Abstract TP026: Stroke Phenotypes and Treatment Response: A Latent Class Analysis of an Acute Stroke Trial

Introduction: The Multi-Arm Optimization of Stroke Thrombolysis (MOST) trial found no benefit of adjunctive argatroban or eptifibatide with intravenous thrombolytic therapy for acute ischemic stroke and reported significantly greater post-stroke disability and higher mortality in the argatroban group. We conducted a secondary analysis to determine whether previously validated latent class analysis (LCA) profiles of the National Institutes of Health Stroke Scale (NIHSS) could account for these worse outcomes. Methods: We analyzed the trial dataset using LCA probabilities derived from individual NIHSS item responses. Outcomes of interest were good functional outcome (modified Rankin Scale of 0-1) and 90-day mortality. Associations between stroke profiles, treatment assignment, and outcomes were assessed using Firth’s penalized logistic regression adjusted for age. Results: Of 514 trial participants, latent class profiles were successfully assigned to 505 patients, with nine patients being excluded (NIHSS incomplete). After adjustment for LCA profiles, treatment with argatroban was not significantly associated with worse functional outcome (OR = 0.54, 95% CI: 0.26, 1.04) but remained significantly associated with higher mortality (OR: 3.58; 95% CI: 1.50–8.45) compared to placebo. Conclusion: Adjustment for stroke phenotypes attenuated the relationship between argatroban and functional outcomes, eliminating statistical significance. However, stroke profiles did not account for the excess mortality observed in the argatroban arm. These findings suggest that stratified stroke phenotypes may help explain observed differences in functional recovery but are insufficient to explain mortality risk. Incorporating such profiles into future stroke trials may improve risk stratification and guide evaluation of treatment effects.