Clin Radiol. 2025 Jun 21. pii: S0009-9260(25)00203-X. [Epub ahead of print]88 106998
AIM: To evaluate the diagnostic performance of a multiparametric diagnostic model integrating intratumoral metabolic heterogeneity parameter and clinical parameters for differentiating neuroblastoma (NB) from ganglioneuroblastoma (GNB).
MATERIALS AND METHODS: This retrospective study included 107 patients (64 NB and 43 GNB) who underwent 2-[18F]-fluoro-D-glucose positron emission tomography/computed tomography (2-[18F]-FDG PET/CT). Baseline characteristics, clinical data, and metabolic parameters of the primary lesions, including peak standardised uptake value, maximum standardised uptake value, mean standardised uptake value, tumour metabolic volume, total lesion glycolysis, and the area under the curve of a cumulative standardised uptake value-volume histogram (AUC-CSH) index, were collected and analysed. Diagnostic performance of the model and predictors was assessed using area under the curve of the receiver operating characteristic curve, with integrated discriminatory improvement, net reclassification improvement, Delong test for performance improvement, and decision curve analysis for clinical utility evaluation.
RESULTS: Among the metabolic parameters, the AUC-CSH index demonstrated the highest diagnostic performance. Multivariate analysis identified the AUC-CSH index, age, and serum neuron-specific enolase level as independent predictors. The multiparametric model integrating these factors significantly outperformed individual metabolic parameters, and its clinical utility was validated by decision curve analysis.
CONCLUSION: A multiparametric diagnostic model integrating intratumoral metabolic heterogeneity parameter derived from 2-[18F]-FDG PET/CT with clinical parameters improves diagnostic performance for differentiating NB from GNB, offering potential for clinical application.