结合PI-RADSv2.1磁共振引导下PSAD和年龄建立预测前列腺癌风险的预测模型

Establishment of the predictive model of the prostate cancer risk combined with PSAD and age guided by PI-RADS v2.1 magnetic resonancewere

目的:探讨结合PI-RADSv2.1磁共振引导下PSAD和年龄建立预测前列腺癌风险的预测模型。方法:将209例行前列腺活检的患者分为实验组(n=135)和验证组(n=74),根据前列腺癌的诊断性能分析前列腺影像报告与数据系统2.1版(PI-RADS v2.1)得分、前列腺特异性抗原(PSA)、游离前列腺特异性抗原与总前列腺特异性抗原比值(f/tPAS)、前列腺特异性抗原密度(PSAD)和年龄,建立逻辑回归分析预测前列腺癌风险评分系统并进行验证。结果:二进制逻辑回归分析结果表明,年龄、PSAD和PI-RADS v2.1得分是影响前列腺癌检测的因素;年龄、磁共振引导下PSAD和PI-RADS v2.1分数与前列腺癌的筛查准确率相关(P<0.05),与PSA及f/tPSA无相关性(P>0.05)。该模型的预测准确性为85.9%。基于对数回归系数建立了前列腺癌预测模型的列线图和校准图,并绘制出了临床决策曲线。ROC曲线评价预测模型表明,PCa预测模型较单独使用PSAD或PI-RADS评分在整个范围内都具有较高的净收益。外部验证证明预测模型有良好的效能(AUC=0.883,95%CI:0.800~0.848)。结论:结合PI-RADS v2.1、PSAD和年龄建立的前列腺癌预测模型具有良好的性能,可以提高前列腺癌筛查的临床准确性,从而减少不必要的前列腺穿刺活检。

Objective:To establish a prediction model of prostate cancer risk combined with PSAD and age guided by PI-RADS v2.1 magnetic resonance.Methods:A total of 209 patients who underwent prostate biopsy were enrolled.There were 135 patients in the model group and 74 patients in the validation group.The PI-RADS v2.1 score,PSA,f/tPSA,PSAD,and age were analyzed in terms of diagnostic performance for prostate cancer.We established a scoring system for predicting prostate cancer risk by logistic regression analysis.The scoring system was verified by using a validation group.Results:The results of binary logistic regression analysis showed that age,PSAD and PI-RADS v2.1 scores were the factors affecting the detection of prostate cancer,and age and MRI guided PSAD and PI-RADS v2.1 scores were correlated with the screening accuracy of prostate cancer(P<0.05).There was no significant difference in PSA and F/TPSA(P>0.05).The prediction accuracy of the model was 85.9%.Based on the logarithmic regression coefficient,the histogram and calibration diagram of the prostate cancer prediction model were established,and the clinical decision-making curve was plotted.The prediction model of ROC curve evaluation showed that the established PCA prediction model had higher net benefits than the PSAD or PI-RADS score alone in the whole range.External validation by the validation group proved that the predictive model had good performance(AUC=0.883,95%CI:0.800-0.848).Conclusion:The establishment of the prostate cancer prediction model with PI-RADS v2.1,PSAD and age has the excellent performance and can improve the clinical accuracy of prostate cancer screening,which can reduce unnecessary prostate biopsy.