External Validation, Molecular Signatures, and Therapeutic Relevance of Pediatric Sepsis-Associated Acute Kidney Injury Subphenotypes

Objective: – Sepsis-associated acute kidney injury (SAKI) is a heterogeneous condition that lacks disease-modifying treatments, and precision medicine approaches are needed. We previously derived two reproducible pediatric SAKI subphenotypes (pSAKI-1 and pSAKI-2) from readily available clinical data. We aimed to externally validate the prognostic relevance of these subphenotypes, evaluate their molecular signatures, and assess for heterogeneity of treatment effect (HTE) across subphenotypes with sepsis therapies. Design: – Secondary analysis of an ongoing multicenter, prospective, observational study of children. Setting: – Ten PICUs in the United States from January 2002 to February 2025. Patients: – Patients 1 week to 18 years old with early (day 1–2) SAKI. Interventions: – None. Measurements and Main Results: – Among 871 patients, 665 (76%) were assigned pSAKI-1 and 206 (24%) to pSAKI-2. On day 1–2, the pSAKI-2 cohort had greater severity of illness, including higher acute kidney injury stage and vasoactive burden, lower platelet counts, and higher lactate values and International Normalized Ratios. These pSAKI-2 patients also had uniformly worse outcomes, including independently higher odds of day 7 severe acute kidney injury (adjusted odds ratio [aOR] 3.2; 95% CI, 2.1–4.7; p < 0.001), death (aOR 2.7; 95% CI, 1.6–4.4; p < 0.001), and fewer PICU-free and vasoactive-free days (p < 0.001). The biomarker signature of pSAKI-2 was characterized by greater inflammation, endothelial dysfunction, and hyperreninemia. On propensity score matched (PSM) analysis, pSAKI-1 patients who received corticosteroids had more day 7 severe acute kidney injury (28% vs. 19%, p = 0.023), 2 fewer PICU-free days (p = 0.04) and greater mortality (10% vs. 3.7%, p = 0.008); no differences were seen in pSAKI-2 patients. Although no HTE was identified on PSM analysis for vasopressin, inverse probability treatment weighting analysis demonstrated a significant interaction between subphenotype-, vasopressin- and vasoactive-free days (p = 0.003). Conclusions: – We externally validated the prognostic relevance of two pSAKI subphenotypes derived from readily available data. These subphenotypes have unique biomarker signatures and differential responses to treatment, representing a potential mechanism for bedside enrichment.