Time-dependent Diffusion MRI for Predicting Response to Induction Chemotherapy in Nasopharyngeal Carcinoma

Purpose To validate the reliability of time-dependent diffusion MRI (td-dMRI)-derived indicators in nasopharyngeal carcinoma (NPC) and determine whether microstructural parameters derived from td-dMRI can help noninvasively predict response to induction chemotherapy. Materials and Methods In this prospective study, participants with locally advanced NPC underwent pretreatment td-dMRI between December 2023 and February 2025. The data were randomly stratified and divided into a training set and an internal test set at a 7:3 ratio. According to Response Evaluation Criteria in Solid Tumors 1.1, all participants were classified as responders or nonresponders. Four imaging microstructural parameters with the limited spectrally edited diffusion-derived microstructural parameters (cell diameter d, cellularity, intracellular volume fraction f_in, and extracellular diffusivity D_ex) and three apparent diffusion coefficient (ADC) values (the ADC with the pulsed gradient spin-echo sequence, the ADC at 20 Hz, and the ADC at 40 Hz) were calculated. Repeatability was evaluated. MRI-derived parameters were compared with pathologic microstructural metrics. In the training set, univariable and multivariable logistic regression analyses were performed to identify parameters associated with induction chemotherapy response; predictive models were constructed. Results In total, 220 participants with NPC were enrolled (median age, 53.0 years [IQR, 48.8-58.0 years]; 155 male participants), including 154 in the training set and 66 in the internal test set. Participants were divided into responders (n = 151) and nonresponders (n = 69). The repeatability study revealed within-subject coefficient of variation values of 4.56%-10.15% and repeatability coefficient values of 0.12-1.71. Moreover, the td-dMRI-derived parameters correlated well with pathologic measurements (r = 0.27-0.67; P < .05). In the training set, univariable and multivariable logistic regression analyses revealed that the tumor-stroma ratio (TSR) and td-dMRI-derived cellularity were independent predictors. The combined model integrating the TSR and cellularity achieved an AUC of 0.82 (95% CI: 0.75, 0.89) in the training set and 0.76 (95% CI: 0.63, 0.87) in the internal test set. Conclusion td-dMRI-derived microstructural parameters, particularly TSR and tumor cellularity, showed good performance in predicting response to induction therapy in individuals with NPC. Keywords: Nasopharyngeal Carcinoma, Response, MRI, Induction Chemotherapy Supplemental material is available for this article. © The Author(s) 2026. Published by the Radiological Society of North America under a CC BY 4.0 license.