Prediction of MYCN amplification status in neuroblastoma using radiomics: a systematic review and meta-analysis.

Abstract

BackgroundMYCN gene amplification is associated with poor prognosis in neuroblastoma (NB) patients; however, its detection relies on the invasive fluorescence in situ hybridization technique. Radiomics can non-invasively predict MYCN gene amplification by extracting high-dimensional features from medical images.PurposeTo systematically review and meta-analyze the performance of radiomics models in predicting MYCN gene amplification status in NB patients.Material and MethodsAs of 18 March 2025, a systematic search was performed for original literature on the prediction of MYCN amplification in NB patients using radiomics models in the following databases: PubMed, Embase, Web of Science, and Cochrane Library. The quality of the literature was assessed using the QUADAS-2 and Radiomics Quality Score (RQS) tools. The meta-analysis was performed using the random-effects model.ResultsThis research ultimately included nine articles (899 patients), from which data could be extracted for both radiomics-only models and combined models that integrate radiomic features with other predictors. The radiomics-only model demonstrated pooled sensitivity of 0.85 (95% confidence interval [CI] = 0.77-0.91) and specificity of 0.86 (95% CI = 0.79-0.90), while the combined model showed a sensitivity of 0.81 (95% CI = 0.75-0.87) and specificity of 0.92 (95% CI = 0.87-0.95). Summary receiver operating characteristic (SROC) curve yielded an area under ROC curve of 0.92 ± 0.02 for the radiomics-only model and 0.94 ± 0.02 for the combined model. No evidence of publication bias was found.ConclusionsRadiomics might be one promising approach for predicting MYCN gene amplification in patients with NB.