Reliability of Functional Magnetic Resonance Imaging in Patients with Brain Tumors: A Critical Review and Meta-Analysis

Abstract

Formal studies that validated functional magnetic resonance imaging (fMRI) against direct cortical stimulation (DCS) have shown inconsistencies. We reviewed the reported data and performed a meta-analysis of studies that had validated fMRI using DCS and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We performed the meta-analysis of the studies that had met our inclusion criteria. The positive predictive values (PPVs), negative predictive values (NPVs), false omission rates (FORs), false discovery rates (FDRs), sensitivity, and specificity were calculated. We also identified the factors that could influence the fMRI findings. Six studies of language activation and two of motor activation were eligible for the present meta-analysis. The PPV of fMRI for the detection of eloquent motor cortex was 59.5%, and the FDR was 40.5%. The NPV was 96% and the FOR was 3.9%. The sensitivity and specificity of fMRI for the detection of the eloquent motor cortex was 91% and 76%, respectively. For language activation, the PPV was 71.7%, the NPV was 80%, the FDR was 28.2%, and the FOR was 19.5%. The sensitivity and specificity was 80% and 71.5%, respectively. The factors that could have caused a mismatch or affected the reliability were analyzed. The results from the present analysis of the available reported evidence suggest that fMRI itself (due to neurovascular uncoupling) or analysis of the findings have limitations regarding reliability when validated against DCS. From the existing studies, our results indicate that using fMRI alone for surgical planning could lead to undesirable outcomes.