Early inflammatory profiling identifies patients at risk for persistent pulmonary dysfunction after COVID-19: a cohort study

Background The lack of robust evidence regarding long-term pulmonary outcomes after COVID-19 hampers the development of effective, evidence-based follow-up strategies for affected patients. Objective To prospectively identify patients at risk of persistent pulmonary impairment after COVID-19 based on clinical and inflammatory profiles, aiming to support individualized follow-up strategies and precision medicine approaches. Methods In this prospective cohort study, adults hospitalized with COVID-19 underwent inflammatory profiling within 24 hours of ICU admission using a 65-plex biomarker panel. Unsupervised k-means clustering was applied to identify distinct inflammatory phenotypes. Diffusing capacity for carbon monoxide (DLCO), total lung capacity (TLC), forced expiratory volume in one second (FEV 1 ), maximal inspiratory pressure (MIP), and functional assessments were evaluated at 6 and 12 months. Linear mixed models were used to identify predictors of long-term pulmonary outcomes. Results Two inflammatory phenotypes were identified: high-inflammatory (Th High ) and low-inflammatory (Th Low ). At 12 months, patients in the Th High group exhibited significantly higher DLCO values compared with those in the Th Low group (β = 9.44; 95% CI, 4.36 to 14.52), independently of demographics, comorbidities, and disease severity. In contrast, FEV 1 was significantly lower in the Th High cluster than in the Th Low cluster (β = −4.05; 95% CI, −7.91 to −0.19). No significant associations were observed between inflammatory phenotypes and TLC or MIP. Conclusions Inflammatory phenotypes during acute COVID-19 are associated with distinct patterns of long-term pulmonary recovery, supporting the use of early risk stratification strategies to identify patients at risk for persistent respiratory dysfunction.