Keywords
CT perfusion - brain glioma - permeability surface area - cerebral blood volume -
cerebral blood flow - mean transit time - time to peak - conjoint factor
Introduction
Brain tumors account for 2% of all malignancies, with glioma being the most common
intra-axial primary central nervous system tumor.[1] Clinical profile of patients with glioma may range from being completely asymptomatic
to a comatose state since gliomas are extremely heterogeneous tumors ranging from
benign slow growing astrocytoma to highly aggressive glioblastoma.[2]
Estimation of the grade of a glioma is imperative in deciding the management plan.
Surgical resection is the mainstay of management wherever feasible. Patients with
high-grade glioma are also administered adjuvant chemoradiation therapy. The chemoradiation
therapy regimen varies considerably with the grade of glioma.[3]
[4] The prognosis is considered to be worse as the grade progresses.[5]
The current gold standard for preoperative grading of gliomas is histopathological
evaluation of the tissue sample obtained by stereotactic biopsy. There are many limitations
involved with this procedure like sampling error, inter-observer bias, and availability
of a wide variety of classification systems. Moreover, stereotactic biopsy being an
invasive procedure is associated with complications such as intratumoral hematoma,
neurological deficits, and associated morbidity.[6]
Since gliomas are very infiltrative tumors, it is helpful to get an overall picture
of the tumor characteristics in vivo, which can be achieved by imaging. Two of the
most important factors for determining the malignancy of gliomas are their ability
to infiltrate the brain parenchyma and to recruit or synthesize vascular networks
for their growth, that is, neoangiogenesis. Malignant brain tumors are characterized
by neoangiogenesis, having a higher proportion of immature and highly permeable vessels.[7]
[8]
Quantitative estimates of various perfusion parameters (imaging biomarkers) help in
assessing the degree of neoangiogenesis and therefore help in estimating the tumor
grade. Traditionally, perfusion imaging of brain tumors was performed with magnetic
resonance (MR) imaging.[6] However, computed tomography perfusion (CTP) can serve as a better alternative to
MR perfusion as it provides a linear relationship between density changes and tissue
concentration of a contrast agent, has better spatial resolution, wider availability,
faster scanning times, and low cost compared with MR perfusion. CTP is not affected
by susceptibility artifact generated by hemorrhage or mineral deposits, a limitation
with MR perfusion. CTP also has the advantage of being easily repeatable unlike invasive
procedures.[9]
[10]
Various perfusion parameters can be obtained by using CTP with a single acquisition,
namely cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT),
time to peak (TTP), and permeability surface area product (PS). CBV, CBF, MTT, and
TTP parameters are related to flow dynamics, whereas microvascular permeability is
assessed by measuring PS. A conjoint factor (PS + CBV) represents the surrogate markers
for microvascular permeability and flow dynamics.[9]
[11]
There is a paucity of literature in evaluating the role of PS in grading gliomas with
no unanimous consensus regarding the most reliable parameter. Hence, this study was
undertaken to evaluate the most reliable perfusion parameter in differentiating high-
and low-grade gliomas.
Materials and Methods
Study Setting
This analytical cross-sectional study was conducted in the Department of Radiodiagnosis
after obtaining ethical clearance from our institutional ethical committee. 33 patients
of brain glioma diagnosed on MR imaging from November 1, 2018 to July 31, 2020 after
signing a written consent were included in the study. Postoperative patients, pregnant
or lactating females, and patients with impaired renal function or contrast allergy
were excluded from the study.
The patients included in the study were between 19 and 77 years of age, with the mean
age of presentation being 32.33 years. There was a male preponderance with 23 males
and 10 females.
CT Perfusion Acquisition
The study was performed on 128-slice dual-source CT scanner (Somatom Definition Flash
Siemens, Erlanger, Germany). The patients were made to lie supine with head still.
Dynamic perfusion study was performed, after injection of 40 mL nonionic contrast
medium, Iomeron (Iomeprol 400 mg of iodine/mL) with flow rate of 5.5 mL/s via an 18-gauge
cannula by using a power injector. Scanning was initiated after a 4 seconds delay
from the start of injection and images were acquired for a total duration of 40 seconds,
with total coverage area of 10 cm. CT parameters used to acquire dynamic data were
0.28 seconds gantry rotation time, 80 kVp, 180 mAs. Multiple time points were obtained
with cycle time of 1.5 seconds.
Post-Processing
Reconstruction and post-processing were done to generate CTP maps using inbuilt syngo
volume perfusion CT neuro (VPCT Neuro) software. Reconstructed section thickness was
5 mm with 3 mm increment. Regions of interest (ROI) for the tumor were hand drawn.
In the presence of multiple tumors, ROIs were drawn for all tumors in the scanning
range. ROIs were also drawn in the background brain parenchyma that acted as control.
Perfusion values for the tumor and background brain parenchyma including CBF, CBV,
MTT, TTP, and capillary PS were calculated and displayed as tables and functional
maps of CBF, CBV, MTT, TTP, and PS. Relative values of the parameters were calculated
and compared to the control values. The tumor grade was predicted, based upon these
relative perfusion values.
After tumor biopsy or resection, histopathological grade was formulated, which was
correlated with the CTP grade. Sensitivity, specificity, predictive values, and diagnostic
accuracy of PS were calculated and compared with other CTP parameters.
Statistical Methods
Categorical variables were presented in number and percentage (%) and continuous variables
were presented as mean ± standard deviation and median. Normality of data was tested
by Kolmogorov–Smirnov test. If the normality was rejected, then nonparametric test
was used.
Statistical tests were applied as follows:
-
Quantitative variables were compared using independent t-test/Mann–Whitney U test (when the data sets were not normally distributed) between
the two groups and paired t-test/Wilcoxon signed rank test was used for comparison between lesion and control.
-
Receiver operating characteristic (ROC) curve was used to find out cut-off point of
parameters for predicting high grade. Delong test was used for comparison of area
under the curve between perfusion parameters.
-
Inter-rater kappa agreement was used to find the strength of agreement between perfusion
parameters findings and actual histopathological examination (HPE) findings. A p-value of 0.05 was considered statistically significant.
The data was entered in MS EXCEL spreadsheet and analysis was done using Statistical
Package for Social Sciences version 21.0.
Results and Observations
In our study, the most common clinical presentation was hemiparesis (45%) followed
by seizures (39%) and headache (30.30%). Most of the gliomas (91%) were located in
the supratentorial compartment, frontal lobe being the most common site (18%) followed
by parietal lobe (15.15%) and frontoparietal lobe (15.15%).
On histopathology, out of the 33 patients, 19 had high-grade glioma (grade III and
IV), while 14 patients had low-grade glioma (grade I and II). Glioblastoma (42.42%)
was the most common tumor phenotype followed by diffuse astrocytoma (24.24%) and pilocytic
astrocytoma (12.12%). CT perfusion images of seven cases representing the spectrum
of brain gliomas obtained in our study have been shown in [Figs. 1] to [7].
Fig. 1 (A) A 20-year-old male presented with pilocytic astrocytoma (World Health Organization
grade I). T2-weighted imaging, (B) histopathology, (C) computed tomography perfusion (CTP) maps, (D) and CTP parameters (D). CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum intensity projection;
MTT, mean transit time; PMB, permeability; TTP, time to peak.
Fig. 2 (A) A 46-year-old male presented with diffuse astrocytoma (World Health Organization
grade II). T2-weighted imaging, (B)histopathology, (C) computed tomography perfusion (CTP) maps, (D) and CTP parameters. CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum
intensity projection; MTT, mean transit time; PMB, permeability; TTP, time to peak.
Fig. 3 (A) A 58-year-old female presented with oligodendroglioma (World Health Organization
grade II). T2-weighted imaging, (B) histopathology, (C) computed tomography perfusion (CTP) maps, (D) and CTP parameters. CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum
intensity projection; MTT, mean transit time; PMB, permeability; TTP, time to peak.
Fig. 4 (A) A 42-year-old male presented with anaplastic oligodendroglioma (World Health Organization
grade III). T2-weighted imaging, (B) histopathology, (C) computed tomography perfusion (CTP) maps, (D) and CTP parameters. CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum
intensity projection; MTT, mean transit time; PMB, permeability; TTP, time to peak.
Fig. 5 (A) A 21-year-old male presented with nonenhancing anaplastic astrocytoma (World Health
Organization grade III). T2-weighted imaging, (B) histopathology, (C) computed tomography perfusion (CTP) maps, (D) and CTP parameters. CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum
intensity projection; MTT, mean transit time; PMB, permeability; TTP, time to peak.
Fig. 6 (A) A 60-year-old male presented with enhancing anaplastic astrocytoma (World Health
Organization grade III). T2-weighted imaging, (B) histopathology, (C)computed tomography perfusion (CTP) maps, (D) and CTP parameters. CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum
intensity projection; MTT, mean transit time; PMB, permeability; TTP, time to peak.
Fig. 7 (A) A 45-year-old male presented with glioblastoma (World Health Organization grade
IV). T2-weighted imaging, (B) histopathology, (C) computed tomography perfusion (CTP) maps, (D) and CTP parameters. CBF, cerebral blood flow; CBV, cerebral blood volume; MIP, maximum
intensity projection; MTT, mean transit time; PMB, permeability; TTP, time to peak.
On comparing the five perfusion parameters between the tumor region and background
normal brain parenchyma, all the five parameters had significantly higher values (p-value < 0.05) in the tumor region.
Subsequently, relative values of all the parameters were calculated by dividing the
values from the tumor and control region, respectively. These relative perfusion values
were compared to differentiate high- and low-grade glioma. Relative PS (rPS; with
p-value < 0.0001) and relative CBV (rCBV; with p-value 0.006) were found to be significant in differentiating between high- and low-grade
glioma ([Table 1]).
Table 1
Comparison of relative value of perfusion parameters between high- and low-grade glioma
|
Normalized value of perfusion parameters
|
High (n = 19)
|
Low (n = 14)
|
p-Value
|
Test performed
|
|
Relative PS
|
|
Mean ± SD
|
40.59 ± 27.53
|
4.58 ± 2.18
|
<0.0001
|
Mann–Whitney U test;7
|
|
Median (IQR)
|
35.16(18.405–56.797)
|
4.79(2.807–5.399)
|
|
Range
|
4.68–94.81
|
1.6–9.01
|
|
Relative CBV
|
|
Mean ± SD
|
2.56 ± 1.06
|
1.57 ± 0.77
|
0.006
|
t-test; 2.94
|
|
Median (IQR)
|
2.45(1.991–3.165)
|
1.51(1.092–1.958)
|
|
Range
|
1.08-5.06
|
0.48–3.45
|
|
Relative PS+ CBV
|
|
Mean ± SD
|
43.15 ± 27.53
|
6.15 ± 2.29
|
<0.0001
|
Mann–Whitney U test;4
|
|
Median (IQR)
|
39.09 (21.052–59.938)
|
5.7 (4.663–7.078)
|
|
Range
|
6.78–96.69
|
3.25–11.46
|
|
|
|
Relative CBF
|
|
Mean ± SD
|
1.75 ± 1.08
|
1.47 ± 1.31
|
0.202
|
Mann–Whitney U test;98
|
|
Median (IQR)
|
1.53(1.058–2.133)
|
1.14(0.941–1.6)
|
|
Range
|
0.61–5.47
|
0.32-5.78
|
|
Relative MTT
|
|
Mean ± SD
|
1.52 ± 0.52
|
1.24 ± 0.38
|
0.1
|
t-test; 1.692
|
|
Median (IQR)
|
1.53(1.176–1.849)
|
1.23(0.996–1.496)
|
|
Range
|
0.49–2.65
|
0.54–1.88
|
|
Relative TTP
|
|
Mean ± SD
|
1.1 ± 0.13
|
1.07 ± 0.15
|
0.519
|
t-test; 0.652
|
|
Median (IQR)
|
1.09 (1.018–1.192)
|
1.05 (0.975–1.116)
|
|
Range
|
0.84–1.36
|
0.84–1.42
|
Abbreviations: CBF, cerebral blood flow; CBV, cerebral blood volume; IQR, interquartile
range; MTT, mean transit time; TTP, time to peak; PS, permeability surface area; SD,
standard deviation.
After drawing the ROC curve for rPS and tumor grade, a sensitivity of 97.74% and specificity
of 100% was calculated with a cut-off value of 9.0065. Positive predictive value was
100%, negative predictive value was 93.3% and diagnostic accuracy was 93.94% for differentiating
high- and low-grade glioma for rPS. rCBV with a cut-off value of 2.0392 had 73.68%
sensitivity, 78.57% specificity, 82.4% positive predictive value, 68.7% negative predictive
value, and 75.76% diagnostic accuracy. The diagnostic accuracy of rCBV was lower than
that of rPS ([Table 2]).
Table 2
ROC curve of perfusion parameters for predicting high-grade glioma
|
High HPE grade
|
Relative PS
|
Relative CBV
|
Relative PS+ CBV
|
Relative CBF
|
Relative MTT
|
Relative TTP
|
|
Area under
the ROC
curve (AUC)
|
0.974
|
0.778
|
0.985
|
0.632
|
0.654
|
0.594
|
|
Standard error
|
0.0273
|
0.0836
|
0.0164
|
0.103
|
0.0969
|
0.104
|
|
95% CI
|
0.849–1.000
|
0.600–0.904
|
0.867–1.000
|
0.446–0.792
|
0.469–0.810
|
0.410–0.761
|
|
p-Value
|
<0.0001
|
0.0009
|
<0.0001
|
0.2026
|
0.1116
|
0.3677
|
|
Cut-off
|
>9.0065
|
>2.0392
|
>11.4573
|
>1.1912
|
>1.3239
|
>1.118
|
|
Sensitivity (95% CI)
|
94.74%
(74–99.9%)
|
73.68%
(49–91%)
|
94.74%
(74–99.9%)
|
68.42%
(43–87%)
|
57.89%
(34–80%)
|
47.37%
(24–71%)
|
|
Specificity (95% CI)
|
100%
(77–100%)
|
78.57%
(49–95%)
|
100%
(77–100%)
|
71.43%
(42–92%)
|
71.43%
(42–92%)
|
78.57%
(49–95%)
|
|
PPV
(95% CI)
|
100%
(82–100%)
|
82.4%
(57–96%)
|
100%
(82–100%)
|
76.5%
(50–93%)
|
73.3%
(45–92%)
|
75%
(43–95%)
|
|
NPV
(95% CI)
|
93.3%
(68–99.8%)
|
68.7%
(41–89%)
|
93.3%
(68–99.8%)
|
62.5%
(35–85%)
|
55.6%
(31–79%)
|
52.4%
(30–74%)
|
|
Diagnostic accuracy
|
93.94%
|
75.76%
|
96.97%
|
66.67%
|
63.64%
|
60.61%
|
Abbreviations: AUC, area under the curve; CBV, cerebral blood volume; CBF, cerebral
blood flow; CI, confidence interval; HPE, histopathological examination; MTT, mean
transit time; NPV, negative predictive value; PPV, positive predictive value; PS,
permeability surface area; ROC, receiver operating characteristic; TTP, time to peak.
Conjoint factor PS + CBV was also compared between high- and low-grade glioma. At
a cut-off of 11.4573, rPS + CBV had 94.74% sensitivity, 100% specificity, 100% positive
predictive value, 93.3% negative predictive value, and diagnostic accuracy of 96.67%
([Table 2]).
On comparing area under the ROC curve for the perfusion parameters, it was found that
PS was significantly better (p-value < 0.05) in predicting the tumor grade from CBV and all other perfusion parameters
([Table 3], [Fig. 8]).
Table 3
Comparison of area under the curve of perfusion parameters with relative PS to predict
high-grade glioma
|
Relative PS
|
Difference between areas
|
95% Confidence interval
|
p-Value
|
|
Relative CBF
|
0.342
|
0.123–0.561
|
0.0022
|
|
Relative CBV
|
0.195
|
0.0223–0.369
|
0.0269
|
|
Relative MTT
|
0.32
|
0.137–0.502
|
0.0006
|
|
Relative TTP
|
0.38
|
0.173–0.586
|
0.0003
|
Abbreviations: CBV, cerebral blood volume; CBF, cerebral blood flow; MTT, mean transit
time; PS, permeability surface area; TTP, time to peak.
Fig. 8 Comparison of area under curve of permeability surface area product with other perfusion
parameters to predict high-grade from low-grade gliomas. CBF, cerebral blood flow;
CBV, cerebral blood volume; IQR, interquartile range; MTT, mean transit time; TTP,
time to peak; PS, permeability surface area; ROC, receiver operating characteristic.
Inter-rater kappa agreement was calculated for all the parameters and PS (k-value:
0.876) had a very good agreement with the tumor grade. Overall concordance rate was
93.94% and overall discordance rate was 6.06% between HPE grade and rPS. CBV (k-value 0.513) had a moderate agreement, while CBF, MTT, and TTP had a fair agreement.
This again supported PS as the most useful parameter ([Table 4]).
Table 4
Inter-rater kappa agreement of relative PS and HPE grade
|
Relative PS
|
HPE grade
|
Total
|
p-Value
|
Kappa
|
|
Low (n = 14)
|
High (n = 19)
|
|
≤9.0065
|
13 (39.39%)
|
1 (3.03%)
|
14 (42.42%)
|
<0.0001
|
0.876
|
|
>9.0065
|
1 (3.03%)
|
18 (54.55%)
|
19 (57.58%)
|
|
Total
|
14 (42.42%)
|
19 (57.58%)
|
33 (100.00%)
|
Abbreviations: HPE, histopathological examination; PS, permeability surface area.
Discussion
CT Perfusion Values between Glioma versus Normal Brain Parenchyma
In our study, all five perfusion parameters, that is, PS, CBV, CBF, MTT, and TTP,
had significantly higher values in the tumor region (in both high- and low-grade gliomas),
when compared to the normal background brain parenchyma. This finding is supported
by the fact that gliomas rely on neovascularization for their blood supply, hence
exhibiting higher perfusion values. Even early low-grade gliomas are well vascularized
by vascular co-option. As the grade progresses, other mechanisms of neoangiogenesis
set into action, that is, angiogenesis, vasculogenesis, vascular mimicry, and glioblastoma–endothelial
cell transdifferentiation, and lead to a substantial increase in the perfusion values.[12]
Our observation was fairly consistent with the findings of Xyda et al[13] who also found PS and CBV significantly higher in the tumor region (in both high-
and low-grade gliomas) as compared to the normal brain parenchyma. However, in contrast
to our study, Ding et al,[14] Maarouf et al,[15] and Kumar et al[16] reported high perfusion values only in high-grade gliomas and not in low-grade gliomas,
when compared to the background normal brain parenchyma.
Comparison of Perfusion Parameters in High- and Low-Grade Gliomas
In our study, both rPS and rCBV were found statistically significant in differentiating
high- and low-grade glioma. CBF, TTP, and MTT were not found statistically significant.
However, rPS had higher sensitivity (97.74%), specificity (100%), and diagnostic accuracy
(93.94%) than rCBV. On comparing area under the ROC curve for the perfusion parameters,
it was found that PS was significantly better (p-value < 0.05) in predicting the tumor grade from CBV and all other perfusion parameters.
The earlier studies[9]
[13]
[14]
[15]
[16]
[17] found PS, CBV, and CBF to be statistically significant in differentiating high-
and low-grade glioma. However, we found only PS and CBV significant and not CBF.
Moreover, in the literature there is no consensus regarding the most sensitive and
specific perfusion parameter in differentiating high- and low-grade glioma. Ding et
al[14] found PS to be the most sensitive (100%) and specific (83.3%) parameter, as observed
in our study. Maarouf et al[15] and Xyda et al[13] found CBV to be the most sensitive and specific parameter, while Kumar et al[16] found CBF to be the most sensitive and specific parameter.
In our study, the cut-off value of rPS for differentiating high- and low-grade glioma
was 9.0006, which was higher than the cut-offs calculated in other studies.[12]
[13]
[14]
[15] The higher cut-off values observed by us could be attributed to higher contrast
concentration and faster injection rate.[18] The contrast concentration and injection rate used in our study was 400 mg/mL at
5.5 mL/s, higher than aforementioned studies.[12]
[13]
[14]
[15] However, Xyda et al[13] demonstrated a lower cut-off value of 2.21 for PS at similar contrast concentration
and injection rate. More studies are required to delineate the effect of contrast
concentration on CT perfusion results in differentiating high- and low-grade glioma.
The sensitivity and specificity for PS to differentiate high- and low-grade glioma
obtained in our study were higher than the studies done by Maarouf et al,[15] Kumar et al,[16] and Saleh et al.[17] Similar sensitivity and specificity profile was observed by Xyda et al[13] as obtained in our study, which probably could be attributed to the higher contrast
concentration.
We calculated inter-rater kappa agreement for all the perfusion parameters. PS with
k-value 0.876 had a very good agreement with the tumor grade on HPE. CBV with k-value 0.513 had a moderate agreement, while CBF, MTT, and TTP had a fair agreement.
This again supported PS as the most useful parameter. Further studies are needed to
corroborate the strength of agreement between CTP and HPE grading as seen in our case.
We also studied the conjoint factor PS + CBV that had a similar diagnostic profile
as PS, with diagnostic accuracy (96.67%) marginally higher than PS (93.94%). PS and
CBV are complimentary factors for tumor vascularity, with PS being a surrogate marker
for microvascular cellular proliferation and CBV being a surrogate marker for microvascular
density.[19] To best of our knowledge, only one study done by Shankar et al[11] has evaluated the conjoint factor (PS + CBV) to differentiate grade III and grade
IV gliomas, and it was found to be statistically significant. However, no study using
conjoint factor to differentiate high- and low-grade glioma has been reported till
date.
Conclusion
Although both rPS and rCBV were helpful in differentiating high- and low-grade glioma,
rPS exhibited best sensitivity and specificity profile and was found significantly
better than the other perfusion parameters. rPS had a very good agreement with the
HPE grade. The conjoint factor (PS + CBV) did not yield any significant diagnostic
advantage over PS. The results of our study suggest that rPS alone can be helpful
in grading brain gliomas, and can add crucial value in the management and prognostication
of these tumors.
