Utility of the Apparent Diffusion Coefficient for Distinguishing Clear Cell Renal Cell Carcinoma of Low and High Nuclear Grade

OBJECTIVE. The purpose of our study was to assess the utility of the apparent diffusion coefficient (ADC) in distinguishing low-grade and high-grade clear cell renal cell carcinoma (ccRCC). MATERIALS AND METHODS. The cases of 57 patients with pathologically proven ccRCC who underwent preoperative MRI, including diffusion-weighted imaging, were retrospectively assessed. ADC values were obtained from ADC maps calculated using b-value combinations of 0 and 400 s/mm(2) and of 0 and 800 s/mm(2) (hereafter referred to as ADC-400 and ADC-800). Lesions were also evaluated for an array of conventional MRI features. A single expert uropathologist reviewed all slides to determine nuclear grade. The utility of ADC for detecting high-grade ccRCC, alone and in combination with conventional MRI features, was assessed using receiver operating characteristic (ROC) analysis and binary logistic regression. RESULTS. ADC-400 and ADC-800 were significantly lower among high-grade than among low-grade ccRCC (2.24 +/- 0.50 mm(2)/s vs 1.59 +/- 0.57 mm(2)/s for ADC-400, p < 0.001; 1.85 +/- 0.40 mm(2)/s vs 1.28 +/- 0.48 mm(2)/s for ADC-800; p < 0.001). The area under the ROC curve for identifying high-grade ccRCC using ADC-400 and ADC-800 was 0.801 and 0.824 respectively (p = 0.606), with optimal thresholds, sensitivity, and specificity as follows: ADC-400: 2.17 mm(2)/s, 88.5%, 64.5% and ADC-800: 1.20 mm(2)/s, 65.4%, 96.0%. Using multivariate logistic regression, only necrosis (p = 0.0229) and perinephric fat invasion (p = 0.0160) were retained among conventional imaging features as independent risk factors for high-grade ccRCC. The accuracy of the logistic regression model for predicting high-grade ccRCC was significantly improved by inclusion of either ADC-400 (p = 0.0143) or ADC-800 (p = 0.015). CONCLUSION. ADC is significantly lower in high-grade ccRCC compared with low-grade ccRCC and increases the accuracy for detecting high-grade ccRCC compared with conventional MRI features alone.