MRI Radiomics Signature of Pediatric Medulloblastoma Improves Risk Stratification Beyond Clinical and Conventional MR...

Area of Science:

• Medical Imaging
• Oncology
• Radiology
• Pediatric Medicine

Background:

• Accurate prognostic evaluation is crucial for tailoring treatment strategies in pediatric medulloblastoma (MB).
• Current risk stratification methods for MB have limitations, with scarce data on radiomics-based models.
• The need for advanced models to improve personalized treatment in pediatric MB is evident.

Purpose of the Study:

• To assess the added value of an MRI radiomics signature for risk stratification in pediatric MB.
• To evaluate the performance of a radiomics signature in predicting overall survival (OS) for pediatric MB patients.
• To compare the predictive accuracy of radiomics, clinical-MRI, and an integrative model for MB prognosis.

Main Methods:

• Retrospective analysis of 111 pediatric MB patients (mean age 5.82 years) using 3T contrast-enhanced T1-weighted MRI.
• Development of a radiomics signature model, a clinical-MRI model, and an integrative model combining both using multivariable Cox regression.
• Performance evaluation using C-index, calibration curves, and decision curve analysis (DCA); statistical significance set at P < 0.05.

Main Results:

• The integrative model demonstrated superior performance in OS evaluation (C-index 0.823) compared to the radiomics signature (C-index 0.702) and the clinical-MRI model (C-index 0.771).
• The integrative model maintained strong predictive accuracy in the validation cohort (C-indices: 0.786, 0.756, 0.721).
• Key features for the clinical-MRI model included remnant tumor, adjuvant treatment, and peritumoral edema; the integrative model showed good calibration and clinical utility.

Conclusions:

• An integrative model incorporating radiomics, clinical, and conventional MRI features provides the most effective OS evaluation for pediatric MB.
• The MRI radiomics signature offers incremental prognostic value beyond conventional clinical and MRI features in pediatric MB.
• This study supports the use of integrative radiomics models for enhanced risk stratification and personalized treatment of pediatric medulloblastoma.