Blood and Urine Transcriptomes Reveal Complementary Host Responses and a Creatinine-Associated Urinary Transcriptional Axis in Sepsis

Background/Objectives: Sepsis is characterized by systemic immune dysregulation and organ dysfunction, but transcriptomic profiles from different biofluids may provide partially overlapping and distinct molecular readouts of the host response. We characterized transcriptomic profiles of whole blood and urinary cellular fractions in sepsis, with particular emphasis on a urinary transcriptional gradient associated with continuous serum creatinine and on whether this association was mirrored in matched blood. Methods: Targeted RNA sequencing was performed on whole blood from 35 patients with sepsis and 10 healthy controls and on urinary cellular fractions from 21 patients with sepsis and 7 controls using the Ion AmpliSeq Transcriptome Human Gene Expression workflow. Differential expression was assessed with DESeq2, with sequencing-adjusted sensitivity models where appropriate. Sensitivity analyses addressed the highest creatinine value, sex and age, and matched blood–urine creatinine-associated effects. Results: Whole-blood analysis identified 2659 differentially expressed genes (DEGs), including a robust core of 1796 genes characterized by neutrophil–myeloid activation and adaptive-immune suppression. The exploratory urine case–control comparison identified 1552 DEGs but was substantially confounded by sequencing chip. Continuous serum creatinine analysis identified 4064 chip-adjusted urinary DEGs, including a robust core of 3774 genes enriched for inflammatory, interferon, hypoxic, autophagic, lysosomal, vesicular, metabolic, and mitochondrial programs. Exclusion of the highest creatinine sample retained 3192/4064 strict DEGs with the same direction and 2926/3774 robust-core genes. In the same 21 patients, adjusted creatinine-associated effects were weakly concordant between blood and urine at the gene level (Pearson r = 0.192; Spearman ρ = 0.199), with five overlapping strict DEGs and no robust-core overlap. Hallmark concordance was greater but moderate (Pearson r = 0.293; Spearman ρ = 0.464), with 14 shared robust pathways. Conclusions: Transcriptomic profiles of whole blood and urinary cellular fractions showed distinct but partially overlapping host-response patterns in sepsis. The serum creatinine-associated urinary transcriptional axis was robust across sensitivity analyses and was only weakly mirrored at the gene level in matched blood, despite partial pathway-level convergence. These findings should be interpreted as serum creatinine-associated transcriptional patterns rather than as evidence of kidney-specific injury or validated diagnostic or prognostic signatures. Independent longitudinal cohorts and orthogonal molecular validation are required.