Sleep Apnea and Cardiac Phenotypes

> Research explainer: This briefing examines verified primary research published 53 days before the briefing date. It is not a same-day research update and does not provide medical advice.

Evidence

This research explainer examines a secondary observational analysis of 435 participants from the RICCADSA cohort. Participants had established coronary artery disease, prior revascularization, preserved left ventricular ejection fraction, and available echocardiographic, sleep, and biomarker data. The analysis evaluated correlates of a study-defined HFpEF-like phenotype, with particular attention to obstructive sleep apnea (OSA), age, and obesity. [pmid:42370761]

The study classified 69.9% of the 435 participants as having the HFpEF-like phenotype. This designation required at least two findings from a five-item composite: elevated filling pressure, left atrial enlargement, increased left ventricular mass index, elevated pulmonary artery systolic pressure, or elevated NT-proBNP. The thresholds specified in the report included E/e′ of at least 15 for elevated filling pressure, pulmonary artery systolic pressure of at least 35 mmHg, and NT-proBNP of at least 125 pg/mL. [pmid:42370761]

OSA was defined categorically as an apnea-hypopnea index (AHI) of at least 15 events per hour. The investigators also assessed AHI and the oxygen desaturation index (ODI) as continuous measures. Their multivariable models included age, sex, obesity, hypertension, diabetes, and OSA status; additional models tested continuous sleep measures and the consequence of excluding body mass index (BMI). [pmid:42370761]

In adjusted analyses, age and obesity were independently associated with the HFpEF-like phenotype. By contrast, categorical OSA was not independently associated with the phenotype after adjustment. Continuous AHI and ODI were also not independently associated in models that included BMI. [pmid:42370761]

A different pattern appeared when BMI was excluded: higher AHI and ODI were significant predictors in those models. The reported odds ratio was 1.019 per unit of AHI (95% confidence interval 1.005–1.033) and 1.030 per unit of ODI (95% confidence interval 1.010–1.050). These estimates describe associations within the specified statistical models; they do not establish a causal effect of sleep-disordered breathing on cardiac structure or function. [pmid:42370761]

Analysis — Obesity-Sensitive Associations

The central interpretive feature is adjustment sensitivity. In this cohort, categorical OSA did not retain an independent association after accounting for the listed covariates, and continuous AHI and ODI likewise were not independently associated when BMI was included. Yet both continuous sleep measures were statistically significant after BMI was removed. [pmid:42370761]

That pattern supports a bounded reading: obesity was an important correlate of the HFpEF-like composite and materially affected how the association between OSA severity and the composite appeared in the models. It does not settle whether obesity is a confounder, a mediator, a shared determinant, or some combination of these roles. The supplied study reports regression results, not an intervention designed to separate those possibilities. [pmid:42370761]

The distinction between categorical OSA and continuous severity metrics also matters. A threshold of AHI at least 15 events per hour did not independently identify the phenotype after adjustment, whereas continuously measured AHI and ODI were associated only in the BMI-excluded models. Therefore, the evidence does not support treating the presence of categorical OSA and the degree of sleep-disordered breathing as interchangeable analytical questions. [pmid:42370761]

The reported prevalence should also be read precisely. It is the prevalence of a study-defined HFpEF-like phenotype among selected, revascularized coronary artery disease participants with preserved ejection fraction—not a prevalence estimate for HFpEF in the general population and not a clinical diagnosis assigned by the composite alone. [pmid:42370761]

Taken together, the analysis places age and adiposity on firmer footing as adjusted correlates in this dataset. The sleep-apnea-severity signal is informative because it changes with BMI adjustment, but its model dependence limits causal interpretation. [pmid:42370761]

Limitations

This was a secondary observational analysis, so its associations cannot demonstrate causation. The cohort was drawn from a single-center Swedish study and was restricted to people with established, revascularized coronary artery disease, preserved ejection fraction, no atrial fibrillation history, and the required sleep, echocardiographic, and biomarker data. Those selection features limit generalization beyond this population. [pmid:42370761]

The outcome was a composite HFpEF-like classification rather than, by itself, a clinical diagnosis of HFpEF. Its components and thresholds shaped who met the definition. Finally, the AHI and ODI findings depended on whether BMI was included in the models. That dependency is a reason to avoid interpreting the results as proof that OSA independently causes cardiac remodeling or the study’s HFpEF-like phenotype. [pmid:42370761]