# Predictive model based on blood cell analysis and coagulation function indicators for neuroblastic tumors staging diagnosis

## TL;DR

This study shows that combining blood cell and coagulation indicators can help predict the stage of neuroblastic tumors in children, improving diagnostic accuracy.

## Contribution

A novel predictive model using blood and coagulation markers for staging neuroblastic tumors and differentiating them from ganglioneuroma.

## Key findings

- Age, RDW-CV, Fib, and Hb were identified as independent predictors of advanced neuroblastic tumor stages.
- A prediction model using age, TT, Mon#, and Hb achieved 78.10% overall accuracy in staging and differentiation.
- ROC analysis showed high diagnostic efficacy with AUCs of 0.867 and 0.941 for key comparisons.

## Abstract

To explore the diagnostic value of integrating blood cell analysis and coagulation function indicators in the staging of neuroblastic tumors, providing a robust basis for clinical decision-making.

A retrospective analysis was conducted on 137 pediatric neuroblastic tumors cases (2017-2024) at the Children’s Hospital Affiliated to Shandong University. Patients were stratified into localized (INSS 1-2, Group 1) and advanced (INSS 3-4, Group 2) stages according to the INSS classification, with mature ganglioneuroma serving as the control group. Univariate and multivariate logistic regression analyses were performed to identify differences in blood cell analysis and coagulation function indicators between groups, complemented by ROC curve analysis to evaluate the efficacy of the models.

The median age of patients with neuroblastic tumor was 23.5 (12–46.75) months (male:female = 1.55:1), which was significantly younger than that of ganglioneuroma patients [72 (53–108) months, p < 0.01]. Multinomial logistic regression identified age, RDW-CV, Fib, and Hb as independent predictors of advanced stages. Older age, higher RDW-CV and Fib levels were positively associated with advanced-stage risk compare to localized stages, while higher Hb showed a negative association. Furthermore, a probability prediction model developed using age, TT, Mon#, and Hb successfully differentiated advanced neuroblastic tumors from ganglioneuroma. The overall accuracy of this prediction model was 78.10%, with specific accuracies of 68.40%, 82.40%, and 80.00% for the localized neuroblastic tumors, advanced neuroblastic tumors, and ganglioneuroma groups, respectively. ROC curves showed AUCs of 0.867 (localized vs. advanced) and 0.941 (advanced vs. ganglioneuroma), indicating high diagnostic efficacy.

The combined analysis of age, RDW-CV, Hb, Mon#, Fib, and TT can effectively assist in the preliminary assessment of whether children with neuroblastic tumors are in an advanced phase or suffering from ganglioneuroma. This method enhances the accuracy and efficiency of clinical diagnosis and serves as a crucial reference for developing disease diagnosis and treatment plans.