Development of an early mortality risk prediction model for pediatric hemophagocytic lymphohistiocytosis

Abstract

Abstract Background Hemophagocytic lymphohistiocytosis (HLH) is a rare but life-threatening disease with rapid progressing and high mortality, which is more commonly seen in children. Objective Our goal was to develop a novel model for predicting early mortality risk in pediatric HLH patients using readily accessible parameters and build a nomogram. Methods We conducted a retrospective analysis of 170 pediatric HLH patients diagnosed at Hunan Children's Hospital between March 1, 2017, and March 1, 2022. These patients were split into a training cohort and a validation cohort. Early mortality was defined as 28-day mortality post-diagnosis. A prediction model with nomogram was developed using binary logistic regression analysis in the training cohort. The model underwent internal and external validation using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis (DCA). Results The final prediction model included 11 predictor variables: glutamic-pyruvic transaminase, albumin, globulin, myohemoglobin, creatine kinase, serum potassium, procalcitonin, serum ferritin, the interval between onset and diagnosis, and the interval between admission and diagnosis. The 28-day mortality prediction AUC was 0.957 in the training cohort and 0.929 in the validation cohort. Utilizing the 28-day mortality prediction for estimating 7-day and 14-day mortality, the AUC values were 0.930 and 0.938, respectively. The calibration plot revealed an adequate fit with 1000 bootstrap resampling and the DCA exhibited great net benefit. Conclusion The study constructed a novel prediction model with nomogram in pediatric HLH, which could contribute to rapid assessment early mortality risk after diagnosis with readily available parameters, providing clinical support to identify patients with a poor prognosis and enhancing their prognostic outcomes.