Intelligent prognostic prediction in patients with hepatoblastoma: a population-based retrospective study

Abstract

Abstract The study meticulously analyzes pertinent risk variables influencing Hepatoblastoma (HB) prognosis, utilizing data from both the SEER database and a singule-center.Encompassing HB patients from 2000 to 2019 within the SEER database, we partitioned the cohort into modeling and internal validation groups. Following rigorous univariate and multivariate analyses of the SEER group, encompassing risk factors with P < 0.05 and incorporating findings from extant literature, a prognostic risk model was erected and subsequently validated. we pursued a follow-up of HB patients initially diagnosed in Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University from January 2002 to August 2022, serving as an external validation group, and proceeded to authenticate the predictive utility, discriminatory prowess, calibration integrity, and clinical utility of the Nomogram through indices such as C-index, ROC curve, calibration curve, and DCA curve. The external validation ascertained the model's aptitude for broad clinical applicability, culminating in the development of an interactive, web-based Dynamic Nomogram. Within both the SEER database and the external validation group, 835 and 61 patients, respectively, satisfied the inclusion-exclusion criteria for study inclusion. In the SEER group, the 1-, 3-, and 5-year survival rates for HB patients were 86.3%, 80.6%, and 78.0%, respectively, with 671 cases serving as the modeling group and 164 as the internal validation group. In the external validation group, the corresponding rates were 78.60%, 75.00%, and 68.50%. Univariate and multifactorial analyses revealed Age, Stage, Chemotherapy, Surgery, and Lung Metastasis as independent prognostic factors influencing overall survival rate in HB. The C-indexes for the modeling group, internal validation group, and external validation group were 0.796, 0.873, and 0.887, respectively. Calibration curves evinced high concordance, while decision analysis curves underscored the pronounced clinical utility of the presented Nomogram.The study to create a Nomogram for predicting the prognosis of patients with HB was validated to have excellent predictive performance, differentiation, calibration, and clinical benefit to help clinicians in decision making.