Integrated Radiomics Model Combining Diffusion Kurtosis Imaging and Dynamic Contrast-Enhanced MRI for Predicting TERT...

The purpose of this study is to investigate the value of a radiomics model based on diffusion kurtosis imaging (DKI) and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for preoperative prediction of telomerase reverse transcriptase (TERT) promoter mutation status in gliomas.

This retrospective study included 126 patients with pathologically confirmed gliomas who underwent TERT promoter mutation testing between January 2020 and June 2025. All patients underwent preoperative multiparametric MRI including DKI and DCE-MRI sequences. Patients were randomly divided into training (n = 88) and validation (n = 38) cohorts at a 7:3 ratio. Radiomics features were extracted from DKI parameter maps (mean kurtosis, mean diffusivity, axial kurtosis, radial kurtosis, axial diffusivity, and radial diffusivity) and DCE-MRI parameter maps (volume transfer constant [Ktrans], extravascular extracellular volume fraction [Ve], rate constant [Kep], and plasma volume fraction [Vp]). Feature selection was performed using the least absolute shrinkage and selection operator (LASSO) regression. DKI-based, DCE-MRI-based, and combined radiomics models were constructed using logistic regression. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis, and comparison between models was performed using the DeLong test. Decision curve analysis was conducted to assess clinical utility.

Significant differences in DKI and DCE-MRI parameters were observed between TERT promoter mutant and wild-type gliomas (p < 0.05). LASSO regression selected 12 optimal features (five from DKI and seven from DCE-MRI) for the combined model. In the training cohort, the DKI-based model, the DCE-MRI-based model, and the combined model achieved areas under the curve (AUCs) of 0.847 (95% confidence interval [CI]: 0.768-0.926), 0.892 (95% CI: 0.821-0.963), and 0.961 (95% CI: 0.927-0.995), respectively. In the validation cohort, the corresponding AUCs were 0.823 (95% CI: 0.691-0.955), 0.869 (95% CI: 0.752-0.986), and 0.943 (95% CI: 0.871-1.000). The combined model demonstrated significantly superior performance compared to single-modality models (p < 0.05), with sensitivity, specificity, and accuracy of 88.9%, 95.0%, and 92.1% in the validation cohort. Decision curve analysis indicated that the combined model provided greater clinical net benefit across threshold probabilities ranging from 0.15 to 0.85.

The integrated multiparametric radiomics model combining DKI and DCE-MRI enables noninvasive preoperative prediction of TERT promoter mutation status in gliomas with high accuracy. The combined approach demonstrates superior predictive performance and clinical utility compared to single-modality imaging, providing valuable imaging biomarkers for molecular stratification and personalized treatment planning in glioma patients.