The prognosis of renal cell carcinoma (RCC) varies greatly, and accurate prognostic stratification is crucial for optimizing clinical management. This study aims to evaluate the feasibility of predictive models based on diffusion relaxation correlation spectroscopic imaging (DR-CSI) in distinguishing RCC patients with different clinical outcomes. A total of 127 RCC patients who underwent DR-CSI were enrolled, cohort 1 (48 patients) for model development, and cohort 2 (79 patients with postoperative follow-up) served for validation. DR-CSI results were analyzed using spectral equipartition method combined with multiple feature selection methods and classifiers, generating models from 2*2 to 9*9. Clinicopathological, conventional MR parameters and SSIGN were used for comparison. Diagnostic and prognostic performance were assessed using AUC, DeLong’s test, Kaplan‒Meier analysis, and multivariable regression. DR-CSI-based models showed excellent interobserver agreement (ICC: 0.86–0.99). In cohort 1, the 6*6 model achieved the highest diagnostic performance for distinguishing metastatic from non-metastatic RCC (AUC = 0.87), significantly outperforming clinicopathological and conventional MR parameters (vs. age, P < 0.001; vs. tumor diameter, P = 0.002; vs. WHO/ISUP grade, P < 0.001; vs. ADC, P = 0.001; vs. T2 value, P < 0.001). In cohort 2, the 6*6 model achieved an AUC of 0.85, which was significantly higher than SSIGN (AUC = 0.73, P = 0.012). This model was also an independent predictor of RCC recurrence (P = 0.005). The DR-CSI-based 6*6 model provides accurate assessment of RCC aggressiveness and shows great promise for prognostic risk stratification, offering a valuable non-invasive tool for clinical decision-making.
Zhu et al. (2026) studied this question.