hoogland PVJM
(1997) Morphology of the human internal vertebral venous plexus:
a cadaver study after intravenous Araldite CY 221 injection. Anat
Rec 249:285–294
Gruner JE, Lapresle J (1962) Etude anatomo-pathologique des
médullopathies d’origine vasculaire. Rev Neurol 106:592–631
Herren RY, Alexander L (1939) Sulcal and intrinsic blood vessels of
human spinal cord. Arch Neurol Psychiatry 41:678–687
Hughes JT (1971) Venous infarction of the spinal cord. Neurology
(Minneap) 21:794–800
Jacobs T (1996) Venae radiculares. Anatomische Untersuchungen zur
venösen Drainage des menschlichen Rückenmarkes. Thesis.
Medizinische Fakultät der Rheinisch-Westfälischen Technischen
Hochschule Aachen
Jellinger K (1966) Zur Orthologie und Pathologie der Rückenmarks-
durchblutung. Springer, Wien/New York
Kadyi H (1889) Über die Blutgefäße des menschlichen Rückenmarks.
Gubrynowicz u Schmidt, Lemberg
Kim RC, Shmith HR, Henbest ML, Choi BH (1984) Nonhemorrhagic
venous infarction of the spinal cord. Ann Neurol 15:379–385
3 Venous Drainage
Krings T, Mull M, Bostroem A, Otto J, Hans FJ, Thron A (2006) Spinal
epidural arteriovenous fistula with perimedullary drainage. Case report
and pathogenetical considerations. J Neurosurg Spine 5:353–358
Kulenkampff C, Matheis H (1961) Zur Problematik der spinalen
Gefäßprozesse. Spinale Thrombophlebitis. Acta Neurochir (Wien)
7(Suppl):379–385
Lanz J, Wachsmuth W (1982) Praktische Anatomie, vol 2/7. Springer,
Berlin/Heidelberg/New York
Lasjaunias P, Berenstein A (1990) Surgical neuroangiography. 3.
Functional vascular anatomy of brain, spinal cord and spine.
Springer Berlin/Heidelberg/New York. pp 80–87
Meijenhorst GCH (1977) Lumbar epidural double-catheter venography
with Metrizamide (Amipaque). Diagn Imaging 48:244–252
Merland JJ, Riché MC, Chiras J (1980) Les fistules arterio-veineuses
intra-canalaires, extra-médullaires à drainage veineux médullaire.
J Neuroradiol 7:271–320
Moes P, Maillot C (1981) Les veines superficielles de la moelle epinière
chez l’homme. Essai de systematisation. Archives D’Anatomie
D’Histologie D’Embryologie normales et expérimentales. Extrait
du tome 64. Edition Alsatia, Paris, pp 5–110
Neumayer E (1966) Spinale Phlebitis. Dtsch Z Nervenheilk 189:
7–103
Oswald K (1961) Untersuchungen über das Vorkommen von
Sperrmechanismen in den Venae radiculares des Menschen. Med
Inaug Diss, Berlin
Otto J (1990) Morphologie des Sperrmechanismus am Duradurchtritt der
Venae radiculares des Menschen. Neuroradiologische und histolo-
gische Befunde. Thesis for doctorate in medicine, University of Aachen
Quast H v (1961) Die Venen der Rückenmarksoberfläche. Gegenbaurs
Morph Jb 102:33–64
Rao KR, Donnenfeld H, Chusid JG, Valdez S (1982) Acute myelopathy
secondary to spinal venous thrombosis. J Neurol Sci 56:107–113
Schroeder JM, Otto J, Thron A (1993) Protective anti-reflux devices for
spinal veins at the radicular nerve entry/exit zone: a correlative
microscopic and microangiographic autopsy study. Schweiz Arch
Neurol Psychiatr 144:29–30
Suh TH, Alexander L (1939) Vascular system of the human spinal cord.
Arch Neurol Psychiat 41:659–677
Tadié M, Hemet J, Aaron C, Bianco C, Creissard P, Huard P (1979) Le
dispositif protecteur anti-reflux des veines de la moelle.
Neurochirurgie 25:28–30
Tadié M, Hemet J, Freger P, Clavier E, Creissard P (1985) Anatomie mor-
phologique et circulaire des veines de la moelle. J Neuroradiol 12:3–20
Théron J, Moret J (1978) Spinal phlebography. Springer, Berlin/
Heidelberg/New York
Thron A (1988) Vascular anatomy of the spinal cord. Neuroradiological
investigations and clinical syndromes, 1st edn. Springer, Wien/
New York
Thron A (2002) Vascular anatomy of the spine. In: Byrne J (ed)
Interventional neuroradiology. Oxford University Press, Oxford
Thron A (2008) Vascular anatomy of the spine and spinal cord. In:
Hurst RW, Rosenblum RH (eds) Interventional Neuroradiology.
Informa Healthcare, New York/London
Thron A, Otto J, Schroeder JM (1990) Functional anatomy of the dural
segment of spinal cord draining veins. A histological and microan-
giographical study. In du Boulay G (ed) Symposium neuroradio-
logicum. Book of abstracts. Springer, London, p 323
Thron A, Krings T, Otto J, Mull M, Schroeder JM (2015) The transdural
course of radicular spinal cord veins – a microangiographical and
microscopical study. Clin Neuroradiol 25(4):361–369
Tureen LL (1938) Circulation of the spinal cord and the effect of vascu-
lar occlusion. Res Publ Ass Res Nerv Ment Dis 18:394–437
Tveten L (1976) Spinal cord vascularity. The venous drainage of the
spinal cord in the rat. Acta Radiol Diagn 17:653–662
van der Kuip M, Hoogland PV, Groen RJ (1999) Human radicular
veins: regulation of venous reflux in the absence of valves. Anat Rec
154:173–180
 Clinical Applications
Only selected clinical examples and case studies are presented
in this book. Mainly for didactic purposes but also with respect
to the clinical impact, we prefer to present cases of spinal arte-
riovenous disorders that are particularly suited for a mental
training in vascular anatomy of spine and spinal cord. Spinal
cord arterial ischemia is discussed in Sect. 2.11. Analysing rare
cases of suspected acute infarction of the spinal cord is impor-
tant but troublesome, and the diagnostic problems encountered
with acute spinal stroke cannot be solved satisfactorily within
the scope of this chapter on clinical applications.
Therefore, this chapter offers the opportunity to analyse
exemplary cases that have been diagnosed (and mostly treated)
© Springer International Publishing 2016
A.K. Thron, Vascular Anatomy of the Spinal Cord: Radioanatomy as the Key to Diagnosis and Treatment,
DOI 10.1007/978-3-319-27440-9_4
4
at our institution (Huffman et al. 1995; Thron and Caplan
2003; Thron and Mull 2004; Krings et al. 2006a; Thron and
Krings 2007; Krings 2010). The selected clinical applications
are restricted to five topics, beginning with normal vascular
anatomy and the interpretation of some normal but unusual
findings observed on clinical tomographic images. The sec-
ond part deals with spinal arteriovenous malformations
(SAVMs), the third with spinal dural arteriovenous fistulas
(SDAVFs) and the fourth vertebrospinal and medullary vascu-
lar tumours. The final fifth part gives advice about how some
of the repeatedly observed pitfalls in procedure and evaluation
of spinal angiography can be recognized and avoided.
137
138
4.1 Applications in Normal Spinal Digital
Subtraction Angiography
and Interpretation of Uncommon
Blood Vessels on Tomographic
Imaging
By selective catheterization of the segmental arteries, the ante-
rior radicular feeders of larger calibre and their continuation
into corresponding segments of the anterior spinal artery can
be demonstrated in the normal spinal cord in a high spatial and
contrast resolution. Visualization of the posterior radicular
feeders is also occasionally possible (Fig. 4.1). Visualization of
longer vessel courses in selective digital subtraction angiogra-
phy (DSA) is hampered not only by technical limitations of
radiology but also by inflow of non-contrasted blood from
other supply territories above and below. This is also why dem-
onstration of the draining veins is difficult using this technique
as long as only small amounts of contrast medium (approx.
2 ml) are injected. The disadvantages of this selective tech-
nique are not encountered with global blood vessel–contrast-
ing techniques such as contrast-enhanced magnetic resonance
angiography (ce-MRA) (Fig. 4.31). Therefore, combining
both techniques means combining the advantages and disad-
vantages of selective and non-selective procedures (Fig. 4.37).
In the thoracolumbar region the great radicular artery can
always be visualized in normal cases provided that a complete,
selective spinal angiography is performed. Visualization of
smaller radiculomedullary feeders (<0.2 mm) in other spinal
Fig. 4.1 Spinal cord–supplying arteries in digital subtraction angiogra-
phy. (a) Typical anterior feeder to the cervicothoracic region originating
from one of the branches of the supreme intercostal artery. The ascend-
ing and descending branches, which are parts of the longitudinal anas-
tomotic tract, called the ‘anterior spinal artery’, have equal calibre. (b)
Anterior radiculomedullary arteries to the upper thoracic region may be
very small (<0.2 mm) and can easily be overlooked (arrows). (c) Films
in the late arterial phase may be helpful (arrows). The continuity of the
anterior spinal artery may even be interrupted in this region. In this case
the blood vessel cannot serve as an anastomotic tract between the cervi-
cal and the thoracolumbar enlargement. (d) Great anterior radiculomed-
ullary artery (Adamkiewicz artery, arrows) supplying the thoracolumbar
enlargement (reproduced from Thron 2008). (e) The unsubtracted
image of d allows a better topical attribution of the segmental arteries,
the extra- and intraspinal anastomoses and the anterior radiculomedul-
lary artery to the osseous elements of the spine. (f) Great anterior radic-
4 Clinical Applications
cord regions can be a problem. Some examples from actual
angiographic equipment are given in Fig. 4.1. In addition to
potential feeders of the spinal cord, small branches to the
homolateral hemivertebra, to the dura and to the nerve roots
always issue from the segmental arteries (Figs. 2.5 and 4.1f).
Attention has to be paid to the fact that the ventral branches of
two intercostal arteries can arise from a common segmental
stem. But the stem of the second segmental artery with the
ramus spinalis of the dorsal branch should then originate sepa-
rately from the aorta (see Fig. 4.1f, as well as Sect. 2.4 and
Fig. 2.5 ). When looking for the feeder(s) of a spinal arteriove-
nous disorder (SAVM or SDAVF) or of spinal cord blood sup-
ply, one must take this situation into consideration.
It is widely assumed that a radicular artery and its con-
tinuation into a descending branch of the anterior or postero-
lateral spinal artery are easily identified by the typical
“hairpin” course. Figure 4.2 exhibits an instructive example
from clinical angiography reminding that the radiculomedul-
lary veins may have an identical configuration.
Figures 4.3, 4.4, and 4.5 demonstrate that the superficial
longitudinal trunks, their intrinsic anastomotic connections
through a midline section of the spinal cord and the continu-
ation of the anterior median vein to the filum terminale or to
a deep sacral root are the largest blood vessels within the
dural sac (1.5–2.0 mm). Additionally they have slow flow
and are therefore together with the veins of the posterior
venous plexus the spinal cord blood vessels most easily
depicted on contrast-enhanced tomographic images.
ulomedullary artery originating from one of two posterior intercostal
arteries that have a common trunk in their origin from the aorta. The
posterior intercostal artery with radiculomeningeal and medullary sup-
ply (hemivertebral blush and radiculomedullary branch) is a true seg-
mental artery (T11). The other one represents only the ventral branch of
the posterior intercostal artery T10. The dorsal branch of this segmental
artery T10 should have a separate origin from the aorta. It has to be
looked for if complete demonstration of radiculomeningeal and radicu-
lomedullary supply is required (Sect. 2.2.2, Fig. 2.5). (g) Posterior
radiculomedullary artery originating from the 12th posterior intercostal
artery in anteroposterior (a.p.) view (arrow). Normal posterior feeders
are generally only depicted well at the levels of the thoracolumbar
enlargement. (h) Lateral view of the subtracted angiographic image. (i)
Lateral view of the unsubtracted image clearly demonstrating the pos-
terior position of the artery in the dural sac and vertebral canal
4.1 Applications in Normal Spinal Digital Subtraction Angiography and Interpretation of Uncommon Blood Vessels 139

a A.p. view b A.p. view c A.p. view

T3
T3
Supreme ICA

d e f A.p. view

T 11

T 10+11

g A.p. view h Lat. view i

T 12
T 12

T 12
140 4 Clinical Applications

a b
A.P. view A.P. view

Fig. 4.2 Digital subtraction angiography of radiculomedullary arteries
and veins. (a) Arterial phase of an angiogram of the great radiculomed-
ullary artery. (b) Venous phase. The radiculomedullary vein is coinci-
dentally observed at the same level. The hairpin course configuration
alone does not help in the differentiation between artery and vein. In the
case of an arteriovenous shunt disorder with early appearing veins, the
time resolution of a dynamic sequence (e.g. in magnetic resonance
angiography or computed tomography angiography) remains important
for the evaluation of the angiogram (Courtesy of Prof. G. Schroth,
Bern, Switzerland)
4.1 Applications in Normal Spinal Digital Subtraction Angiography and Interpretation of Uncommon Blood Vessels 141

a b
Ant.
Ant.

T7

T5

Fig. 4.3 (a) Venous midline anastomosis in a midsagittal microangio- fistula. The configuration is typical for this type of midline anastomosis
gram (arrow) (Reproduced from Thron 1988). (b) Lateral view of clini- (arrow) (Courtesy of Prof. G. Schroth, Bern Switzerland)
cal spinal angiography at thoracic levels in a case of dural arteriovenous
142 4 Clinical Applications

a b c
a b c

Fig. 4.4 Large transmedullary midline anastomoses depicted in a clin- allows no differentiation between a superficial or intrinsic venous mid-
ical and a post-mortem examination. (a) Midsagittal section of contrast- line anastomosis (arrow). (c) Microangiogram of a midsagittal section
enhanced magnetic resonance imaging (T1-weighted) (arrow). (b) from the specimen shown in b (arrow) (a Courtesy of Prof. D. Petersen,
X-ray of a contrast-injected spinal cord specimen in lateral view, which Luebeck, Germany; b, c Reproduced from Thron 1988)
4.2 Applications in SAVMs
a b
Fig. 4.5 Vein of the filum terminale as continuation of the anterior
median spinal vein in a clinical and a post-mortem examination. (a)
Midsagittal section of a T1-weighted contrast-enhanced magnetic reso-
nance image showing the superficial anterior (arrow) and posterior
midline veins. (b) Microangiogram of a midsagittal section demonstrat-
ing the anterior median and the terminal vein and intrinsic central veins
(Reproduced from Thron 1988)
143
4.2 Applications in SAVMs
4.2.1 Definition and Classification
Arteriovenous malformations (AVMs) are direct or abnormal
connections between arteries and veins without an inter-
posed capillary bed. In the case of intradural SAVMs, they
are supplied by arteries supplying the spinal cord paren-
chyma (radiculomedullary arteries) and drained by spinal
cord veins. Localization can be within the medulla spinalis
or perimedullary on the surface. The arteriovenous transit of
blood is fast and the resulting shunt volume is high. They are
regarded as inborn lesions.
Vascular malformations of the spinal canal and its menin-
ges are rare diseases that lead to damage of the spinal cord if
not treated. Haemorrhage and circulatory disorders are the
most frequent causes of damage, depending on the type of
vascular disease. Space-occupying effects play a minor role.
Following initial MRI, spinal angiography has to be per-
formed with an adequate technique to define the exact nature
of the lesion. This is a critical point because the type of lesion
and the anatomical vascular details for both the lesion and
the spinal cord are fundamental to the decision about the
most appropriate therapy, which may be endovascular, neu-
rosurgical, combined or attentive (Thron and Caplan 2003;
Thron and Mull 2004; Krings et al. 2004, 2006a).
Topographically, vascular malformations of the head and
spine can be found in extra- and intradural compartments.
The dura mater may be involved in rare cases of these inborn
diseases. But the dura mater itself is a place in which acquired
shunts between arteries and veins occur. Those from the cra-
nial meninges have been known for a long time. There, the
established shunts lead arterial flow from meningeal branches
commonly into the cranial dural sinuses. This makes the
diagnosis easy. In the spinal canal dural sinuses do not exist,
and the “hidden” arteriovenous shunt directly into the pial
spinal cord veins may look like a true arteriovenous malfor-
mation if important details of vascular anatomy are
disregarded.
Classification schemes for vascular malformations can be
based on many different criteria. Parameters from histopa-
thology, angiography, endovascular accessibility or genetic
predispositions (Rodesch et al. 2002) may be selected for
classification purposes. A generally accepted classification
does not exist.
From an anatomical and clinical point of view it seems
reasonable to establish a classification of spinal vascular
malformations, which is similar to that commonly used for
the brain and which separates the inborn malformations from
the acquired lesions (Thron and Caplan 2003; Krings et al.
2005a, 2006a).
144
A classification of inborn spinal vascular malformations
comprises
• Arteriovenous malformations (AVMs) and
• Cavernomas.
Capillary telangiectases are another form of inborn
lesions frequently encountered in the brain and brain stem,
but to our knowledge they have not been reported to occur in
the spinal cord parenchyma.
Depending on the size and angioarchitecture of the nidus
(which means the area of histologically undefined blood ves-
sels between supplying artery and draining vein) and haemo-
dynamics of a spinal AVM, the following angiographically
defined subtypes have been introduced:
• Spinal AVM subtypes
– Fistula type 1–3 (perimedullary f.; 1 = small, 2 = medium,
3 = giant)
– Glomus type (or nidus type; conglomeration of abnor-
mal blood vessels)
– Juvenile type (spinal AV metameric syndrome (SAMS)/
Cobb syndrome)
AVMs of perimedullary fistula type are simple direct tran-
sitions of a dilated superficial artery into an arterialized vein.
A frequent effect of the shunt is a dilated venous pouch or
aneurysm at the recipient venous side (Figs. 4.6, 4.7, 4.8, 4.9,
and 4.10). It may be very small (Fig. 4.6, type 1) or large and
space-occupying (Fig. 4.10, type 3).
Djindjian et al. in 1977 were the first to describe perimed-
ullary fistulas. Merland and his group distinguished three
types, according to the vessel size, the shunt volume and the
drainage pattern (Riché et al. 1983; Gueguen et al. 1987).
A spinal AVM of fistula type 3 (giant or macro-fistula) is
now known as a neurovascular disorder typically associated
with hereditary haemorrhagic telangiectasia (HHT)
(Mont’Alverne et al. 2003; Krings et al. 2005b, 2005c). The
macro-fistula consists of several arterial feeders, entering a
solitary large venous pouch through a single hole (Fig. 4.10).
The shunt volume is high and the draining vein is corre-
spondingly enlarged.
AVMs of glomus type represent a frequent type which is
characterized by a conglomeration of greatly dilated vascular
masses (nidus) like in most cerebral AVMs. It may be located
superficially on the surface of the spinal cord or deep within
the cord parenchyma or may extend to both compartments.
Owing to the numerous anastomoses between the spinal cord
arteries, the nidus is always supplied by several arteries or
branches derived from the anterior or posterior spinal arter-
ies (Figs. 4.11, 4.12, 4.13, 4.14, 4.15, 4.16, 4.17, and 4.18).
4 Clinical Applications
AVMs of juvenile type comprise extensive AVMs and
especially the metameric AVMs, involving all the structures
of a given metamere, soft tissue, bone, dura and spinal cord
(SAMS, Cobb syndrome, Klippel–Trenaunay, Parks–Weber
syndrome) (Fig. 4.20).
In clinical practice, many spinal AVMs are complex vas-
cular malformations. Mixed composition of fistulous and
plexiform parts within the nidus (Fig. 4.19) is a frequent
observation of superselective angiography.
It is important to keep in mind that—corresponding to the
intracranial situation—AVMs of the intradural compartment
of the spinal canal are supplied by spinal cord–supplying
anterior and/or posterior radiculomedullary arteries and
drained by radiculomedullary veins. In cases of dural arterio-
venous fistulas the spinal cord–supplying arteries are not
involved in the arteriovenous shunt.
Extradural AVMs or vascular tumours with possible intra-
spinal epidural extension are supplied and drained exclu-
sively by extradural vessels (Fig. 4.41).
Only in the different forms of metameric angiomatosis
segmental extradural and meningeal branches as well as spi-
nal cord supplying arteries are invoved.
4.2.2 Angiographical Findings
Cases with intradural AVMs of the spinal cord correspond in
principle to those of the brain, exhibiting a high variability in
size, supply, shunt volume and drainage of the malformations.
Several feeding vessels are usually involved which can be rec-
ognized as medullary arteries by their typical course. As a
result of the AV shunts, their calibre can be extremely enlarged
(Figs. 4.15, 4.19, and 4.20). The convolution of abnormal ves-
sels, continuing as draining veins with enlarged calibre and
thickened walls, can form an intra- and/or extramedullary mass
(Figs. 4.15 and 4.18). AVMs typically show a rapid and high
shunt into the draining vessels. Low shunt volume and diffuse
staining are more suggestive of vascular tumours (Di Chiro and
Doppman 1969; Vogelsang 1983) (Figs. 4.42, 4.44, and 4.45).
In exceptional cases combinations of haemangioblasto-
mas and AVMs may occur. Aneurysms of the angioma feed-
ers are an extremely rare finding. Instead, aneurysmatic
dilations of the veins are more frequent (Djindjian 1978a;
Cogen and Stein 1983) (Figs. 4.8 and 4.9). Venous drainage
can variably proceed via all longitudinal systems and radicu-
lar paths of drainage.
The angiographical appearance of purely extradural
AVMs of the spinal canal is variable. These vertebral or
soft tissue haemangiomas may or may not extend into the
epidural space. Since the X-ray image of a vertebral hae-
mangioma is rather characteristic (Figs. 4.20 and 4.41), yet
4.2 Applications in SAVMs
the spinal angiographies are diversified, Djindjian et al.
(1981) differentiated three groups (A–C). In the most fre-
quent case (type A) the supplying segmental arteries are not
markedly enlarged and there is a heterogeneous, intensive
vascular blush that exceeds the area of the hemivertebra.
This persists into the late venous phase; early draining
veins are not observed. Similar to an epidural tumour, intra-
spinal extension of the malformation may exert a compres-
sion effect on the spinal cord and its vessels. The
combination with previously mentioned forms of spinal
vascular malformations rarely occurs, with a rate of 4 %
(Djindjian et al. 1981). This also applies to the combination
of segmental angioma of the skin, vertebra and spinal cord,
also known as Cobb syndrome or metameric angiomatosis
(Fig. 4.20). Characteristic of the other forms of vertebral
haemangioma is a weak, very irregular blush (type B), or a
total lack of enhancement (type C) in the angiogram.
According to angiographical criteria, vertebral haemangio-
a c
Fig. 4.6 SAVM of perimedullary fistula type 1 at the level of the
medulla oblongata (C1/2). (a) DSA of the left vertebral artery.
Descending branch of the V4 segment, non-fusion of bilateral tracts
(arrow) with a small aneurysmatic formation at the fistula site. Draining
vein (white arrowhead). (b, c) CTA with multidirectional view on the
145
mas are therefore often of the capillary venous type.
However, if there is dilation of supplying vessels, as seen in
Fig. 4.41, or early venous drainage, at least arteriovenous
components are present. Such forms are more appropriate
for endovascular therapy.
4.2.3 Conclusions
Intradural and extradural AVMs should be clearly distin-
guished from the dural AVFs draining via intrathecal veins.
While the pathophysiology of neurological deficits in
cases of intradural AVMs includes steal phenomena
(Kaufman et al. 1970), dural AV anomalies are pathogeneti-
cally effective only on the basis of increased spinal venous
pressure (Aminoff et al. 1974; Kendall and Logue 1977;
Symon et al. 1984; Hassler et al. 1989; Hurst et al 1995),
because no arteries supplying the spinal cord are involved.
d
vascular malformation allows better evaluation of the anatomical archi-
tecture of the lesion. (d) T2w MRI, midsagittal section. Severe damage
caused by haemorrhage in the region of the (venous) aneurysm at the
fistula site and infarction of the cervical spinal cord
146 4 Clinical Applications

a A. p. view b c A. p. view

T 12 T 11
T 12

L1

T 12

L2 L2
L1

d e f
Lat. view

T 12

T 12

T 12

L12 L2
4.2 Applications in SAVMs
a b a. p. view
Fig. 4.8 SAVM of perimedullary fistula type 2 at T12/ L1 level. (a)
Great anterior radiculomedullary artery originating from L2 (small
arrow). Venous ectasia or aneurysm at the fistula point (large arrow).
(b) Drainage into a midline vein first downwards, then upwards (arrow-
heads) without radicular outflow between L1 and T9. With exception of
the venous aneurysm at the fistula point, location, supply and flow
direction of this SAVM are very similar to the case presented in Fig. 4.7
(Reproduced from Thron 1988)
Fig. 4.7 SAVM of perimedullary fistula type 1 at low thoracic level
(T12/L1). (a) Anterior radiculomedullary artery originating from the
L2 level. Fistula point at T12/L1 level (black arrowhead) into irregular
venous convolutions (white arrowhead) on the posterolateral surface.
(b–e) Drainage is first directed downwards, finally upwards (c) without
147
A. p. view
T 11
Fig. 4.9 SAVM of fistula type 2 at T11 level. Radiculomedullary artery
(upper small arrow) originating from the T11 posterior intercostal
artery and anterior spinal artery which can be followed caudally to the
AV shunt (lower small arrow). Venous aneurysm at the shunt site (large
arrow) and irregularly dilated draining veins (arrowhead). Goal of
therapy in all these cases is occlusion of the aneurysm and shunt with
preservation of vessel patency in the course of the anterior spinal artery
radicular outflow between L1 and T10. (f, g) T2w MRI in midsagittal
(f) and axial sections (g) show remnants of haemorrhage at the fistula
point (arrow). The pencil-shaped hyperintensity of the medulla corre-
sponds to severe multisegmental myelopathic damage
148 4 Clinical Applications

a b c

d e f
4.2 Applications in SAVMs
a Lat. view
b Oblique view
R R
Fig. 4.11 SAVM of glomus or nidus type located in the medulla
oblongata. (a) Unsubtracted image of the right vertebral artery angio-
gram in lateral view. (b–d) DSA of both vertebral arteries in oblique
projection. Supply of the AVM is derived from vertebral artery and
Fig. 4.10 SAVM of fistula type 3 (giant), typically associated with
hereditary haemorrhagic teleangiectasia (HHT). (a) DSA of the left
vertebral artery in a.p. view depicts a large venous pouch. Several feed-
ers from both sides (not shown) enter the pouch through a single hole
(arrow). (b) Midsagittal section of T2w MRI shows the space-occupy-
149
c Oblique view
d Oblique view
R L
mainly posterior inferior cerebellar artery (PICA) branches to the
medulla oblongata on the right side (arrowhead). An additional anterior
feeder to the AVM is the descending branch from the right V4 segment
(arrow) (Courtesy of Prof. L. Solymosi, Wuerzburg, Germany)
ing pouch before treatment. High shunt volume results in enlarged
draining veins. (c) Occlusion of the single hole with coils through the
main feeder from the left vertebral artery. (d–f) DSA and MRI follow-
up with controlled thrombotic occlusion of venous aneurysm and drain-
ing vein
150
a b
Fig. 4.12 SAVM of glomus or nidus type in the upper cervical spinal
cord parenchyma. (a, b) Anterior (open arrows) and posterior (small
arrow) medullary feeders from the right vertebral artery share in the
supply of the vascular lesion. Early draining vein (white arrowheads).
(c) Intrinsic position of the AVM rather posterior than anterior. Anterior
4 Clinical Applications
c d
(open arrow) and posterior (small arrow) supply is better demonstrated
on the lateral view. The early draining vein is the posterior midline vein
(white arrowheads). (d) T2w MRI in a midsagittal section shows rem-
nants of haemorrhage in the anterior part of the spinal cord at the C2
level (Courtesy of Prof. L. Solymosi, Wuerzburg, Germany)
4.2 Applications in SAVMs 151

a c e g i

b d f h j

Fig. 4.13 SAVM of glomus or nidus type in the cervical spinal cord
parenchyma at the C5 level (large arrow). (a, b) DSA of the right
vertebral artery, early arterial phase. Posterior supply from the radicu-
lar level C7 (small open arrow). Early opacification of the anterior
median vein (white arrowhead). (c–f) DSA of the left thyrocervical
trunk. Anterior supply from the radicular level C8 (small closed
arrow). Early filling of the anterior median vein (white arrowhead)
and drainage to the internal vertebral venous plexus. (g, h) DSA of the
left vertebral artery. Posterior supply from the radicular C7 level
(small open arrow), similar to the right side and early draining vein
(white arrowhead). (i, j) T1w and T2w mediosagittal image with
characteristic appearance of an intramedullary AVM and surrounding
areas of oedema or gliosis. Concerning differential diagnosis of spinal
haemangioblastomas please compare the inhomogeneous appearance
of the AVM nidus with the tumour opacification shown in Figs. 4.43,
4.44, and 4.45
152 4 Clinical Applications

a b

Fig. 4.14 SAVM of glomus type and low shunt volume at mid-cervical
level. (a) DSA of the right vertebral artery. Exclusive supply through an
anterior radiculomedullary artery. Faint and inhomogeneous opacifica-
tion of the nidus (large arrow) via small branches of the anterior spinal
artery (arrow). Apparently low shunt volume with no very early drain-
ing veins. Collaterals from the vertebral artery to the deep neck arteries
depict a lower posterior radiculomedullary artery not involved in AVM
supply (arrowhead). (b) The lateral view visualizes the very thin
ascending branch (small arrow) not shown in (a). (c) Superselective
injection near the nidus (large arrow) depicts the ascending branch
(small arrow) and small veins draining to the epidural plexus (white
arrowhead)
4.2 Applications in SAVMs
a b
e f
Fig. 4.15 SAVM of glomus or nidus type and high shunt volume at
low cervical level. (a) DSA of the thyrocervical trunk in a non-
subtracted image demonstrating the nidus at the medullary C6/7 level.
(b, c) The anterior feeder is a radiculomedullary artery entering with the
C8 root. The intense opacification of the nidus, early draining veins
(white arrowheads) into the anterior venous system and immediate out-
flow to the epidural plexus (open arrow) indicate high shunt volume
and possibly intranidal fistulous parts. (d) T2w mediosagittal section to
153
c d
g h
facilitate topical orientation. (e–h) DSA of the left vertebral artery dem-
onstrating essential posterior supply from the C7 radiculomedullary
artery. A different posterior part of the nidus is filled and drainage is
directed mainly to the posterior median vein (white arrowheads).
Different arterial supply of parts of an AVM as shown by angiography
does not imply that the nidus consists of separated compartments. It is
often possible to occlude the complete AVM from a single source of
supply
154
a b
Fig. 4.16 SAVM of monocompartmental glomus type at the T3 level.
(a) A posterior radiculomedullary artery at T4 on the right side is the
main supply to the nidus (arrow). The anterior median vein is the main
drainage route (arrowhead). (b) Contralateral injection at this level
passes the nidus and visualizes posterior feeders from both sides (thin
arrows). Only upon closer inspection can a third radiculomedullary
artery be identified, originating from T4 on the left side
4 Clinical Applications
c d
(thick arrow). (c) In the course of embolization, performed from T4 on
the right side, a left-sided control injection reveals the posterior feeder
(thin arrow) and the anterior spinal artery, now clearly depicted above
and below a not yet occluded part of the nidus (thick arrows). (d) T1w
MRI demonstrating the thrombosed AVM at T3 level, but also (tran-
siently symptomatic) thrombosis of the AVM-draining anterior median
vein (see a), requiring intermittent anticoagulation
4.2 Applications in SAVMs 155

a b

Fig. 4.17 Very similar case to Fig. 4.16 with the glomus type AVM again was monocompartimental and accessible through the main poste-
suspended between two posterior spinal arteries from the T11 level (a) rior feeder (T 11 on the right)
and only limited anterior spinal artery supply from T8 (b). This AVM
156
a b
d e
Fig. 4.18 SAVM of glomus type at the cone. (a) ce-MRA. (b–e) The
great anterior radiculomedullary artery with the arcade around the cone
(b, c) and posterior feeders originating from T11 on the left and L1 on
the right (arrows, d, e) are the main feeders to the different parts of the
nidus (white arrow). The arteries have typical configurations for ante-
4 Clinical Applications
c
f
rior and posterior radiculomedullary feeders. f Improved visualization
of the posterior superficial arterial network including transverse inter-
communications in the course of embolization. Compare the demon-
stration of posterior tracts (black arrows), supplied from T11, in (e) and
(f) (Reproduced from Thron 2008)
4.2 Applications in SAVMs 157

a c e g

b d f h

Fig. 4.19 SAVM at the T10 level without defined nidus. (a) ce-MRA. enlarged compared to the massive dilatation of the superficial spinal
(b) T2w midsagittal MRI shows cord atrophy and numerous distended cord veins. This suggests a rather fistulous composition of the malfor-
blood vessels, mainly arterialized veins all over the thoracic region. mation and possibly drainage obstruction
(c–h) The anterior (c, d) and posterior feeders (e–h) are only moderately
158
a b
d e
Fig. 4.20 Spinal arteriovenous metameric syndrome (SAMS) with
predominant extradural manifestation. (a) X-ray film of the cervical
spine in lateral view showing cystic lesions of the cervical vertebrae 3
to 5 and narrowing of intervertebral spaces. (b) Frontal (c) sagittal
section of T1w MRI. A mass of flow-void structures (arrow) in extra-
and intradural location involves several cervical segments with more
4 Clinical Applications
c
f g
infiltrating than space occupying effect. (d, g) Lateral view of both ver-
tebral arteries in an early angiographic phase. Intense opacification of
the AVM from vertebral artery branches of both sides. (e, f) AVM sup-
ply from segmental branches of the deep cervical and ascending cervi-
cal arteries (Reproduced from Thron 1988)
4.2 Applications in SAVMs 159

A. p. veiw

+"

d
A

a
A
Lat. veiw

V
L3 V

V
V
A
A
e

d-
V
g
A
A

S1

c v
f
V A A
d
v

e A
A
shunt A
f V

c g
g v

Fig. 4.21 Spinal AVM of the filum terminale presenting as a complex
malformation. (a) Selective angiography of the 6th intercostal artery in
a.p. and lateral view. The anterior spinal artery is unusually large for that
spinal level (arrow, a). (b) The blood vessel can be followed downwards
to the L5/S1 level. There, a second blood vessel returns the blood flow in
cranial direction (open arrow). (c) Schematic drawing following evalua-
tion of serial histological sections from the excised shunt area at L5/S1
as indicated in (b). (d–g) The complexity of the vascular AVM is much
higher than could be expected from the limited spatial and contrast reso-
lution of the angiogram performed in 1992. A large, abnormally thick-
walled artery (a) and an arterialized dilated vein communicate just
below the level of e. Below that level, the size of this artery drops drasti-
cally (c, f, g). A second abnormal artery and another small vein are pres-
ent with continuous course (small A and B) and without change of
calibre in this part. Below the shunt level, the large vein ends as a blind
sac as demonstrated in c and g. The complex architecture with numerous
involved blood vessels and the additional content of fat and fibrous tissue
are in accordance with a true vascular midline hamartoma including an
AVM (c–g Courtesy of Dr. Wilhelm Kueker, Institute of Neuropathology
and Department of Neuroradiolgy, University Hospital Aachen)
160 4 Clinical Applications

A. p. view

T 10 A. p. view

L2

Lat. view Lat. view

A. p. view

L4
L4 L4
L4
L4

a c d e f

Fig. 4.22 AVM of the filum terminale at the L4/5 level. Analysis of
supply. (a) Great anterior radiculomedullary artery (arrows) originating
from the T10 posterior intercostal artery. The enlarged and tortuous ves-
sel can be followed downwards to the intersegmental L4/5 level as a
dilated artery of the filum terminale. At this point reversal of flow
through another blood vessel, the terminal vein, can be observed (curved
arrow). (b) At the level of L2 on the left a posterior radicular artery can
be demonstrated (arrowhead), which enforces the posterolateral spinal
tract and contributes in this way to the opacification of the terminal vein
(arrow). The anatomical connection is the arcade between anterior and
posterior tracts around the cone. (c) At the L4 level on the right another
source of AVM supply can be observed. Opacification of the terminal
vein can again be demonstrated (open arrow). The anatomical connec-
tion must be a meningeal and nerve root–supplying radicular artery L4
(arrow) connected to the filum terminale and thus to the shunt area
(curved arrow). (d) Lateral view of T10 DSA which shows only the
haemodynamically dominating shunt supply from the anterior spinal
artery (arrow) to the terminal vein (open arrow). (e) Unsubtracted angi-
ographic image. (f) T2w MRI in a midsagittal section showing the
abnormal vascular structures in the region of the cauda equina
4.3 Applications in Spinal Dural AV Fistulas
4.3 Applications in Spinal Dural AV
Fistulas
4.3.1 History and Definition
The studies on assumed SAVMs by Kendall and Logue (1977)
followed by re-evaluations by Merland et al. (1980) had fur-
nished evidence that one type of SAVM, the most frequent
one (>70 %), had to be reconsidered. The vast majority of the
malformations previously classified as dorsal retromedullary
angiomas (Djindjian et al. 1963, 1975) had been shown to be
pathological arteriovenous shunts with a fistulous arteriove-
nous communication lying on or within the meningeal cover-
ings of the spinal cord (Fig. 4.24). Definition: SDAVFs are
supplied by meningeal (dural) branches of segmental arteries
and drained by spinal cord veins. Contrary to fistulous types
of spinal intradural AV malformations, spinal cord–supply-
ing medullary arteries do not participate in the supply of
dural AVFs. There is a striking similarity in the angiographi-
cal characteristics of dural arteriovenous fistulas of the spinal
dura mater and those of the cranial dura mater. But a wide-
spread network of dilated meningeal arteries penetrating into
the sinus, as for example in arteriovenous fistulas of the pos-
terior cranial fossa, cannot occur spinally as no comparable
dural blood conductors exist. It should be reminded that “one
of the important differences between the meningeal cover-
ings of brain and spinal cord is that the spinal dura mater is
a single-layered membrane. It lacks the periostal component
of the cranial dura mater. The inner layer of the cranial dura
mater is continuous at the foramen magnum with the spinal
dural sheath, which is seperated from the vertebral perios-
tum by an epidural space”, which is an actual and not only
a potential space (Nolte 1999). In clinically symptomatic
cases, the spinal dural AV shunt is located in a region where a
radiculomedullary vein pierces the spinal dura, which forms
their wall for a short stretch (Tadié et al. 1979, 1985; Thron
et al. 2015). The primary venous recipient of the meningeal
artery branch should either be a radiculomedullary vein or
the internal vertebral venous plexus.
A radiculomeningeal arterial branch persists at every seg-
mental level (Chap. 2) but radiculomedullary arteries persist
only at certain levels due to the embryonic desegmentation
process. In the evaluation of symptomatic cases the arterial
shunt volume is almost always directed to the superficial spi-
nal cord veins, and not to the epidural internal vertebral venous
plexus even though the two venous compartments are con-
nected to each other (Sect. 3.7). With increasing experience
we were able to realize a mixed involvement on the angio-
grams more often (Figs. 4.38 and 4.39). Explanations could
be that drainage to the plexal epidural compartment as a pri-
mary drainage route is asymptomatic and tends to occlude.
And in case it is a concomitant drainage route to radiculomed
veins it might be found thrombotically occluded at the time of
diagnosis. (Krings et al. 2006a, Thron et al. 2015).
Figure 4.27 demonstrates on histological serial sections
direct transition of the dilated radiculomeningeal branch into an
161
arterialized vein within the dura mater spinalis. In the histologi-
cally examined case of Fig. 4.26 the meningeal arterial feeder
courses in the wall of the recipient radiculomedullary vein,
before entering it’s lumen. This type of AV shunt has been
called a “fistula” because of the abrupt transition of a vein into
an artery without interposition of a convolution of abnormal
blood vessels. The recipient vein passes into normal paths of
spinal drainage (Figs. 4.23 and 4.24).
From a therapeutic (endovascular or neurosurgical) point of
view SDAVFs could be classified together with arteriovenous
malformations as arteriovenous disorders of the spinal canal and
its meninges. Indeed, they are the most frequent type of this dis-
order. But they should be classified as a separate entity. Like
their cranial counterparts, SDAVFs are acquired lesions, although
their aetiology and possible predisposing factors are not yet
clear. They are pathogenetically different from AVMs and affect
a different age and sex population. They are the clinically most
serious type of spinal arteriovenous disorders characterized by
progressive paraparesis and sphincter disturbances (Aminoff and
Logue 1974; Koenig et al. 1989; Thron et al. 1987, 2003)
4.3.2 Localization
The localization of dural AV shunts corresponds to the dural
exit of radiculomedullary veins. In 69 % of cases this is joint
or near (<1 mm) the exit/entry of a nerve root in the dural
sleeve lateral of the root (Otto 1990). In 31 % of cases radic-
ulomedullary veins pass the dura at greater distance indepen-
dently from the nerve roots. In 121 of our own clinical cases
of SDAVF diagnosed until 2003 more than 75 % of fistulas
were found in the thoracolumbar region (Thron et al. 2003).
But all segments of the spinal axis including the pelvic region
and the skull base are potential sources of dural AVFs with
spinal drainage (Figs. 4.36 and 4.37). Nevertheless, it is sur-
prising to note the rarity of published cases of cervically
localized typical dural AVFs. In only one case from a series
of 129 SDAVFs did we observe double spinal arteriovenous
fistulas at the levels L1 on the right and L2 on the left side
(Krings et al. 2004).
4.3.3 Aetiological and Pathophysiological
Aspects
The term dural arteriovenous fistula is more appropriate for
this most probably acquired disease than malformation or
angioma, because these last two terms suggest inborn lesions.
The epidural or dural vascular convolutions resemble arterio-
venous malformations in histological sections (Fig. 4.27).
The high age at manifestation suggests at least an acquired
triggering mechanism like hypertension, inflammation,
thrombosis or microtrauma. A possible morphological basis
of these pathological shunts could be the AV anastomoses of
the dura in the vicinity of the root sheaths, described by Vuia
and Alexaniu (1969), or the “pelotons vasculaires”, cluster-
162
like vascular formations between arteries and veins men-
tioned by Manelfe et al. (1972) in their study on dural vascular
supply.
The occurrence of physiological arteriovenous communica-
tions in the dura mater cranialis and spinalis is mentioned in
old anatomical textbooks (Langer 1877) and in an anatomical
study by Hammersen (1964). In our morphological and micro-
scopic examinations of spinal dural AV shunts performed in
the 1980s (Fig. 4.25), we observed a shunt-supplying artery
running in the broadened wall of a radiculomedullary vein
before opening into the lumen of the vein (Fig. 4.26). This
could be assessed as an indication that the establishment of an
abnormal arteriovenous shunt pathology might occur between
a meningeal vasa vasorum and the blood vessel wall of the
spinal cord drainig vein. In the transdural course of radicular
spinal cord veins the regular layers of the venous blood vessels
are replaced by meningeal tissue. This may be accompanied by
unusually configured structural arrangements like in the bulge
or nodular type in which the narrowed and sometimes split
lumen is fibrotically encapsulated (Thron et al. 2015).
Aetiological concepts and pathophysiological consider-
ations about impaired venous return from the spinal cord
should revisit the anatomy and the haemodynamic role of the
internal vertebral venous plexus, neglected in the past.
Arteriovenous fistulous connections with these epidural
venous spaces and secondary reflux into radiculomedullary
veins are extremely rare observations (Krings et al. 2006b).
But these observations may be important for aetiological and
pathophysiological considerations, which differ from those
which were discussed until now (Geibprasert et al. 2008;
Krings and Geibprasert 2009).
4.3.4 Pathogenesis of Spinal Cord Damage
While the pathophysiology of neurological deficits in cases of
intradural AVMs is complex and includes steal phenomena
(Kaufman et al., 1979), dural AVFs are pathogenetically effec-
tive only on the basis of increased venous pressure, since arter-
ies supplying the spinal cord are not involved (Aminoff et al.
1974; Kendall and Logue 1977; Symon et al. 1984; Thron
et al. 1987; Hassler et al. 1989, 1994; Hurst et al. 1995).
Mechanical factors such as compression of the cord by dilated,
intrathecal veins are excluded by intraoperative findings.
In Fig. 4.24 intraoperative Doppler recordings and mea-
surements of pressure in the spinal cord veins, close and
more distant to the fistula site are presented. Measurements
were performed before and after occlusion of the menin-
geal feeder. The results strongly support the pathogenetical
concept that an increase in venous pressure and a dimin-
ished arteriovenous pressure gradient causes venous con-
gestion, which means an intramedullary oedema due to
impaired venous drainage (Hassler and Thron 1994). In the
long-term chronic hypoxia leads to necrotizing myelopa-
4 Clinical Applications
thy, a severe damage of the spinal cord, described by Foix
and Alajouanine in 1926. The course of the disease is pro-
gressive. Irreversible paraplegia is the final stage, if the dis-
ease is not diagnosed and treated at an early stage (Koenig
et al. 1989).
4.3.5 Angiographical Findings
Leptomeningeal branches of a segmental spinal artery empty
abruptly into a radicular artery in the vessel segment cross-
ing the dura mater. Flow is now reversed against the spinal
cord and continues through superficial subarachnoid venous
trunks (Fig. 4.28). Normally, these small radiculomeningeal
rami only supply nerve roots and dura and are not displayed
angiographically. Although they do not have significant
calibres, the shunt leads to an arterialization and dilation of
the recipient veins. These veins, which are the physiologi-
cal superficial veins of the spinal cord, get dilated and elon-
gated by reversed arterial flow and produce the impression
of a true intradural AVM (Figs. 4.23 and 4.24). This explains
initial radiological misinterpretations as a retromedullary
angioma resulting in temporarily misguided treatment of
this disorder.
Ascending or descending dural (leptomeningeal) branches
of neighbouring segmental arteries can participate in the
shunt, because anastomotic connections in the network of
dura-supplying arteries are abundant. The localized charac-
ter of the shunt is based on the fact that it usually connects to
a single vessel on the inner side of the dura. This can be dif-
ferent if a short common stem of two or more radicular veins
is affected (Figs. 3.35b and 3.38). The estimated shunt vol-
ume is extremely variable. The progressive filling of draining
veins may exhibit a relatively slow flow even in cases with
apparently large shunt volume. Drainage can be observed
running cranially and/or caudally at the same time over quite
a distance (Figs. 4.32 and 4.33). Even repeated changes in
upwards/downwards flow direction can be seen associated
with a change of the longitudinal trunks that were used
(Fig. 4.35). In an initial evaluation of our first 20 cases
(Thron 1988, p. 86) the shunted blood was received and
transported more often by the dorsal veins. However, a
switch to the ventral side (or vice versa), using the external
and intrinsic venous anastomoses could also be observed
(Figs. 4.34 and 4.35). There was a tendency of fistulas at low
levels (S3–T11) to drain upwards and of high thoracic levels
to go downwards. The fistulas at intermediate thoracic levels
all drained in both directions. A fistula at the lumbosacral
level primarily involves the terminal vein or a sacral radicu-
lar vein (Figs. 4.37 and 4.39).
Often, outflow to the extradural plexus is not apparent over
a long distance and drainage can be observed from the lumbar
region to the cervical or cranial area (Fig. 4.35). Merland et al.
(1980) were the first who suspected a second angiographic
4.3 Applications in Spinal Dural AV Fistulas
characteristic of SDAVFs: outflow via radicular veins of
neighbouring segments seemed to be impaired or missing,
when compared with the large number of radiculomedullary
veins which had been counted in normal anatomy.
The explanation might be an overload of the venous drain-
age capacity by the shunt volume. But this contrasts with the
even larger shunt volume in cases of SAVMs. More likely is an
individually restricted drainage capacity in the thoracic region,
caused by a primary or secondary veno-occlusive disease.
4.3.6 Diagnostic Management
The dural fistulas are rare, but they are the most frequent and
the most devastating form of a spinal arteriovenous disorder
in an unselected population (Koenig et al. 1989; Thron and
Caplan 2003). Therefore they have to be regarded as one of
the greatest challenges for neuroradiologists.
Meanwhile, MRI has become an early screening method
with reliable findings in case of clinical suspicion. The
results of ce-MRA can serve as an indicator of the shunt site
(Backes et al. 2004; Mull et al. 2007). This allows more
localized, not necessarily complete angiography along the
spinal axis, which is helpful considering that the average
patient age is over 60 years. Should the angiograms be
negative in the suspected region, completion must be
attempted, since arterialization of the spinal veins can occur
from the sacral end of the dural sac into the terminal vein
(Merland et al. 1980; Burguet et al., 1985) (Fig. 4.37) or
from fistulas localized at the skull base (Woimant et al. 1982;
Reinges et al. 2001) (Fig. 4.36). In this context it should be
reminded that the segmental (radicular und dural) supply to
the fifth lumbar level is derived from the iliolumbar artery, a
branch of the internal iliac artery. When indicated, it should
be injected selectively. Even in case of complete but negative
selective spinal and cranial angiography, repetition should
be taken into account if the suspected diagnosis of a dural
fistula is strongly supported by clinical data and imaging
results. In our opinion, the serious clinical consequences of
a missed diagnosis of SDAVF justify or even require this
management. Following this strategy we were increasingly
successful over the years in quite a lot of cases.
But with increasing sensitivity of screening methods, espe-
cially ce-MRA, the number of these cases will drop consider-
ably and be restricted to singular cases of low volume shunts.
4.3.7 The Question of Chronic Myelopathies
in the Absence of AV Shunt Disorders
The experiences gained by selective spinal angiography in the
late 1970s and the 1980s have taught clinicians in the neuro-
disciplines that the superficial spinal cord–draining veins may
play an important pathogenetic role in vascular diseases of the
163
spinal cord, like in SAVMs and different forms of spinal arte-
riovenous fistulae (SAVF), mainly the dural ones (SDAVFs).
Aboulker et al. (1977) and others (Merland et al. 1980; Tadié
et al. 1985; Thron 1988) attempted to explain a group of
chronic myelopathies by venous intraspinal hypertension
caused by deficient venous outflow or disturbed reflux control
from potentially increased pressure in the epidural plexus
(Sect. 3.7, 3.9.4). But clinical experience gained over the last
30 years has not provided new evidence for this assumption.
Among investigators of spinal cord blood vessels it was
generally accepted that spinal cord veins cannot be filled
with blood or any other fluid by injection from the outside of
the dura mater. The nature of this reflux-impeding mecha-
nism has been a matter of dispute. Studies by Tadie et al.
(1979, 1985) excluded the presence of valves and demon-
strated instead an anti-backflow system resulting from nar-
rowing and zigzagging of the vein while crossing the dura
mater. We confirmed these results and presented a second
arrangement in the dural segment of radiculomedullary
veins, which might be appropriate to modulate flow (Thron
et al. 2015). van der Kuip et al. (1999) and Groen et al.
(1997) had no problems to achieve retrograde filling of the
radicular veins in a post-mortem study following thorough
preparation of the cadavers with infusion of saline contain-
ing a thrombolytic agent. They concluded that reflux could
be considered to be a physiological phenomenon.
4.3.8 Open Questions
One of the open questions in the probably acquired dural AV
shunt, is the surprising fact that blood flow is almost always
directed against the spinal cord and not into the compart-
ments of the internal vertebral venous plexus.
But the predominant compartment structure of the blood
pools of the internal vertebral venous plexus in the microan-
giograms (Figs. 3.42–3.44) underline the hypothesis that in
the vast majority of SDAVFs an initially arterialized compart-
ment of the epidural plexus becomes spontaneously occluded.
But in the meantime some clinical cases of all these varia-
tions have been reported (Krings et al. 2006b; Figs. 4.34,
4.38, and 4.39). To our knowledge, only a few reports exist
about clinically symptomatic arteriovenous fistulas with the
flow primarily directed to the epidural plexus (Chen et al.
1977; Cognard et al. 1995 and Pirouzmand et al. 1997). All
of these cases, except one (Chen et al. 1977), had secondary
obvious reflux from the epidural plexus to the superficial spi-
nal cord veins, as demonstrated by angiography. This is a
strong argument for the paramount impact of AV shunts to
generate clinically significant impairment of venous outflow.
The clinical impact of disturbed reflux control is uncertain.
The functional role of the probably predominant epidural
venous plexus for the spinal cord blood circulation remains
poorly understood.
164 4 Clinical Applications

Fig. 4.23 Photographs of

findings at surgery. (a) View on Epidural Intradural
the posterior surface of the spinal space space
cord. The elongated and very
tortuous blood vessel is the
regular posterior median spinal
vein, secondarily enlarged as a
result of arterialized flow. (b)
Enlargement of the veins (open
arrow) starts at the inner surface
of the dura mater with a Dura
circumscribed dilatation at the mater
point where the vein crosses the
dura (arrow). In the short
segment of transdural exit, the
vein is entered by small
radiculomeningeal arteries
(Courtesy of Prof. Grote,
Department of Neurosurgery,
University of Tübingen,
Germany) (b Reproduced from
Thron 1988) Transdural
exit

a b
4.3 Applications in Spinal Dural AV Fistulas 165

Fig. 4.24 Illustration of operative findings in a case of spinal dural ferent distances from the fistula point are demonstrated before and after
AVF. Doppler recordings of leptomeningeal feeder and draining vein as shunt occlusion (70 mmHg versus 45 mmHg) (Courtesy of Prof. Hassler,
well as measurements of venous pressure in the spinal cord veins at dif- Department of Neurosurgery, University of Tübingen, Germany)
166 4 Clinical Applications

Fig. 4.25 Dural segment of an

arterialized vein (aV).
Histological section of the
excised shunt area in case of
spinal dural AVF. Oblique
passage of the arterialized vein
through the dura mater (D). Only I
marginal narrowing of the blood aV
vessel in its transdural course
(arrows). The entry of the artery
into the vein is not shown on this
section (van Gieson staining,
original magnification ×40) D
(Courtesy of Prof. Roggendorf,
Department of Neuropathology, D
University of Tübingen,
Germany) aV

a b
Artery

Vein
Venous wall

Artery

Dura mater

Dura mater

c d
Arteriovenous
shunt

Shunt

aV
aV

Fig. 4.26 SDAVF with an artery running in the broadened wall of the
arterialized vein before entering it. Histological sections of an excised
shunt area. (a) A small artery is enclosed in the broadened wall of an
arterialized vein (arrow) (previous vasa vasorum?). (b) In the area
before the shunting artery opens, the composition of the vessel wall
becomes irregular with splitting of the internal elastic lamina and
marked fibrosis of the tunica media (arrow). The arterialized vein (aV)
shows a considerable increase in elastic fibres and smooth muscle cells.
(c, d) The lumen of the artery has opened into that of the vein. Elastic
fibres from the arterial wall now form the bottom of the resulting
groove (arrow) (van Gieson staining) (Courtesy of Prof. Roggendorf,
Department of Neuropathology, University of Tübingen, Germany)
4.3 Applications in Spinal Dural AV Fistulas 167

a b

A
E E
A
D
G
D G
D
V
V

I I

Fig. 4.27 (a–c) Transition of a dural artery (A) in the dural segment of
a radiculomedullary vein (V). Histology of an excised shunt in a case of
spinal dural AVF. Series of histological sections with staining of the
internal elastic lamina reveal the direct transition of a dural artery into
the dural segment of an abnormal radicular vein (small arrow in b;
large arrow in c). The artery is characterized by the internal elastic
lamina (small arrow in c). The arterialized vein shows a thickened wall
with scattered elastic fibres (open arrow in c). A glomus-like formation
(G) is situated close to the communication point (small arrow in b,
large arrow in c). E epidural space, I intradural space, D dura mater.
Elastica van Gieson staining. (Courtesy of Prof. Roggendorf,
Department of Neuropathology, University of Tübingen, Germany) (c
Reproduced from Thron 1988)
168 4 Clinical Applications

a b

Fig. 4.28 DSA of SDAVF at T7 level with high shunt volume. (a) A
network of dilated radiculomeningeal arterial branches (arrow) sup-
plies the arteriovenous shunt. (b) Drainage passes the radiculomedul-
lary vein to the longitudinal midline trunk and is directed up- and
downwards (white arrowheads). An intradural spinal AVM can be
excluded because no spinal cord–supplying arteries are involved in this
arteriovenous shunt disorder. The shunt site has to be looked for at dural
level of T7. DSA digital subtraction angiography, SDAVF spinal dural
arteriovenous fistula
4.3 Applications in Spinal Dural AV Fistulas 169

Fig. 4.29 DSA of SDAVF at T5 level. In this case two different longi-
tudinal veins are opacified simultaneously (white arrows) at the shunt
site (black arrow). This is certainly caused by the presence of a short
common stem before the radicular veins exit the dural sac, as shown in
the morphological and radio-anatomical examples of Figs. 3.32a, 3.35b,
and 3.38. The diagnosis of SDAVF (and not SAVM) is based on the fact
that a spinal cord–supplying artery is obviously not involved. DSA digi-
tal subtraction angiography, SDAVF spinal dural arteriovenous fistula,
SAVM spinal arteriovenous malformation (Reproduced from Thron
2008)
170
a b
Fig. 4.30 DSA of SDAVF at the L3 level. (a, b) Subtracted images in
a.p. view. (c) Unsubtracted in lateral view. (a) The blood vessel that is
opacified first (white arrow, question mark) resembles a radiculomedul-
lary artery. It runs upwards to the lumbar enlargement and shows a hair-
pin configuration. (b) On later images it branches into typical venous
tracts (white arrows). Furthermore, a focal change of calibre in the
course of the superficial vessels, indicating an arteriovenous shunt, does
not exist. Therefore, the presence of a SAVM of perimedullary fistula
type can be ruled out. All of intradural blood vessels are veins and the
4 Clinical Applications
c
shunt has to be localized in the dura mater at L3 (black arrow).
(c) Ventral position of this lumbar vein L3 on the surface of the spinal
cord (white arrow). Compare with the anatomical examples given in
Figs. 3.11a and 3.31. The configuration of a radiculomedullary vein can
be very similar to that of an artery and in case of an AV shunt careful
analysis and high time resolution of dynamic angiographic series are
helpful. DSA digital subtraction angiography, SDAVF spinal dural arte-
riovenous fistula, SAVM spinal arteriovenous malformation (Reproduced
from Thron 2008)
4.3 Applications in Spinal Dural AV Fistulas
a b c d
Fig. 4.31 Contrast-enhanced spinal magnetic resonance angiography
(ce-MRA) (a, b) and selective spinal digital subtraction angiography
(DSA) (c, d) in the screening and diagnostic evaluation of suspected
SDAVFs can occur principally at each segmental level of the
spine from the base of the skull down to the end of the spinal
canal. ce-MRI has turned out to be an imaging technique not
only helpful to support or exclude the suspected diagnosis of the
presence of a spinal AV shunt but also to localize the lesion at
least roughly, if not exactly, like in the presented example. The
technique is very sensitive in depicting the early filling of veins
and of venous flow in general. Thus, selective spinal DSA can
171
spinal arteriovenous disorders, exemplified with a spinal dural
arteriovenous fistula (SDAVF) at T10 level
be focused on the region of interest and complete the diagnostic
evaluation with details concerning topographical and differen-
tial diagnostic aspects required for assured diagnosis and safe
treatment. It offers the advantage of selectivity and of high tem-
poral and spatial resolution, often necessary to differentiate the
many forms of slow or fast flowing AV shunt lesions and of
feeder identification. A questionable alternative to MRA may be
offered by spinal CT techniques.
172
a b
Fig. 4.32 Selective spinal angiography of a spinal dural AVF at the
T12 level with abnormal drainage to cervical and cranial levels. (a) The
dural AV shunt is located in the intervertebral foramen T12/L1 (arrow).
Ascending drainage via a radiculomedullary vein (arrowheads) and the
superficial veins of the spinal cord. (b) No significant radicular outflow
4 Clinical Applications
c
in the thoracic region. (c) The draining veins can be followed up to the
cervical and craniocervical region. Only at mid-cervical levels, the
anteromedian vein (arrowhead) partly empties into the internal verte-
bral venous plexus (arrows) (Reproduced from Thron 1988)
4.3 Applications in Spinal Dural AV Fistulas 173

a b c d e

Fig. 4.33 Selective spinal angiography of a spinal dural AVF with degree in the ascending direction. (d) The serial angiogram at 19 s visual-
abnormal drainage preferentially in caudal direction. (a) Arterial shunt izes opacification of the terminal vein in the lumbar region. (e) The most
occurs through the radiculomeningeal branch of the fifth intercostal impressive finding on myelography is the dilated terminal vein (arrows)
artery. (b, c) Drainage can be observed in the descending and to a lesser far below the shunt localization (Reproduced from Thron 1988)
174
b c
Fig. 4.34 Digital subtraction angiography of SDAVF at the T12 level
showing large anteroposterior venous anastomoses directing blood flow
to different drainage routes. (a, b) Subtracted and unsubtracted image
of the shunt region. The vertical circumscribed contrast accumulation
between the meningeal feeder and the draining vein (white arrowhead)
Fig. 4.35 SDAVF at the L2 level with repeating reversal of upward and
downward flow direction and typical MRI appearance of “necrotizing
myelopathy” as long-term damage of this disease. (a–f) Digital subtrac-
tion angiography in a.p. (a, b) and lateral view (c–f). The early flow
direction is upwards (a, b). At the T11/12 level a proportion of this flow
passes through an intramedullary anteroposterior anastomosis of typi-
cal configuration to go downwards (c–e). At the L1 level there is again
reversal of flow direction of this flow component. Finally (e, f), blood is
distributed to anterior and posterior superficial spinal cord veins drain-
4 Clinical Applications
d
probably corresponds to a minimal involvement of parts of the epidural
plexus (arrow). (c, d) Tortuous superficial longitudinal trunks (white
arrowheads) interconnected by large presumably midsagittal anasto-
moses (arrows). Thus drainage can be directed to different routes
ing cranially (white arrowheads). (g) T1w ce-MRI with typical
enhancement of dilated, relatively slow flowing superficial spinal cord
veins. The spinal cord volume may be increased in subacute cases.
Intrinsic central enhancement of the spinal cord parenchyma indicates
subacute–chronic hypoxic damage caused by disruption of the “blood–
spinal cord barrier”. Unfortunately, neoplasm is an occasional misdiag-
nosis resulting from these signs. (h) T2w MRI with flow-void vascular
structures in the cerebrospinal fluid around the spinal cord and central
parenchymal hyperintensity due to oedema and infarction
4.3 Applications in Spinal Dural AV Fistulas 175

a b g h

c d e f
176
a b
Fig. 4.36 Dural AVF at the skull base near to the foramen magnum
with perimedullary drainage. (a) X-ray in lateral view, unsubtracted
image. Injection of contrast medium into the left ascending pharyngeal
artery and opacification of a vein at the posterior C0/C2 level are visible
(white arrow). (b) Subtracted image in lateral view. Posterior menin-
geal branches of the artery communicate with a short segment of the
4 Clinical Applications
c d
marginal sinus (black arrow) before connection with the posteromedian
spinal vein is evident (white arrowhead). (c) Downwards flow within
the posterior subarachnoid space can be seen (white arrowheads) far
below the cervical region. (d) T2 w mediosagittal MRI shows signs of
parenchymal damage (hyperintensity) and abnormal blood vessels as
dark flow-void dots within the bright cerebrospinal fluid
4.3 Applications in Spinal Dural AV Fistulas 177

a b c d e

Fig. 4.37 Fistulous AV shunt at S1 level between a meningeal sacral
branch from the internal iliac artery to the terminal vein. (a, b) T2w and
ceT1w MRI. The MRI aspect is like in SDAVF with involvement of the
terminal vein. (c) ce-MRA. The localization of the lesion at S1 (arrow)
and the direction of flow in the terminal vein (white arrowhead) are not
unambiguously clear on tomographic or angiographic MR images. The
shunt might be localized everywhere between low thoracic and sacral
levels and drainage could go downwards, upwards or in both directions.
Therefore, the high resolution of conventional angiograms in space and
time remains important. (d, e) The main feeder to the AV shunt (arrow)
as demonstrated by selective DSA is a dural branch from the internal
iliac artery (small arrow). Draining veins (white arrowheads).
Additional small meningeal branches to the AV shunt are not shown
178
A. p. view A. p. view
T 11
a b
Fig. 4.38 SDAVF with involvement of the epidural internal vertebral
venous plexus. (a) DSA image in the early arterial phase visualizes the
AV communication (small black arrow) with draining vein (white
arrowhead) but also a small accumulation of contrast medium in a ver-
tical line (white arrow), presumably in epidural localization (cf.
Figs. 4.3 and 4.10a). (b) In the late arterial phase progressive filling of
the superficial spinal cord veins and the epidural compartment (white
4 Clinical Applications
Lat. view
c d
arrows) are clearly demonstrated. Filling of both venous systems seems
to occur simultaneously, but with higher flow rate into the intrathecal
veins than in the epidural venous compartment. (c, d) Subtracted and
unsubtracted early image of DSA in lateral view showing the shunt
localization at a posterior radiculomedullary vein (white arrowhead)
and early opacification of the epidural component (white arrow)
4.3 Applications in Spinal Dural AV Fistulas 179

Fig. 4.39 Spinal epidural AVF A. p. view A. p. view Lat. view

with secondary remote reflux into a b c
perimedullary veins. (a) DSA of
the third lumbar artery on the
right depicts a shunt to anterior L1
parts of the internal vertebral
venous plexus on the right side
(white arrowhead). Collateral
filling of L2 on the left shows the
great anterior radiculomedullary
artery originating from this level
and side (arrow) as an incidental
finding. (b) Progressive filling of
other parts of the anterior internal
vertebral venous plexus first
downwards, then again upwards
L3 L3
(white arrowheads) results in
entering of a radiculomedullary
vein on the opposite side (white
arrow). (c) The lateral view of a
late venous image demonstrates
the extension of involvement of
the anterior epidural plexus
(arrowhead). The irregular
lacunar spaces do not form a
homogenous or completely
symmetrical venous drainage d e f
system with transverse
communications at each level.
The perimedullary draining veins
indicate that a postulated
reflux-impeding mechanism is not
effective or is non-existent in this
case. (d, e) T1w and T2w MRI
with typical signs of SDAVF with
congestive and necrotizing
vascular myelopathy. (f) ce-MRA
shows epidural (arrowhead) and
intrathecal abnormal findings
(arrow) without context
L3

L1
180 4 Clinical Applications

L1

b c d

L1
a e f g

Fig. 4.40 Anterior spinal artery and SDAVF originating from the same
segmental artery (L1 on the left). Comparison of findings in ce-MRA
(a, b, e) and DSA (c, d, f, g). The arrows point to the normal anterior
spinal artery in a coronal (b) and oblique sagittal target MRA image (e).
The arrowheads mark the arterialized posterior veins of the SDAVF in
all figure parts. The corresponding DSA images in a.p. view are
presented in c and d. DSA comparison in lateral view in an early and
late angiographic phase is given in f and g (a, b, e–g Reproduced from
Mull et al. 2007)
4.4 Applications in Vertebrospinal and Medullary Vascular Tumours
4.4 Applications in Vertebrospinal and
Medullary Vascular Tumours
Purely extradural AVMs of the spine may have quite variable
angiographic presentations. These vertebral or soft tissue
haemangiomas may or may not extend into the epidural
space. Similar to a progressively expanding epidural tumour,
intraspinal extension of some of these malformations may
exert a serious and clinically relevant compression effect on
the spinal cord (Fig. 4.41). This may require endovascular
devascularization combined with surgical decompression.
The X-ray image of vertebral haemangiomas is rather char-
acteristic (Figs. 4.20a and 4.41a), yet the spinal angiogra-
phies are diversified. Djindjian et al. (1981) differentiated
three groups (A–C). In the most frequent case (type A) the
supplying segmental arteries are not markedly enlarged and
there is a heterogeneous, intensive vascular blush that
exceeds the area of the hemivertebra. This persists into the
late venous phase; early draining veins are not observed. The
combination with previously mentioned forms of spinal vas-
cular malformations rarely occurs, with a rate of 4 %
(Djindjian et al. 1981). This also applies to the combination
of segmental angioma of the skin, vertebra and spinal cord,
also known as Cobb syndrome or metameric angiomatosis.
Characteristic of the other forms of vertebral haemangioma
is a weak, very irregular blush (type B), or a total lack of
enhancement (type C) in the angiogram. According to angio-
graphical criteria, vertebral haemangioma therefore often
belong to the capillary venous type.
The non-expanding vertebral haemangiomas are a fre-
quent incidental finding at autopsies and spinal MRI.
The angiographical appearance of the other intra- and
extramedullary tumours (ependymomas, neurinomas,
meningiomas) and of vertebrogenic tumours (sarcomas,
chondromas, osteochondromas, aneurysmatic osseous cysts,
181
giant cell tumours, metastases) does not facilitate the differ-
ential diagnosis of the tumour type in the majority of cases
(Voigt et al. 1978; Djindjian et al. 1981). Nevertheless, indi-
cation of the benign or malignant nature of a lesion may be
provided by demonstration of pathological vessels, arterio-
venous shunts as well as by the extent and type of tumour
blush (Fig. 4.42).
In addition, spinal angiography supplies preoperative
information on the degree of vascularization and the source
of arterial supply of the tumour, plus the very important indi-
cation as to origin and possible displacement of anterior
radicular arteries supplying the spinal cord. The risk of a
“vascular catastrophe” during surgery can be diminished and
the feasibility of preoperative or palliative embolization can
be estimated.
Although MRI and spinal CT have decisively improved the
diagnostics of vertebrospinal masses, spinal angiography can be
helpful and even mandatory as a supplementary diagnostic
method. Compared to the glomerular types of AVMs, low shunt
volume and diffuse staining are more suggestive of vascular
tumours like haemangioblastoma. This is a vascular tumour
characteristic as part of the inherited von Hippel–Lindau syn-
drome. These vascular tumours of intraparenchymal localization
in the spinal cord (15 %), medulla oblongata (5 %) and cerebel-
lum (80 %) are supplied by brain- or spinal cord–supplying arter-
ies of the region in which they become manifest (Figs. 4.43, 4.44,
and 4.45). Therefore, the potentials of preoperative embolization
are limited if damage to the spinal cord prior to surgery should be
avoided. Spinal cord–supplying arteries can be studied in angi-
ographies of these tumours because they are prominent, but pre-
operative embolization carries a separate risk. The neurosurgeon
at our hospital did not demand preoperative embolization. He
surgically isolated the angiographically well-described lesion
before taking it out and followed his own advice:” do not cut in
it before!”.
182
R A. p. view
T9
a c
Fig. 4.41 Vertebral haemangioma with epidural extension.
Significant progressive space-occupying effect on the spinal cord. (a)
Myelography of the thoracic region. The X-ray demonstrates the typi-
cal honey comb appearance of the involved bony vertebral structures.
Circular compression of the dural sac and its content with subtotal
block is evident, but no signs of an intradural vascular malformation.
(b) Selective angiography of the ninth intercostal artery on the right
4 Clinical Applications
T 10
shows descending supply with multiple feeders and an early homoge-
neous blush. The left ninth intercostal artery participated in the supply
to a lesser degree (not shown). (c) Selective angiography of the tenth
intercostals artery with a more inhomogeneous blush, clearly exceed-
ing the normal hemivertebral area of supply. Again, additional supply
could be demonstrated from the contralateral segmental artery (not
shown) (Reproduced from Thron 1988)
4.4 Applications in Vertebrospinal and Medullary Vascular Tumours
a b
Fig. 4.42 Intraspinal neurinoma demonstrated by digital subtraction
angiography. (a) A.p. view of the circular hypervascularized mass (large
open arrow). It is supplied by an anterior and posterior radicular artery,
both emerging from the tenth intercostal artery on the right. The postero-
lateral spinal artery (small arrow) carries the main tumour supply result-
ing in an increase of vessel lumen. Both arteries (small and large closed
a A. p. b c
T 10
T 11
Fig. 4.43 Solitary haemangioblastoma demonstrated by digital sub-
traction angiography. Several spinal cord arteries from different levels
and both sides share in the supply of the vascular tumour at the T9 level.
Posterior spinal arteries from T10 and T11 of the right and T10 of the
183
c
arrows) can be discriminated by the position of the top of the hairpin
curve and the position of the descending branch above the mass. At
tumour level both arteries are considerably displaced to the left border.
(b) Late phase with tumour blush and draining veins (open arrow).
Drainage is in time. (c) Lateral view. Better discrimination of position and
calibre of anterior and posterior spinal arteries (large and small arrow)
d e
T9
T 10 T 10
Late phase
left side contribute to the tumour blush (a, b, d) as well as the anterior
spinal artery from T9 on the left side (c). The strongest and largest dif-
fuse opacification of the mass is observed via the segmental feeder T10
on the left. The shunt volume is lower compared to AVMs (e)
184 4 Clinical Applications

A. p. view A. p. view

T3

T6

a b c

Fig. 4.44 Multiple haemangioblastomas in von Hippel–Lindau syn- vascular tumour nodules are distinctly visualized. (c) T1w ce-MRI in a
drome. (a, b) Digital subtraction angiography at the levels T6 on the midsagittal section shows the scattered bright tumour formations
right and T3 on the left side. Together with the opacification of either accompanied by extensive dark perinodular cystic formations
posterior (a) or anterior (b) spinal cord–supplying arteries, the numerous
4.4 Applications in Vertebrospinal and Medullary Vascular Tumours
A. p. view A. p. view
C5
a b
Fig. 4.45 Cervical haemangioblastoma demonstrated by digital subtrac-
tion angiography in early and late arterial phase. Differential diagnosis of
a cervical arteriovenous malformation (AVM). (a, c) Intensive opacifica-
tion of the vascular tumour supplied by an anterior feeder from the verte-
bral artery (black open arrow) appears very early. (b, d) In late arterial
phase, an early draining anteromedian vein in the lateral view indicates a
185
Lat. view Lat. view
c d
rather high shunt volume (d, white arrowheads) In a.p. view a short time
later (b) advanced outflow of the anterior median vein (white arrow-
heads) to the internal vertebral venous plexus (white open arrows) has
occurred. Differential diagnosis of glomerular types of AVMs on the
basis of angiographic criteria alone can only refer to the diffuse, more or
less homogeneous “tumour-like” opacification (cf. Fig. 4.15)
186
4.5 Potential Pitfalls in Spinal
Angiographic Examinations
and Evaluations
Standards of image acquisition in spinal DSA and potential
pitfalls in the anatomical evaluation of blood vessel anatomy
mainly in spinal arteriovenous disorders should be men-
tioned and critically commented on.
1. Imaging of feeders in two planes. Correct image inter-
pretation requires sufficient information from the pro-
vided images. This sounds self-evident. But in our
experience spinal angiographies are often performed in
only an a.p. view and sometimes even without presenting
an unsubtracted image. If ever possible, one of additional
planes from the diagnostically important series should be
a definitely lateral one. Prior to a therapeutic interven-
tion, but even for diagnostic purposes, it is essential to
identify the feeders of a spinal AVM from both the ante-
rior and posterior circulation. On the basis exclusively of
an a.p. view this can be difficult because of space-occu-
pying effects or scoliosis. Figures 4.15 and 4.42 illustrate
the different configuration of the hairpin curve in ante-
rior and posterior radiculomedullary arteries. This is
additionally helpful. Nevertheless, it is essential to have
a lateral projection for an assured identification. This
also facilitates identification of a spinal cord–supplying
artery originating from the same pedicle (Fig. 4.40).
2. Selective injections with good intra-arterial placement
of the catheter tip. Another technical aspect for not
being able to localize the feeder and shunt site of a sus-
pected SDAVF is missing selectivity of the angiographic
vessel injections. An exemplifying case is the pelvic
dural fistula shown in Fig. 4.37. In case of an SDAVF at
the L5 level neither the global injection of the abdominal
aorta nor the demonstration of the pelvic vessels from
the injected internal iliac artery demonstrated the fistula.
Only selective injection into the iliolumbar artery was
successful. Today, in the majority of cases, ce-MRA is
helpful in localizing at least the region of the AV shunt,
so that selective or superselective injections can be
restricted to this area. But selectivity is needed to iden-
tify the therapeutic target.
3. The diagnostic challenge of “spinal varicosis”. Varicose
convolutions, more frequent on the posterior than on the
anterior surface of the spinal cord, are not unusual
(Figs. 3.31, 3.32, and 3.33). Even extreme varicosities of
radiculomedullary veins like in Fig. 3.32d may be
observed as incidental findings. The term “spinal varico-
sis” was frequently used in the older neuropathology
literature and in the evaluation of myelograms. Even
today, accentuation of plexiform veins on tomography
4 Clinical Applications
images may cause confusion with regard to the differen-
tiation between a normal and a pathological finding. In
cases of doubt, especially if combined with clinical find-
ings that substantiate suspicion of an AV shunt disorder
additional adequate examinations are necessary.
4. Distended vein of the filum terminale or of a sacral
radicular vein (Figs. 4.5, 4.21, 4.22, 4.30, 4.33, and
4.37). As pointed out in Sect. 3.2.1 a sacral radicular
vein may assume the function of the terminal vein,
which may be an impressively strong normal blood ves-
sel (Fig. 4.5). Enlargement of this important drainage at
the lower end of the vertebral canal or increased flow is
noticeable in modern MRI and ce-MRA examinations.
But until now, the underlying disease of a suspected AV
disorder has to be clarified by selective angiography.
Flow can be directed downwards from an AVM on the
myelon or from a dural shunt. Flow can be directed
upwards from an AVM at the filum terminale or from a
dural shunt at lumbosacral levels. Both situations have to
be checked thoroughly. It is important to understand that
the AV shunts of the filum terminale are true pial AVMs
with main but not exclusive supply from spinal cord–
supplying arteries. In contrast, DAVFs at sacral levels
are exclusively supplied by leptomeningeal arteries. The
feeders may be derived from different sources. Therefore,
the contribution of the anterior spinal artery is the essen-
tial question. If the anterior spinal artery participates in
supply, AVM of the filum terminale is the diagnosis and
the concept of treatment is in favour of open surgery.
5. The diagnostic challenge of a pseudo-bimetameric trunk.
Selective spinal angiography is not complete as long as
segmental supply to the spine at each level has not been
achieved. (Figs. 2.5 and 4.1). A pseudo-bimetameric trunk
is not a combination of two segmental arteries, but the
interconnection of one segmental (posterior intercostal)
artery and two neighbouring ventral branches of the poste-
rior intercostal arteries. These ventral intercostal branches
have nothing to do with supply of the spine and spinal
cord. The situation is easily comprehensible after closer
inspection of Fig. 2.2, which has been modified for that
reason from earlier schemes (Thron 2008). Therefore one
has to look for an independent dorsal branch of the poste-
rior intercostal artery originating from the aorta, which is
missing so far. The challenge is that catheterization of this
small segmental branch from the aorta may be difficult.
But otherwise, spinal, radiculomeningeal and radiculom-
edullary arterial branches will be missed in this situation.
6. Demonstration of all potential feeders. For therapy plan-
ning the most appropriate feeder has to be selected for a
safe and efficient treatment.
7. Restricted field of view. If an anterior or posterior spinal
artery seems to be too large for the region to be supplied
References
or if it extends below the level of the cone it should be
followed caudally. This is the only way not to miss the
small AVMs of the filum terminale. A field of view cov-
ering the whole length of the vertebral column is impor-
tant especially if abnormal findings are obvious
(Fig. 4.19).
8. Distinction artery vs vein. On the way from the spinal
cord surface to the dural outlet, the radiculomedullary
vein takes a hairpin course similar to the arterial config-
uration Figs. 3.10, 3.11, and 4.2. Therefore, it can be
mistaken for an artery, particularly when an AVM with
early venous filling is present.
9. The risk of very small radiculomedullary arteries.
Anterior radiculomedullary arteries may be very small
in the upper and mid thoracic region (down to less than
0.2 mm, see Table 1.2), but nevertheless they may be
indispensable for the regional supply (Fig. 4.1b, c).
This makes embolizations in this region more hazard-
ous. Small feeders may be hidden behind the nidus of
an AVM (Fig. 4.16). Tiny feeders may easily be over-
looked or not be demonstrated prior to an intended
embolization. Reasons can be limited spatial and con-
trast resolution of the angiographic equipment or other
factors leading to decreased image quality. High flow
to an AV shunt from the same segmental source, as
shown on Fig. 4.40, might be a cause of haemodynamic
origin. One of our few serious complications from glue
embolization of a fistula in the upper thoracic region in
the early 1990s was most probably caused by this
constellation.
10. Differentiation of SDAVF vs SAVM? (Figs. 4.28 and
4.29) SDAVFs may look very similar to AVMs and have
been misjudged as such for many years. This has
remained a problem especially with large shunts enter-
ing the short common stem of two or more radiculomed-
ullary veins (vein complexes) coming from different
parts of the medullary surface as illustrated and described
in Sect. 3.7. But as long as no typical radiculomedullary
artery is involved, the AV shunt must be located at the
level of the dura mater.
11. Differentiation of SDAVF or SAVM by fistula type?
(Fig. 4.30) Discrimination between a spinal dural AVF
and the fistulous type of a perimedullary AVM may
again be difficult if the recipient arterialized vein looks
like an anterior radiculomedullary artery taking a hair-
pin curve on the ventral surface of the spinal cord as
demonstrated in Fig. 4.30. This event is not rare in
SDAVFs at lower lumbar levels when the arterialized
vein is one of the large lumbar veins shown in Figs. 3.11,
3.33e, and 3.34. The sharp bending course does not dis-
criminate between arteries and veins. Furthermore, the
terminal vein or comparably large veins of the cauda
187
equina are connected to the ventral midline vein of the
lower spinal cord. To avoid misinterpretation, careful
analysis of the vessel anatomy in the region of the inter-
vertebral foramen is important as well as a look at the
further course of the vessel on the spinal cord surface on
later images (Fig. 4.30b). If you are unable to identify an
AV shunt (fistula) in perimedullary veins on the spinal
cord, ask yourself whether the intradural blood vessels
as a whole could not be veins, exhibiting reversed flow
from a dural shunt located at low lumbar or sacral level.
12. Same pedicle for arteriovenous shunt disorder and spi-
nal cord–supplying artery. Spinal cord–supplying arter-
ies and the meningeal feeders of a dural shunt
occasionally arise from the same pedicle as demon-
strated in Fig. 4.40. They are better discriminated on lat-
eral views, supporting the relevance of acquiring two
perpendicular planes. In case of treatment the surgical
approach might be the safer one.
References
Aboulker J, Bar D, Mursault L, Khoudja F, Redondo A, Gard L, Nahum
H (1977) L’hypertension veineuse intra-rachidienne par anomalies
multiples du systeme cave: une cause majeure de souffrance medul-
laire. Chirurgie 103(1003–10):15
Aminoff MJ, Barnard RO, Logue V (1974) The pathophysiology of spi-
nal vascular malformations. J Neurol Sci 23:255–263
Aminoff MJ, Logue V (1974) Clinical features of spinal vascular mal-
formations. Brain 97:197–210
Backes W, Nijenhuis R, Mull M, Thron A, Wilmink J (2004) Contrast-
enhanced MRAngiography of the spinal arteries: current possibili-
ties and limitations. Riv di Neuroradiologia 17:282–292
Burguet JL, Dietemann JL, Wackenheim A, Kehr P, Buchheit F (1985)
Sacral meningeal arteriovenous fistula fed by branches ofthe hypo-
gastric arteries and drained through medullary veins. Neuroradiology
27:232–237
Chen CJ, Huang CC, Hsu YY, Hsu WC (1997) Small isolated paraspi-
nal arteriovenous fistula. AJNR Am J Neuroradiol 18:359–361
Cogen P, Stein BM (1983) Spinal cord arteriovenous malformations with
significant intramedullary components. J Neurosurg 59:471–478
Cognard C, Semaan H, Bakchine S, Miaux Y, Thibault S, Sola Martinez
MT et al (1995) Paraspinal arteriovenous fistula with perimedullary
venous drainage. AJNR Am J Neuroradiol 16:2044–2048
Di Chiro G, Doppman J (1969) Differential angiographic features of
hemangioblastomas and arteriovenous malformations of the spinal
cord. Radiology 93:25–30
Djindjian R, Dumesnil M, Faure C, Tavernier C (1963) Angiome
medullaire dorsal. Etude clinique et arteriographique. Rev Neurol
108:432–434
Djindjian M, Djindjian R, Ray A, Hurth M, Houdart R (1977) Intradural
extra-medullary spinal arteriovenous malformation fed by the ante-
rior spinal artery. Surg Neurol 8:85–94
Djindjian R, Houdart R, Hurt M, Cophignon J, Rey A, Thurel J (1975)
Embolisation dans les angiomes de la moelle. J Neuroradiol 2:73–172
Djindjian R (1978) Angiography in angiomas of the spinal cord. In: Pia
HW, Djindjian R (eds) Spinal angiomas. Springer, Berlin/
Heidelberg/New York, pp 98–136
188
Djindjian R, Merland JJ, Djindjian M, Stoeter P (1981) Angiography of
spinal column and spinal cord tumors. Thieme, Stuttgart/New York
Foix C, Alajouanine T (1926) La myélite nécrotique subaigue. Rev
Neurol 33:1–42
Geibprasert S, Pereira V, Krings T, Jiarakongmun P, Toulgoat F,
Pongpech S, Lasjaunias P (2008) Dural arteriovenous shunts: a new
classification of craniospinal epidural venous anatomical bases and
clinical correlations. Stroke 39:2783–2794
Groen RJ, Groenewegen HJ, van Alphen HA, Hoogland PV (1997)
Morphology of the human internal vertebral venous plexus: A
cadaver study after intravenous Araldite CY 221 injection. Anat Rec
249:285–294
Gueguen B, Merland JJ, Riché MC, Rey A (1987) Vascular malforma-
tions of the spinal cord: intrathecal perimedullary arteriovenous fis-
tulas fed by medullary arteries. Neurology 37:969–979
Hammersen F (1964) Über das feinere Gefäßmuster in der Dura mater
encephali des Menschen. Anat Anz (Erg-H) 113:375–377
Hassler W, Thron A (1994) Flow velocity and pressure measurements
in spinal dural arteriovenous fistulas. Neurosurg Rev 17:29–36
Hassler W, Thron A, Grote EH (1989) Hemodynamics of dural arterio-
venous fistulas. J Neurosurg 70:360–370
Huffmann BC, Gilsbach JM, Thron A (1995) Spinal dural arteriovenous
fistulas: a plea for neurosurgical treatment. Acta Neurochir (Wien)
135:44–51
Hurst RW, Kenyon LC, Lavi E, Raps EC, Marcotte P (1995) Spinal
dural arteriovenous fistula: the pathology of venous hypertensive
myelopathy. Neurology 45:1309–1313
Kaufman HH, Ommaya AK, Chiro G (1970) Compression vs. “steal”,
The pathogenesis of symptoms in arteriovenous malformations of
the spinal cord. Arch Neurol 23:173–178
Kendall BE, Logue V (1977) Spinal epidural angiomatous malforma-
tions draining into intrathecal veins. Neuroradiology 13:181–189
Koenig E, Thron A, Schrader V, Dichgans J (1989) Spinal arteriovenous
malformations and fistulae: clinical, neuroradiological and neuro-
physiological findings. J Neurol 236:260–266
Krings T, Geibprasert S (2009) Spinal dural arteriovenous fistulas.
AJNR Am J Neuroradiol 30:639–648
Krings T (2010) Vascular malformations of the spine and spinal cord.
Anatomy, classification, treatment. Clin Neuroradiol 20:5–24
Krings T, Mull M, Reinges MHT, Thron A (2004) Double spinal dural
arteriovenous fistulas: case report and review of the literature.
Neuroradiology 46:238–242
Krings T, Mull M, Gilsbach JM, Thron A (2005a) Spinal vascular mal-
formations. Eur RadioI 15:267–278
Krings T, Ozanne A, Chng SM, Alvarez H, Rodesch G, Lasjaunias P
(2005b) Neurovascular phenotypes in hereditary hemorrhagic telan-
giectasia patients according to age. Review of 50 consecutive
patients aged 1 day to 60 years. Neuroradiology 47:712–720
Krings T, Chng SM, Ozanne A, Alvarez H, Rodesch G, Lasjaunias P
(2005c) Hereditary hemorrhagic telangiectasia in children: treat-
ment of neurovascular malformations. Neuroradiology 47:946–954
Krings T, Lasjaunias P, Reinges MH, Mull M, Hans FJ, Thron A (2006a)
Spinal vascular malformations. Diagnostic and therapeutic manage-
ment. Clin Neuroradiol 16:217–227
Krings T, Mull M, Bostroem A, Otto J, Hans FJ, Thron A (2006b)
Spinal epidural arteriovenous fistula with perimedullary drainage.
Case report and pathogenetical considerations. J Neurosurg Spine
5:353–358
van der Kuip M, Hoogland PV, Groen RJ (1999) Human radicular
veins: regulation of venous reflux in the absence of valves. Anat Rec
154:173–180
Langer K (1877) Über die Blutgefäße des Knochens des Schädeldaches
und der harten Hirnhaut. Denkschr Akad Wiss Wien Math-Nat Kl
37:217–241
Lazorthes G, Poulhes J, Bastide G, Rouleau J, Chancollé AR (1958) La
vascularisation artérielle de la moelle. Recherches anatomiques et
4 Clinical Applications
applications à la pathologie médullaire et à la pathologie aortique.
Neuro-Chirurgie 4:3–19
Manelfe C, Lazorthes G, Roulleau J (1972) Artères de la dure-mère
rachidienne chez l’homme. Acta Radiol 13:829–841
Merland JJ, Riché MC, Chiras J (1980) Les fistules arterio-veineuses
intra-canalaires, extra-médullaires à drainage veineux médullaire.
J Neuroradiol 7:271–320
Mont’Alverne F, Musaccio M, Talentino V, Belzile F, Riquelme C,
Tornade A (2003) Giant spinal perimedullary fistula in hereditary
hemorrhagic telangiectasia: diagnosis, endovascular treatment and
review of the literature. Neuroradiology 45:830–836
Mull M, Nijenhuis RJ, BackesWH KT, Wilmink JT, Thron A (2007)
Value and limitations of contrast-enhanced MR Angiography in spi-
nal arteriovenous malformations and dural arteriovenous fistulas.
AJNR Am J Neuroradiol 28:1249–1258
Nolte J (1999) The human brain. An introduction to its functional anat-
omy. 4th ed. Mosby, St Louis
Otto J (1990) Morphologie des Sperrmechanismus am Duradurchtritt
der Venae radiculares des Menschen. Neuroradiologische und his-
tologische Befunde. Thesis for doctorate in medicine. University of
Aachen, Aachen
Pirouzmand F, Wallace MC, WiIIinsky R (1997) Spinal epidural arte-
riovenous fistula with intramedullary reflux. Case report.
J Neurosurg 87:633–635
Reinges HT, Thron A, Mull M, Huffmann B, Gilsbach JM (2001) Dural
arteriovenous fistulae at the foramen magnum. J Neurol 248:
945–950
Riché MC, Scialfa G, Gueguen B, Merland JJ (1983) Giant extramedul-
lary arteriovenous fistula supplied by the anterior spinal artery:
treatment by detachable balloons. AJNR Am J Neuroradiol
4:391–394
Rodesch G, Hurth M, Alvarez H, Tadie M, Lasjaunias PL (2002)
Classification of spinal cord arteriovenous shunts: proposal for a
reappraisal – the Bicetre experience with 155 consecutive patients
treated between 1981 and 1999. Neurosurgery 51:374–379
Symon L, Kuyama H, Kendall B (1984) Dural arteriovenous malforma-
tions of the spine. Clinical features and surgical results in 55 cases.
J Neurosurg 60:238–247
Tadié M, Hemet J, Aaron C, Bianco C, Creissard P, Huard P (1979) Le
dispositif protecteur anti-reflux des veines de la moelle. Neuro-Chir
25:28–30
Tadié M, Hemet J, Freger P, Clavier E, Creissard P (1985) Anatomie
morphologique et circulaire des veines de la moelle. J Neuroradiol
12:3–20
Thron A (1988) Vascular anatomy of the spinal cord. Neuroradiological
investigations and clinical syndromes. Springer, Wien/New York
Thron A (2008) Vascular anatomy of the spine and spinal cord. In:
Hurst RW, Rosenwasser RH (eds) Interventional Neuroradiology.
Informa Health Care, New York/London
Thron A, Caplan LR (2003) Vascular malformations and interventional
neuroradiology of the spinal cord. In: Brandt T, Caplan LR, Dichgans
J, Diener HC, Kennard C (eds) Neurological disorders. Course and
treatment, 2nd edn. Academic, Amsterdam, pp 517–528
Thron A, Krings T (2007) Interventionelle therapie spinaler vaskulärer
malformationen. In: Brandt T, Dichgans J, Diener HC (eds) Therapie
und Verlauf neurologischer Erkrankungen, 5th edn. Kohlhammer,
Stuttgart
Thron A, Mull M (2004) Treatment options in spinal nidus-type
AVMs. Embolization with particles. Riv di Neuroradiologia 17:
377–382
Thron A, Koenig E, Pfeiffer P, Rossberg C (1987) Dural vascular anom-
alies of the spine-an important cause of progressive
radiculomyelopathy. In: Cervos-Navarro J, Ferszt R (eds) Stroke
and microcirculation. Raven, New York
Thron A, Mull M, Gilsbach J (2003) Der arteriovenöse Kurzschluss in
der Dura mater des Spinalkanals. Eine behandelbare, wenig
References
bekannte Ursache einer Querschnittslähmung des älteren Patienten.
Dtsch Arzteblatt 100:1132–1137
Thron A, Krings T, Otto J, Mull M, Schroeder JM (2015) The transdural
course of radicular spinal cord veins. A microangiographical and
microscopical study. Clin Neuroradiol 25:361–369
Vogelsang H (1980) Angiographische Untersuchungen von Wirbelsäule,
Spinalkanal und Rückenmark. In: Diethe1m K, Wende S (eds)
Handb Med Radiologie, vol XIV/I: Neuroradiologie. Springer,
Berlin/Heidelberg/New York
189
Voigt K, Hoogland PH, Stoeter P, Djindjian R (1978) Diagnostic value
and limitations of selective spinal angiography in different lesions
of the vertebral bones. Radiol Clin 47:73–90
Vuia O, Alexaniu M (1969) Arteriovenous shunt in the spinal cord cir-
culation. Acta Neurol Scand 45:216–223
Woimant F, Merland JJ, Riché MC, De Liège P, Bertrand HJ, Bacri D,
Haguenau M, Pepin B (1982) Syndrome bulbo-médullaire en rap-
port avec une fistule artério-veineuse meningée du sinus lateral à
drainage veineux médullaire. Rev Neurol 138:559–566
Index

A Blood-deviating system, 134

Anatomical and haemodynamical aspects Blood vessels, anatomical description of, 1
of arterial blood supply
longitudinal trunks, 79–81
spinal cord ischemia and infarct, 81–82 C
of venous drainage Capillaries, 39, 41, 43
circulatory disturbances, 135–136 Capillary telangiectases, 144
intrinsic veins, 133 Caudal continuation, 113
intrinsic venous anastomoses, 134–135 Cavernomas, 144
superficial and radiculomedullary veins, 135 C4–C5 level, microangiogram of, 33
Angiograms, 161 ce-MRA. See contrast-enhanced magnetic resonance angiography
Angiographical appearance, 181 (ce-MRA)
Anterior and posterior radiculomedullary veins, Central (sulcal) arteries, 20, 27, 49, 87
122, 135 Central grey matter, 20, 23
Anterior radiculomedullary arteries, 15, 17 Centro-anterolateral anastomoses, 45, 47
Anterior spinal artery, 9, 11, 17, 19, 20, 30, 31, 50, 64 Centro-dorsolateral anastomoses type 1, 100
Anti-backflow control system, 124 Centro-posterolateral anastomoses, 45, 47
Arcade of cone, 20, 24, 68 Cervical haemangioblastoma, 185
Arterial blood supply Cervical spinal cord, 49–58, 108–114
anatomical and haemodynamical aspects Conjoint and non-functional radicular veins, 123
longitudinal trunks, 79–81 Conjoint radiculomedullary veins, 127
spinal cord ischemia and infarct, 81–82 Contrast-enhanced magnetic resonance angiography (ce-MRA),
arterioarterial anastomoses 138, 171
intrinsic anastomoses, 45 Coronal microangiograms, 40, 54, 61, 114
superficial anastomoses, 45–48 Corona vasorum, 27
cervical spinal cord, 49–58
embryology, 9
intrinsic arteries, 27 D
arterial territories, 42–44 Digital subtraction angiography (DSA), 127, 138–139, 171, 174,
capillaries, 39, 41, 43 183, 184
central arteries, 28–38 Dura mater, 3, 15, 122, 123, 143
peripheral system (vasocorona), 39, 40
radicular and radiculomedullary arteries, 14–19
spine and spinal cord E
extra-and intraspinal extradural anastomoses, 10–13 Ependymomas, 181
segmental spinal branch, 10–13 Extradural AVMs, 144, 145, 181
spinal angiography, 14 Extradural intraspinal and extraspinal venous systems,
superficial arteries, 20–26 131–133
thoracolumbar spinal cord and cauda equina, 68–78
upper-and mid-thoracic spinal cord, 59–67
Arterioarterial anastomoses F
intrinsic anastomoses, 45 Facultative vasoafference regions, 79
superficial anastomoses, 45–48 Fibrocartilaginous embolism, 82
Arteriovenous disorders, 3, 161 Filum terminale, 12, 17, 19, 20, 121, 143, 159, 186
Arteriovenous malformations (AVMs), 81, 143, 144, 160 Fistula type, spinal AVM subtypes, 144, 161
Artery of filum terminale, 20 Foix–Alajouanine disease, 135

B G
Batson’s plexus, 123 Glomus type, spinal AVM subtypes, 144, 152, 153
Blind ending radicular veins, 123 Great radicular artery, 68, 80, 81

© Springer International Publishing 2016 191

A.K. Thron, Vascular Anatomy of the Spinal Cord: Radioanatomy as the Key to Diagnosis and Treatment,
DOI 10.1007/978-3-319-27440-9
192 Index

H Posterior radiculomedullary arteries, 15, 18, 68

Hodgkin’s disease, 58 Posterior radiculomedullary veins, 110, 122, 135
Posterior spinal artery, 9, 18, 20, 186
Posterior venous plexus, 110, 115
I Predominant terminal vein, 116
Inborn spinal vascular malformations, 144 Pseudo-bimetameric supply of T9 and T8, 14
Internal vertebral venous plexus, 123–125, 130–131, 133
Intradural compartment, AVMs of, 144
R
Intraspinal neurinoma, 183
Radial and central veins, 89
Intrinsic anastomoses, 45
Radial (peripheral) veins, 87
Intrinsic arteries, 27–41
Radicular veins, 94, 96, 122, 124, 135
Intrinsic veins, 85–93
Radiculomedullary vein(s)
Intrinsic venous anastomoses, 101, 103, 106, 111, 134–135
complex, 123
dural segment of, 123
and transdural course, 122–130
J
Radio-anatomical images, 115
Juvenile type, spinal AVM subtypes, 144

S
L SAVF. See Spinal arteriovenous fistulae (SAVF)
Large transmedullary midline anastomoses, 142 SAVMs. See Spinal arteriovenous malformations (SAVMs)
L2/3 level, microangiogram of transverse section, 30 SDAVFs. See Spinal dural arteriovenous fistulas (SDAVFs)
Longitudinal venous trunks, 94, 110, 122, 127 Small peripheral veins, 87
Lumbar artery compression, 82 Solitary haemangioblastoma, 183
Lumbosacral spinal cord, 74, 76 Spinal angiographic examinations, pitfalls in, 186–187
Lumbosacral vein patterns, 118 Spinal arteriovenous fistulae (SAVF), 161, 163
Spinal arteriovenous malformations (SAVMs)
applications in
M angiographical findings, 144–160
Marginal branches, 87 definition and classification, 143–144
Meningiomas, 181 complex malformation, 159
Microangiograms with predominant extradural manifestation, 158
cervical spinal cord, 108 Spinal cord parenchyma, 15, 85, 86, 95, 100, 135, 143, 144, 150
of intrinsic and superficial veins, 90, 91 Spinal digital subtraction angiography, 138–143
of large midline anastomoses, 106–107 Spinal dural arteriovenous fistulas (SDAVFs), 123, 165
lumbar level, 71 aetiological and pathophysiological aspects of, 161–162
of median (S2) coronal slices, 55–56 angiographical findings in, 162–163
radicular veins, dural segment of, 129 chronic myelopathies, 163
thoracic level, 71, 77 diagnostic management, 163
upper thoracic spinal cord C8–T3, 92 digital subtraction angiography of, 165
Midline anastomoses type 2, 100 histological sections of, 166
Monocompartmental glomus type, 154 histology, 167
Multiple haemangioblastomas, 184 history and definition, 161
Myelon, 28 internal vertebral venous plexus, 183–184
localization, 161
spinal cord damage, pathogenesis of, 162
N spinal cord–supplying artery and radiculomeningeal supply, 180
Neurinomas, 181 Spinal radicular arteries, 15
Neurological deficits, pathophysiology of, 145, 162 Spinal varicosis, 123, 186
Non-functional radicular veins, 123, 130 Spinal vascular system
anatomical methods and materials, 2
contrast medium and injection technique, 4
O injection techniques, 2–3
Obligatory vasoafference regions, 79 microangiography, 5
preservation, 5
radiological examination techniques
P clinical spinal angiography, 3–4
Paucisegmental type, 15, 19, 79 CT and magnetic resonance angiography, 4
Perimedullary fistula type, AVMs, 144, 147–148 post-mortem macro-and microangiography, 3
Peripheral arteries, 27 X-ray techniques, 4–5
Peripheral system, 39, 42, 47 Sulcal (central) arteries, 20, 27, 49, 87
Pial venous networks of lower degree, 94–99 Superficial anastomoses, 45–48, 134
Pia mater, 20, 87, 94, 135 Superficial arteries, 20–26, 45
Plurisegmental type, 15, 79, 123 Superficial veins
Posterior inferior cerebellar artery (PICA), 4, 9, 51, 149 longitudinal venous trunks, 94–96
Posterior median spinal veins, 87, 91, 94, 135 pial venous networks, 94, 96–99
Index
T Venous drainage
Terminal vein, 94, 143
Territories, arterial supply, 44
Thoracolumbar levels, 13, 15, 20, 75, 121
Thoracolumbar region, 3, 10, 15, 21, 68, 161
Tractus arteriosus medulla spinalis, 20
Transmedullary midline anastomoses, 108, 142
Transmedullary venous anastomoses, 100–107
Transparenchymal anastomoses, 45, 112
T3–T4 level, anterolateral extradural spinal tumour, 67
V longitudinal venous trunks, 94–96
Variable vasoafference region, 79
Varicose convolutions, 94, 98, 114,
115, 119, 123, 186
Vascular malformations, classification schemes for,
1, 143–145, 181
Vascular territories, 2, 34, 43
Vasocorona, 20, 22–27, 39. See also Peripheral system
Vasodeficiency regions, 79
193
anatomical and haemodynamical aspects
circulatory disturbances, 135–136
intrinsic veins, 133
intrinsic venous anastomoses, 134–135
superficial and radiculomedullary veins, 135
cervical spinal cord, 108–114
direction of, 85–86
extradural intraspinal and extraspinal venous systems, 131–133
intrinsic veins, 87–93
radiculomedullary veins and transdural course, 122–130
superficial veins
pial venous networks, 94, 96–99
thoracolumbar and sacral spinal cord, 114–121
transmedullary venous anastomoses, 100–107
Venous midline anastomosis, 141, 142
Ventral fissure, 29, 94
Vertebral haemangioma, 144, 145, 181, 182
Vertebrospinal vascular tumours, 181–185
von Hippel–Lindau syndrome, 181, 184