摘要目的:评估多参数MRI（mpMRI）前列腺影像报告与数据系统第2.1版（PI-RADS v2.1）对临床有意义前列腺癌（csPCa）的诊断价值。方法:双向性收集2015年6月至2020年12月因前列腺特异性抗原升高在广西医科大学第一附属医院行前列腺mpMRI的561例患者，根据病理结果分为csPCa组（276例）和非csPCa组（285例）。由低、高年资影像科医师各1名依据PI-RADS v2.1评分标准进行评分。测量前列腺体积并计算前列腺特异性抗原密度（PSAD）。对扩散加权成像及动态增强MRI进行图像后处理，测量主要病灶的定量参数，包括表观扩散系数（ADC）、转运常数（K trans）、速率常数（K ep），采用Mann-Whitney U检验比较2组间各参数的差异。采用logistic回归分析筛选csPCa的最佳预测因子，建立多参数预测模型。绘制受试者操作特征曲线评估PI-RADS v2.1及预测模型诊断csPCa的效能，通过DeLong检验比较曲线下面积（AUC）。 结果:csPCa组的高年资医师PI-RADS评分、PSAD、K trans、K ep高于非csPCa组，ADC值低于非csPCa组，差异有统计学意义（ Z=-16.69、-12.49、-3.43、-4.67、13.91， P均<0.001）。logistic回归分析显示高年资医师PI-RADS评分（OR=3.064，95%CI 2.428~3.866， P<0.001）、PASD（OR=1.554，95%CI 1.170~2.064， P=0.002）、ADC值（OR=0.095，95%CI 0.032~0.288， P<0.001）是csPCa的最佳预测因素，构成多参数预测模型。低、高年资医师PI-RADS评分及预测模型诊断csPCa的AUC分别为0.861（95%CI 0.830~0.892）、0.895（95%CI 0.868~0.922）、0.923（95%CI 0.898~0.944），两两间差异均有统计学意义（低、高年资医师PI-RADS评分间 Z=3.24， P=0.001；低年资医师PI-RADS评分与预测模型间 Z=5.54， P<0.001；高年资医师PI-RADS评分与预测模型间 Z=4.20， P<0.001）。 结论:基于mpMRI，低、高年资医师应用PI-RADS v2.1对csPCa均有较高的诊断效能，多参数联合模型对csPCa的诊断效能最佳。

abstractsObjective:To evaluate the diagnostic performance of the prostate imaging reporting and data system version 2.1 (PI-RADS v2.1) based on multiparametric MRI (mpMRI) in the detection of clinically significant prostate cancer (csPCa).Methods:A total of 561 patients who underwent prostate mpMRI in the First Affiliated Hospital of Guangxi Medical University from June 2015 to December 2020 due to elevated prostate specific antigen were collected ambispectively. The patients were divided into csPCa group (276 cases) and non-csPCa group (285 cases) according to pathological findings. Prostate were scored according to the PI-RADS v2.1 scoring standard by a junior and a senior radiologist. The prostate volume was measured and the prostate specific antigen density (PSAD) was calculated. The diffusion-weighted imaging and dynamic contrast-enhanced MRI images were processed to measure the quantitative parameters of the index lesion, including apparent diffusion coefficient (ADC), volume transfer constant (K trans) and rate constant (K ep) values. The Mann-Whitney U test was used to compare the difference in parameters between the two groups. The predictors of csPCa were screened by logistic regression analysis. Predictive model of multi-parameter was established. The receiver operator characteristic curves were used to evaluate the efficacy of PI-RADS v2.1 and the model in diagnosing csPCa, and the comparisons of area under the curve (AUC) were conducted by DeLong test. Results:Compared with non-csPCa group, the patients in csPCa group had higher PI-RADS score of senior physician, PSAD, K trans and K ep value, lower ADC value ( Z=-16.69, -12.49, -3.43, -4.67, 13.91, all P<0.001). The PI-RADS scores of senior physician (OR=3.064, 95%CI 2.428-3.866, P<0.001), PSAD (OR=1.554, 95%CI 1.170-2.064, P=0.002) and ADC value (OR=0.095, 95%CI 0.032-0.288, P<0.001) were the predictors of csPCa. The AUC of junior, senior physician PI-RADS and combined prediction model were 0.861 (95%CI 0.830-0.892), 0.895 (95%CI 0.868-0.922) and 0.923 (95%CI 0.898-0.944). The pairwise difference was statistically significant (the PI-RADS score between the junior and senior physicians Z=3.24, P=0.001, the difference between the PI-RADS score of junior physician and prediction model Z=5.54, P<0.001, the difference between the PI-RADS score of senior physician and prediction model Z=4.20, P<0.001). Conclusion:Based on mpMRI, the application of PI-RADS v2.1 by junior and senior radiologists has the high diagnostic efficacy for csPCa, and the multi-parameter model has the best diagnostic efficacy for csPCa.