Predicting c-KIT Inhibitor Efficacy in Patient-Derived Models of Sinonasal Mucosal Melanomas through Integrated Histogram Analysis of Whole-Tumor DKI, IVIM, and DCE-MRI.

Abstract

Purpose: To evaluate whole-tumor histogram analysis of diffusion kurtosis imaging (DKI), intravoxel incoherent motion (IVIM), and dynamic contrast-enhanced MRI (DCE-MRI) in predicting the efficacy of imatinib, a c-KIT inhibitor, for treating patient-derived models derived from sinonasal mucosal melanomas (MM).

Experimental design: This study included 38 patients with histologically confirmed sinonasal MM, who underwent DKI, IVIM, and DCE-MRI. Patient-derived tumor xenograft models and precision-cut tumor slices were established to evaluate tumor response to imatinib. Whole-tumor histogram analysis was conducted on imaging parameters, and logistic regression models were applied to determine the predictive value of these metrics in differentiating responders from nonresponders.

Results: Among the 38 patients with sinonasal MM, 12 were classified as responders and 26 as nonresponders based on patient-derived tumor xenograft and precision-cut tumor slice model responses to imatinib. The DKI model revealed significant differences in mean, median, 10th percentile, and 90th percentile values of Dk and K between responders and nonresponders (P < 0.05). The IVIM model indicated significant differences in 10th percentile and mean values of D, with kurtosis f being a strong predictor. The DCE-MRI model, using the 90th percentile Ktrans metric, demonstrated robust predictive performance, achieving an AUC of 0.89, with 80.77% specificity and 91.67% sensitivity. The combined logistic model integrating DKI, IVIM, and DCE-MRI metrics produced the highest predictive accuracy, with an AUC of 0.90.

Conclusions: Whole-tumor histogram analysis of DKI, IVIM, and DCE-MRI offers a noninvasive method for predicting the efficacy of c-KIT inhibitors in sinonasal MMs, presenting valuable implications for guiding targeted treatment in this rare cancer type.