Photo Credit: Adobe

The following is a summary of “Early prediction of progression-free survival of patients with locally advanced nasopharyngeal carcinoma using multi-parametric MRI radiomics,” published in the March 2025 issue of the BMC Cancer by Jian et al.

Accurate prognostic prediction is essential for guiding personalized treatment strategies in patients with locoregionally advanced nasopharyngeal carcinoma (LANPC). While conventional MRI-based radiomic models have been widely used for prognostication, the potential incremental value of incorporating functional MRI remains underexplored. This study aimed to develop and validate an advanced radiomic model that integrates functional MRI to enhance prognostic accuracy in patients with LANPC. A total of 126 patients (training dataset: n = 88; validation dataset: n = 38) with LANPC were retrospectively included in the study. Radiomic features were extracted from multiple MRI sequences, including T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), contrast-enhanced T1WI (cT1WI), and diffusion-weighted imaging (DWI). 

To identify features associated with progression-free survival (PFS), Pearson correlation analysis and recursive feature elimination or relief were applied. Five machine learning algorithms with cross-validation were compared to determine the optimal approach for developing single-layer and fusion radiomic models. Additionally, clinical and combined models were constructed using multivariate Cox regression analysis. The clinical model based solely on TNM staging demonstrated limited predictive performance, with a C-index of 0.544 in the validation dataset. 

In contrast, the fusion radiomic model, which integrated DWI-, T1WI-, and cT1WI-derived imaging features, achieved superior prognostic accuracy, yielding a C-index of 0.788 for predicting PFS. This model also outperformed individual MRI-based models, including the DWI-based (C-index = 0.739), T1WI-based (C-index = 0.734), cT1WI-based (C-index = 0.722), and T1WI plus cT1WI-based (C-index = 0.747) models. Furthermore, the fusion radiomic model demonstrated strong predictive capability for distant metastasis-free survival (C-index = 0.786) and overall survival (C-index = 0.690). 

Notably, the incorporation of TNM staging into the fusion radiomic model did not enhance its predictive performance, underscoring the independent prognostic value of radiomic features derived from functional MRI. 

In conclusion, this study highlights the potential of a fusion radiomic model integrating DWI-derived imaging features to significantly improve survival prediction in patients with LANPC. These findings suggest that functional MRI enhances the predictive accuracy of conventional MRI-based radiomic models, providing a more reliable prognostic tool for clinical decision-making and individualized treatment planning.

Source: bmccancer.biomedcentral.com/articles/10.1186/s12885-025-13899-2