[Phenotype based precision typing for risk stratification and prognostic evaluation in pediatric sepsis induced acute kidney injury].

Objective: To identify distinct clinical subphenotypes of pediatric sepsis-induced acute kidney injury (SI-AKI) based on clinical characteristics and evaluate the role of this classification system in predicting clinical outcomes. Methods: A retrospective cohort study was conducted. Clinical data were collected from 484 children who were diagnosed with SI-AKI in Children's Hospital of Nanjing Medical University from November 2017 to October 2024. The cohort was randomly divided into a training set and a validation set at a 7∶3 ratio using stratified sampling. A consensus clustering algorithm was employed in an unsupervised manner to derive SI-AKI phenotypes in the training set, and the resulting model was applied to the validation set for performance evaluation. The Kruskal-Wallis H test and the χ² test were used to compare differences in clinical characteristics, organ dysfunction and prognosis among the subtypes. Kaplan-Meier survival analysis and the Log-rank test were used to compare differences in mortality risk among the distinct phenotypes. Results: A total of 484 children with SI-AKI (267 males and 217 females), aged 42 (8, 109) months, were included in this study. The cohort was divided into a training set (339 cases) and a validation set (145 cases). Based on the training set, 3 clinical subphenotypes were identified. The model demonstrated highly consistent phenotypic distribution when applied to the verification set. The overall population comprised the stable type (184 cases), simple shock type (203 cases), and multi-organ dysfunction (MODS) type (97 cases). Statistically significant differences were observed among the 3 phenotypes in terms of age, BMI, Phoenix sepsis score (PSS), heart rate, respiratory rate, and prothrombin time (all P<0.001). The 28 d mortality rate of the MODS type was significantly higher than that of the simple shock type and the stable type (58% (56/97) vs. 14% (28/203) vs. 16% (29/184), respectively; χ²=80.35, P<0.001). Furthermore, the rates of invasive mechanical ventilation and CRRT use were highest in the MODS type (invasive mechanical ventilation: 95% (92/97) vs. 63% (128/203) vs. 59% (109/184), χ²=40.88, P<0.001; CRRT: 30% (29/97) vs. 12% (24/203) vs. 19% (35/184), χ²=12.56, P<0.001). Conclusions: Three distinct SI-AKI phenotypes with different clinical characteristics and prognoses were identified and validated. Among them, the multi-organ dysfunction type is associated with the most severe clinical presentation and the poorest prognosis.