Original Paper
Eur Neurol 2011;66:277–282
DOI: 10.1159/000332037
Preadmission Therapeutic Anticoagulation
Reduces Cerebral Infarct Volume in Patients with
Nonvalvular Atrial Fibrillation
Mari Matsumoto a, b Shuhei Okazaki a, b Manabu Sakaguchi a, b
Nobuyuki Ohara e Shigetaka Furukado a, b Keiko Nagano f Yasuyuki Kuwagata c
Takeshi Shimazu c Toshiki Yoshimine a, d Kazuo Kitagawa a, b
a
Stroke Center, b Department of Neurology, c Department of Traumatology and Acute Critical Medicine, and
d
Department of Neurosurgery, Osaka University Graduate School of Medicine, e Osaka Neurological Institute, and
f
  Division of Stroke Medicine, Osaka National Hospital, Osaka, Japan
Key Words
Ischemic stroke ⴢ Stroke volume ⴢ Atrial fibrillation ⴢ
Warfarin ⴢ Anticoagulation
Abstract
Aim: We investigated the influence of preadmission antico-
agulation on infarct volume in patients with nonvalvular atri-
al fibrillation (NVAF). Methods: Data were collected on con-
secutive ischemic stroke patients with NVAF admitted to
Osaka University Hospital between 2004 and 2011. Patients
were divided into 3 groups: the no-anticoagulation group,
the subtherapeutic anticoagulation group [admission pro-
thrombin time international normalized ratio (PT-INR) !1.6],
and the therapeutic anticoagulation group (PT-INR 61.6). In
analyses of neurological outcome, we excluded patients
with a modified Rankin Scale (mRS) score of 11 before onset.
Results: Of the 68 patients, 45 were classified into the no-
anticoagulation group, 9 into the subtherapeutic group, and
14 into the therapeutic group. The median value of infarct
volume was 60 (interquartile range 9–176), 142 (64–184), and
8 (3–46) ml in each group, respectively. Infarct volume in the
therapeutic group was significantly smaller than in the sub-
therapeutic group (p = 0.010), and tended to be smaller than
© 2011 S. Karger AG, Basel
0014–3022/11/0665–0277$38.00/0
Fax +41 61 306 12 34
E-Mail karger@karger.ch Accessible online at:
www.karger.com www.karger.com/ene
Received: April 29, 2011
Accepted: August 13, 2011
Published online: October 21, 2011
in the no-anticoagulation group (p = 0.086). National Insti-
tute of Health Stroke Scale score at admission, and mRS score
at discharge were significantly reduced in the therapeutic
group compared with those in the other groups (p = 0.028
and p = 0.017, respectively). Conclusion: Therapeutic antico-
agulation reduces infarct volume and improves neurological
outcome after ischemic stroke in patients with NVAF.
Copyright © 2011 S. Karger AG, Basel
Introduction
Atrial fibrillation occurs in approximately 6% of indi-
viduals aged 165 years [1] and is one of the most common
risk factors for ischemic stroke. Although the benefit of
warfarin therapy in patients with atrial fibrillation is well
established, its potential effect on infarct size and out-
come has not been fully evaluated. Recent studies have
reported that appropriate anticoagulation has a favorable
effect on the prognosis of acute ischemic stroke combined
with atrial fibrillation [2, 3], but the mechanism of this
remains unclear.
Although it is highly likely that a better outcome is
associated with a smaller lesion size, only a few studies
Mari Matsumoto, MD
Department of Neurology
Osaka University Graduate School of Medicine
2-2 Yamadaoka, Suita, Osaka 565-0871 (Japan)
Tel. +81 6 9879 3576, E-Mail m-marine @ medone.med.osaka-u.ac.jp
have evaluated infarct size in patients treated with or
without appropriate anticoagulation [4, 5]. Furthermore,
these studies included patients not only with nonvalvu-
lar atrial fibrillation (NVAF) but also with atherothrom-
bosis [4] and cardiac valvular diseases [4, 5]. As the size
and characteristics of thrombi depend on the level of
blood coagulation and underlying diseases [6], a combi-
nation of the several diseases that have the ability to
cause ischemic stroke would make it difficult to clarify
the mechanisms of the beneficial effects of appropriate
anticoagulation.
In this study, we examined the influence of preadmis-
sion anticoagulation on infarct volume and neurological
outcome in patients with NVAF in order to focus on the
effect of anticoagulation on thrombus formation induced
by NVAF in the left atrium.
Methods
Study Population
The study subjects were recruited from consecutive patients
admitted to the stroke care unit of Osaka University Hospital,
Osaka, Japan, from January 2004 to February 2011. Patients were
included if they had NVAF and had been admitted within 48 h of
the onset of ischemic stroke. Patients were excluded if they had a
history of cardiac diseases other than NVAF, received thrombo-
lytic therapy, diagnosed with lacunar infarct or branch atheroma-
tous disease, or had carotid artery stenosis of 650% according to
the European Carotid Surgery Trial criteria [7].
The patients were divided into 3 groups: the no-anticoagula-
tion group, the subtherapeutic anticoagulation group (admission
prothrombin time international normalized ratio (PT-INR) of
!1.6), and the therapeutic anticoagulation group (PT-INR of
61.6). We selected a cutoff value of PT-INR 1.6 in accordance
with the lowest value of PT-INR recommended by the guideline
of the Japanese Circulation Society for elderly patients with NVAF
to prevent stroke [8].
In the analyses of neurological outcome and the length of hos-
pital stay, we excluded patients whose modified Rankin Scale
(mRS) score was 11 before stroke onset, and compared National
Institutes of Health Stroke Scale (NIHSS) score at admission,
mRS score at discharge, and the length of hospital stay between
the therapeutic anticoagulation group and the others.
Risk Factors
Information pertaining to the patients’ medical history of
cerebrovascular and coronary diseases, current medications, and
smoking habits was obtained from the patients’ clinical records.
The patients were categorized as having cardiovascular disease if
they had a history of coronary heart disease (myocardial infarc-
tion, angina, history of coronary artery bypass surgery, or coro-
nary artery angioplasty) or cerebrovascular disease (stroke or
transient ischemic attack). Hypertension was defined as a casual
blood pressure of 6140/90 mm Hg or the current use of antihy-
pertensive agents. Diabetes mellitus was defined as a fasting blood
glucose level of 67.0 mmol/l, a glycosylated hemoglobin A1c
278 Eur Neurol 2011;66:277–282
(HbA1c) concentration of 65.8%, or the use of glucose-lowering
agents. Dyslipidemia was defined as a fasting total serum choles-
terol level of 65.7 mmol/l, a TG level of 61.7 mmol/l, an HDL
cholesterol level of ^1.1 mmol/l, or the use of cholesterol-lower-
ing agents. Smoking status was evaluated based on self-reports.
Brain Imaging Protocol
Brain computed tomography (CT) examinations were per-
formed on a LightSpeed VCT, a Discovery CT 750HD (General
Electric Medical Systems, Milwaukee, Wisc., USA) or an Aquilion
ONE (Toshiba Medical Systems, Otawara, Japan). The section
thickness was 4 mm in the posterior fossa and 8 mm in superior
brain regions (120 kV, 240 mAS).
CT performed at 2–5 days after stroke onset was used to eval-
uate infarct volume. All CTs were retrospectively evaluated by 2
neurologists (M.M. and S.O.) blinded to the patients’ clinical in-
formation. We used Image J version 1.43 software (National In-
stitute of Mental Health, Bethesda, Md., USA) on a desktop PC at
a window level of 35 HU and a width of 60 HU for the tracing and
calculated the area of the ischemic lesion in each slice. Each region
was traced manually. We multiplied the thickness of each slice
and then used integration to calculate the volume of the infarc-
tion. We previously found that the inter-examiner reliability of
this technique was high (intraclass correlation coefficient = 0.95).
Differences in analysis were resolved by consensus. Hemorrhagic
transformation was diagnosed as a demarcated hyperdense area
on CT together combined with neurological worsening involving
a greater than 2-point increase in the NIHSS score compared with
the admission NIHSS score.
Statistical Analysis
For the analyses of baseline characteristics, categorical vari-
ables were compared using the ␹2 test, continuous variables were
examined using the Kruskal-Wallis test followed by Steel-Dwass
test. Bivariate relations were tested using Spearman’s correlation.
All numeric variables are expressed as the median and interquar-
tile range (IQR). A level of p ! 0.05 was considered statistically
significant. All statistical analyses were performed using the JMP
9.01 software program (SAS Institute Inc., Cary, N.C., USA).
Results
Of the 124 consecutive patients with atrial fibrillation
admitted to the stroke care unit of Osaka University Hos-
pital from January 2004 to February 2011, 56 patients
were excluded as they had a history of post-valve replace-
ment (n = 11), implantation with a left ventricular assist
device (n = 9), nonbacterial thrombotic endocarditis (n =
3), stroke during pacemaker insertion (n = 2), cardiac val-
vular disease (n = 1), or infective endocarditis (n = 1);
were receiving thrombolytic therapy (n = 22); diagnosed
with lacunar infarct or branch atheromatous disease (n =
5), or were suffering from carotid artery stenosis (n = 2).
Consequently, 68 patients were enrolled in this study.
Forty-five patients were classified into the no-anticoagu-
Matsumoto et al.  
 Infarction with hemorrhagic transformation
600 Infarction without hemorrhagic transformation

500 p = 0.086

Lesion volume (ml)

400
p = 0.010
300
Fig. 1. Comparison of lesion volume ac-
cording to admission PT-INR. Each dot 200
depicts the ischemic volume and admis-
100
sion PT-INR of a patient. The black dots
indicate patients who displayed hemor- 0
rhagic transformation. The dashed line 1.0 1.5 2.0 2.5 3.0 WF(–) WF(+) WF(+)
shows PT-INR = 1.6. WF(–) = No pread- PT-INR PT-INR <1.6 PT-INR ≥1.6
mission warfarin use; WF(+) = preadmis- n = 45 n=9 n = 14
sion warfarin use.

Table 1. Baseline characteristics according to the antithrombotic medication status at admission

Characteristics WF(–) WF(+) PT-INR <1.6 WF(+) PT-INR ≥1.6 p value

(n = 45) (n = 9) (n = 14)

PT-INR 1.1 (1.1–1.2) 1.4 (1.3–1.5) 1.9 (1.7–2.0)

Age, years 74 (65–83) 75 (68–78) 73 (70–76) 0.898
Female sex, % 38 22 29 0.534
Plasma glucose, mmol/l 6.5 (5.8–9.7) 7.7 (6.3–8.0) 5.6 (5.3–8.8) 0.386
Systolic blood pressure, mm Hg 168 (147–188) 155 (146–161) 135 (122–169) 0.041
Diastolic blood pressure, mm Hg 93 (79–105) 85 (83–105) 64 (60–89) 0.003
NIHSS score 13 (4–21) 20 (11–23) 12 (2–16) 0.087
Infarct volume, ml 60 (9–176) 142 (64–184) 8 (3–46) 0.015
Risk factors
Hypertension, % 78 89 64 0.374
Diabetes mellitus, % 17 11 36 0.248
Dyslipidemia, % 39 33 57 0.406
Fixed atrial fibrillation, % 51 67 57 0.676
Prior cardiovascular disease, % 11 33 7 0.150
Prior TIA/stroke, % 36 44 71 0.062
Preadmission use of medications
Statins, % 17 22 31 0.537
ACE inhibitors/ARB, % 30 44 36 0.697
Antiplatelets, % 38 56 14 0.108

All numeric variables are expressed as median and interquartile range (IQR). ACE = Angiotensin-converting enzyme; ARB = an-
giotensin II type 1 receptor blocker; WF(–) = no preadmission warfarin use; WF(+) = preadmission warfarin use.

lation group, 9 into the subtherapeutic group (a PT-INR
of !1.6), and 14 into the therapeutic group (a PT-INR of
61.6). The median duration from stroke onset to a blood
examination in the subtherapeutic group (11.2, IQR 4.5–
38 h) was similar to the therapeutic group (9.8, IQR 1.7–
27.3 h).
Table 1 shows the baseline characteristics of the study
cohort. Systolic and diastolic blood pressure were lower
in the therapeutic group. Figure 1 shows the relationship
between an admission PT-INR value and infarct volume.
The median value of infarct volume was 60 (IQR 9–176),
142 (64–184) and 8 (3–46) ml in the no-anticoagulation,
the subtherapeutic and the therapeutic groups, respec-
tively. Infarct volume in the therapeutic group was sig-
nificantly smaller than in the subtherapeutic group (p =
0.010), and tended to be smaller than in the no-anticoag-

Anticoagulation Reduces Infarct Volume Eur Neurol 2011;66:277–282 279

in NVAF
Table 2. Bivariate relations with infarct volume on CT significantly reduced in the therapeutic anticoagulation
group compared with those in the other groups (no and
Covariate CT lesion volume, ml p subtherapeutic anticoagulation group). The length of
value
present absent hospital stay was also significantly shorter in the thera-
peutic group [12 (6–15) vs. 23 (14–31) days, p = 0.007].
Female sex 122 (7–222) 26 (6–122) 0.107
Mortality was not different between the 2 groups (0 vs.
Risk factors
Hypertension 38 (5–142) 109 (8–242) 0.127 9%, p = 0.305).
Diabetes mellitus 26 (8–216) 52 (5–163) 0.828
Dyslipidemia 65 (2–182) 33 (11–142) 0.919
Fixed atrial fibrillation 60 (14–201) 19 (1–131) 0.092 Discussion
Prior coronary artery
disease 43 (9–187) 52 (6–163) 0.878
Prior TIA/stroke 58 (5–194) 37 (10–150) 0.734 Our results show that appropriate anticoagulation
Preadmission use of medications therapy reduces infarct size, improves neurological out-
Statins 52 (1–210) 47 (11–147) 0.599 comes, and shortens the length of hospital stay after the
ACE inhibitors/ARB 20 (1–130) 62 (12–184) 0.124 onset of ischemic stroke in patients with NVAF. The
Antiplatelets 29 (5–136) 53 (8–177) 0.505
smaller infarct size observed in the therapeutic antico-
Admission PT-INR ≥1.6 8 (3–46) 82 (14–179) 0.012
Admission PT-INR r = –0.19 0.324 agulation group in our study is in line with previous stud-
Age r = 0.29 0.243 ies showing that appropriate anticoagulation therapy im-
Plasma glucose r = 0.26 0.020 proved neurological outcomes in patients with NVAF [2,
Systolic blood pressure r = 0.19 0.249 3] and reduced the infarct size in all types of cerebral in-
Diastolic blood pressure r = 0.32 0.034 farction [4, 5].
All numeric variables are expressed as median and interquar- We recruited only patients with NVAF in this study,
tile range (IQR). TIA = Transient ischemic attack; ACE = angio- therefore the smaller infarct size observed was largely as-
tensin-converting enzyme; ARB = angiotensin II type 1 receptor cribed to the smaller thrombi [9] produced in the left atri-
blocker. um in patients with a PT-INR of 61.6. The same cutoff
value, a PT-INR of 1.6, has been shown to be useful for
preventing stroke recurrence [8, 10, 11] and reducing in-
farct size [4], which supports the hypothesis that thera-
ulation group (p = 0.086). The number of patients with peutic anticoagulation inhibits thrombus enlargement in
hemorrhagic transformation was not different between 3 the left atrium even though thrombi occur. In addition,
groups (p = 0.323; fig. 1). When the data were dichoto- thrombi that have embolized in the cerebral artery might
mized using an INR threshold of 2.0, although only 4 be prone to recanalization in patients with a PT-INR of
patients were classified into the therapeutic group, the 61.6.
results were similar; the infarct volume in patients with On the other hand, we found no linear correlation be-
PT-INR 62.0 (5 ml, 1–35 ml) tended to be smaller as tween admission PT-INR value and infarct volume. In
compared with patients who had PT-INR of !2.0 (55 ml, addition, infarct volume of the subtherapeutic group
6–161 ml; p = 0.137), or as compared with patients who tended to be greater than that of the no-anticoagulation
had no anticoagulation therapy (60 ml, 9–176 ml; p = group. Because medication adherence of patients with
0.090). Table 2 shows the associations between admission subtherapeutic anticoagulation is often poor, a transient
characteristics and infarct volume. There was no signifi- hypercoagulable state may emerge in the subtherapeutic
cant correlation between an admission PT-INR value and group from the difference in plasma half-life between
infarct volume, but a plasma glucose level and diastolic warfarin-dependent pro-coagulation factors II and X,
blood pressure showed a significant association with in- and warfarin-dependent anticoagulant proteins C and S
farct volume (table 2). [12]. Consequently, thrombi in the subtherapeutic group
In analyses of the neurological outcome, 12 patients might become even greater than those in the no-antico-
who had a disability (mRS 62) before stroke onset were agulation group, and the thrombi might cause greater
excluded. Subsequently, 56 patients were subjected to brain infarct.
analysis of their functional prognosis. The NIHSS score Our results also show that hemorrhagic transforma-
at admission [3 (2–16) vs. 14 (4–21), p = 0.028] and the tion after ischemic stroke is rare after an optimal PT-INR
mRS score at discharge [1 (0–3) vs. 3 (1–5), p = 0.017] were level has been achieved with warfarin. A smaller infarct

280 Eur Neurol 2011;66:277–282 Matsumoto et al.

 
size and an optimal PT-INR level could be the main rea-
sons why few patients in the therapeutic anticoagulation
group in our study suffered hemorrhagic transformation
[13]. However, intense anticoagulation therapy is known
to increase the risk of intracranial hemorrhage [14].
Therefore, we could not exclude the possibility that large
brain infarcts are prone to hemorrhagic transformation
in patients with an optimal PT-INR level.
Previous studies have reported that high blood pres-
sure [15] and a high plasma glucose level [16] are associ-
ated with poor outcome after ischemic stroke. Studies us-
ing animal models of stroke have also shown that both
hypertension and hyperglycemia are associated with in-
creased infarct volumes [17, 18]. Hyperglycemia decreas-
es reperfusion to the ischemic tissue and aggravates lac-
tate acidosis leading to infarct expansion [17]. Hyperten-
sion also impairs endothelial function and impedes
collateral circulation after occlusion of cerebral major ar-
tery [18]. In our study, blood pressure was lower in the
therapeutic group and plasma glucose and diastolic blood
pressure showed a significant correlation with infarct
volume. These findings are in line with those of previous
studies. However, severe stroke is often accompanied by
a generalized stress reaction that involves the activation
of the hypothalamic-pituitary-adrenal axis, resulting in
increased levels of cortisol and catecholamines [17].
Therefore, high blood pressure and high plasma glucose
level might be partially due to the large infarct volume
itself.
There are several limitations to this study. First, evalu-
ation of vessel occlusion was not investigated in our study.
Knowing the precise site of vessel occlusion would help
us to estimate thrombus size and the timing of recanali-
zation under optimal anticoagulation. Secondly, as this
was a general retrospective study, we could not unify the
timing of CT (CT was performed at 2–5 days after onset),
and edema progression during this period may have in-
fluenced infarct size. Third, the small sample size did not
allow us to analyze the results by using multivariate re-
gressive model. Further studies are needed to clarify the
interaction among admission PT-INR, infarct volume,
and other clinical factors, such as blood pressure or plas-
ma glucose levels. Lastly, because we substituted the ad-
mission PT-INR values for pre-stroke PT-INR values, we
might underestimate the real pre-stroke PT-INR values.
However, the median duration from onset to a blood test
in our study was sufficiently short relative to the half-life
of warfarin. Thus, we believe that this temporal differ-
ence hardly affected our results.
In conclusion, our results show that anticoagulation
therapy not only prevents ischemic stroke but is also ef-
fective in reducing infarct size and relieving neurological
deficits after cerebral embolism. A prospective study in-
volving magnetic resonance or CT angiograms in large
numbers of patients receiving warfarin therapy or the
newly developed anticoagulant dabigatran is necessary to
clarify the mechanism of optimal anticoagulation thera-
py for reducing infarct size in acute ischemic stroke pa-
tients.

References
1 Feinberg WM, Blackshear JL, Laupacis A,
Kronmal R, Hart RG: Prevalence, age distri-
bution, and gender of patients with atrial fi-
brillation. Analysis and implications. Arch
Intern Med 1995;155:469–473.
2 Hylek EM, Go AS, Chang Y, Jensvold NG,
Henault LE, Selby JV, Singer DE: Effect of
intensity of oral anticoagulation on stroke
severity and mortality in atrial fibrillation.
N Engl J Med 2003;349:1019–1026.
3 O’Donnell M, Oczkowski W, Fang J, Kearon
C, Silva J, Bradley C, Guyatt G, Gould L,
D’Uva C, Kapral M, Silver F, Investigators of
the Registry of the Canadian Stroke Net-
work: Preadmission antithrombotic treat-
ment and stroke severity in patients with
atrial fibrillation and acute ischaemic stroke:
an observational study. Lancet Neurol 2006;
5:749–754.
4 Ay H, Arsava EM, Gungor L, Greer D, Sin-
ghal AB, Furie KL, Koroshetz WJ, Sorensen
AG: Admission International Normalized
Ratio and acute infarct volume in ischemic
stroke. Ann Neurol 2008; 64:499–506.
5 Wakita M, Yasaka M, Minematsu K, Yama-
guchi T: Effects of anitcoagulation on infarct
size and clinical outcome in acute cardioem-
bolic stroke. Angiology 2002; 53:551–556.
6 Chung I, Lip GY: Virchow’s triad revisited:
blood constituents. Pathophysiol Haemost
Thromb 2003;33:449–454.
7 European Carotid Surgery Trialists’ Collab-
orative Group: Randomised trial of endar-
terectomy for recently symptomatic carotid
stenosis: final results of the MRC European
Carotid Surgery Trial (ECST). Lancet 1998;
351:1379–1387.
8 JCS Joint Working Group: Guidelines for
pharmacotherapy of atrial fibrillation (JCS
2008). Circ J 2010;74:2479–2500.
9 Puetz V, Dzialowski I, Hill MD, Subramani-
am S, Sylaja PN, Krol A, O’Reilly C, Hudon
ME, Hu WY, Coutts SB, Barber PA, Watson
T, Roy J, Demchuk AM, Calgary CTA Study
Group: Intracranial thrombus extent pre-
dicts clinical outcome, final infarct size and
hemorrhagic transformation in ischemic
stroke: the clot burden score. Int J Stroke
2008;3:230–236.
10 Yamaguchi T: Optimal intensity of warfarin
therapy for secondary prevention of stroke
in patients with nonvalvular atrial fibrilla-
tion: a multicenter, prospective, randomized
trial. Japanese Nonvalvular Atrial Fibrilla-
tion-Embolism Secondary Prevention Co-
operative Study Group. Stroke 2000; 31:817–
821.

Anticoagulation Reduces Infarct Volume Eur Neurol 2011;66:277–282 281

in NVAF
11 Yasaka M, Minematsu K, Yamaguchi T: Op-
timal intensity of international normalized
ratio in warfarin therapy for secondary pre-
vention of stroke in patients with non-valvu-
lar atrial fibrillation. Intern Med 2001; 40:
1183–1188.
12 Harrison L, Johnston M, Massicotte P,
Crowther M, Moffat K, Hirsh J: Comparison
of 5-mg and 10-mg loading doses in initia-
tion of warfarin therapy. Ann Intern Med
1997;126:133–136.
13 Hornig CR, Dorndorf W, Agnoli AL: Hem-
orrhagic cerebral infarction – a prospective
study. Stroke 1986; 17:179–185.
282 Eur Neurol 2011;66:277–282
14 Stroke Prevention in Atrial Fibrillation In-
vestigators: Bleeding during antithrombotic
therapy in patients with atrial fibrillation.
Arch Intern Med 1996;156:409–416.
15 Leonardi-Bee J, Bath PM, Phillips SJ, Sand-
ercock PA, IST Collaborative Group: Blood
pressure and clinical outcomes in the Inter-
national Stroke Trial. Stroke 2002; 33: 1315–
1320.
16 Bruno A, Biller J, Adams HP Jr, Clarke WR,
Woolson RF, Williams LS, Hansen MD:
Acute blood glucose level and outcome from
ischemic stroke. Trial of ORG 10172 in Acute
Stroke Treatment (TOAST) Investigators.
Neurology 1999;52:280–284.
17 Kruyt ND, Biessels GJ, Devries JH, Roos YB:
Hyperglycemia in acute ischemic stroke:
pathophysiology and clinical management.
Nat Rev Neurol 2010;6:145–155.
18 Oyama N, Yagita Y, Sasaki T, Omura-Mat-
suoka E, Terasaki Y, Sugiyama Y, Sakoda S,
Kitagawa K: An angiotensin II type 1 recep-
tor blocker can preserve endothelial func-
tion and attenuate brain ischemic damage in
spontaneously hypertensive rats. J Neurosci
Res 2010;88:2889–2898.
Matsumoto et al.  
Copyright: S. Karger AG, Basel 2011. Reproduced with the permission of S. Karger AG, Basel. Further
reproduction or distribution (electronic or otherwise) is prohibited without permission from the copyright
holder.