Diagnostic performance of DWI for differentiating primary central nervous system lymphoma from glioblastoma: a systematic review and meta-analysis

Xiaoyang Lu,Weilin Xu,Yuyu Wei,Tao Li,Liansheng Gao,Xiongjie Fu,Yuan Yao,Lin Wang
DOI: https://doi.org/10.1007/s10072-019-03732-7
2019-01-31
Neurological Sciences
Abstract:ObjectiveThe purpose of this meta-analysis was to evaluate the diagnostic performance of diffusion-weighted imaging (DWI) for differentiating primary central nervous system lymphoma (PCNSL) from glioblastoma (GBM).Materials and methodsA thorough search of the databases including PubMed, EMBASE, and Cochrane Library was carried out and the data acquired were up to November 1, 2017. The quality of the studies involved was evaluated using QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies, revised version). Multiple analytic values including sensitivity (SEN), specificity (SPE), positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and the summary receiver operating characteristic (SROC) curve were calculated and pooled for the statistical analysis. The subgroup analysis was also performed to explore the heterogeneity.ResultsEight retrospective studies (461 patients with 461 lesions) were included. The pooled SEN, SPE, PLR, NLR, and DOR with 95% confidence interval (CI) were 0.82 [95% CI 0.70–0.90], 0.84 [95% CI 0.75–0.90], 4.96 [95% CI 3.20–7.69], 0.22 [95% CI 0.13–0.37], and 22.85 [95% CI 10.42–50.11], respectively. The area under the curve (AUC) given by SROC curve was 0.90 [95% CI 0.87–0.92]. The subgroup analysis indicated the slice thickness of the images (> 3 mm versus ≤ 3 mm) was a significant factor affecting the heterogeneity. No existence of significant publication bias was confirmed with Deeks’ test.ConclusionsDWI showed moderate diagnostic performance for differentiating primary central nervous system lymphoma (PCNSL) from glioblastoma (GBM). Moreover, it is of clinical significance using DWI combined with conventional MRI to differentiate PCNSL from GBM.
neurosciences,clinical neurology
What problem does this paper attempt to address?