Development and validation of a nomogram based on common biochemical indicators for survival prediction of children with high-risk neuroblastoma: A valuable tool for resource-limited hospitals

Abstract

Abstract Background Despite multiple attempts have been made to develop risk stratification within high-risk neuroblastoma (NB) patients (age of diagnosis ≥ 18 month-old with metastatic NB), the definition of “ultra high-risk NB” is still lack of consensus, and indicators for identifying this subgroup are still unclear. This study aimed to develop a nomogram based on easy-to-obtain blood-derived biofactors for identifying ultra high-risk NB patients with highest risk of death within 3 or 5 years. Methods One hundred sixty-seven NB patients who treated at Sun Yat-sen University Cancer Center between 2015 and 2023 were recruited and clustered randomly into training and validation cohorts (116 and 51 cases, respectively). Univariate and multivariate Cox analysis were performed in training set to screen independent prognostic indicators for constructing nomogram model of predicting 1-, 3- and 5-year overall survival (OS). The discrimination power of the nomogram in training and validation sets were assessed by concordance index (C-index) and calibration plot. Based on the risk score obtained from nomogram model, the prognostic accuracy of 1-, 3- and 5-year OS rates in training and validation cohorts were further evaluated using the area under receiver operating characteristic (ROC) curves (AUC). Results Through univariate and multivariate Cox analysis, independent prognostic indicators, including serum lactate dehydrogenase (LDH) and albumin (ALB), were identified in training set, and used to establish a nomogram model. The model showed good discrimination power with C-index in training cohort being 0.706 (95%CI: 0.633—0.788). According to the cut-point calculated based on the established nomogram, patients with a nomogram score > 34 points could be stratified to ultra high-risk NB subgroup, and this subgroup had poorer OS than those in non-ultra one (p < 0.001). AUC values of ROC curves for 3- and 5-year OS rates in the training set were 0.758 and 0.756, respectively. Moreover, based on the cut-point score (34 points) developed in training set, The model also showed good discrimination power with C-index of 0.773 (95%CI: 0.664—0.897) and powerful prognostic accuracy of AUC for 3- and 5-year OS rates being 0.825 and 0.826, respectively, in validation cohort. Conclusions We developed a simple-to-use nomogram based on common laboratory indicators to identify the subgroup of ultra high-risk NB before treatment, providing these children even from developing countries or regions access to intensified multimodal treatments earlier and thus improving their long-term outcome.