Overview and Diagnostic Accuracy of Near Infrared Spectroscopy in Carotid Endarterectomy: A Systematic Review and Meta-analysis

Carotid endarterectomy is recommended for the prevention of ischaemic stroke due to carotid stenosis. However, the risk of stroke after carotid endarterectomy has been estimated at 2% - 5%. Monitoring intra-operative cerebral oxygenation with near infrared spectroscopy (NIRS) has been assessed as a strategy to reduce intra- and post-operative complications. The aim was to summarise the diagnostic accuracy of NIRS to detect intra-operative ischaemic events, the values associated with ischaemic events, and the relative contribution of external carotid contamination to the NIRS signal in adults undergoing carotid endarterectomy.EMBASE, MEDLINE, Cochrane Centre Register of Controlled Trials, and reference lists through May 2019 were searched.Non-randomised and randomised studies assessing NIRS as an intra-operative monitoring tool in carotid endarterectomy were included. Studies using NIRS as the reference were excluded. Risk of bias was assessed using the Newcastle Ottawa Scale, RoB-2, and QUADAS-2.Seventy-six studies were included (n = 8 480), under local (n = 1 864) or general (n = 6 582) anaesthesia. Seven studies were eligible for meta-analysis (n = 524). As a tool for identifying intra-operative ischaemia, specificity increased with more stringent NIRS thresholds, while there was unpredictable variation in sensitivity across studies. A Δ20% threshold under local anaesthesia resulted in pooled estimates for sensitivity and specificity of 70.5% (95% confidence interval, CI, 54.1 - 82.9) and 92.4% (95% CI 85.5 - 96.1) compared with awake neurological monitoring. These studies had low or unclear risk of bias. NIRS signal consistently dropped across clamping and recovered to pre-clamp values upon de-clamp in most studies, and larger decreases were observed in patients with ischaemic events. The contribution of extracranial signal to change in signal across clamp varied from 3% to 50%.NIRS has low sensitivity and high specificity to identify intra-operative ischaemia compared with awake monitoring. Extracranial signal contribution was highly variable. Ultimately, data from high quality studies are desperately needed to determine the utility of NIRS.