Discrimination between Glioma Grades II and III Using Dynamic Susceptibility Perfusion MRI: A Meta-Analysis

Abstract

Background: DSC perfusion has been evaluated in the discrimination between low-grade and high-grade glioma but the diagnostic potential to discriminate beween glioma grades II and III remains unclear.

Purpose: Our aim was to evaluate the diagnostic accuracy of relative maximal CBV from DSC perfusion MR imaging to discriminate glioma grades II and III.

Data sources: A systematic literature search was performed in PubMed/MEDLINE, Embase, Web of Science, and ClinicalTrials.gov.

Study selection: Eligible studies reported on patients evaluated with relative maximal CBV derived from DSC with a confirmed neuropathologic diagnosis of glioma World Health Organization grades II and III. Studies reporting on mean or individual patient data were considered for inclusion.

Data analysis: Data were analyzed by using inverse variance with the random-effects model and receiver operating characteristic curves describing optimal cutoffs and areas under the curve. Bivariate diagnostic random-effects meta-analysis was used to calculate diagnostic accuracy.

Data synthesis: Twenty-eight studies evaluating 727 individuals were included in the meta-analysis. Individual data were available from 10 studies comprising 190 individuals. The mean difference of relative maximal CBV between glioma grades II and III (n = 727) was 1.76 (95% CI, 1.27-2.24; P < .001). Individual patient data (n = 190) had an area under the curve of 0.77 for discriminating glioma grades II and III at an optimal cutoff of 2.02. When we analyzed astrocytomas separately, the area under the curve increased to 0.86 but decreased to 0.61 when we analyzed oligodendrogliomas.

Limitations: A substantial heterogeneity was found among included studies.

Conclusions: Glioma grade III had higher relative maximal CBV compared with glioma grade II. A high diagnostic accuracy was found for all patients and astrocytomas; however, the diagnostic accuracy was substantially reduced when discriminating oligodendroglioma grades II and III.

Fig 1.

PRISMA flow chart of included and excluded studies.

Fig 2.

Forest plot of mean differences between glioma grades II and III with a random-effects model.

Fig 3.

Funnel plot of 28 included studies (n = 727 patients) illustrated by open circles with the effect estimate mean difference (MD) of rCBVmax plotted on the horizontal axis, the standard error (SE) of the MD plotted on the vertical axis, and a triangular 95% confidence region. The study distribution is symmetric without apparent publication bias.

Fig 4.

ROC curve on individual patient data from 10 studies including all patients (n = 190).

Fig 5.

Paired forest plots for individual patient data (n = 190). A, Forest plot of sensitivity. B, Forest plot of specificity.

Fig 6.

Graph showing a point estimate of the pair of sensitivity and false-positive rates plotted together with a confidence region, without extrapolation beyond the range of the original data (individual patient data, n = 190).