B20-06 Determinants of Divergent Cardiovascular Outcomes in Obstructive Sleep Apnea: Myocardial Infarction vs. Stroke/Transient Ischemic Attack in the Sleep Apnea Cardiovascular Endpoints (SAVE) Trial

Abstract Rationale Obstructive sleep apnea (OSA) is associated with atherosclerosis and cardiovascular disease (CVD). While prior studies have largely focused on composite cardiovascular outcomes in OSA, this approach may mask the differential association between specific OSA metrics and distinct cardiovascular events (e.g., stroke/TIA vs. MI). Therefore, using data from the Sleep Apnea Cardiovascular Endpoints (SAVE) randomized controlled trial, with adjustment for continuous positive airway pressure (CPAP) assignment, we examined the relationships between individual OSA metrics and differential risks of stroke/TIA versus MI. Methods Investigation was performed using Cox proportional hazards models with mixed effects and predictor*event type interaction terms. Univariate analysis identified polysomnography variables exhibiting differential associations between stroke/TIA compared with MI risk in the SAVE cohort, defined by interaction term p 0.05. Among clusters of correlated variables identified via Pearson correlation coefficient, a clinically representative polysomnography variable was chosen for multivariate analysis. Multivariate analysis was performed with models adjusting for age, sex, body mass index (BMI), prior cardiovascular event, history of diabetes/hypertension, and randomization to CPAP therapy. Sensitivity analysis was performed within the subgroup of patients randomized to no CPAP therapy. Results 2,687 patients from SAVE were included in the analysis, with 1,346 patients randomized to CPAP treatment and 1,341 randomized to usual care (no CPAP). There were a total of 81 MI events and 154 stroke/TIA events. Univariate analysis identified 14 variables associated with significantly differential effects on MI versus Stroke/TIA (Figure 1a). The apnea-hypopnea index (AHI), median desaturation-resaturation time ratio, and time spent with oxygen saturation below 90% (T90%) were selected as representative variables for inclusion in the multivariate analysis. In multivariate analysis, only T90% remained significant for differential effects on the risk of MI versus stroke/TIA (p = 0.005), where every 10% point increase in T90% was associated with 15.7% higher hazard for MI (95% CI = 2.4%, 30.7%, p = 0.019), but not stroke/TIA (95% CI=-21.0, 7.8%, p = 0.31, Figure 1b). Among patients assigned to usual care, sensitivity analysis demonstrated a consistent, though nonsignificant, association between higher T90% and increased MI risk. Conclusions In the SAVE cohort, T90%, but not AHI, demonstrated a significantly differential impact on the risk of MI versus stroke/TIA, with higher T90% levels associated with an increased risk of MI. Overall, these findings underscore the importance of looking beyond both composite cardiovascular outcomes and traditional metrics such as the AHI to better characterize OSA and its distinct cardiovascular consequences, which may reflect differing underlying pathobiological mechanisms. This abstract is funded by: NIH grant(s): 1R01HL168897-01A1; R01Hl175992; 5R01HL153805; 5R01HL143221