Diffusion-Tensor MRI at 3 T: Differentiation of Central Gland Prostate Cancer From Benign Prostatic Hyperplasia

Abstract

The purpose of this article is to retrospectively evaluate the utility of diffusion-tensor imaging (DTI) at 3 T in differentiating central gland prostate cancer from benign prostatic hyperplasia (BPH). Eighty consecutive patients (57 with central gland cancer and 23 without central gland cancer) were included in this study. All patients underwent T2-weighted imaging and DTI at 3 T, followed by surgery. For predicting central gland cancer, experienced and less-experienced radiologists independently analyzed T2-weighted imaging and combined T2-weighted imaging and DTI, respectively. Apparent diffusion coefficient (ADC) and fractional anisotropy (FA) values were measured for central gland cancers and BPH foci of stromal and glandular hyperplasia. Statistical analyses were performed using McNemar test, linear mixed model, receiver operating characteristic (ROC), and kappa statistics. For predicting central gland cancers, the area under the curve (Az) of combined T2-weighted imaging and DTI for the experienced (0.915) and less-experienced reader (0.753) was superior to that of T2-weighted imaging (0.723 vs 0.664; p<0.001). The mean ADC and FA values were 0.77×10(-3) mm2/s and 0.35, respectively, for central gland cancers, 1.22×10(-3) mm2/s and 0.26, respectively, for stromal hyperplasia foci, and 1.59×10(-3) mm2/s and 0.21, respectively, for glandular hyperplasia foci, and the values differed significantly. For differentiating central gland cancer from stromal hyperplasia foci and glandular hyperplasia foci, Az values of ADC versus FA were 0.989 and 1.0 versus 0.818 and 0.916, respectively, and the difference was statistically different. DTI at 3 T is useful for distinguishing central gland cancers from BPH foci, with significantly different ADC and FA values. Furthermore, ADC showed greater diagnostic accuracy than FA in differentiating central gland cancers from stromal and glandular hyperplasia foci.