Serum Exosomal Multi-Omic Signatures Stratify Glucose Tolerance in Cystic Fibrosis and Reveal Partial Therapeutic Reprogramming by CFTR Modulators

Cystic fibrosis-related diabetes (CFRD) affects up to 60% of adults with CF and contributes to poorer clinical outcomes, including accelerated lung decline and increased mortality. CFRD is often diagnosed late, with limited mechanistic insight and few tools for early detection. We profiled serum-derived exosomes from 186 individuals, 173 with CF, across two independent cohorts (Australia and Denmark), stratified by oral glucose tolerance test (OGTT) into normal (NGT), impaired (IGT), and CFRD groups. In a longitudinal subset, matched samples were collected before and after CFTR modulator therapy. Exosomes were isolated via size-exclusion chromatography and validated by NTA and TEM. Label-free proteomics and small RNA sequencing were used to profile exosomal cargo. Multi-analyte classifiers were identified using machine learning, with internal cross-validation. Exosomal profiles captured a continuum of metabolic dysfunction, with distinct signatures in CFRD including elevated PTPN1, MYO5A, and VWF (insulin resistance/hepatic dysfunction), and reduced 14-3-3ζ (β-cell dysfunction). miRNA profiles reinforced these trends, with CFRD exosomes enriched in miR-375-3p, miR-122-5p, and miR-1260a/b. CFTR modulator therapy partially reversed proteomic and transcriptomic markers of insulin resistance and hepatic dysfunction but failed to regulate β-cell-associated signatures. Machine learning models achieved high classification performance (AUC = 0.83), identifying robust multi-omic panels predictive of glucose tolerance state. This study provides the first comprehensive exosome-based multi-omics and machine learning framework for CFRD. Our findings show that serum exosomes hold promise as diagnostic and therapeutic biomarkers for early detection and monitoring of CFRD in precision CF care.