Histogram Analysis of Diffusion Kurtosis Magnetic Resonance Imaging in Differentiation of Pathologic Gleason Grade of Prostate Cancer

Qing Wang,Hai Li,Xu Yan,Chen-Jiang Wu,Xi-Sheng Liu,Hai-Bin Shi,Yu-Dong Zhang
DOI: https://doi.org/10.1016/j.urolonc.2015.05.005
IF: 2.954
2015-01-01
Urologic Oncology Seminars and Original Investigations
Abstract:Objective: To investigate diagnostic performance of diffusion kurtosis imaging with histogram analysis for stratifying pathologic Gleason grade of prostate cancer (PCa). Materials and methods: This retrospective study was approved by the institutional review board, and written informed consent was waived. A total of 110 patients pathologically confirmed as having PCa (diameter>0.5 cm) underwent preoperative diffusion-weighted magnetic resonance imaging (b value of 0-2,100 s/mm(2)) at 3T. Data were postprocessed by monoexponential and diffusion kurtosis models for quantitation of apparent diffusion coefficients (ADCs), apparent diffusion for Gaussian distribution (D-app), and apparent kurtosis coefficient (K-app). The measurement was based on an entire-tumor histogram analysis approach. The ability of imaging indices for differentiating low-grade (LG) PCa (Gleason score [GS] <= 6) from intermediate-/high-grade (HG: GS>6) PCa was analyzed by receiver operating characteristic regression. Results: There were 49 LG tumors and 77 HG tumors at pathologic findings. HG-PCa had significantly lower ADCs, lower diffusion kurtosis diffusivity (D-app), and higher kurtosis coefficient (K-app) in mean, median, 10th, and 90th percentile, with higher D-app in skewness and kurtosis than LG-PCa (P< 0.05). The 10th ADCs, the 10th D-app, and the 90th K-app showed relatively higher area under receiver operating characteristic curve (Az), Youden index, and positive likelihood ratio in stratifying aggressiveness of PCa against other indices. The 90th K-app showed relatively higher correlation (rho>0.6) with ordinal GS of PCa; significantly higher Az, sensitivity, and specificity (0.889, 74.1%, and 93.9%, respectively) than the 10th D-app did (0.765, 61.0%, and 79.6%, respectively; P<0.05); and higher Az and specificity than the 10th ADCs did (0.738 and 71.4%, respectively; P<0.05) in differentiating LG-PCa from HG-PCa. Conclusions: It demonstrated a good reliability of histogram diffusion kurtosis imaging for stratifying pathologic GS of PCa. The 90th K-app had better diagnostic performance in differentiating LG-PCa from HG-PCa. (c) 2015 Elsevier Inc. All rights reserved.
What problem does this paper attempt to address?