C21-07 Tirzepatide Outcome Prediction Using Sleep Apnea Endotyping in SURMOUNT-OSA

Abstract Rationale Obstructive sleep apnea (OSA) is highly prevalent and associated with adverse cardiovascular outcomes. In SURMOUNT-OSA, tirzepatide-induced weight loss in patients with obesity and OSA yielded pronounced improvements in OSA severity (per apnea-hypopnea index, AHI) after 52 weeks and is now approved for clinical use. Given the heterogeneity in OSA pathophysiology, treatment effects are unlikely to be uniform. Identifying physiological subgroups that derive the greatest benefit could inform patient selection, support precision medicine treatment approaches, and accelerate clinical implementation of tirzepatide for OSA. Methods We conducted a pre-specified secondary analysis of SURMOUNT-OSA including 383 participants who provided primary endpoint data (apnea-hypopnea index, AHI) at 52 weeks (210 tirzepatide, 163 placebo); Participants from Study 1 (not on PAP therapy) and Study 2 (on PAP therapy) were pooled for statistical power. Strong response was defined as greater than two-thirds reduction in AHI from baseline (tirzepatide median: 69.5%). Physiological endotypic traits—capturing key OSA mechanisms—were estimated from baseline polysomnography, including upper-airway collapsibility, ventilatory control sensitivity (“loop gain”), pharyngeal dilator muscle compensation, and arousal threshold. For model development, traits from two-thirds of the participants were combined with patient factors (age, body mass index BMI, baseline AHI, very severe OSA AHI90/hr) to predict a strong treatment response at 52 weeks (logistic regression). For validation, in the remaining third of participants, we tested whether tirzepatide (vs placebo) was more effective in the strong-response endotype vs moderate-response endotype subgroups (mixed-model treatment-by-subgroup interaction). Results Participant characteristics: 265M:118F, median IQR age=5041-57 years, BMI=3734-41 kg/m2, AHI=4525-72 events/hr. Using training data, we identified a strong-response endotype (younger, lower BMI, higher loop gain, more severe collapsibility, lower arousal threshold) who showed a 77% AHI reduction from baseline, meaningfully greater than the 53% reduction in the moderate-response endotype. In holdout validation, participants in the strong-response endotype showed a 78% AHI reduction from baseline (52 41, 62 vs. placebo, estimate 95% CI), significantly greater than the 53% reduction observed in the moderate-response endotype (29 9, 44 vs. placebo; difference: 24 5, 38%, P = 0.015), confirming subgroup validity. Although less profound, the moderate-response endotype had clinically-significant mean reductions in AHI. Conclusion A strong-response endotype—characterized by younger age, less severe obesity, greater collapsibility, greater ventilatory control sensitivity—achieved profound OSA improvements with tirzepatide in SURMOUNT-OSA. This represents the first validated endotype-based prediction model in OSA, establishing a potential framework for more personalized data-driven expectations for the implementation of pharmacologic treatment with tirzepatide in patients with OSA. This abstract is funded by: Eli Lilly and Company