The benefit of deferred carotid revascularization in patients with moderate-severe disabling cerebral ischemic stroke.

ABSTRACT

OBJECTIVE: Symptomatic carotid artery stenosis needs revascularization within 2 weeks by carotid endarterectomy (CEA) to reduce the risk of symptom recurrence; however, the optimal timing of intervention is yet to be defined in patients with large-volume cerebral ischemic lesion (LVCIL) and modified Rankin scale (mRS) score ≥3. The aim of this study was to determine the most appropriate timing for CEA in patients with a recent stroke and LVCIL. METHODS: Data from patients with symptomatic carotid stenosis with LVCIL and mRS score of 3 or 4 from 2007 to 2017 were considered. Patients were submitted to CEA if they had a stable clinical condition and life expectancy >1 year. LVCIL was defined as a cerebral ischemic lesion of volume >4000 mm3. Perioperative stroke and death were evaluated by stratifying for timing of CEA by χ2 test and multiple logistic regression. Patients with similar characteristics (LVCIL and mRS score of 3 or 4) unfit for CEA served as the control group for recurrence of stroke at 1-year follow-up. RESULTS: In an 11-year period, of a total 4020 CEAs, 126 (2.9%) were performed in patients with a moderate stroke and LVCIL occurring in the same admission. The patients' median age was 69 years (interquartile range [IQR], 10 years); 72% (91) were male, with mRS score of 3 (IQR, 1) and LVCIL volume of 20,000 mm3 (IQR, 47,000 mm3). The median time elapsed from symptoms to CEA was 7 weeks (IQR, 8 weeks). Overall perioperative stroke/death was 7.3% (eight strokes and one death). By selective timing evaluation of the postoperative events, CEA performed within 4 weeks was associated with a significantly higher rate of stroke/death compared with patients operated on after 4 weeks 11.9% (8/67) vs 1.7% (1/59; P = .03). By logistic regression, CEA within 4 weeks was an independent (from sex, cerebral ischemic lesion volume, dyslipidemia, and carotid stenosis) predictor of postoperative stroke/death (odds ratio, 8.2; 95% confidence interval, 1.01-73). In the same period, 101 patients were considered unfit for CEA for dementia (n = 22), severe comorbidities (n = 55), or short (<1-year) life expectancy (n = 24), and 43 (43%) survived at 1 year. At 1 year, the perioperative/recurrent stroke after CEA vs patients unfit for CEA was similar (6.2% vs 13.9%; P = .11), but CEA performed after 4 weeks led to significantly lower perioperative/recurrent stroke (1.7% vs 13.9%; P = .02). CONCLUSIONS: The surgical risk of CEA in patients with a recent moderate-severe ischemic stroke and LVCIL is high. However, if the intervention is delayed >4 weeks, its benefit seems significant.