Comparison Of Magnetic Resonance Spectroscopy And Diffusion Weighted Imaging In The Differentiation Of Glioma Recurrence From Radiation Necrosis: Meta-Analysis

Wen-Fei Li,Yu-Chen Zhang,Chen Niu,Li-Ping Guo,Feng-Li Liang,Xiao Ling,Xuan Niu,Ming Zhang
2016-01-01
Abstract:Aim: The purpose of this study was to compare the diagnostic accuracy of diffusion-weighted magnetic resonance imaging (DWI) and magnetic resonance spectroscopy (MRS) for the diagnosis of differentiating glioma recurrence from radiation necrosis through a meta-analysis. Methods: We performed a meta-analysis of all available studies of the diagnostic performance of DWI and MRS for differentiating glioma recurrence from radiation necrosis. PubMed, Web of Science, and Chinese Biomedical databases (CBM) were searched for initial studies. Pooled sensitivity (SEN), specificity (SPE), negative likelihood ratio (NLR), positive likelihood ratio (PLR), and diagnostic odds ratio (DOR) were calculated. Results: 5 studies involving 112 patients (116 lesions) met all inclusion and exclusion criteria. For DWI, the pooled sensitivity of Apparent Diffusion Coefficient (ADC) was 0.80 and specificity was 0.81. Overall, PLR was 3.90, NLR was 0.28 and DOR was 17.53. For MRS, the pooled sensitivity of Cho/Cr was 0.86 and specificity was 0.79. PLR was 3.37, NLR was 0.22, and DOR was 21.04. The pooled sensitivity of Cho/NAA was 0.75 and specificity was 0.88, PLR was 4.57, NLR was 0.27, and DOR was 25.83. Without statistically significant differences except for the fact that the pooled sensitivity of Cho/NAA and Cho/Cr, and Cho/NAA also displayed higher area under the curve and Q *index compared with ADC (P<0.05). Conclusion: Both DWI and MRS were accurate tools for detecting glioma recurrence. Although MRS seemed to be superior to DWI, the latter could also be used to differentiate glioma recurrence from radiation necrosis when the former was unavailable. However, MRS and DWI could play different roles in differentiating glioma recurrence from radiation necrosis. Because of significant publication bias, pooled diagnostic measures might be overvalued. Large-sample randomized controlled studies were needed to establish its value for differentiating glioma recurrence from radiation necrosis.
What problem does this paper attempt to address?