Background: The management of the respiratory post-acute sequelae in critical COVID-19 constitutes an ongoing medical challenge. MicroRNAs (miRNAs) inform on the mechanistic pathways driving diverse pathological conditions and constitute useful biomarkers when combined with clinical data.

Aim: To integrate miRNA profiling performed during ICU stay with clinical predictors of respiratory sequelae in critical COVID-19.

Methods: Multicenter study including 491 critically ill patients from the CIBERESUCICOVID trial (NCT04457505). Lung diffusion impairment was defined as a diffusing capacity for carbon monoxide (DLCO) <80% after hospital discharge (2-12 months follow-up). Sixteen miRNAs, previously associated with COVID-19 severity by our group, were quantified using qPCR in plasma samples collected within the first 48 hours upon ICU admission. Stepwise logistic regression models were used to integrate candidate miRNAs with a clinical model previously constructed by our group.

Results: Ninety patients from 16 hospitals completed the follow-up. No differences were observed between survivors with and without follow-up. miR-16-5p, miR-148a-3p and miR-323a-3p were selected in the multivariable model. The incorporation of miRNAs to the clinical model significantly improved the discrimination (AUC from 0.78 to 0.89) and classification (IDI of 0.208 and a NRI of 0.876) of the sequelae. Selected miRNAs target metabolic pathways (FoxO and PI3K-Akt) in enrichment analyses.

Conclusions: Circulating miRNAs emerge as biomarkers for the early prediction of respiratory sequelae after severe COVID-19 infection and provide possible therapeutic options.