The value of MRI-based radiomic nomograms in differential diagnosis and metastasis prediction of rhabdomyosarcoma and neuroblastoma in children

Abstract

Abstract Background: Rhabdomyosarcoma (RMS) and neuroblastoma (NB) are highly malignant soft tissue sarcoma with tendency to metastasize. Due to the similarities in clinical manifestations and imaging features between RMS and NB, they are often misdiagnosed, which resulted in improper treatment progression of the mass. On the other hand, the treatment paradigm for patients with metastasis RMS/NB and non-metastasis RMS/NB is different. Preoperative abdominal magnetic resonance imaging (MRI) can provide valuable information for differential diagnosis and metastasis prediction to support surgical decisions. This study aimed to develop MRI-based whole-volume tumor radiomic signatures for differential diagnosis and metastasis prediction. Methods: We retrospectively sampled 40 patients (21 patients with RMS and 19 patients with NB). Using least absolute shrinkage and selection operator (LASSO) regression and stepwise logistic regression, a classification model and a metastasis prediction model based on MRI radiomic signatures were constructed. Nomograms were established by integrating the MRI information for better classification and prediction. Harrell's concordance index (C-index) and time-dependent receiver operating characteristic (ROC) curves were used as performance evaluating metrics. Results: The nomograms consisting of radiomic signatures demonstrated good discrimination and calibration in classification (area under the curve [AUC]=89.97%) and metastasis prediction (AUC=82.25%). The calibration curve and GiViTI calibration belt value analysis indicated that the radiomic nomograms can be used in clinical practice. Conclusions: MRI-based whole-tumor radiomic signatures have excellent performance for differential diagnosis and metastasis prediction in pediatric RMS and NB. Radiomic nomograms may aid in preoperative risk assessment and guide personalized treatment strategies for pediatric soft tissue sarcomas.