S170. Optimal baseline selection approach for somatosensory evoked potential monitoring during cerebral aneurysm surgery

Abstract

Optimal baseline selection is imperative for identifying significant somatosensory evoked potential (SSEP) changes and warning the surgical team of impending stroke during cerebral aneurysm surgery. In this study, we aim to evaluate the diagnostic accuracy of SSEP changes for perioperative stroke from different baseline measurements and the optimal baseline selection approach. We retrospectively reviewed the intraoperative neurophysiological records of 100 patients who underwent intracranial aneurysm clipping at our institution between 2009 and 2013. First, 25 patients who incurred a new onset perioperative stroke were identified and then 75 age- and sex- matched controls were selected. Quantitative analysis was done to evaluate SSEP amplitude and latency changes during various surgical events. Diagnostic accuracy was calculated as the area under the receiver operating characteristic curve (AUC) for maximum SSEP amplitude and latency changes from baselines set at pre-incision, incision, and dural opening. SSEP amplitude changes were associated with an increased stroke risk from all baselines: AUC (95% CI), pre-incision: 0.81 (0.71–0.91), incision: 0.80 (0.71–0.90), dural opening: 0.77 (0.65–0.90). Latency changes were only found to associated with stroke from dural opening baseline: 0.68 (0.55–0.81). The current alarm criteria of 50% change in amplitude from a pre-incision baseline had a sensitivity of 0.92 and specificity of 0.51. Using a later baseline such as dural opening improves specificity (0.80) but lowers sensitivity (0.60). Using a pre-incision baseline until dural opening and then using the dural opening baseline was found to have the best compromise of sensitivity (0.88) and specificity (0.71). A 10% latency change from a pre-incision baseline was not a good predictor for stroke. SSEP amplitude changes are predictive of stroke. A 50% amplitude decrease predicts stroke. Latency changes have limited utility in predicting stroke.