Department of Urology, Beijing Friendship Hospital, Capital Medical University, No. 95, Yongan Road, Xicheng District, Beijing, People's Republic of China.
National Clinical Research Center for Digestive Diseases, Beijing Friendship Hospital, Capital Medical University, No. 95, Yongan Road, Xicheng District, Beijing, People's Republic of China.
Int J Clin Oncol. 2020 Jan;25(1):175-186. doi: 10.1007/s10147-019-01524-9. Epub 2019 Aug 31.
To determine whether patients can avoid systematic prostate biopsy (PBx) if their Prostate Imaging Reporting and Data System version 2 (PI-RADs v2) score is ≤ 3 and how we clinicians make decisions that can maximize benefit.
We reviewed our prospectively maintained database of consecutive men who received transrectal ultrasound-guided 24-core biopsy as well as pre-biopsy multi-parametric magnetic resonance imaging (mp-MRI). Of the 1276 men who were performed PBx in our institution from 2012 to July 2018, 491 patients conformed to the criteria. Negative predictive value (NPV) of negative mp-MRI (defined as PI-RADs < 3) combined prostate-specific antigen density (PSAD) were calculated. Models based on PI-RADs v2 were developed to predict the absence of clinically significant prostate cancer (CSPCa) and prostate cancer (PCa). Nomograms as well as receiver operating curves (ROC) were established to estimate the discrimination. Calibration curves were used to assess the concordance between predictive value and true risk. Decision curves were made to measure the overall net benefit.
Prostate cancer and CSPCa detection rates were 21.6%, 7.3% and 36.7%, 23.4% in PIRADs v2 < 3 cohort and PIRADs v2 = 3 cohort, respectively. Men with biopsy-proved CSPCa had higher prostate-specific antigen (PSA), lower prostate volume (PV) and higher PSAD (all p < 0.05 in the two cohorts) than patients with clinically insignificant prostate cancer (CIPCa) or negative results. NPV of negative mp-MRI for detection of PCa was much higher when the PSAD was less than 0.15 (p < 0.001) and 0.2 for CSPCa (p = 0.007). According to multivariate analysis, we developed the model comprising Age, PSAD and PI-RADs v2 to predict the absence of CSPCa and PCa. The area under the curve (AUC) of the model for non-CSPCa was 0.75 (95% CI 0.68-0.80, PSAD cutoff 0.20), better than 0.71 (95% CI 0.65-0.80, PSAD cutoff 0.15). As for model for non-PCa, the AUC was 0.76 (95% CI 0.70-0.80, PSAD cutoff 0.15), higher than 0.71(95% CI 0.67-0.78, PSAD cutoff 0.20). Internally validated calibration curves showed that the model might overestimated the risk of the absence of CSPCa when the threshold was between 53 and 72%, and if the threshold was between 72 and 87%, it might underestimate the risk. As for the absence of PCa, the model might overestimate the risk between 52 and 76%. Decision curves showed that a better clinical net benefit was met when the threshold was 55% for non-PCa and 70% for non-CSPCa.
NPV of negative mp-MRI for detection of CSPCa and PCa was improved with decreasing PSAD. The nomograms based on PI-RADs v2, age and PSAD showed internally validated high discrimination and calibration for the absence of PCa and CSPCa. When the predictive value was greater than 70% for the absence of CSPCa and 55% for the absence of PCa, we could avoid unnecessary PBx to maximize net benefit.
确定如果患者的前列腺影像报告和数据系统（PI-RADS v2）评分≤3，他们是否可以避免进行系统性前列腺活检（PBx），以及我们的临床医生如何做出可以最大化获益的决策。
我们回顾了我们前瞻性维护的连续接受经直肠超声引导 24 针活检以及术前多参数磁共振成像（mp-MRI）的男性数据库。在我们机构从 2012 年至 2018 年 7 月进行 PBx 的 1276 名男性中，491 名患者符合标准。计算了阴性 mp-MRI（定义为 PI-RADs<3）联合前列腺特异性抗原密度（PSAD）的阴性预测值（NPV）。基于 PI-RADs v2 建立了预测无临床显著前列腺癌（CSPCa）和前列腺癌（PCa）的模型。建立了列线图和接收者操作特征曲线（ROC）来估计区分度。使用校准曲线来评估预测值与真实风险之间的一致性。使用决策曲线来衡量整体净获益。
PI-RADs v2<3 队列和 PI-RADs v2=3 队列的前列腺癌和 CSPCa 检出率分别为 21.6%、7.3%和 36.7%、23.4%。活检证实的 CSPCa 患者的前列腺特异性抗原（PSA）更高，前列腺体积（PV）更小，PSAD 更高（两个队列中的所有差异均<0.05），而临床意义不显著的前列腺癌（CIPCa）或阴性结果的患者则更低。当 PSAD 小于 0.15（p<0.001）和 0.2 时，阴性 mp-MRI 对 PCa 的 NPV 要高得多（p=0.007 对 CSPCa）。根据多变量分析，我们建立了包含年龄、PSAD 和 PI-RADS v2 的模型，用于预测 CSPCa 和 PCa 的缺失。非 CSPCa 模型的曲线下面积（AUC）为 0.75（95%CI 0.68-0.80，PSAD 截断值为 0.20），优于 0.71（95%CI 0.65-0.80，PSAD 截断值为 0.15）。对于非 PCa 模型，AUC 为 0.76（95%CI 0.70-0.80，PSAD 截断值为 0.15），高于 0.71（95%CI 0.67-0.78，PSAD 截断值为 0.20）。内部验证的校准曲线表明，当阈值在 53%至 72%之间时，该模型可能高估了 CSPCa 缺失的风险，而当阈值在 72%至 87%之间时，该模型可能低估了风险。对于 PCa 的缺失，该模型可能高估了 52%至 76%之间的风险。决策曲线表明，当阈值为非 PCa 时为 55%，非 CSPCa 时为 70%时，可获得更好的临床净获益。
随着 PSAD 的降低，阴性 mp-MRI 对 CSPCa 和 PCa 的检测 NPV 得到改善。基于 PI-RADS v2、年龄和 PSAD 的列线图显示了对 PCa 和 CSPCa 缺失的内部验证高区分度和校准。当 CSPCa 缺失的预测值大于 70%，PCa 缺失的预测值大于 55%时，我们可以避免不必要的 PBx 以最大化净获益。
