Abstract
Pediatric hemophagocytic lymphohistiocytosis (HLH) lacks reliable risk models for personalized treatment. We aimed to develop a survival nomogram to address this gap. We retrospectively analyzed 404 pediatric HLH patients from multiple centers. Independent prognostic variables identified by Cox regression were incorporated into a nomogram. Five predictors of mortality were identified: central nervous system involvement, gastrointestinal bleeding, WBC < 2.3 × 10⁹/L, Hb < 80 g/L, and IL-10 > 69pg/mL. The nomogram demonstrated strong predictive accuracy, with validation cohort AUCs of 0.83, 0.78, and 0.79 for 1-, 2-, and 3-year overall survival (OS). Calibration and decision curve analysis confirmed accuracy. Using an X-tile-determined cutoff of 261, patients were stratified into low- and high-risk groups with significantly different 3-year OS (training cohort: 94.0% vs. 47.2%; validation cohort: 88.2% vs. 43.2%; p < 0.0001). This multicenter study established and externally validated a nomogram-based model that enables early risk stratification and may inform individualized therapy for pediatric HLH. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s00277-025-06676-5.