Data-driven progression trajectories across clinical subtypes of amyotrophic lateral sclerosis

Background Amyotrophic lateral sclerosis (ALS) exhibits substantial clinical heterogeneity; however, conventional classifications based on the predominant site of involvement may not fully capture patterns of disease progression. This study aimed to identify data-driven subtypes of ALS and reconstruct their temporal progression using item-level scores from the ALS Functional Rating Scale–Revised (ALSFRS-R). Methods Baseline data from 3067 participants in the Pooled Resource Open-Access ALS Clinical Trials database were analyzed. Subtype and Stage Inference (SuStaIn) was applied to ALSFRS-R item scores to identify subtypes and infer their progression trajectories. Internal consistency was evaluated using complexity and sensitivity analyses. Longitudinal stability was assessed at 180 and 365 days. Associations with functional decline and survival were examined using adjusted statistical models. Results Three clinically interpretable subtypes were identified, characterized by predominant upper-limb, lower-limb, or bulbar involvement. These subtypes exhibited distinct, consistent progression patterns across the analyses. Subtype stability was moderate short-term but limited longer-term; stage progression was largely monotonic. The bulbar subtype showed more rapid functional decline at both 180 and 365 days. Higher SuStaIn stage was associated with increased mortality. In an exploratory stage-stratified analysis, poorer survival was observed in the bulbar subtype than in the upper-limb subtype within the lower-stage group. Conclusions This study reconstructed ALS subtype-specific functional progression trajectories and inferred disease stage using a widely accessible clinical scale. SuStaIn-derived subtypes characterized patterns of functional heterogeneity, and higher stage was associated with increased mortality. Further prospective comparative validation is needed to establish the prognostic and clinical utility of this framework.