The prognostic evaluation value of lung ultrasound score combined with serum miR-224-5p in acute respiratory distress syndrome

This study investigated whether the combined application of the lung ultrasound score (LUS score) and serum miR-224-5p enhances the prognostic prediction accuracy in patients with acute respiratory distress syndrome (ARDS). Patients with ARDS were stratified into the survivor and non-survivor groups according to their 28-day survival status post-diagnosis. Logistic regression analysis was performed to identify prognostic risk factors. Receiver operating characteristic (ROC) curve was applied to evaluate the predictive efficacy of two-variable combined model (LUS score + serum miR-224-5p). In vitro, BEAS-2B cells were treated with LPS. Cell viability was evaluated by the CCK-8 assay. Caspase-3 activity assay, MDA content measurement, and SOD activity assay were conducted to evaluate cell apoptosis and oxidative stress. The transcriptional and secretory levels of inflammatory cytokines (IL-6/IL-8/IL-1β) were determined by RT-qPCR and ELISA, respectively. An elevated LUS score was identified as a risk factor for unfavorable prognosis, while miR-224-5p served as an independent protective factor. Integrating the LUS score with miR-224-5p markedly improved prognostic predictive accuracy, yielding a combined model with an AUC of 0.8924 (95%CI: 0.8370–0.9479), 83.33% sensitivity and 84.78% specificity. In cells induced by LPS, miR-224-5p significantly enhanced cell viability, increased the activity of SOD, and reduced the synthesis of MDA and inflammatory factors. As the target of miR-224-5p, SETD7 (SET domain-containing lysine methyltransferase 7) weakened the protective effect of miR-224-5p. The combination of LUS score and miR-224-5p can effectively improve the accuracy of prognostic assessment in patients with ARDS. miR-224-5p mitigates LPS-induced bronchial epithelial cell injury via targeted regulation of SETD7. However, given that this was a single-center retrospective study, the prognostic model requires validation in multicenter cohorts.