Multi-scale analysis of the HIF1A hypoxia-inflammation axis in severe COVID-19

Objective and design: To test whether the hypoxia-inducible factor 1-alpha (HIF1A) axis links transcriptional, cellular, biomarker and host-genetic features of severe COVID-19, in a retrospective multi-scale observational study of public cohorts and one hospital cohort. Material or subjects: 35,268 hospitalized patients (Mexican national registry); 1,100 samples from five bulk RNA-seq cohorts; 25,773 peripheral neutrophils from six individuals; 326 hospitalized patients (Sao Paulo, Brazil); 93 hospitalized patients (University Clinical Center Kragujevac, Serbia). Treatment: None. Methods: Within-cohort effect sizes for a 10-gene HIF1A panel; single-cell pathway scoring; logistic regression and gradient boosting with SHAP; TaqMan genotyping of HIF1A rs11549465 and rs41508050; cross-validation with bootstrap confidence intervals. Results: HIF1A (mean Cohen d +0.39) and LDHA (+0.76) were higher in severe disease in all five cohorts. Neutrophils of all three patients who failed extubation scored higher than those of both weaned patients and over-expressed S100A8/A9. C-reactive protein dominated ICU prediction; laboratory variables raised the AUC from 0.636 to 0.794. In Kragujevac, a clinical model predicted severe outcome (AUC 0.84, 95% CI 0.75 to 0.92). rs11549465 CT/TT carriers had lower mortality (1/11 vs 28/78; p = 0.095). Conclusions: HIF1A pathway activation is reproducible across cohorts and localizes to an alarmin-rich neutrophil state; host-genetic effects require larger cohorts.