Identification and experimental validation of diagnostic and prognostic genes CX3CR1, PID1 and PTGDS in sepsis and ARDS using bulk and single-cell transcriptomic analysis and machine learning.
Summary
Across multiple public datasets, WGCNA and machine learning converged on CX3CR1, PID1, and PTGDS as shared diagnostic/prognostic genes for sepsis and ARDS, with external validation and experimental confirmation (RT-qPCR, H&E). Immune infiltration and single-cell analyses mapped cell-type specificity and suggested drug candidates.
Key Findings
- Identified 242 shared DEGs between sepsis and ARDS; modules linked to poor prognosis and ARDS
- Three key genes (CX3CR1, PID1, PTGDS) selected by WGCNA and ML (LASSO, RF, Boruta) showed high AUCs and external validation
- Single-cell and immune infiltration analyses mapped gene localization; RT-qPCR and H&E confirmed differential expression in PBMCs and mouse models
Clinical Implications
The three-gene panel could inform diagnostic scoring and prognostic assessment once prospectively validated; predicted drugs offer a starting point for repurposing studies.
Why It Matters
Provides a reproducible, multi-omic framework identifying tractable biomarkers for sepsis–ARDS, enabling risk stratification and hypothesis-driven target discovery.
Limitations
- Retrospective datasets with potential batch effects; sample sizes per cohort not detailed here
- Clinical implementation requires prospective validation and standardized assays
Future Directions
Prospective, multicenter validation with standardized platforms; functional studies of CX3CR1, PID1, PTGDS and testing predicted drugs.
Study Information
- Study Type
- Case-control
- Research Domain
- Diagnosis
- Evidence Level
- III - Retrospective case-control analyses with external validation and experimental confirmation
- Study Design
- OTHER