Intraoperative evoked potential monitoring for detecting cerebral injury during adult aneurysm clipping surgery: a systematic review and meta-analysis of diagnostic test accuracy

Objectives We aim to evaluate the diagnostic test accuracy (DTA) of intraoperative evoked potential (EP) monitoring to detect cerebral injury during clipping of cerebral aneurysms. Design Systematic review. Data sources Major electronic databases including MEDLINE, EMBASE, LILACS. Eligibility criteria We included studies that reported the DTA of intraoperative EP monitoring during intracranial aneurysm clipping procedures in adult patients. Data extraction and synthesis After quality assessment, we performed a meta-analysis using the bivariate random effects model, and calculated the possible range of DTA point estimates using a new best-case/ worst-case scenario approach to quantify the impact of rescue intervention on DTA. Results A total of 35 studies involving 4011 patients were included. The quality of the primary studies was modest and the heterogeneity across studies was high. The pooled sensitivity and specificity for predicting postoperative neurological deficits for the somatosensory evoked potential (SSEP) monitoring was 59% (95% CI: 39% to 76%; I-2: 76%) and 86% (95% CI: 77% to 92%; I-2: 94%), for motor evoked potential (MEP) monitoring was 81% (95% CI: 58% to 93%; I-2: 54%) and 90% (95% CI: 86% to 93%; I-2: 81%), and for combined SSEP and MEP monitoring was 92% (95% CI: 62% to 100%) and 88% (95% CI: 83% to 93%). The best-case/ worst-case range for the pooled point estimates for sensitivity and specificity for SSEP was 50%-63% and 81%-100%, and for MEP was 59%-74% and 93%-100%, and for combined SSEP and MEP was 89%-94% and 83%-100%. Conclusions Due to the modest quality and high heterogeneity of the existing primary studies, it is not possible to confidently support or refute the diagnostic value of EP monitoring in cerebral aneurysm clipping surgery. However, combined SSEP and MEP appears to provide the best DTA for predicting postoperative stroke. Contrary to popular assertion, the modest sensitivity of SSEP monitoring is not explained by the use of rescue intervention.