Plasma miRNome Profiling Reveals Subphenotype-Specific Mortality Prediction Models in Elderly COVID-19 Patients

Le résumé fourni par la source

Background: Elderly COVID-19 patients, who exhibit considerable clinical heterogeneity, present a high mortality risk. Circulating microRNAs (miRNAs) are promising biomarkers for predicting outcomes, but their prognostic value in this population remains unclear. Aim: To assess whether miRNA profiling provides prognostic information in elderly COVID-19 patients. Methods: Multicenter study including hospitalized COVID-19 patients aged ≥ 65 years (n=763, COVID-Ponent, NCT04824677 ). Subphenotypes were identified using sociodemographic factors, clinical and infection-timing data via elbow criteria-based cluster analysis. Plasma miRNome was profiled using qPCR. VSURF was applied to develop clinical-molecular models for 90-day mortality prediction. Functional analyses were performed using GO, Reactome, and KEGG databases. Results: The mean age was 79 years, with 44.0% women. The 90-day mortality rate was 26.7%. Thirteen candidate miRNAs were identified during the screening phase (n=39), but no association with 90-day mortality was found in the derivation cohort (n=340). Three subphenotypes (eCOVID-1, -2 and -3) with distinct clinical features and mortality risks (22.0%-34.2%) were identified. Differential miRNA expression patterns and molecular mechanisms were found across subphenotypes. In eCOVID-2 (eldest, most comorbid) miR-106b-3p was the strongest predictor, selected in a VSURF model with PaO₂/FiO₂, age, and CRP. Metabolic and immune processes were linked to the fatal outcome in eCOVID-2. Conclusions: Plasma miRNAs provide complementary prognostic value when combined with clinical variables in specific elderly COVID-19 subphenotypes.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.