"Phenotyping obstructive sleep apnea: a pathway to precision medicine".

PURPOSE: The apnea-hypopnea index (AHI) has long dominated the diagnosis of obstructive sleep apnea (OSA); however, mounting evidence indicates that AHI alone cannot capture the disorder's heterogeneity. This review synthesizes current evidence on polysomnography-based phenotyping and endotyping into a practical framework for precision OSA management.

METHODS

PubMed/MEDLINE and Scopus were searched from inception through December 2025 using terms related to OSA, phenotyping, endotyping, and polysomnographic parameters. Randomized trials, cohort studies, cluster analyses, and machine learning studies reporting phenotype data or phenotype-stratified outcomes were included.

RESULTS

Symptom-based cluster analyses have identified reproducible clinical phenotypes with differential treatment outcomes and cardiovascular risk. The PALM framework identifies non-anatomic endotypic traits amenable to targeted therapies. Multidimensional polysomnographic features, including hypoxic burden, event duration, arousal dynamics, positional dependency, and sleep architecture, provide clinically relevant information beyond AHI. Analytic strategies spanning conventional statistics to machine learning enable phenotype-guided therapy selection. Practical implementation using metrics already available from standard polysomnography is feasible without additional equipment.

CONCLUSIONS

The convergence of polysomnographic characterization with molecular profiling and advanced computational methods may eventually help redefine OSA from a syndrome defined by breathing pauses into mechanistically distinct disorders amenable to targeted interventions, although prospective validation remains necessary.