Molecular subtyping and immune microenvironment heterogeneity in pediatric influenza-associated prolonged multiple organ dysfunction syndrome.

Background: Pediatric influenza infections that progress to prolonged multiple organ dysfunction syndrome (PMODS) carry a high mortality rate. The underlying molecular heterogeneity, particularly involving dysregulated autophagy pathways and the immune microenvironment, remains poorly characterized, hindering the development of targeted interventions. This study aimed to integrate transcriptomic profiling and machine learning to dissect autophagy-related gene (ARG) dysregulation, characterize the immune microenvironment, and identify clinical biomarkers associated with PMODS severity. Methods: We analyzed the publicly available transcriptomic dataset GSE236877, comprising 191 pediatric samples from influenza patients: 38 with PMODS or who died, 27 who recovered from MODS (RM), and 126 who never developed MODS (NM). Differential expression analysis of ARGs was performed. Unsupervised consensus clustering was used to identify molecular subtypes within the PMODS group. Immune cell infiltration was quantified using CIBERSORT. A Random Forest (RF) machine learning algorithm was employed to prioritize key discriminatory genes, whose correlations with clinical parameters were subsequently assessed. Results: Compared to NM samples, PMODS cases exhibited significant upregulation of Conclusions: This study reveals significant molecular heterogeneity within pediatric influenza-associated PMODS, delineating two distinct subtype-specific mechanisms: C1 hyperinflammation