Near-infrared spectroscopy for cerebral monitoring during carotid endarterectomy : symptomatic and asymptomatic patients might have different thresholds

Objectives This study evaluated the reliability of near-infrared spectroscopy (NIRS) monitoring during carotid endarterectomy (CEA) in patients affected by asymptomatic and symptomatic carotid stenosis. To identify the optimal cut-off value for regional oxygen saturation (rSO2) decrease in the two groups of patients. Methods We retrospectively reviewed NIRS data of 372 CEAs performed under general anesthesia from March 2007 to October 2014. Among them, 80 (21.5%) were treated for symptomatic disease. Ten (2.7%) postoperative neurologic events were registered (5 in asymptomatic and 5 in symptomatic group). Mean rSO2 preclamp values were collected and compared with the lowest rSO2 values during carotid cross-clamp. We considered different duration cutoffs, based on decrease lasting 1, 1.5, 2, 2.5, and 3 minutes. Receiver operating characteristic (ROC) curve analysis was performed to determine the best threshold value of rSO2 decrease in order to identify neurologic distress. Results No significant difference was found between the diverse duration of rSO2 decrease. ROC curve analysis was significant in asymptomatic patients, with an area under the curve (AUC) at 3 minutes of 0.75 (95% confidence interval [CI], 0.53-0.98). An optimal cutoff value of −17% was found in this group. Sensitivity was 80% (95% CI, 40%-100%) and specificity was 76.31% (95% CI, 70.9%-81.37%). Positive (PPV) and negative (NPV) predictive values were 5.48% and 99.54%, respectively. NIRS monitoring was not significant in symptomatic patients. In this group, we found an AUC of 0.39 (95% CI, 0.12-0.66) along with a threshold value of −9% (sensitivity, 100%; specificity, 24%; PPV, 8.1%; NPV, 100%). Conclusions Our results suggest that NIRS is a reliable method for cerebral monitoring during CEA in asymptomatic patients. A cutoff value of −17% might be safe, with a high NPV and acceptable sensitivity and specificity. For symptomatic patients, having in mind that our results did not reach statistical significance, we advise a −9% reduction as a threshold and therefore a wider use of shunt.