Yu Yi-Ning, Zhang Sheng-Jian
Department of Radiology, Fudan University Cancer Hospital, Department of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China.
Zhonghua Nan Ke Xue. 2019 Dec;25(12):1106-1112.
To assess the value of stretched-exponential and mono-exponential diffusion-weighted imaging (DWI) in predicting the aggressiveness of PCa.
This retrospective study included 36 cases of PCa with 48 lesions in the peripheral zone diagnosed by DWI with b-values of 0, 500, 1000, and 2000 s/mm2. We reconstructed the apparent diffusion coefficient (ADC), distributed diffusion coefficient (DDC) and α maps on the post-processing workstation, performed a histogram analysis on the largest slice of PCa on T2WI and Spearman's rank-order analysis on the correlation of the histogram variables with Gleason grade grouping (GG). Then, we assessed the values of the histogram variables in differentiating low-grade from high-grade PCa using the receiver operating characteristic (ROC) curve.
The percentile and mean ADCs and DDCs were correlated with GG (ρ: 0.392－0.641) but not the α value, skewnesses and kurtosises (ρ: 0.055－0.266). High-grade PCa exhibited significantly lower 10th-, 25th-, 50th- and 75th-percentile and mean ADCs (490 ± 141, 591 ± 137, 695 ± 137, 781 ± 139 and 888 ± 135 mm2/s) and DDCs (420 ± 146, 534 ± 167, 666 ± 182, 787 ± 190 and 912 ± 175 mm2/s) than low-grade PCa (ADCs: 636 ± 74, 727 ± 86, 825 ± 85, 907 ± 85 and 975 ± 117 mm2/s; DDCs: 542 ± 80, 666 ± 93, 806 ± 108, 910 ± 110 and 1023 ± 105 mm2/s), but there were no statistically significant differences between low- and high-grade PCa in the α value (0.67 ± 0.042 vs 0.64 ± 0.036), kurtosises (ADC 0.105 vs 0.078; DDC －0.027 vs －0.401) or skewnesses (ADC －0.042 vs 0.067; DDC －0.058 vs 0.162). Both 10th-percentile ADCs and DDCs showed a higher efficiency than the mean ones in differentiating high- from low-grade PCa, though with no statistically significant difference (P > 0.05).
Histogram variables DDCs and ADCs, rather than the α value, can be used to predict the aggressiveness of PCa, even more efficiently at the 10th percentile than on the mean.
评估拉伸指数和单指数扩散加权成像（DWI）在预测前列腺癌（PCa）侵袭性方面的价值。
本回顾性研究纳入了36例经DWI诊断为外周带PCa且有48个病灶的患者，DWI的b值为0、500、1000和2000 s/mm2。我们在图像后处理工作站上重建了表观扩散系数（ADC）、分布扩散系数（DDC）和α图，对T2WI上PCa最大层面进行直方图分析，并对直方图变量与Gleason分级分组（GG）的相关性进行Spearman等级分析。然后，我们使用受试者工作特征（ROC）曲线评估直方图变量在区分低级别与高级别PCa方面的价值。
ADC和DDC的百分位数及均值与GG相关（ρ：0.392 - 0.641），但α值、偏度和峰度与GG不相关（ρ：0.055 - 0.266）。高级别PCa的第10、25、50和75百分位数及平均ADC（490±141、591±137、695±137、781±139和888±135 mm2/s）和DDC（420±146、534±167、666±182、787±190和912±175 mm2/s）显著低于低级别PCa（ADC：636±74、727±86、825±85、907±85和975±117 mm2/s；DDC：542±80、666±93、806±108、910±110和1023±105 mm2/s），但低级别与高级别PCa在α值（0.67±0.042对0.64±0.036）、峰度（ADC 0.105对0.078；DDC -0.027对-0.401）或偏度（ADC -0.042对0.067；DDC -0.058对0.162）方面无统计学显著差异。第10百分位数的ADC和DDC在区分高级别与低级别PCa方面均显示出比均值更高的效能，尽管无统计学显著差异（P>0.05）。
直方图变量DDC和ADC而非α值可用于预测PCa的侵袭性，第10百分位数的预测效能甚至高于均值。
