Collaborative Pediatric Critical Care Research Network, n/a
.
National Institute of Child Health and Human Development
Ethnicity, United States
Disclosure information not submitted.
Zhenjiang Fan, BS
Mr
School of Computer Sciences, University of PIttsburgh, United States
Disclosure information not submitted.
Joseph Carcillo, MD
Professor
Children's Hospital of Pittsburgh of UPMC, United States
Disclosure information not submitted.
Title: Four computable 24-hour pediatric sepsis phenotypes have different inflammation profiles
INTRODUCTION: Thrombocytopenia Associated Multiple Organ Failure and hyperinflammatory Macrophage Activation Syndrome are predominate causes of late pediatric sepsis mortality. Though treatable with anti-inflammatory regimens including methylprednisolone, intravenous immune globulin (IVIG), and plasma exchange, these conditions are often missed or diagnosed late. Using machine learning we derive computable 24-hour sepsis phenotypes to facilitate enrollment in early precise anti-inflammatory trials to reduce late mortality from these two conditions.
METHODS: Computable phenotypes were derived at 24 hours among 404 septic children from 9 intensive care units using consensus k-means clustering of twenty-five bedside clinical and laboratory variables including C-reactive protein and ferritin.
RESULTS: Machine learning identified four distinct sepsis phenotypes (PedSep-A, B, C, and D). On average, PedSep-A had the least inflammation (C-reactive protein 7.3 mg/dL, ferritin 125 ng/mL), younger age, and respiratory failure (n = 135; 2% mortality); PedSep-B had more inflammation (C-reactive protein 13.2 mg/dL, ferritin 225 ng/ mL), and multiple organ failure with respiratory failure, shock, and a Glasgow Coma Scale score < 7 (n = 102, 12% mortality); PedSep-C had elevated ferritin (C-reactive protein 15.2 mg/dL, ferritin 405 ng/mL), lymphopenia, and shock (n = 110; mortality 10%); and PedSep-D has hyperferritinemia (C-reactive protein 13.1 mg/dL ferritin 610 ng/mL), thrombocytopenia, and multiple organ failure manifest as respiratory, cardiovascular, hepatic, renal, hematologic, and/or central nervous system failures (n = 56, 34% mortality). Patients in PedSep-D were most likely to develop Thrombocytopenia Associated Multiple Organ Failure (Odds Ratio 37.67 [95% CI 16.19-87.61], p < 0.0001) and Macrophage Activation Syndrome (Odds Ratio 37.24 [95% CI 13.10-105.86], p < 0.0001). PedSep-D patients also showed improved survival odds with combination therapy comprised of methylprednisolone plus IVIG (Multivariate Logistic Regression Odds Ratio 23.31 [95% CI 1.43-760], p = 0.04).
CONCLUSIONS: Membership in PedSep-D identifies an optimal population for early enrollment in pediatric sepsis trials targeting Thrombocytopenia Associated Multiple Organ Failure and Macrophage Activation Syndrome. Funded by R01 GM108618.