Clinical Neurology and

Neurosurgery
Manuscript Draft

Manuscript Number: CNN-D-17-1147R1

Title: Preoperative Grading of Glioma Using Dynamic Susceptibility

Contrast MRI: Relative Cerebral Blood Volume Analysis of Intra-tumoural
and Peri-tumoural Tissue

Article Type: Full Length Article

Keywords: Glioma grading; DSC-MRI; rCBV; peri-tumoral tissue

Abstract: Objectives: To assess the usefulness of intra- tumoral and

peri-tumoral relative cerebral blood volume (rCBV) in preoperative glioma
grading.

Patients and Methods: 21 patients with histopathologically confirmed

glioma were included. Imaging was achieved on a 1.5T MRI scanner. Dynamic
susceptibility contrast (DSC) MRI was performed using T2* weighted
gradient echo-planner imaging (EPI). Multiple regions of interest (ROIs)
have been drawn in the hotspots regions, the highest ROI has been
selected to represent the rCBV of each intra-tumoral and peri-
tumoral regions. Based on histopathology, tumours were subdivided into
low grade and high grade. Receiver operating characteristic analysis
(ROC) of rCBV, of both intra-tumoral and peri-tumoral regions, was
performed to find cut-off values between high and low-grade tumors. The
resulting sensitivity, specificity, positive predictive value, negative
predictive value, and accuracy were calculated.

Results: Based on the histopathology, high-grade glioma (HGG) represented

76.2% whereas low-grade glioma (LGG) represented 23.8%. Both intra-
tumoral and peri-tumoral rCBV of HGG were significantly higher than those
of LGG. A cut-off value >2.9 for intra-tumoral rCBV provided
sensitivity, specificity, and accuracy of 80%, 100%, and 85.7%
respectively to differentiate between HGG and LGG. Additionally, the cut-
off value >0.7 for peri-tumoral rCBV provided sensitivity, specificity,
and accuracy of 100%, 66.6%, and 90.5% respectively to differentiate
between HGG and LGG.

Conclusion: rCBV of each of intra-tumoral and peri-tumoral rCBV are

significantly reliable for the preoperative distinction between HGG and
LGG. Combined intra-tumoral and peri-tumoral rCBV provides overall better
diagnostic accuracy and helps to decrease the invasive intervention for
non-surgical candidates.
Title Page

Preoperative Grading of Glioma Using Dynamic Susceptibility Contrast MRI: Relative

Cerebral Blood Volume Analysis of Intra-tumoural and Peri-tumoural Tissue

Radwa K. Solimana,
Sara A. Gamala,
Abdel-Hakeem A. Essab,
Mostafa H. Othmana

a
Diagnostic radiology department, Faculty of Medicine, Asyut University, Assiut, Egypt,
b
Neurosurgery department, Faculty of Medicine, Assiut University, Asyut, Egypt,

*Corresponding Author
Dr. Radwa Kamel A. Soliman
Diagnostic radiology department, Faculty of Medicine, Asyut University Hospital, Asyut/ Egypt
P.O: 71515
Email: Radwa.soliman@yahoo.co.uk

Dicslosure : The authors declare that they have no conflicts of interest concerning this article.

Funding : This research did not receive any specific grant from funding agencies in the public, commercial, or
not-for-profit sectors.

Ethics: This work was approved by ethical committee of Assiut University Hospitals.
*Revision Note

Cover Letter (Revision Notes)

Date: 24 / 12/ 2017
Dear editor in chief,
Clinical Neurology and Neurosurgery Journal,

I am writing to submit the revised manuscript of our full article “Preoperative Grading of Glioma Using Dynamic Susceptibility Contrast
MRI: Relative Cerebral Blood Volume Analysis of Intra-tumoural and Peri-tumoural Tissue” for re-consideration for publication in the
Journal of Clinical Neurology and Neurosurgery, after revision.

Comment Author’s Response

Reviewer #1

1. A great idea to be done in Egyptian patients with brain We thank the reviewer for this comment.
Gliomas to try to make the grading by radiology
instead of the ordinary pathology grading

2. The number of the patients is small Unfortunately, this is one of our limitation. It was
mentioned at the end of the discussion. We are aiming
to increase the population number in our future work
3. There is no need to put the number and sex of the This statement has been modified. It has been
patients in the start of patients and methods and to removed from patients and methods section and only
repeat them again in the results. It is better to put them left in results.
in the results.
4. -Regarding the histopathology, as it is the gold- -Reference for the classification of the brain tumor to
standard to diagnose the brain tumors with their grades. low and high grade have been addressed. (patients and
I can not find in the patients and methods the reference methods, section 2.5).
for your classification of the brain tumor to low and
high grade.
-Number of specimens received either from sterotactic
-Did you make the pathology result from sterotactic biopsy or after open surgical operation, has been
specimen or after open surgical operation? added in the patients and methods section (2.5).

-The sharing of the pathology department has been

-Where is the sharing of the pathology department in added in the patients and methods section (2.5) and in
this manuscript? the acknowledgments.

-The types of gliomas that present in the pathology,

-What are the types of gliomas that present in the has been added in the in the results section (3.2) and
pathology? also were shown in (Table1)

-(Table1) has been added for the pathological types of

-There is no table for the types of different gliomas that different gliomas
present in the manuscript

5. There is no clear comparison between the different Comparison between the different types of brain
types of brain gliomas and the radiology methods. As gliomas and the radiology method (DSC-MRI values),
your results showed specificity and sensitivity to the has been added in the results section (3.3) and was
 radiological method of grading it had to be compared also added in (Table1)
with gold-standard histopathology results with grading

6. Regarding spelling and grammar, after the arrow is the We thank the reviewer, so much, for this meticulous
correction. correction.
Everything has been corrected and highlighted in
yellow in the manuscript
 of intra-tumour and peritumuoral--> intra-tumor  All” tumour “have been corrected to “tumor”in
and peritumoral the manuscript
 histopathological confirmed glioma-->  It has been corrected.
histopathologically
 rCBV of each intra-tumoural and peritumoural  It has been corrected.
regions. Based on histopathology, tumours -->
peritumoral regions. Based on histopathology,
tumors
 between high and low-grade tumours.-->  It has been corrected.
tumors  It has been corrected.
 Both intratumoural--> low-grade glioma......
intramural  It has been corrected.
 A cut-off value >2.9 for intratumoural-->
intratumoral  It has been corrected.
 peri-tumoural rCBV provided sensitivity,
specificity,and -->specificity, and  It has been corrected.
 common primary brain tumour which show
annual increase [1]. --> tumor which shows  It has been corrected.
 information about the tumour, including-->
about a tumour,  It has been corrected.
 characterization, tumour location--> tumor
location,  It has been corrected.
 about the blood brain barrier destruction. -->
blood-brain  It has been corrected.
 it is of a limited diagnostic accuracy,--> it is a
limited diagnostic accuracy  It has been corrected.
 the morphological characters of low and high
grade gliomas [2]. --> high-grade  It has been corrected.
 aspects of the tumour would-->aspects of a
tumour
 results in cellular hypoxia and hypoglycaemia  It has been corrected.
that is in turn stimulate--> hypoglycemia that is,
in turn, stimulate
 the peri-tumoural tissue is a feature of highly  It has been corrected.
malignant tumour [12], --> of a highly
malignant tumor  It has been corrected.
 Only few studies have been investigated -->
Only a few  It has been corrected.
 intra-tumoural rCBV in distinguishing low and
high grade gliomas--> high-grade gliomas.  It has been corrected.
 Then, to further investigate the value of
peritumoural rCBV --> peritumoral  It has been corrected.
 of both intar-tumoral and --> intra-tumoral
 It has been corrected.
 examined in supine position with standard
circularly --> the supine position with the
standard
 It has been corrected.
 axial FALIR (6000/120 ms) TR/TE spin-->
FLAIR
 It has been corrected.
 After intravenous administration of
Gadolinium- DTPA, contrast enhanced-->
contrast-enhanced
 It has been corrected.
 transferred to Phillips' workstation to be
analysed using --> analyzed
 It has been corrected.
 Spatial and temporal smoothing were applied,--
> smoothing was applied
 It has been corrected.
 reduce the effect of motion artefact. Gamma-->
artifact.  It has been corrected.
 adjacent to the tumour border, --> the tumor
border,  It has been corrected.
 most hyperaemic areas as shown on the colour
maps.--> color  It has been corrected.
 not to include an intra-tumoural vessel or areas
of haemorrhage,--> of hemorrhage,  It has been corrected.
 Similarly, ROI were also drawn--> ROI was
also  It has been corrected.
 the intra-tumour or in the peri-tumoural region--
> intra-tumor or in the peri-tumoral  It has been corrected.
 were classified according WHO classification --
> according to WHO  It has been corrected.
 significantly differentiate LGG and HGG, with
p value of 0.001. --> the p-value  It has been corrected.
 rCBVt at cut-off value higher than 2.9 --> at a
cut-off  It has been corrected.
 few were miss classified, most of--> miss-
classified  It has been corrected.
 low and high grade gliomas, indicating--> high-
grade  It has been corrected.
 differentiate LGG and HGG, with p value of
0.002 --> the p-value  It has been corrected.
 Therefore, the peritumoral CBV can be
beneficial diagnostic marker--> the beneficial  It has been corrected.
diagnostic
 cut-off value > 0.7 best differentiated low from  It has been corrected.
high grade gliomas with --> best-differentiated
low from high-grade gliomas  It has been corrected.
 identified as high grade ones. These--> high-
 grade  It has been corrected.
 On the other hand, Bulakbasi et al, reported-->
Bulakbasi et al reported  It has been corrected.
 In other words, on combination of --> on a
combination
 throughout the literatures. The heterogeneous
nature of the tumours--> throughout the kinds  It has been corrected.
of literature. The heterogeneous nature of the
tumors
 measurement, using the hot spot method-->
hotspot  It has been corrected.
 Apart from the tumour heterogeneity --> the
tumor heterogeneity  It has been corrected.
 post processing and, subsequently--> post-
processing  It has been corrected.
 MRI protocol of brain tumours. In this regard,
the American society of Functional--> tumors
..........Society  It has been corrected.
 the technique and post processing would be
more achievable.--> post-processing
 In this study, we applied a preload dose for the -  It has been corrected.
-> applied a preloading dose
 linear fit and leakage modelling [25,33]. On the  It has been corrected.
post processing, ---> post-
processing..........modeling.......post-processing,  It has been corrected.
 In addition, there were no definite match---->
there was
 Conclusion: Our results confirm that intra-  It has been corrected.
tumoural rCBVt, is significantly correlate to-->
is significantly correlated to
 our results showed that, combination --> that
 It has been corrected.
 combination
 low and high grade gliomas Therefore---> high-  It has been corrected.
grade gliomas,
 we recommend the integration of both intra-  It has been corrected.
tumoural and peritumoural--> peritumoral

Thank you for receiving our revised manuscript and re-considering it for review. We appreciate your time and looking forward to your
response.

Kind regards,

Radwa K. Soliman.
MSc, EDiNR, Doctorate, Radiology
Email: Radwa.Soliman@yahoo.co.uk
*Highlights (for review)

Highlights:
 Intar-tumoral and peri-tumoural rCBV (rCBVt and rCBVp) were measured to evaluate
the glioma grade.

 Each of rCBVt and rCBVp significantly distinguished between low and high-grade
glioma.

 Integration of both rCBVt and rCBVp provided better overall diagnostic accuracy for
the preoperative glioma grading.
*Manuscript
Click here to view linked References

1
2 Abstract:
3
4
5
Objectives: To assess the usefulness of intra-tumor and peri-tumoral relative cerebral blood
6
7 volume (rCBV) in preoperative glioma grading.
8
9
10
11 Patients and methods: 21 patients with histopathologically confirmed glioma were included.
12
13 Imaging was achieved on a 1.5T MRI scanner. Dynamic susceptibility contrast (DSC) MRI
14
15 was performed using T2* weighted gradient echo-planner imaging (EPI). Multiple regions of
16 interest (ROIs) have been drawn in the
17
18 hotspots regions, the highest ROI has been selected to represent the rCBV of each intra-
19
20 tumoral and peri-tumoral regions. Based on histopathology, tumors were subdivided into low
21
22 grade and high grade. Receiver operating characteristic analysis (ROC) of rCBV, of both
23
24 intra-tumoral and peri-tumoral regions, was performed to find cut-off values between high
25
26 and low-grade tumors. The resulting sensitivity, specificity, positive predictive value,
27 negative predictive value, and accuracy were calculated.
28
29
30
31 Results: Based on the histopathology, high-grade glioma (HGG) represented 76.2% whereas
32
33 low-grade glioma (LGG) represented 23.8%. Both intra-tumoral and peri-tumoral rCBV of
34
35 HGG were significantly higher than those of LGG. A cut-off value >2.9 for intra-tumoral
36
37
rCBV provided sensitivity, specificity, and accuracy of 80%, 100%, and 85.7% respectively
38 to differentiate between HGG and LGG. Additionally, the cut-off value >0.7 for peri-tumoral
39
40 rCBV provided sensitivity, specificity, and accuracy of 100%, 66.6%, and 90.5% respectively
41
42 to differentiate between HGG and LGG.
43
44
45
46 Conclusion: rCBV of each of intra-tumoral and peri-tumoral rCBV are significantly reliable
47
48
for the preoperative distinction between HGG and LGG. Combined intra-tumoral and peri-
49 tumoral rCBV provides overall better diagnostic accuracy and helps to decrease the invasive
50
51 intervention for non-surgical candidates.
52
53
54
55 Key words:
56 Glioma grading – DSC-MRI - rCBV – peri-tumoral tissue
57
58
59
60
61
62
63
64
65
 1
2 1-Introduction:
3
4 Gliomas are the most common primary brain tumor which shows annual increase [1].
5
Preoperative grading is essential in order to select the optimal treatment plan and to get the
6
7 best outcome.
8
9
10
11 Conventional MRI provides comprehensive morphological information about a tumor,
12
13 including the soft tissue characterization, tumor location, size and extension as well as the
14
15 associated edema and the mass effect. In addition, post-contrast enhancement provides
16 information about the blood-brain barrier destruction. Despite being comprehensive, it is a
17
18 limited diagnostic accuracy, as there is some degree of overlap between the morphological
19
20 characters of low and high-grade gliomas [2]. Therefore, evaluation of other physiological
21
22 aspects of a tumor would be of a beneficial value in determining the glioma grade. Neo-
23
24 angiogenesis is one important factor is this concern [3]. The rapid progression of the
25
26 malignant cells, which exceed the limits of diffusion of nutrients from the native capillaries,
27 results in cellular hypoxia and hypoglycaemia that is, in turn, stimulate the production of the
28
29 vasoactive endothelial growth factor (VEGF). Such factor promotes the formation of
30
31 abnormal new blood vessels, hence the neo-angiogenesis [4,5].
32
33
34
35 Relative cerebral blood volume (rCBV) measurement can be derived from DSC- MRI
36
37
perfusion. It reflects the vascular proliferation, consequently, the degree of the
38 neovascularization. Previous studies have focused mainly on investigating the intra-tumoral
39
40 rCVB in determining glioma grades [6,7,8,9,10,11].
41
42 The assumption that, cellular infiltration along with neovascularization in the peri-tumoral
43
44 tissue is a feature of highly malignant tumor [12], has been investigated previously in several
45
46 studies to differentiate HGG from metastasis [13,14,15,16,17]. Such a feature can be evident
47
48
in HGG but not in LGG. Hence It can be employed in determining the grade of glioma. Only
49 a few studies have been investigated the peri-tumoral rCBV in differentiation different
50
51 glioma grades [18,19,20,21]
52
53
54
55 Therefore, we aim in this study to, first, assess the usefulness of intra-tumoral rCBV in
56
57 distinguishing low and high-grade gliomas. Then, to further investigate the value of peri-
58
tumoral rCBV in this concern. Finally, to evaluate the impact of integration of both intra-
59
60
61
62
63
64
65
 tumoral and peri-tumoral rCBV on the overall diagnostic accuracy in the preoperative
1
2 grading of glioma.
3
4
5
6
2-Patients and methods:
7
8
9 2.1: Patient population and MRI scan: Patients have been examined using a
10
11 superconducting 1.5 Tesla MRI scanner (Phillips, Achieva, Netherland). They were
12
13 examined in the supine position with standard circularly polarized head coil. Conventional
14
15 MRI and DSC-MRI data were obtained in the same session.
16
17
18 2.2: Conventional MRI protocol: using the following sequences. Axial T1WI (550/8.7 ms)
19
20 TR/TE spin echo, coronal T2WI (5000/96 ms) TR/TE spin echo and axial FLAIR (6000/120
21
22 ms) TR/TE spin echo, 5 mm section thickness, 230 x 230 Field of view (FOV) and 256 x 256
23
24 matrix size. After intravenous administration of Gadolinium- DTPA, contrast-enhanced 3D
25
26 T1WI was obtained.
27
28
29 2.3: DSC-MRI protocol: A preloading dose of 0.05 mmol/kg Gd- DTPA was injected, via
30
an 18-gauge intravenous catheter, to correct T1 leakage effect, 5-10 minutes earlier, before
31
32 the start of the examination. DSC-MRI was then performed using T2* weighted gradient
33
34 EPI. Dynamic scans were acquired with 1.7 seconds intervals after intravenous bolus
35
36 injection of 0.1 mmol/kg Gd- DTPA, at a rate of 5ml/s, followed by a 20-ml saline flush. EPI
37
38 acquisition parameters were as follows: (TR 1800 ms, TE 40 ms, flip angle 90, FOV 24 cm x
39
40 24 cm, slice thickness 5 mm and acquisition matrix of 128x128).
41
42
43 2.4: Post-processing: All Imaging data were transferred to Phillips’ workstation to be
44
45
analyzed using NeuroT2* Perfusion software package. The software package employs a
46 standard algorithm for deconvolution of the tissue signal intensity in order to generate CBV
47
48 maps from DSC- MRI raw data. Spatial and temporal smoothing was applied, whenever
49
50 required, to improve the resolution and to reduce the effect of motion artifact. Gamma variate
51
52 fit was also applied to minimize the recirculation.
53
54
55 Multiple ROIs were drawn in the intra-tumoral and along the peri-tumoral regions,
56
57 immediately adjacent to the tumor border, to measure the CBV values. These ROIs were
58
59 drawn on the hotspot areas, i.e., most hyperaemic areas as shown on the color maps. Only the
60
61
62
63
64
65
 ROI that showed the maximum value for each parameter, has been selected to represent the
1
2 intra-tumoral and peri-tumoral CBV (CBVt and CBVp). Caution was taken, while drawing
3
4 ROIs, not to include an intra-tumoral vessel or areas of hemorrhage, calcification or necrosis.
5
6 Similarly, ROI was also drawn in the contralateral normal brain tissue at the level of the
7
8 lesion. The relative ratios (rCBVt and rCBVp) were, then, calculated by dividing the
9
10 maximum value in either the intra-tumor or in the peri-tumoral region by the value in the
11
12
contralateral normal brain tissue.
13
14
15
16
17 2.5: Histopathological classification: Seventeen patients underwent stereotactic biopsy, and
18 four patients underwent open surgery. Histopathological assessment of the specimens was
19
20 performed in the pathology department in our hospital. Brain tumors were classified
21
22 according to WHO classification into four grades (from I to IV). For statistical purposes, they
23
24 were further classified into LGG (WHO grade I and grade II) and HGG (WHO grade III and
25
26 grade IV).
27
28
29
30
31
32
33 2.6: Statistical analysis: Data analysis was performed using the SPSS statistical software
34
35 Package, version 16. The mean ± standard deviations of each parameter (rCBVt and rCBVp),
36
37 for both high and low-grade gliomas, were calculated.
38
39
40
41 Nonparametric Mann–Whitney U tests were used to evaluate the statistically significance
42
43 differences between HGG and LGG for rCBVt and rCBVp. P-values less than 0.05 were
44
45 considered statistically significant.
46
47
48 ROC analysis was then performed to identify the optimum cut-off value for each parameter
49
50 that best discriminate LGG from HGG. The corresponding sensitivity, specificity, positive
51
52 practice and negative predictive values (PPV and NPV) as well as the accuracy were then
53
54 obtained.
55
56
Multivariate linear regression analysis was further performed to determine the best
57
58 combination of rCBVt and rCBVp to differentiate low-grade from high-grade gliomas.
59
60
61
62
63
64
65
 1
2 3-Results:
3
4
5
Examples of both LGG and HGG are given in (Figure1) and (Figure2) respectively.
6
7
8
9 3.1: Study population:
10
11 Our study population consisted of 21 patients (10 female and 11 male) with brain gliomas.
12
13 DCS-MRI was performed in attempt for preoperative grading of gliomas. Patients' age
14
15 ranged from 20 to 75 years.
16
17
18 3.2: Histopathological diagnosis:
19
20 Based on WHO classification, histopathological evaluation of the specimens revealed the
21
22 following: one patient had a grade I pilocytic astrocytoma, two patients had grade II diffuse
23
24 astrocytoma and two other patients had grade II oligodendroglioma. Whereas, four patients
25
26 had grade III anaplastic astrocytoma and twelve patients had grade IV glioblastoma, as
27 shown in (Table1). Of all the gliomas, grade I and II (LGG) represented 23.8%, while grade
28
29 III and IV (HGG) represented 76.19%, as shown in (Figure 3). Glioblastoma reached the
30
31 highest proportion and represented %57.1% of our population.
32
33
34
35 3.2 Comparison between the pathological types of gliomas and DSC-MRI perfusion
36
37
parameters (both rCBVt and rCBVP):
38 For grade I pilocytic astrocytoma, the rCBVt was 1.2 while the rCBVp was 0.7. Whereas, for
39
40 grade II astrocytomas (diffuse astrocytoma and oligodendroglioma), the mean rCBVt was
41
42 2±0.59, while the mean rCBVp was 0.6±0.19. Regarding grade III anaplastic astrocytoma, the
43
44 mean rCBVt was 3.5±1.09, while the mean rCBVp was 1.1±0.4. Finally, grade IV
45
46 glioblastoma, the mean rCBVt was 4.4±1.8, whereas the mean rCBVp was 1.4±0.6 (also
47
48
shown in Table1).
49
50
51
52
53
54
55 3.4: Comparison of both rCBVt and rCBVP between high and low-grade gliomas:
56
57 The mean rCBVt of LGG was 1.90 ± 0.70 and of HGG was 4.20 ± 1.83 with significant
58
59 difference (P=0.001). Whereas, the mean rCBVp of LGG was 0.68 ± 0.19 and of HGG was
60
61
62
63
64
65
 1.37 ± 0.66 with significant difference (P=0.002) as shown in (Table 2).
1
2
3
4 3.5: Correlation between both rCBVt and rCBVP values and glioma grades.
5
6 Both rCBVt and rCBVP, revealed positive significant correlation, (r=0.631, P=0.002) and
7
8 (r=0.643, P=0.002) respectively, to the glioma grade, as shown in (Table 3).
9
10
11 3.6: The findings of glioma grading using ROC analysis for both rCBVt and rCBVP:
12
13 The results of the ROC curve analysis are presented in (Table 4). Cut-off values for both
14
15 rCBVt and rCBVP, were calculated, giving the sensitivity, specificity, PPV, NPV and
16
17 accuracy for distinguishing high grade gliomas from low grades ones.
18
19 At ROC curve analysis, the area under the curve (AUC) of rCBVt used to distinguish
20
21 between LGG and HGG was 0.944, with a cut-off value > 2.9 to differentiate the two grades
22
23 at a sensitivity of 80% and specificity of 100%, PPV of 100%, NPV of 66.7% and accuracy
24
25 of 85.7%, (Figure 4a). On the other hand, for the rCBVP, a cut-off value > 0.7 and AUC of
26
27 0.939 with sensitivity of 100%, specificity of 66.7%, PPV of 88.2%, NPV of 100% of and
28
29 accuracy of 90.5% best discriminated the high- and low grade gliomas (Figure 4b).
30
31
32 3.7: Multivariate linear regression analysis of the combined parameters:
33
34 Combination of Both rCBVt and rCBVp, at cut-off values of >2.9 and > 0.7 respectively,
35
36 resulted in a further higher sensitivity, specificity and overall accuracy than those of each
37
38 parameter separately as shown in (Table 5).
39
40
41
42
43
44
45 4-Discussion:
46
47
48
49 Preoperative grading is essential to select the optimal treatment plan for glioma.
50
51
Conventional MRI usually provide an excellent soft tissue characterization, yet with a
52 relatively limited accuracy to define the tumor grade. Since DSC MRI, mainly reflects the
53
54 tumor vascularity and neo-angiogenesis, it is expected to be of a great help in preoperative
55
56 glioma grading.
57
58
59
60
61
62
63
64
65
 In the present study, we assessed the usefulness of intra-tumoral rCBV in determination of
1
2 glioma grade. In addition, we exploited the hypothesis that malignant infiltration might
3
4 extend beyond the tumor border in high grade tumors, to distinguish between the low and
5
high-grade gliomas, by investigating the efficacy of rCBV in the surrounding peri-tumoral
6
7 region. Eventually, we further assessed the impact of integration of these parameters together
8
9 on the overall diagnostic accuracy.
10
11
12
13 Our population included 21 patients, age ranges from 20 to 75 years old. As in clinical
14
15 setting, the incidence of HGG (76.2%) was higher compared to that of LGG (23.8%) and
16 HGG was higher in male (59.1%) than in females (42.9%). Glioblastoma was the comments
17
18 among all gliomas and represented 57.1 %. These results are in accordance with the
19
20 estimated global incidence of glioma [1,2,22].
21
22
23
24 In the present study, rCBVt was able to significantly differentiate LGG and HGG, with the p-
25
26 value of 0.001. Additionally, it showed significant positive correlation (r=0.631, P=0.002)
27
28 with the tumor grade. This finding is in agreement with many previous reports
29
30
[6,7,8,9,10,11,23,24,25,26,27,28], emphasising the crucial role of intra-tumoral CBV, being
31 an indicator for the tumor neovascularization, in distinguishing the glioma grade.
32
33
34
35 Our results showed that rCBVt at a cut-off value higher than 2.9 best differentiated between
36
37 LLG and HGG and provided 80% sensitivity,100% specificity and 90% diagnostic accuracy.
38
39 The 100% specificity, reflecting a high true negative rate or, in other words, all LGGs were
40
41 correctly diagnosed. The relatively high sensitivity means that, though few were miss-
42
43 classified, most of the HGGs, were correctly diagnosed. These results are in agreement with
44 previous studies [6,23,24,29,30], where they revealed a relatively high both sensitivity and
45
46
specificity of rCBVt in the distinction between the low and high-grade gliomas, indicating the
47
48 solid reliability of rCBV in this concern.
49
50
51
52 Interestingly, on the peri-tumoral region, our results revealed that rCBVp was able to
53
54 significantly differentiate LGG and HGG, with the p-value of 0.002 and also shows a
55
56 significant positive correlation (r =0.643, P = 0,.002) with the glioma grade. These results are
57
58 in agreement with different reports [18,19,20,21,31], and further support that the malignant
59
60 infiltration, along with neo-angiogenesis, extends to the peri-tumoral tissue in HGG but not
61
62
63
64
65
 evident in LGG. Therefore, the peri-tumoral CBV can be the beneficial diagnostic marker in
1
2 the distinction between LGG and HGG.
3
4
5
6 A rCBVp at cut-off value > 0.7 best-differentiated low from high-grade gliomas with 100%
7
8 sensitivity, even higher than that of rCBVt, further confirm the importance of rCBVp
9
10 application in this concern. Conversely, the corresponding specificity was relatively low
11
12
(66%) indicating low true negative rate and some LGG falsely identified as high-grade ones.
13 These results are comparable to those reported by Sever et al., in which they revealed a 0.675
14
15 cut-off value for rCBVp, to distinguish between grade II and grades III+IV, with high
16
17 sensitivity (96.2%), and relatively low specificity (60%) [19]. On the other hand, Bulakbasi et
18
19 al reported 1.9 cut-off value for rCBVp can distinguish the two grades with sensitivity
20
21 (68.1%) specificity (100%) [18].
22
23
24
25 Nevertheless, our results revealed that the integration between, rCBVt and rCBVp using
26
27 linear regression analysis, can reduce the false grading, as seen in (Table 4), At rCBVt >2.9
28
29 and rCBVp > 0.7, the sensitivity, the specificity, and the diagnostic accuracy were 100%
30
31
32
each. In other words, on a combination of rCBVt and rCBVp at the above-mentioned cut-off
33
34
values, all HGG and LGG cases were correctly classified in our population.
35
36
37 It worth mention that, despite the consensus that rCBV is a powerful tool in determining the
38
39 grade, there was a considerable variation in the cut-off values throughout the kinds of
40
41 literature. The heterogeneous nature of the tumors together with the method of the analysis,
42
43 are partially responsible for these variations. In the present study, we reported a high cut-off
44 value (2.9) for CBVt to differentiate low and high-grade gliomas. This high cut-off value is
45
46 most probably because we relied on the highest CBV, measurement, using the hotspot
47
48 method. This was in accordance with other studies [18,19,28], which reported high cut off-
49
50 values (from 2.9 to 3.9) to discriminate between high and low glioma and also relied on the
51
52 ROIs representing the highest CBV value in their analysis. Some authors, however, reported
53
54 lower cut-off values (1.5 to 2.2) for the intra-tumoral CBV, albeit using the hotspot method
55 analysis [6,7,24]. Apart from the tumor heterogeneity along with the method of analysis, it is
56
57 undoubtedly true, that there are many other factors can contribute to these variations. Of
58
59 those, are the magnetic field strength, the acquisition parameters and the leakage correction
60
61
62
63
64
65
 methods as well as the post-processing tools. Standardization of the technique, method of
1
2 analysis, post-processing and, subsequently, rCBV values are, therefore, required, so that,
3
4 DSC-MRI can be feasibly applied in the routine MRI protocol of brain tumors. In this regard,
5
the American Society of Functional Neuroradiology (ASFNR) has introduced guidelines and
6
7 recommendations to optimize the protocol [32]. Yet, further standardization of the technique
8
9 and post-processing would be more achievable.
10
11
12
13 In this study, we applied a preloading dose for the optimum correction of the T1WI leakage
14
15 effect, as has been suggested previously [33,34,35,36]. Some authors, however, used a small
16 flip angle and or longer TE to minimize the T1WI leakage effect [23,37,38], this, however,
17
18 may reduce the signal to noise ratio. Whereas others mainly relied on the post-processing
19
20 correction e.g. base line subtraction [39,40], or linear fit and leakage modeling [25,33]. On
21
22 the post-processing, we further applied the gamma variate fit to minimize the recirculation
23
24 effect [41,42]. Combination of preload dose and post-processing correction has been
25
26 suggested to obtain even better correction of the leakage effect [35,36,43]. In addition, we
27 applied spatial and temporal smoothing, whenever necessary, to improve the spatial and
28
29 temporal resolution.
30
31
32
33 The small size of our population considers as a limitation of our study. In addition, there was
34
35 no definite match between the peri-tumoral tissue and the histopathological correspondence.
36
37
38
39
40 5-Conclusion:
41
42
43
44 Our results confirm that intra-tumoral rCBVt, is significantly correlated to the degree of
45
46 malignancy and significantly distinguished between LGG and HGG. Additionally,
47
48
assessment of peri-tumoral rCBVp provided an additional significant role in this respect and
49 revealed even higher sensitivity than that of the intra-tumoral rCBVt. Furthermore, our results
50
51 showed that combination of both rCBVt, and rCBVp, provided further higher sensitivity,
52
53 specificity and overall diagnostic accuracy (100%), in distinguishing low and high-grade
54
55 gliomas Therefore, we recommend the integration of both intra-tumoral and peri-tumoral
56
57 rCBV in the preoperative grading of glioma.
58
59
60
61
62
63
64
65
 Acknowledgments
1
2 We would like to express our gratitude to the staff of the Pathology Department, with special
3
4 thanks to Dr. Ahmed A. Sattar Essa, for performing the histopathological assessment. We
5
6 gratefully acknowledge the staff of the Departments of Diagnostic Radiology and
7
8 Neurosurgery, for their support. We further thank our patients who consented to the study.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 1
2
3
4
5
6
References:
7
8
9 1. Schwartzbaum JA, Fisher JL, Aldape KD, Wrensch M. Epidemiology and molecular
10 pathology of glioma. Nat Clin Pract Neurol. 2006;2(9):494-503; quiz 1 p following 16
11 2. Upadhyay N, Waldman AD. Conventional MRI evaluation of gliomas. Br J Radiol. 2011;84
12 Spec No 2:S107-11.
13
3. Brem S, Cotran R, Folkman J. Tumor angiogenesis: a quantitative method for histologic
14
15 grading. J Natl Cancer Inst. 1972;48(2):347-56.
16 4. Plate KH, Breier G, Weich HA, Risau W. Vascular endothelial growth factor is a potential
17 tumour angiogenesis factor in human gliomas in vivo. Nature. 1992;359(6398):845-8.
18 5. Shweiki D, Itin A, Soffer D, Keshet E. Vascular endothelial growth factor induced by hypoxia
19 may mediate hypoxia-initiated angiogenesis. Nature. 1992;359(6398):843-5.
20
6. Shin JH, Lee HK, Kwun BD, Kim J-S, Kang W, Choi CG, et al. Using Relative Cerebral
21
22 Blood Flow and Volume to Evaluate the Histopathologic Grade of Cerebral Gliomas:
23 Preliminary Results. American Journal of Roentgenology. 2002;179(3):783-9.
24 7. Law M, Yang S, Wang H, Babb JS, Johnson G, Cha S, et al. Glioma grading: sensitivity,
25 specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic
26 imaging compared with conventional MR imaging. AJNR Am J Neuroradiol.
27
2003;24(10):1989-98.
28
29 8. Di Costanzo A, Pollice S, Trojsi F, Giannatempo GM, Popolizio T, Canalis L, et al. Role of
30 perfusion-weighted imaging at 3 Tesla in the assessment of malignancy of cerebral gliomas.
31 La radiologia medica. 2008;113(1):134-43.
32 9. Morita N, Wang S, Chawla S, Poptani H, Melhem ER. Dynamic susceptibility contrast
33 perfusion weighted imaging in grading of nonenhancing astrocytomas. J Magn Reson
34
Imaging. 2010;32(4):803-8.
35
36 10. Caulo M, Panara V, Tortora D, Mattei PA, Briganti C, Pravata E, et al. Data-driven grading of
37 brain gliomas: a multiparametric MR imaging study. Radiology. 2014;272(2):494-503.
38 11. Katsaros VC, Nikiforaki K, Manikis G, Marias K, Liouta E, Boskos C, et al., Whole tumor MR
39 Perfusion histogram analysis in the preoperative assessment of patients with gliomas,
40 Hellenic Journal of Radiology, 2016;2; 30-38.
41
12. Claes A, Idema AJ, Wesseling P. Diffuse glioma growth: a guerilla war. Acta
42
43 Neuropathologica. 2007;114(5):443-58.
44 13. Hakyemez B, Erdogan C, Gokalp G, Dusak A, Parlak M. Solitary metastases and high-grade
45 gliomas: radiological differentiation by morphometric analysis and perfusion-weighted MRI.
46 Clin Radiol. 2010;65(1):15-20.
47 14. Blasel S, Jurcoane A, Franz K, Morawe G, Pellikan S, Hattingen E. Elevated peritumoural
48
rCBV values as a mean to differentiate metastases from high-grade gliomas. Acta
49
50 Neurochirurgica. 2010;152(11):1893-9.
51 15. Lemercier P, Maya SP, Patrie JT, Flors L, Leiva-Salinas C. Gradient of Apparent Diffusion
52 Coefficient Values in Peritumoral Edema Helps in Differentiation of Glioblastoma From
53 Solitary Metastatic Lesions. American Journal of Roentgenology. 2014;203(1):163-9.
54 16. Law M, Cha S, Knopp EA, Johnson G, Arnett J, Litt AW. High-grade gliomas and solitary
55
metastases: differentiation by using perfusion and proton spectroscopic MR imaging.
56
57 Radiology 2002;222(3):715–21.
58 17. Chiang IC, Kuo YT, Lu CY, Yeung KW, Lin WC, Sheu FO, et al. Distinction between high-
59 grade gliomas and solitary metastases using peritumoral 3-T magnetic resonance
60 spectroscopy, diffusion, and perfusion imagings. Neuroradiology 2004;46(8):619–27.
61
62
63
64
65
 18. Bulakbasi N, Kocaoglu M, Farzaliyev A, Tayfun C, Ucoz T, Somuncu I. Assessment of
1 diagnostic accuracy of perfusion MR imaging in primary and metastatic solitary malignant
2 brain tumors. AJNR Am J Neuroradiol. 2005;26(9):2187-99.
3 19. Server A, Orheim TED, Graff BA, Josefsen R, Kumar T, Nakstad PH. Diagnostic
4
examination performance by using microvascular leakage, cerebral blood volume, and blood
5
6 flow derived from 3-T dynamic susceptibility-weighted contrast-enhanced perfusion MR
7 imaging in the differentiation of glioblastoma multiforme and brain metastasis.
8 Neuroradiology. 2011;53(5):319-30.
9 20. Svolos P, Tsolaki E, Kapsalaki E, Theodorou K, Fountas K, Fezoulidis I, et al. Investigating
10 brain tumor differentiation with diffusion and perfusion metrics at 3T MRI using pattern
11
recognition techniques. Magnetic Resonance Imaging.31(9):1567-77.
12
13 21. Akbari H, Macyszyn L, Da X, Wolf RL, Bilello M, Verma R, et al. Pattern analysis of dynamic
14 susceptibility contrast-enhanced MR imaging demonstrates peritumoral tissue heterogeneity.
15 Radiology. 2014;273(2):502-10
16 22. Ostrom QT, Gittleman H, Farah P, Ondracek A, Chen Y, Wolinsky Y, et al. CBTRUS
17 statistical report: Primary brain and central nervous system tumors diagnosed in the United
18
States in 2006-2010. Neuro Oncol. 2013;15 Suppl 2:ii1-56
19
20 23. Knopp EA, Cha S, Johnson G, Mazumdar A, Golfinos JG, Zagzag D, et al. Glial neoplasms:
21 dynamic contrast-enhanced T2*-weighted MR imaging. Radiology. 1999;211(3):791-8.
22 24. Hakyemez B, Erdogan C, Ercan I, Ergin N, Uysal S and Atahan S. High-grade and low-
23 grade gliomas: differentiation by using perfusion MR imaging. Clinical Radiology 2005; 60;(4)
24 493–502
25
25. Boxerman JL, Schmainda KM, Weisskoff RM. Relative Cerebral Blood Volume Maps
26
27 Corrected for Contrast Agent Extravasation Significantly Correlate with Glioma Tumor
28 Grade, Whereas Uncorrected Maps Do Not. American Journal of Neuroradiology.
29 2006;27(4):859-67.
30 26. Hilario A, Ramos A, Perez-Nunez A, Salvador E, Millan JM, Lagares A, et al. The added
31 value of apparent diffusion coefficient to cerebral blood volume in the preoperative grading of
32
diffuse gliomas. AJNR Am J Neuroradiol. 2012;33(4):701-7.
33
34 27. Van Cauter S, De Keyzer F, Sima DM, Sava AC, D'Arco F, Veraart J, et al. Integrating
35 diffusion kurtosis imaging, dynamic susceptibility-weighted contrast-enhanced MRI, and
36 short echo time chemical shift imaging for grading gliomas. Neuro Oncol. 2014;16(7):1010-
37 21.
38 28. Saberi M, Faeghi F, Ghanaati H, Miri M, Rostamzadeh A, Khodakarim S & Naleini F.
39
Grading of Glioma Tumors by Analysis of Minimum Apparent Diffusion Coefficient and
40
41 Maximum Relative Cerebral Blood Volume, Caspian Journal of Neurological Sciences,
42 2016;2:(4):42-53.
43 29. Young R, Babb J, Law M, Pollack E, Johnson G. Comparison of region-of-interest analysis
44 with three different histogram analysis methods in the determination of perfusion metrics in
45 patients with brain gliomas. J Magn Reson Imaging. 2007;26(4):1053-63.
46
30. Spampinato MV, Smith JK, Kwock L, Ewend M, Grimme JD, Camacho DL, et al. Cerebral
47
48 blood volume measurements and proton MR spectroscopy in grading of oligodendroglial
49 tumors. AJR Am J Roentgenol. 2007;188(1):204-12.
50 31. Fan GG, Q L Deng, Wu ZH, and Guo QY. Usefulness of diffusion/perfusion-weighted MRI in
51 patients with non-enhancing supratentorial brain gliomas: a valuable tool to predict tumour
52 grading? The British Journal of Radiology, 2006;79(944):652–658
53
32. Welker K, Boxerman J, Kalnin A, Kaufmann T, Shiroishi M, Wintermark M, et al. ASFNR
54
55 recommendations for clinical performance of MR dynamic susceptibility contrast perfusion
56 imaging of the brain. AJNR Am J Neuroradiol. 2015;36(6):E41-51.
57 33. Schmainda KM, Rand SD, Joseph AM, Lund R, Ward BD, Pathak AP, et al. Characterization
58 of a first-pass gradient-echo spin-echo method to predict brain tumor grade and
59 angiogenesis. AJNR Am J Neuroradiol. 2004;25(9):1524-32
60
61
62
63
64
65
 34. Paulson ES, Schmainda KM. Comparison of dynamic susceptibility-weighted contrast-
1 enhanced MR methods: recommendations for measuring relative cerebral blood volume in
2 brain tumors. Radiology. 2008;249(2):601-13.
3 35. Hu LS, Baxter LC, Pinnaduwage DS, Paine TL, Karis JP, Feuerstein BG, et al. Optimized
4
preload leakage-correction methods to improve the diagnostic accuracy of dynamic
5
6 susceptibility-weighted contrast-enhanced perfusion MR imaging in posttreatment gliomas.
7 AJNR Am J Neuroradiol. 2010;31(1):40-8.
8 36. Boxerman, J.L., et al., The Role of Preload and Leakage Correction in Gadolinium-Based
9 Cerebral Blood Volume Estimation Determined by Comparison with MION as a Criterion
10 Standard. AJNR. American journal of neuroradiology, 2012. 33(6):1081-1087.
11
37. Cha, S., et al., Intracranial mass lesions: dynamic contrast-enhanced susceptibility-weighted
12
13 echo-planar perfusion MR imaging. Radiology, 2002. 223(1):11-29.
14 38. Bell, L.C., et al., Optimization of DSC MRI Echo Times for CBV Measurements Using Error
15 Analysis in a Pilot Study of High-Grade Gliomas. American Journal of Neuroradiology, 2017.
16 38(9):1710-1715
17 39. Cha, S., et al., Dynamic contrast-enhanced T2*-weighted MR imaging of tumefactive
18
demyelinating lesions. AJNR Am J Neuroradiol, 2001. 22(6):1109-16.
19
20 40. Wetzel, S.G., et al., Relative cerebral blood volume measurements in intracranial mass
21 lesions: interobserver and intraobserver reproducibility study. Radiology, 2002. 224(3):797-
22 803.
23 41. Rosen BR, Belliveau JW, Vevea JM, et al. Perfusion imaging with NMR contrast agents.
24 Magn Reson Med 1990;14 (2):249–65
25 42. Law, M., et al., Comparison of cerebral blood volume and vascular permeability from
26 dynamic susceptibility contrast-enhanced perfusion MR imaging with glioma grade. AJNR
27
Am J Neuroradiol, 2004. 25(5):746-55.
28
29 43. Lev, M.H., et al., Glial Tumor Grading and Outcome Prediction Using Dynamic Spin-Echo
30 MR Susceptibility Mapping Compared with Conventional Contrast-Enhanced MR:
31 Confounding Effect of Elevated rCBV of Oligodendroglimoas. American Journal of
32 Neuroradiology, 2004. 25(2): 214-221.
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Table(s)

Table 1): Different types of glioma according WHO classification as well as, comparison
between these pathological types of glioma and the values of DSC-MRI perfusion parameters.

Number Histopathological diagnosis of WHO grading DSC-MRI perfusion

the tumor of the tuomrs parameters results

rCBVT rCBVP

Mean ± SD Mean ± SD

1 Piloycytic astrocytoma Grade I 1.2 0.7

2 Diffuse astrocytoma Grade II 2±0.59 0.6±0.19

2 Oligodendroglioma Grade II

4 Anaplastic astrocytoma Grade III 3.5±1.09 1.1±0.4

12 Glioblastoma Grade IV 4.4±1.8 1.4±0.6

(Table 2): Mean ± SD of both rCBVt and rCBVP in low and high-grade glioma

DSC-MRI Histopathology
Mann-Whitney Test
perfusion
Low- grade glioma High-grade glioma P-value

rCBVt:
0.001*
Mean ± SD 1.90±0.70 4.20 ± 1.83

rCBVP:
0.002*
Mean ± SD 0.68 ± 0.19 1.37 ± 0.66

(Table 3): Correlation between both rCBVt and rCBVP and glioma grades:

DSC-MRI perfusion R P. value

 rCBVt 0.631 0.002**

 rCBVP 0.643 0.002**

Table 4): ROC analysis of each of rCBVt and rCBVP for the distinction between low and high-
grade gliomas

DSC-MRI Cut-off Sensitivity Specificity PPV NPV Accuracy AUC

perfusion

rCBVt >2.9 80.00 100.00 100.0 66.7 85.7 0.944

rCBVP >0.7 100.00 66.67 88.2 100.0 90.5 0.939

Table 5): Combination of both rCBVt and rCBVP for the distinction between low and high-
grade gliomas

DSC-MRI perfusion Cut-off Sensitivity Specificity Accuracy

rCBVt and rCBVP >2.9 and >0.7 100.00 100.00 100.00
Figure(s)

1) Example for Low grade glioma

a B C
A A A
a a a

d e
D A
8 A
A a
A
a

Figure 1: a) Axial FLAIR, b) Axial T1WI, c) post contrast axial T1WI, showing a right frontal SOL, eliciting low
signal intensity (SI), to the grey matter, on T1WI, and high SI on FLAIR with no enhancement on post contrast
T1WI. Associated minimal perilesional edema and mild mass effect evident by effacement of the adjacent sulci.
(d & e) DSC-MRI: d) source EPI and e) corresponding colour map, with multiple ROIs were drawn in the intra-
tumoral, peri-tumoral tissue and in the contralateral brain tissue: rCBVt =1.7 and rCBVp =0.6. Pathology: Diffuse
astrocytoma (WHO grade II).

b
b
 2) Example for high grade glioma

Figure 2: a) Axial FLAIR, b) Axial T1WI, c) post contrast Axial T1WI, showing a right frontal SOL, eliciting mainly
low SI on T1WI, and intermediate SI on FLAIR with minimal enhancement on post contrast T1WI. Associated
marked perilesional edema and mass effect evident by compression upon the ipsilateral lateral ventricle with
midline shift. (d & e) DSC-MRI: d) source EPI and e) corresponding colour map, with multiple ROIs were drawn in
the intra-tumoral, peri-tumoral tissue and in the contralateral brain tissue: rCBVt =5 and rCBVp =1.3. Pathology:
Glioblastoma multiform (WHO grade IV).
3) Incidence of each grade:

Figure 3: Incidence of LGG and HGG in our population

4) ROC for both rCBVt and rCBVP

Figure 4: a- ROC for rCBVt b- ROC for rCBVp