e u r o p e a n j o u r n a l o f p a e d i a t r i c n e u r o l o g y 1 7 ( 2 0 1 3 ) 3 0 2 e3 0 7

Official Journal of the European Paediatric Neurology Society

Original article

Differentiation between high and low grade tumours

in paediatric patients by using apparent diffusion
coefficients

Luciana Porto a,*, Alina Jurcoane a, Dirk Schwabe b, Matthias Kieslich c, Elke Hattingen a
a
Neuroradiology Department of the Johann Wolfgang Goethe University, Frankfurt/Main, Germany
b
Paediatric Haematology/Oncology Department of the Johann Wolfgang Goethe University, Frankfurt/Main, Germany
c
Neuropaediatric Department of the Johann Wolfgang Goethe University, Frankfurt/Main, Germany

article info abstract

Article history: Objective: This study was performed to confirm the hypothesis that pre-operative apparent
Received 8 October 2012 diffusion coefficient (ADC) can be used to distinguish between “low grade” and “high grade”
Received in revised form tumours in paediatric patients.
3 December 2012 Material and methods: ADC values were retrospectively evaluated in thirty-six paediatric
Accepted 9 December 2012 brain tumours. Twenty-one children with low grade brain tumours (12 WHO I astrocytomas,
1 giant cell tumour, 1 pilomyxoid astrocytoma, 4 WHO II astrocytomas, 2 craniopharyngiomas
Keywords: and 1 ganglioglioma) and 15 children with high grade brain tumours (6 medulloblastomas, 3
Brain tumours WHO III ependymomas, 1 PNET, 1 malignant rhabdoid tumour, 1 malignant germ cell
Children tumour, 1 WHO III astrocytoma, 1 WHO IV astrocytoma, 1 rhabdomyosarcoma metastasis)
DWI were included in this study. Minimum and mean ADC values were compared between low
ADC grade and high grade tumours and cut-off values were evaluated.
Results: The cut-off values to differentiate low and high grade paediatric brain tumours
were 0.7
103 mm2/s and 1.0
103 mm2/s for minimum ADC and average ADC values
respectively. All but one high grade infratentorial ependymoma showed significantly lower
ADC values than low grade brain tumours in children.
Conclusion: Combining the information obtained from conventional MR imaging with the
ADC values may increase the accuracy of pre-operative differentiation between low grade
and high grade paediatric tumours. Cut-off values can help to discern low from high grade
tumours. However, it has to be considered that there is a substantial overlap between
tumour types previously described in the literature.
ª 2012 European Paediatric Neurology Society. Published by Elsevier Ltd. All rights
reserved.

1. Introduction type is therefore challenging, and paediatric oncologists

and radiologists would welcome an image modality which
Brain tumours in children are rare and include a wide would help to differentiate between low and high grade
range of histologies. To gain experience in a specific tumour tumours.

* Corresponding author. Institut für Neuroradiologie, Klinikum der Johann Wolfgang Goethe-Universität, Schleusenweg 2-16, D-60528
Frankfurt am Main, Germany. Tel.: þ49 69 6301 5462; fax: þ49 69 6301 7176.
E-mail address: luciana.porto@kgu.de (L. Porto).
1090-3798/$ e see front matter ª 2012 European Paediatric Neurology Society. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ejpn.2012.12.002
 e u r o p e a n j o u r n a l o f p a e d i a t r i c n e u r o l o g y 1 7 ( 2 0 1 3 ) 3 0 2 e3 0 7
Diffusion-weighted imaging (DWI) has been reported to
provide information that adds to conventional MRI in brain
tumours.1e3 DWI measurements further provide the oppor-
tunity to evaluate quantitative diffusion values. The most
common value is the apparent diffusion coefficient (ADC)
which has been described to be prognostic for the survival
in patients with brain tumours.4 However, fewer publications
have evaluated diffusion characteristics in the paediatric
population and some have exclusively analysed cerebellar
tumours.5e9 Furthermore, there has been no paediatric pub-
lished data investigating cut-off values between low and high
grade tumours.
Our goal was to analyse the ADC values in two different types
of paediatric tumours e “low grade” and “high grade” e and
try to define a cut-off value between these two groups.
2. Material and methods
2.1. Subjects
We retrospectively queried the radiology reports of brain MR
imaging examinations performed between June 2008 and
June 2012 using the keywords “brain tumour in the paediatric
population”. Inclusion criteria for subsequent analysis were
the final histopathological diagnosis of the tumour, according
to World Health Organization (WHO) 2007,10 the availability of
initial MR images before any treatment, and DWI and ADC
images. Only lesions measuring more than 1 cm in diameter
were included. Assessment was obtained from records.
Every year ca. 35 new paediatric cases of brain tumour are
sent to and undergo treatment at the Institute of Paediatric
Haematology/Oncology, Johann Wolfgang Goethe University,
Frankfurt am Main. However, only 36 children with tumours
had a complete MRI including ADC maps. In the other 3/4 of
the patients, MRI was performed only after treatment and/or
the first examination carried out elsewhere did not include
ADC maps.
Twenty-one children had low grade brain tumours
(12 WHO I astrocytomas, 1 giant cell tumour, 1 pilomyxoid
astrocytoma, 4 WHO II astrocytomas, 2 craniopharyngiomas
and 1 ganglioglioma) and 15 children high grade brain tumours
(6 medulloblastomas, 3 WHO III ependymomas e one infra-
tentorial and two supratentorial, 1 PNET, 1 malignant rhabdoid
tumour, 1 malignant germ cell tumour, 1 WHO III astrocytoma,
1 WHO IV astrocytoma, 1 rhabdomyosarcoma metastasis).
2.2. Imaging studies
MR imaging of the brain was performed on a 3 T whole body
system (Magnetom Verio, Siemens Medical Solutions, Erlan-
gen, Germany) with an 8-channel phased array head coil.
The protocol included axial T2-weighted (w) images, Fluid
attenuated inversion recovery (FLAIR), T2*-w images and T1-w
images before and after administration of the intravenous
contrast agent. Further axial echo planar diffusion-weighted
images (DWI) were performed in all patients (TR/TE 4900/88;
130
130 matrix; 220
220 mm field of view) using b-values of
0, 500, and 1000 s/mm,2 using identical orientation for all axial
images. Diffusion gradients were applied in the z, y and x
303
directions. ADC maps (mm2/sec) were automatically calcu-
lated by the built-in software of the scanner.
2.2.1. Regions of interest (ROIs)
Different ROI settings and ADC analyses were performed as
described below.
1. The core of the tumour was defined based on contrast-
enhanced T1-w images (cystic, necrotic and calcified
areas were excluded) and then manually marked onto ADC
maps (see Figs. 3C and 4C). Evaluation was only performed
on the mean ADC value of the solid enhancing area of the
tumour. In the case of non-enhancing tumours, measure-
ments were based on T2-w images (the homogeneous solid
area was selected).
2. Different ROI settings were placed within the solid part of
the tumour. The lowest value was reported as the “minimal
ADC value” (see Figs. 3D and 4D)
3. As a control, ADC measurements were made of the normal,
contra-lateral, normal-appearing brain tissue.
According to the World Health Organization (WHO),10 histo-
logically diagnosed WHO III and IV tumours are defined as “high
grade” and WHO I and II tumours were defined as “low grade”.
2.3. Statistics
The statistical analysis was done with R Statistics (http://
www.R-project.org/). ADC values (minimum and average)
were compared between the groups (low versus high grade
tumours and WHO I astrocytomas versus medulloblastomas)
by means non-parametric testing for two independent groups
(ManneWhitney U Test). The proposed cut-off values selected
were ADC values lying between the extremes of the two
investigated groups (excluding outliers) in the interval where
the whiskers of the two boxplots did not overlap.
3. Results
The results are summarised below and shown in Figs. 1 and 2.
The ADC values (both ADC Min and ADC Average, p < 0.001)
of low grade tumours in children were higher than those of the
high grade tumours. Between these two groups (low versus
high), there was one overlap in an individual tumour, the
only intratentorial ependymoma in our series (marked with an
arrow in Fig. 1). Excluding this case, the cut-off values of
1.0
103 mm2/s for ADC Average and 0.7
103 mm2/s for ADC
Min could differentiate between low and high grade tumours.
WHO I astrocytomas could be differentiated from medul-
loblastomas by using both ADC Min ( p ¼ 0.001) and ADC
Average ( p < 0.001). For a cut-off value of 1.0
103 mm2/s
(ADC Average) and 0.7
103 mm2/s (ADC Min) there was no
overlap in individual tumour ADC values between WHO I
astrocytomas and medulloblastomas.
In the normal-appearing white matter there were no
significant differences between the ADC values of the patients
with low grade tumours and those with high grade ( p ¼ 0.68).
This also applied when comparing the medulloblastomas
with the WHO I astrocytomas ( p ¼ 0.33).
304 e u r o p e a n j o u r n a l o f p a e d i a t r i c n e u r o l o g y 1 7 ( 2 0 1 3 ) 3 0 2 e3 0 7

Fig. 1 e Low and high grade tumours in paediatric patients. Boxplot depicting ADC minimum and ADC average in low and
high grade tumours in paediatric patients. For low and high grade tumours are shown.

Fig. 2 e Medulloblastomas and WHO I astrocytomas in paediatric patients. Boxplot depicting ADC minimum and ADC
average in medulloblastomas and WHO I astrocytomas in paediatric patients.
 e u r o p e a n j o u r n a l o f p a e d i a t r i c n e u r o l o g y 1 7 ( 2 0 1 3 ) 3 0 2 e3 0 7 305

Fig. 3 e MR images of a child with a malignant rhabdoid tumour. A: T2-weighted image shows a heterogeneous mass with foci
of cysts (white arrow). B: T1-weighted axial image after contrast shows a strong and heterogeneous enhancement. C: ADC-
maps. The mean ADC value was measured only at the enhancing core of the tumour, i.e. cystic areas (white arrow) were not
included. D: ADC-maps. Multiple isolated ROI settings were placed and the lowest value reported as the minimal ADC value.

The image aspect, i.e. the subjective evaluation was
mostly consistent with the ADC values of the tumours. Low
grade tumours were predominantly hyperintense, and some
were isointense to slightly hyperintense compared to the
brain parenchyma on ADC maps. In comparison, high grade
tumours were, in all but one case, hypointense and in some
cases slightly isointense. The exception was the infratentorial
WHO III ependymoma, which was hyperintense on ADC maps
compared with brain parenchyma.
4. Discussion
Diffusion-weighted sequences measure Brownian motion
of water molecules which is notably influenced by the
microstructure of the tissue under investigation. The calcu-
lation of ADC maps not only eliminates the influence of
T2-relaxation times on signal of DW images, but also gives the
opportunity of quantifying the random motion of water
molecules. Diffusion is restricted in dense tumours with
high cellularity (accordingly with small extracellular space)
and high nuclear-to-cytoplasmatic ratio. Corresponding to the
WHO classification of tumours of the CNS (central nervous
system),11 high grade tumours typically show higher cellu-
larity and higher nuclear-to-cytoplasmatic ratio than low
grade tumours. In the literature, restricted diffusion has
been reported in paediatric cerebral high grade tumours.8,12
Furthermore, it has been implied that the measure of ADC
values may permit a better characterisation of tumour
grading.8,12
306 e u r o p e a n j o u r n a l o f p a e d i a t r i c n e u r o l o g y 1 7 ( 2 0 1 3 ) 3 0 2 e3 0 7

Fig. 4 e MR images with a typical presentation of a pilocytic astrocytoma, cystic tumour with a mural nodule. ADC
measurements should only be performed at the solid part of the tumour. A: On T2-weighted image the solid component of
the tumour appears hyperintense (black arrow). B: T1-weighted axial image after contrast shows a homogeneous
enhancement of the solid part of the tumour (white arrow). C: ADC-maps. It is important not to include the cystic component
of the tumour in the area where the mean ADC value will be measured. D: ADC-maps. The same is true when evaluating the
minimal ADC value.

Adult studies have shown a significant overlap between
the measurements of ADC values in high and low grade
tumours.13,14 It is therefore likely that other components
of tumour histology besides tumour cellularity may also play
a significant role, such as the tumour matrix, and fibrous or
gliotic tissues.13 In brief, although tumour cellularity is the
major determinant of ADC values of brain tumours, it is
probably not the only one13 and the histologic variability may
play a significant role.
Previous studies in children suggested that the assessment
of ADC values of enhancing solid tumour is probably a reliable
technique for pre-operative differentiation between cerebellar
tumours in paediatric patients9; Rumboldt et al.9 showed that
low grade posterior fossa tumours, i.e. WHO I astrocytomas
had significantly higher ADC values than medulloblastomas
and ependymomas. Their cut-off values of >1.4
103 mm2/s
for WHO I astrocytomas and <0.9
103 mm2/s for medullo-
blastomas were 100% specific.9 Similar results for untreated
posterior fossa tumours were found by Schneider et al.,6,7
i.e. ADC values were significant lower in medulloblastomas
compared to WHO I astrocytomas and ependymomas.
Our results corroborate these previous findings and a clear
distinction between WHO I astrocytomas and medulloblas-
tomas was possible. The cut-off value of ADC Average was
1.0
103 mm2/s; this result was a 100% specific with no overlap.
It should be noted that ependymomas seem to be the
“problematic” tumour, as distinguishing classic (WHO grade
II) and anaplastic ependymomas (WHO grade III) is difficult for
pathologists. Jaremko et al.5 found an overlap between the
ADC of medulloblastomas and ependymomas. Because of the
 e u r o p e a n j o u r n a l o f p a e d i a t r i c n e u r o l o g y 1 7 ( 2 0 1 3 ) 3 0 2 e3 0 7
wide range of histologic ependymomas, diffusion character-
istics also have an overlapping with other tumour types.5
Accordingly, the only infratentorial ependymoma WHO III in
our series was the clear exception with high ADC values
(see Fig. 1). In fact, the ependymoma’s cellularity in posterior
fossa is usually between that of WHO I astrocytomas and
medulloblastomas, but several variants exist, including
a hypercellular tumour.
It is a fact that different studies5,7,9 have showed that ADC
measurement can be highly useful in the diagnosis of paedi-
atric posterior fossa tumours, correctly distinguishing WHO I
astrocytomas from medulloblastomas and ependymomas
when combined with simple structural features (such as
tumour location and morphology). However, one must be
aware that a true overlap between diffusion characteristics of
WHO I astrocytomas, ependymoma, and medulloblastoma
does exist.5 Technical difficulties, such as small tumour
size, tumour bleeding or calcifications can lead to the misin-
terpretation of ADC values. For example, ADC values can be
increased within the necrotic area of the tumour; on the other
hand it may appear reduced in a calcified part of the tumour.
Because of these problems, ADC measurements should be
restricted to the enhancing solid area of the tumour (see Figs. 3
and 4) without clear bleeding, calcification or necroses. In case
of a non-enhancing, low grade tumour the area was defined
on T2 weighted images.
The relatively small number of patients that could
be included in this study was an important limitation. Only
around 25% of the children with brain tumours who are
treated in our institution could be included. A second limita-
tion was the small number of less common tumours. A new
prospective study with a bigger number of patients will be
important to determine the role of the ADC in paediatric
patients with brain tumours; infratentorial ependymomas
particularly deserve further attention.
Our results generally showed a negative correlation
between the tumour grade and ADC values in paediatric brain
tumours. We demonstrated that ADC is helpful in character-
ising and splitting paediatric brain tumours into low and high
grades. However, one should be aware that exceptions, such
as the infratentorial ependymomas, do exist.
Our goal was not to determine an absolute value or specific
ADC cut-off for the different brain tumours in children, but to
estimate a rough value to be able to determine low and
high grade paediatric brain tumours. The known overlap5,9
between tumour types limits the prediction of the tumour
grade in individual cases. This overlap is not only due to
technical difficulties in ADC measurement, but also to the
intrinsic discrepancy in tumour pathology, especially in the
case of ependymomas.
5. Conclusion
The combination of routine image interpretation, with T2
imaging evaluation, and ADC measurements has a high
predictive value in paediatric brain tumours. Furthermore,
DWIs and ADCs can provide information useful for the diag-
nosis of brain tumours in children that cannot be obtained
with conventional MR imaging.
307
The proposed respective cut-off values of 0.7
103 mm2/s
and 1.0
103 mm2/s for ADC Minimum and ADC Average
help to differentiate between low and high grade paediatric
brain tumours. However, while these values can help to
determine low and high grade tumours, they should never be
used alone due to the considerable overlap between tumour
types previously described in the literature.
Disclosure statement
No competing financial interests exist.
references
1. Arvinda HR, Kesavadas C, Sarma PS, et al. Glioma grading:
sensitivity, specificity, positive and negative predictive values
of diffusion and perfusion imaging. J Neurooncol
2009;94:87e96.
2. Hygino da Cruz Jr LC, Vieira IG, Domingues RC. Diffusion MR
imaging: an important tool in the assessment of brain
tumors. Neuroimaging Clin N Am 2011;21:27e49.
3. Jolapara M, Patro SN, Kesavadas C, Saini J, et al. Can diffusion
tensor metrics help in preoperative grading of diffusely
infiltrating astrocytomas? A retrospective study of 36 cases.
Neuroradiology 2011;53:63e8.
4. Pope WB, Lai A, Mehta R, et al. Apparent diffusion coefficient
histogram analysis stratifies progression-free survival in
newly diagnosed bevacizumab-treated glioblastoma. AJNR
Am J Neuroradiol 2011;32:882e9.
5. Jaremko JL, Jans LB, Coleman LT, et al. Value and limitations
of diffusion-weighted imaging in grading and diagnosis of
paediatric posterior fossa tumours. AJNR Am J Neuroradiol
2010;31:1613e6.
6. Schneider JF, Viola A, Confort-Gouny S, et al. Infratentorial
paediatric brain tumours: the value of new imaging
modalities. J Neuroradiol 2007;34:49e58.
7. Schneider JF, Confort-Gouny S, Viola A, et al.
Multiparametric differentiation of posterior fossa tumours in
children using diffusion-weighted imaging and short
echo-time 1H-MR spectroscopy. Magn Reson Imaging
2007;26:1390e8.
8. Kan P, Liu JK, Hedlund G, et al. The role of diffusion-weighted
magnetic resonance imaging in paediatric brain tumours.
Childs Nerv Syst 2006;22:1435e9.
9. Rumboldt Z, Camacho DL, Lake D, et al. Apparent diffusion
coefficients for differentiation of cerebellar tumours in
children. AJNR Am J Neuroradiol 2006;27:1362e9.
10. Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, editors. WHO
classification of tumours of the central nervous system. Lyon:
IARC; 2007.
11. Thurnher MM. 2007 World Health Organization classification
of tumours of the central nervous system. Cancer Imaging
2009;2:9.
12. Gauvain KM, McKinstry RC, Mukherjee P, et al. Evaluating
paediatric brain tumour cellularity with diffusion-tensor
imaging. AJR Am J Roentgenol 2001;177:449e54.
13. Kono K, Inoue Y, Nakayama K, et al. The role of diffusion-
weighted imaging in patients with brain tumours. AJNR Am J
Neuroradiol 2001;22:1081e8.
14. Provenzale JM, Mukundan S, Barboriak DP. Diffusion-
weighted and perfusion MR imaging for brain tumour
characterization and assessment of treatment response.
Radiology 2006;239:632e49.