Cutoff Points, Sensitivities, and Specificities of Intraoperative Motor-Evoked Potential Monitoring Determined Using Receiver Operating Characteristic Analysis.

Abstract

Although intraoperative motor-evoked potential (MEP) monitoring is widely performed during neurosurgical operations, evaluating its results is controversial. The cutoff point of MEP monitoring should be determined not only to predict but also to prevent postoperative neurologic deficits. MEP monitoring was performed during 484 neurosurgical operations for patients without definitive preoperative motor palsy including 325 spinal operations, 102 cerebral aneurysmal operations, and 57 brain tumor operations, all monitored by transcranial stimulation, and 34 brain tumor operations monitored under direct cortical stimulation. To exclude the effects of muscle relaxants on MEP, the compound muscle action potential (CMAP), measured immediately after transcranial stimulation or direct cortical stimulation at supramaximal stimulation of the peripheral nerve, was used for normalization. The cutoff points, sensitivity, and specificity of MEP recorded during neurosurgery were examined by receiver operating characteristic (ROC) analyses and categorized according to the type of operation and stimulation. In spinal operations under transcranial stimulation, amplitude reduction of 77.9% and 80.6% as cutoff points for motor palsy with and without CMAP normalization, respectively, provided a sensitivity of 100% and specificity of 96.8% and 96.5%. In aneurysmal operations under transcranial stimulation, cutoff points of 70.7% and 69.6% offered specificities of 95.2% and 95.7% with and without CMAP normalization, respectively. The sensitivities for both were 100%. In brain tumor operations under direct stimulation, cutoff points were 83.5% and 86.3% with or without CMAP normalization, respectively, and the sensitivity and specificity for both were 100%. An amplitude decrease of 80% in brain tumor operations, 75% in spinal operations, and 70% in aneurysmal operations should be used as the cutoff points.