Multidomain perioperative analysis of neurological deficits following adult Moyamoya revascularisation: A retrospective cohort study.

BACKGROUND: Moyamoya disease (MMD) presents distinctive perioperative challenges owing to impaired cerebrovascular autoregulation and exhausted vascular reserve. Perioperative predictors of neurological deficit remain undercharacterized. METHODS: This single-center retrospective cohort study analyzed 24 consecutive MMD patients undergoing revascularization (2017-2022). A multidomain dataset encompassing Suzuki stage, surgical technique, intraoperative hemodynamics, perioperative laboratories, and anesthetic variables was analyzed using ROC analysis and exploratory penalized logistic regression with leave-one-out cross-validation. RESULTS: Thirty-nine procedures were performed (23 Stage I, 16 Stage II; all primary MMD). Stage I comprised direct STA-MCA bypass (74%), combined (17%), and indirect techniques (8.6%); Suzuki stage III predominated (46%). Neurological deficits occurred in 14 of 23 patients (61%) after Stage I: 8 (57%) resolved before discharge, 4 (29%) persisted, and 2 (14%) were fatal; rates fell to 31% at Stage II. Neither surgical technique nor Suzuki stage predicted deficits (P = 1.000; P = 0.368). Anesthesia duration was the strongest discriminator (AUC 0.73; threshold >480 minutes), followed by preoperative hemoglobin (AUC 0.68; <14.3 g dl-1), though neither reached univariate significance. The two-variable model achieved leave-one-out AUC 0.64 (95% CI 0.39-0.86); adding disease-severity covariates reduced discrimination. EtCO2 variability correlated with functional outcome (P = 0.031) and deficits with ICU stay >2 days (P = 0.040). CONCLUSIONS: Longer anesthesia duration and lower preoperative hemoglobin were the most consistent associates of neurological deficit; surgical technique and disease severity were not. Both are potentially modifiable perioperative targets, but associations are hypothesis-generating and require prospective multicenter validation.