Cerebral Oximetry During Awake Carotid Endarterectomy

Background. Selective shunting during carotid endarterectomy requires a reliable intraoperative test of cerebral ischaemia. Under regional anaesthesia the awake neurological test remains the accepted reference standard, but it cannot be used under general anaesthesia. Near-infrared spectroscopy has been evaluated in many cohorts, and a systematic review has identified two recurring methodological limitations: rSO2 thresholds were frequently derived from the study data rather than pre-specified, and masking of the index test from the clinical team was seldom reported. Both limitations inflate apparent diagnostic accuracy. This study addresses both by prospective registration, pre-specification of thresholds, and complete masking of the oximeter.

Design. Prospective, single-centre, consecutive-sampling cohort study with masking of the index test, incorporating a diagnostic accuracy analysis reported according to STARD 2015 and a prognostic analysis reported according to STROBE. There is no assignment of any intervention; the oximeter is used for measurement only and contributes to no clinical decision.

Anaesthetic technique. All participants receive combined ultrasound-guided superficial cervical plexus block and carotid sheath block with 1% lidocaine, maximum 5 mg/kg. Routine sedation is prohibited. Supplemental oxygen is standardised at 3 L/min by nasal cannula with end-tidal carbon dioxide sampling. Invasive arterial pressure is monitored continuously via a radial artery catheter.

Index test. Bilateral frontal rSO2 is recorded continuously using a single oximeter model throughout the study (Masimo O3 regional oximetry on the Masimo Root platform), sampled at a minimum of one value every 5 seconds, from before the block until 10 minutes after declamping. The oximeter is positioned outside the field of view of the clinical team, the display is covered with an opaque shield and audible alarms are disabled. Only an independent investigator who takes no part in the clinical care of that patient may view the device, solely to verify signal quality, and is prohibited from communicating any value to the team. Trend data are downloaded after surgery. The device clock and the anaesthesia monitor clock are synchronised at the start of each operating day against a single reference and the residual offset is recorded for every case.

The baseline is the mean rSO2 over a two-minute stable awake period immediately before cross-clamping. The primary index measure is the relative decline in ipsilateral rSO2 from this baseline to the nadir within the first three minutes after cross-clamping, computed over an identical time window in every participant, or until shunt insertion if this occurs earlier.

Pre-specified thresholds. Primary: relative decline of 20% or more in ipsilateral rSO2. Secondary: relative decline of 12% or more; relative decline of 10% or more; absolute decline of 10 percentage points or more; absolute value below 50%. A data-derived optimal threshold based on the Youden index will be reported as an exploratory analysis with bootstrap correction for optimism.

Reference standard. A standardised awake neurological test comprising verbal responsiveness, naming, contralateral hand grip graded by the Medical Research Council scale, continuous squeezing of an audible device, and level of consciousness. It is performed by a trained anaesthetist following a written script who is masked to all rSO2 data, at baseline and at 30 seconds and 1, 2 and 3 minutes after cross-clamping, and every 2 minutes thereafter until declamping. A positive test is any new focal deficit, dysarthria, aphasia, unresponsiveness or seizure. Shunt placement is determined solely by this test. Inter-observer agreement is assessed independently by two observers in the first 20 participants.

Prognostic exposure. Cumulative cerebral desaturation burden is defined as the area under the ipsilateral rSO2 curve below the line corresponding to a 20% relative decline from baseline, over the whole cross-clamp period, expressed in percentage-minutes and computed from the raw trend data. The area-under-the-curve index reported by the oximeter itself, which quantifies depth and duration below a user-defined low alarm limit fixed at 50% for all participants and read at clamping and at declamping, is recorded as a secondary, device-derived measure; agreement between the two measures will be reported.

Exploratory measures. Where available from the device, changes in oxygenated haemoglobin, deoxygenated haemoglobin and total haemoglobin concentration indices, and the difference between rSO2 and peripheral oxygen saturation, will be recorded and analysed as exploratory measures of the components contributing to cross-clamp desaturation. The contralateral-corrected index, defined as the ipsilateral relative decline minus the contralateral relative decline, will also be examined.

Postoperative cognitive dysfunction is defined as a fall of 2 or more points in the Montreal Cognitive Assessment total score between baseline and one month, with alternate versions of the instrument used at the two time points to limit practice effect. No non-surgical comparison group is enrolled, and reliable change indices are therefore not derived; this is acknowledged as a limitation of the cognitive outcome.

Covariates. Mean arterial pressure, heart rate, peripheral oxygen saturation, end-tidal carbon dioxide, respiratory rate, temperature, vasopressor administration, preoperative haemoglobin, degree of ipsilateral and contralateral stenosis by NASCET criteria, contralateral occlusion, completeness of the circle of Willis on computed tomography or magnetic resonance angiography read by a radiologist masked to rSO2, previous stroke or transient ischaemic attack, diabetes, hypertension, smoking, American Society of Anesthesiologists physical status, baseline Montreal Cognitive Assessment score, clamp duration, stump pressure where measured, and surgeon.

Statistical analysis. Diagnostic accuracy at each pre-specified threshold will be reported as sensitivity, specificity, predictive values, likelihood ratios and overall accuracy with 95% Wilson confidence intervals, together with the area under the receiver operating characteristic curve and its 95% confidence interval. The prognostic analysis will treat desaturation burden as a continuous exposure in a single pre-specified model containing burden, age and previous stroke or transient ischaemic attack, estimated by modified Poisson regression with robust standard errors to yield adjusted risk ratios; no stepwise selection will be used. Pre-specified sensitivity analyses will exclude participants converted to general anaesthesia, participants who received rescue sedation, and participants with less than 90 per cent valid oximetry signal during clamping. Two-sided alpha is 0.05. Secondary and exploratory outcomes are labelled as such and will not be interpreted as confirmatory.

Sample size. The target of 150 participants in the surgical cohort is based on estimating sensitivity with a pre-specified precision, assuming an expected sensitivity of 0.85 for the primary threshold, a half-width of 0.13 and an anticipated proportion of positive awake tests of 25 per cent, requiring 29 positive tests, with inflation for anticipated attrition; and on the requirement for at least 27 events of the composite neurological outcome for the pre-specified three-variable prognostic model. A blinded sample size re-estimation will be performed after 75 participants: the overall proportion of positive awake tests will be recalculated without unmasking any rSO2 data and without any comparative analysis, and the target sample size revised if necessary to retain at least 30 positive tests.