268 Phenotypic Clusters and Prognostic indicators in Ruptured Brain Arteriovenous Malformations: A Machine Learning Analysis

Le résumé fourni par la source

INTRODUCTION: Ruptured brain arteriovenous malformations (AVMs) present with significant clinical heterogeneity, making patient management challenging. Current methods lack the precision needed for effective patient stratification. This study employs k-prototype clustering, recursive partitioning, and t-distributed Stochastic Neighbor Embedding (t-SNE) to classify patients with ruptured AVMs into clinically relevant subgroups and identify key prognostic indicators for optimal long-term outcomes. METHODS: We retrospectively assessed a single-center database of adult and pediatric patients with catheter angiogram-confirmed ruptured AVMs. K-prototype clustering delineated distinct patient subgroups. Recursive partitioning analysis identified critical prognostic thresholds associated with optimal outcomes, defined as a modified Rankin Scale score of ≤ 2 at last follow-up. t-SNE plots provided dimensionality reduction for visual validation of these subgroups. RESULTS: The study included 418 patients with ruptured AVMs, followed for a mean of 74 months. Two primary clusters emerged: Cluster 1 (optimal outcomes) and Cluster 2 (suboptimal outcomes). Cluster 1 (n=255) comprised younger patients (mean age 17.9 years), predominantly female (53.7%), with larger AVMs (mean 2.55 cm), lower mean Spetzler-Martin Supplementary (SMSup) scores (4.48), and a higher incidence of deep venous drainage (64.3%), though less frequent initial neurologic deficits (49.4%). Cluster 2 (n=163) consisted of older patients (mean age 50.6 years), predominantly male (58.3%), with smaller AVMs (mean 2.20 cm), higher mean SMSup scores (5.29), and more frequent neurologic deficits at presentation (56.4%). Optimal outcomes were significantly more prevalent in Cluster 1 (86.3% vs. 67.5%). Recursive partitioning identified age as a critical prognostic factor, with optimal outcomes associated with patients younger than 41 years and unfavorable outcomes more likely in those older than 68 years. CONCLUSIONS: This analysis identified two distinct subgroups of ruptured AVM patients. Younger age, larger AVMs with deep venous drainage, and lower SMSup scores were associated with more favorable long-term outcomes. These findings aid understanding of outcomes and therapeutic decision-making.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.