Development and validation of a novel nomogram for predicting long-term rebleeding risk among patients with hemorrhagic moyamoya disease: a 10-year multicenter retrospective cohort study

OBJECTIVE The aim of this study was to develop and validate a predictive nomogram model for long-term rebleeding events in patients with hemorrhagic moyamoya disease (HMMD). METHODS In total, 554 patients with HMMD from the Fifth Medical Center of the Chinese PLA General Hospital (5-PLAGH cohort) were included and randomly divided into training (390 patients) and internal validation (164 patients) sets. An independent cohort from the First Medical Center and Eighth Medical Center of Chinese PLA General Hospital (the 1-PLAGH and 8-PLAGH cohort) was used for external validation (133 patients). Univariate Cox regression analysis and least absolute shrinkage and selection operator (LASSO) regression algorithm were used to identify significant factors associated with rebleeding, which were used to develop a nomogram for predicting 5- and 10-year rebleeding. RESULTS Intraventricular hemorrhage was the most common type of cerebral hemorrhage (39.0% of patients in the 5-PLAGH cohort and 42.9% of the 1-PLAGH and 8-PLAGH cohort). During the mean ± SD follow-up period of 10.4 ± 2.9 years, 91 (16.4%) patients had rebleeding events in the 5-PLAGH cohort. The rebleeding rates were 12.3% (68 patients) at 5 years and 14.8% (82 patients) at 10 years. Rebleeding events were observed in 72 patients (14.3%) in the encephaloduroarteriosynangiosis (EDAS) surgery group, whereas 19 patients (37.3%) experienced rebleeding events in the conservative treatment group. This difference was statistically significant (p < 0.001). We selected 4 predictors (age at onset, number of episodes of bleeding, posterior circulation involvement, and EDAS surgery) for nomogram development. The concordance index (C-index) values of the nomograms of the training cohort, internal validation cohort, and the external validation cohort were 0.767 (95% CI 0.704–0.830), 0.814 (95% CI 0.694–0.934), and 0.718 (95% CI 0.661–0.775), respectively. The nomogram at 5 years exhibited a sensitivity of 48.1% and specificity of 87.5%. The positive and negative predictive values were 38.2% and 91.3%, respectively. The nomogram at 10 years exhibited a sensitivity of 47.1% and specificity of 89.1%. The positive and negative predictive values were 48.5% and 88.5%, respectively. CONCLUSIONS EDAS may prevent rebleeding events and improve long-term clinical outcomes in patients with HMMD. The nomogram accurately predicted rebleeding events and assisted clinicians in identifying high-risk patients and devising individual treatments. Simultaneously, comprehensive and ongoing monitoring should be implemented for specific patients with HMMD throughout their entire lifespan.