Gliomas account for approximately a third of nervous system tumors and the majority of malignancies. The incidence is between 5 and 6 per 100,000.1–3 4 5–7
However, an abnormal MRI signal may represent either tumor recurrence (TR) or radiation injury (RI), and it is important to distinguish between them.8–10 9 10 9
Other MRI techniques, including diffuse-weighted imaging (DWI), perfusion-weighted imaging (PWI), and MRI spectroscopy (MRS), are often used for analyzing cancer and necrotic tissue, and these may be better able to differentiate TR and RI lesions.11–23
Here, we aimed to evaluate the effectiveness of dynamic susceptibility contrast-enhanced (DSC) PWI (DSC-PWI) for distinguishing between TR and RI in patients with glioma.
MATERIALS AND METHODS
Relevant articles in the PubMed, Embase, and Cochrane Library databases were identified up to October 2021. The research strategy was: ((((((perfusion weighted imaging) OR (PWI)) OR (perfusion MR)) OR (MR perfusion)) OR (dynamic susceptibility contrast-enhanced)) AND (glioma)) AND ((recurrence) OR (recurrent)) AND ((radiation injury ) OR (radiation necrosis)). This meta-analysis was registered at https://inplasy.com/
Studies eligible for inclusion were: (a ) studies assessing the differential diagnosis of TR and RI in patients with glioma; (b ) PWI was used as the diagnostic tool; (c ) DSC was used as the PWI technique; and (d ) studies in which sensitivity and specificity were provided. Reviews and case reports, as well as studies dealing with animals, with sample sizes less than 20, or lacking English titles and abstracts, were excluded.
Extraction of Data and Quality Evaluation
Data were extracted by 2 independent investigators, and disputes were decided by a third investigator. The data included information on first author, publication year, country, study design, blinding status, number of patients, tumor numbers, World Health Organization (WHO) grading, postoperative radiation dose, field strength of MRI, reference standards, cutoff values, and diagnostic performance (true-positive, false-positive, true-negative, and false-negative results).
The likelihood of bias in included studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool.24
Definitions
Tumor recurrence was defined as the positive result, and RI was defined as the negative result. Tumor recurrence was diagnosed according to the histopathology of the lesion and whether there was growth of the lesion and deterioration of the patient's condition during the follow-up period. The RI diagnosis depended on histopathology and whether the size of the lesion decreased and the patient's condition improved during the follow-up period. High-grade glioma was defined as a tumor with a WHO grade of III or greater.19
Meta-analysis
RevMan v5.3 and Stata v12.0 were used for meta-analysis. The sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and relative cerebral blood volume (rCBV) values from the individual studies were pooled. A high likelihood of correct diagnosis for either TR or RI was represented by PLR greater than 5 or NLR less than 0.2. Summary receiver operating characteristic (SROC) curves were produced; the diagnosis was judged to be reliable when the area under the curve (AUC) was greater than 80%.
Heterogeneity was assessed by I 2 tests, with I 2 values greater than 50% suggesting significant heterogeneity. Random-effects models were used for significant heterogeneity, while fixed-effects models were used for significant homogeneity. A P value less than 0.05 indicated significance. The sources of heterogeneity were investigated using sensitivity, subgroup, and meta-regression analyses.
Publication bias was evaluated by the Deeks funnel plot and the Egger test, with a P value less than 0.05 indicating bias.
RESULTS
Baseline Information
We found 128 studies on the initial search. After step-by-step selection, 13 studies were finally included in the analysis (Fig. 1 , Table 1 ). These studies included 513 patients with 522 lesions. Of the 528 lesions, 329 were TRs and 193 were RIs. All studies were retrospective in nature. Six of the studies were from Asian institutions, while 7 were produced by European or American institutions. All studies used the rCBV value from PWI as the diagnostic tool. The details of the field strength, reference standards, WHO grades, and cutoff values of rCBV are shown in Table 1 .
FIGURE 1: Flowchart diagram of our meta-analysis. Figure 1 can be viewed online in color at
www.jcat.org .
TABLE 1 -
Characteristics of Studies Included in Meta-analysis
Studies
Year
Country
Patients, n
Tumors, n
Who Grade
RD, Gy
FS
RS
Cutoff
Barajas et al
11
2009
United States
57
66
IV
59.4–60
1.5
S, B, F
1.75
Cha et al
12
2014
Korea
35
35
IV
Not given
3
S, B, F
1.8
Di Costanzo et al
13
2014
Italy
29
29
IV
60
3
S, F
Not given
Fink et al
14
2012
United States
40
40
II-IV
54–65
3
S, B, F
2.08
Hojjati et al
15
2018
United States
22
22
IV
58.62
3
S, F
3.32
Hu et al
16
2009
United States
40
40
III/IV
63
3
S, B, F
0.71
Ozsunar et al
17
2016
Turkey
32
32
II-IV
Not given
1.5
S, B, F
1.3
Qiao et al
18
2019
China
42
42
III/IV
Not given
Not given
S, B, F
1.83
Seeger et al
19
2016
Germany
40
40
III/IV
60
1.5
S, B, F
2.25
Sha et al
20
2013
China
52
52
II-IV
40–60
3
S, B, F
1.375
Wang et al
21
2018
China
69
69
II-IV
40–60
3
S, F
2.26
Xu et al
22
2011
China
35
35
II-IV
48–68.8
3
S, B
2.15
Young et al
23
2013
United States
20
20
IV
59.4–60
1.5/3
S, B, F
2.4
B indicates biopsy; F, follow-up; FS, field strength; RD, radiation dose; RS, reference standard; S, surgery.
The raw data detailing the true-positive, false-positive, true-negative, and false-negative results are shown in Table 2 . Furthermore, 9 studies compared the rCBV values between TR and RI groups.11–14,18–22
TABLE 2 -
Raw Data of Diagnostic Performance of Studies Included in This Meta-analysis
True Positive
False Positive
False Negative
True Negative
rCBV Values
Recurrence
RI
Barajas et al
11
36
6
10
14
2.38
1.57
Cha et al
12
9
4
2
20
2.15
1.4
Di Costanzo et al
13
18
1
3
7
1.73
0.86
Fink et al
14
26
1
4
9
3.62
1.31
Hojjati et al
15
18
1
0
3
Not given
Not given
Hu et al
16
22
0
2
16
Not given
Not given
Ozsunar et al
17
19
3
3
7
Not given
Not given
Qiao et al
18
26
1
7
8
2.68
1.33
Seeger et al
19
19
4
4
13
3.91
1.73
Sha et al
20
21
0
9
22
2.43
0.67
Wang et al
21
24
3
11
31
3.39
1.39
Xu et al
22
16
3
4
12
4.36
1.28
Young et al
23
16
1
0
3
Not given
Not given
Risk of Bias
The QUADAS-2 tool was used to assess the possibility of bias (Figs. 2 A, B). Of the 13 included studies, 9 did not indicate whether patients were enrolled in a consecutive manner.11,13–20
FIGURE 2: Representation of the methodological quality (A) graph and (B) summary. Figure 2 can be viewed online in color at
www.jcat.org .
Diagnostic Performance
Nine studies compared rCBV values between the TR and RI groups.11–14,18–22 P < 0.00001; Fig. 3 ). The heterogeneity was significant (I 2 = 88%). The sensitivity analysis showed that removing individual studies had no impact on the detected heterogeneity. The Egger test found no evidence of publication bias (P = 0.367), nor was publication bias detected by the Deeks funnel plot asymmetry test (P = 0.496).
FIGURE 3: The pooled rCBV results between patients with TR and RI. Figure 3 can be viewed online in color at
www.jcat.org .
The pooled values for sensitivity, specificity, PLR, and NLR were 83% (95 CI, 77%–88%; Fig. 4 A), 85% (95 CI, 77%–91%; Fig. 4 B), 5.60 (95 CI, 3.61–8.70; Fig. 4 C), and 0.20 (95% CI, 0.14–0.27; Fig. 4 D), respectively. The heterogeneity of sensitivity (I 2 = 33.18%), specificity (I 2 = 24.01%), PLR (I 2 = 0.00%), and NLR (I 2 = 6.68%) was not significant. The AUC value is 0.91 (95% CI, 0.88–0.93; Fig. 4 E). The SROC curve lacked a shoulder, suggesting the absence of a threshold effect. The Fagan plot is seen in Figure 4 F. The posttest PLR and NLR probabilities were estimated at 58% and 5%, respectively, when the pretest probability was 20%. The accuracy of diagnosis was found to be high.
FIGURE 4: The pooled results (A) sensitivity, (B) specificity, (C) PLR, (D) NLR, (E) SROC, and (F) Fagan diagram in this meta-analysis. Figure 4 can be viewed online in color at
www.jcat.org .
Meta-Regression and Subgroup Analyses
The meta-regression analysis was conducted based on the factors of MRI field strength, tumor WHO grade, and country (Table 3 ). The sensitivity and specificity were not affected by any of these factors. Table 4 lists the results of the subgroup analyses. The 3.0 T MRI, high-grade glioma, and Europe/America patient subgroups showed simultaneous PLR greater than 5 and NLR less than 0.2. Different rCBV cutoff value subgroups did not show simultaneous PLR greater than 5 and NLR less than 0.2.
TABLE 3 -
Results of Meta-Regression
Sensitivity
Specificity
Estimate
Coefficient
P
Estimate
Coefficient
P
Field strength
1.5 T
1
1
3.0 T
0.83 (0.66–0.93)
1.58
0.93
0.96 (0.88–0.99)
3.09
0.05
WHO grades
II-IV
1
1
High grade only
0.89 (0.79–0.95)
2.14
0.23
0.78 (0.54–0.91)
1.25
0.40
Countries
Asia
1
1
Europe/America
0.91 (0.83–0.95)
2.30
0.06
0.80 (0.56–0.92)
1.37
0.54
TABLE 4 -
Subgroup Analysis
Studies, n
Sensitivity (95% CI)
Specificity (95% CI)
PLR (95% CI)
NLR (95% CI)
AUC
Field strength
1.5 T
4
84% (76%–90%)
73% (59%–83%)
3.06 (1.95–4.83)
0.22 (0.14–0.35)
0.86
3.0 T
8
84% (74%–90%)
90% (82%–94%)
8.13 (4.71–14.04)
0.18 (0.12–0.29)
0.94
WHO grade
II–IV
5
78% (69%–86%)
89% (76%–95%)
7.00 (3.19–15.13)
0.24 (0.17–0.35)
0.88
High grade only
8
87% (79%–92%)
84% (73%–91%)
5.33 (3.06–9.31)
0.16 (0.09–0.27)
0.92
Countries
Asia
6
75% (66%–82%)
90% (81%–95%)
7.18 (3.76–13.71)
0.28 (0.28–0.39)
0.84
Europe/America
7
89% (81%–94%)
85% (72%–93%)
5.89 (2.89–11.97)
0.13 (0.07–0.25)
0.93
Cutoff value
<2
6
80% (73%–86%)
89% (72%–97%)
7.58 (2.57–22.32)
0.22 (0.16–0.31)
0.82
≥2
6
88% (74%–95%)
82% (70%–90%)
4.98 (2.88–8.67)
0.15 (0.07–0.32)
0.90
DISCUSSION
The current meta-analysis assessed the efficacy of DSC-PWI for the differential diagnosis of TR and RI in patients with glioma. Tumor recurrence is usually associated with higher rCBV values than seen with RI. Dynamic susceptibility contrast-enhanced PWI is effective for evaluating the physiological status and hemodynamics of the blood vessels.25,26 27 I 2 = 88%) was also detected. It is possible that this heterogeneity may have been caused by the use of different MRI instruments, the assignment of different WHO grades, or different radiation doses administered after surgery.
The pooled sensitivity and specificity indicated that DSC-PWI are useful methods for differentiating between TR and RI in patients with glioma. Moreover, the diagnostic accuracy should be evaluated via the PLR, NLR, and SROC values. The pooled PLR (5.60) and NLR (0.20) values in this meta-analysis demonstrated that a higher rCBV value (more than the cutoff) may indicate TR, while a lower rCBV value (less than the cutoff) may indicate RI. The SROC value (0.91) indicated that the overall diagnostic accuracy of DSC-PWI was high.
The heterogeneity of the sensitivity, specificity, PLR, and NLR was low. These findings indicate that these pooled diagnostic results are stable. Considering the different MRI field strengths, 3.0 T MRI was found to be better able to diagnose both TR and RI. Furthermore, we found that DSC-PWI was better able to differentiate between TR and RI in patients with high-grade glioma. High-grade gliomas have more blood supply and thus greater numbers of arteries relative to low-grade tumors.28 27,28
In addition to PWI, other functional MRI techniques, such as DWI and MRS, are also used for the differential diagnosis of TR and RI in patients with glioma.14,19,29 14 29 P = 0.70). These results may indicate that PWI is superior to DWI when differentiating TR and RI. Although the DWI procedure is more rapid than that of PWI as it does not require the administration of gadolinium contrast, the investment of the additional time in PWI would be both beneficial and worthwhile.14
Magnetic resonance imaging spectroscopy is also a useful method for differentiating TR and RI in patients with glioma.28,30 31 19
Both DSC and dynamic contrast-enhanced techniques provide similar measurements of angiogenesis, which can help in diagnosis of brain tumors.32 K trans , which could be useful as biomarkers for angiogenesis. However, it cannot provide the same spatial coverage and temporal resolution compared with DSC imaging.32 32
This meta-analysis has several limitations. First, all the included studies were retrospective in nature. Therefore, there is a relatively high risk of bias, and additional prospective studies would be necessary to verify and expand these results. Second, the different MRI parameters, which included the instrument manufacturers, field strength, repetition time, echo time, field of view, and matrix, were not the same among these studies. These findings further increase the risk of bias. Although we performed the subgroup analysis based on the 1.5 and 3.0 T MRI, the heterogeneity of each subgroup still cannot be eliminated. Third, the reference standards and cutoff values were not the same among these studies, and the use of different standards and cutoffs may influence the diagnostic accuracy.
In conclusion, it was found that the use of DSC-PWI is effective for the differential diagnosis of TR and RI in patients with glioma.
REFERENCES
1. Soni N, Ora M, Mohindra N, et al. Diagnostic performance of PET and perfusion-weighted imaging in differentiating tumor recurrence or progression from radiation necrosis in posttreatment gliomas: a review of literature.
AJNR Am J Neuroradiol . 2020;41:1550–1557.
2. Li Z, Zhou Q, Li Y, et al. Mean cerebral blood volume is an effective diagnostic index of recurrent and
radiation injury in glioma patients: a meta-analysis of diagnostic test.
Oncotarget . 2017;8:15642–15650.
3. Ostrom QT, Gittleman H, Liao P, et al. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2007-2011.
Neuro Oncol . 2014;16 suppl 4(suppl 4):iv1–iv63.
4. Auffinger B, Thaci B, Nigam P, et al. New therapeutic approaches for malignant glioma: in search of the Rosetta stone.
F1000 Med Rep . 2012;4:18.
5. Kim M, Jung SY, Park JE, et al. Diffusion- and perfusion-weighted MRI radiomics model may predict isocitrate dehydrogenase (IDH) mutation and tumor aggressiveness in diffuse lower grade glioma.
Eur Radiol . 2020;30:2142–2151.
6. Aquino D, Gioppo A, Finocchiaro G, et al. MRI in glioma immunotherapy: evidence, pitfalls, and perspectives.
J Immunol Res . 2017;2017:5813951.
7. Tsiouris S, Bougias C, Fotopoulos A. Principles and current trends in the correlative evaluation of glioma with advanced MRI techniques and PET.
Hell J Nucl Med . 2019;22:206–219.
8. Parvez K, Parvez A, Zadeh G. The diagnosis and treatment of pseudoprogression, radiation necrosis and brain tumor recurrence.
Int J Mol Sci . 2014;15:11832–11846.
9. Bobek-Billewicz B, Stasik-Pres G, Majchrzak H, et al. Differentiation between brain tumor recurrence and
radiation injury using perfusion, diffusion-weighted imaging and MR spectroscopy.
Folia Neuropathol . 2010;48:81–92.
10. Verma N, Cowperthwaite MC, Burnett MG, et al. Differentiating tumor recurrence from treatment necrosis: a review of neuro-oncologic imaging strategies.
Neuro Oncol . 2013;15:515–534.
11. Barajas RF Jr., Chang JS, Segal MR, et al. Differentiation of recurrent glioblastoma multiforme from radiation necrosis after external beam radiation therapy with dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging.
Radiology . 2009;253:486–496.
12. Cha J, Kim ST, Kim HJ, et al. Differentiation of tumor progression from pseudoprogression in patients with posttreatment glioblastoma using multiparametric histogram analysis.
AJNR Am J Neuroradiol . 2014;35:1309–1317.
13. Di Costanzo A, Scarabino T, Trojsi F, et al. Recurrent glioblastoma multiforme versus
radiation injury : a multiparametric 3-T MR approach.
Radiol Med . 2014;119:616–624.
14. Fink JR, Carr RB, Matsusue E, et al. Comparison of 3 Tesla proton MR spectroscopy, MR perfusion and MR diffusion for distinguishing glioma recurrence from posttreatment effects.
J Magn Reson Imaging . 2012;35:56–63.
15. Hojjati M, Badve C, Garg V, et al. Role of FDG-PET/MRI, FDG-PET/CT, and dynamic susceptibility contrast perfusion MRI in differentiating radiation necrosis from tumor recurrence in glioblastomas.
J Neuroimaging . 2018;28:118–125.
16. Hu LS, Baxter LC, Smith KA, et al. Relative cerebral blood volume values to differentiate high-grade glioma recurrence from posttreatment radiation effect: direct correlation between image-guided tissue histopathology and localized dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging measurements.
AJNR Am J Neuroradiol . 2009;30:552–558.
17. Ozsunar Y, Mullins ME, Kwong K, et al. Glioma recurrence versus radiation necrosis? A pilot comparison of arterial spin-labeled, dynamic susceptibility contrast enhanced MRI, and FDG-PET imaging.
Acad Radiol . 2010;17:282–290.
18. Qiao Z, Zhao X, Wang K, et al. Utility of dynamic susceptibility contrast perfusion-weighted mr imaging and
11 C-methionine PET/CT for differentiation of tumor recurrence from
radiation injury in patients with high-grade gliomas.
AJNR Am J Neuroradiol . 2019;40:253–259.
19. Seeger A, Braun C, Skardelly M, et al. Comparison of three different MR perfusion techniques and MR spectroscopy for multiparametric assessment in distinguishing recurrent high-grade gliomas from stable disease.
Acad Radiol . 2013;20:1557–1565.
20. Sha L, Fan G, Cao Q, et al. Value of MR perfusion weighted imaging in diagnosis of glioma recurrence and cerebral radiation injuries.
J Dalian Med Univ . 2013;35:551–560.
21. Wang YL, Chen S, Xiao HF, et al. Differentiation between radiation-induced brain injury and glioma recurrence using 3D pCASL and dynamic susceptibility contrast-enhanced perfusion-weighted imaging.
Radiother Oncol . 2018;129:68–74.
22. Xu JL, Shi DP, Dou SW, et al. Distinction between postoperative recurrent glioma and delayed
radiation injury using MR perfusion weighted imaging.
J Med Imaging Radiat Oncol . 2011;55:587–594.
23. Young RJ, Gupta A, Shah AD, et al. MRI perfusion in determining pseudoprogression in patients with glioblastoma.
Clin Imaging . 2013;37:41–49.
24. Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies.
Ann Intern Med . 2011;155:529–536.
25. Law M, Young RJ, Babb JS, et al. Gliomas: predicting time to progression or survival with cerebral blood volume measurements at dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging.
Radiology . 2008;247:490–498.
26. Jain R, Ellika SK, Scarpace L, et al. Quantitative estimation of permeability surface-area product in astroglial brain tumors using perfusion CT and correlation with histopathologic grade.
AJNR Am J Neuroradiol . 2008;29:694–700.
27. Ellika SK, Jain R, Patel SC, et al. Role of perfusion CT in glioma grading and comparison with conventional MR imaging features.
AJNR Am J Neuroradiol . 2007;28:1981–1987.
28. Falk Delgado A, Falk Delgado A. Discrimination between primary low-grade and high-grade glioma with
11 C-methionine PET: a bivariate diagnostic test accuracy meta-analysis.
Br J Radiol . 2018;91:20170426.
29. Liu C, Hu X, Zheng L. Application analysis of functional magnetic resonance imaging in postoperative recurrence and
radiation injury of gliomas.
Ying Xiang Yan Jiu Yu Yi Xue Ying Yong . 2019;3:159–160.
30. Elias AE, Carlos RC, Smith EA, et al. MR spectroscopy using normalized and non-normalized metabolite ratios for differentiating recurrent brain tumor from
radiation injury .
Acad Radiol . 2011;18:1101–1108.
31. Zhang H, Ma L, Wang Q, et al. Role of magnetic resonance spectroscopy for the differentiation of recurrent glioma from radiation necrosis: a systematic review and meta-analysis.
Eur J Radiol . 2014;83:2181–2189.
32. Nguyen TB, Cron GO, Perdrizet K, et al. Comparison of the diagnostic accuracy of DSC- and dynamic contrast-enhanced MRI in the preoperative grading of astrocytomas.
AJNR Am J Neuroradiol . 2015;36:2017–2022.
