Comparative Effectiveness of Management by Surgical Resection vs Observation for Cerebral Cavernous Malformations: A Matched Propensity Score Analysis

Abstract

BACKGROUND: Despite increasing incidental identification of cerebral cavernous malformations (CCMs), uncertainty remains about how to best manage CCM patients. OBJECTIVE: To compare the outcomes of patients with surgically resected vs observed CCMs after adjusting for indication bias through propensity score matching. METHODS: Single-center retrospective chart review was performed on individuals treated for CCMs between 1995 and 2016 at Washington University and Barnes-Jewish Hospital. Participants underwent surgical resection (n = 54) or were observed without surgery (n = 203). Propensity score matching was used to match observed to surgical cases on characteristics at first observation or initial surgery. Time to first focal neurological deficit and mortality postinitial visit or postsurgery was estimated using a Cox proportional hazards model with robust sandwich estimators. Differences in the presence of specific follow-up symptoms were estimated using exact logistic regression. RESULTS: After propensity score matching, 34 matched pairs were created. The median (IQR) follow-up duration was 1.5 (0, 5) years for surgical cases and 3 (0, 5) years for observed cases. After matching, no significant differences were found for time to first focal neurological deficit (P = .37, hazard ratio [HR]: 0.76, 95% CI: 0.41, 1.39) or mortality (P = .91, HR: 0.85, 95% CI: 0.06, 12.24). No significant differences were found between the groups for seizure (P = .15, odds ratio: 0.33, 95% CI: 0.06, 1.34) or hemorrhage (P = 1.00, odds ratio 1.00, 95% CI: 0.07, 13.80). CONCLUSION: In this propensity scored–matched analysis, there were no significant differences between surgical or observed cases in time to focal neurological deficit, mortality, recurrent hemorrhage, or seizure.