Department of Urology, Erasmus University Medical Center, Rotterdam, the Netherlands.
Department of Urology, The Netherlands Cancer Institute, Amsterdam, the Netherlands.
J Urol. 2020 Sep;204(3):503-510. doi: 10.1097/JU.0000000000001012. Epub 2020 Mar 9.
We developed a model predicting the probability of detecting prostate cancer recurrence outside the prostatic fossa on prostate specific membrane antigen positron emission tomography/computerized tomography in patients with biochemical recurrence after radical prostatectomy.
We retrospectively included 419 consecutive patients with biochemical recurrence (prostate specific antigen less than 2.0 ng/ml) after radical prostatectomy who underwent Ga-prostate specific membrane antigen-11 positron emission tomography/computerized tomography to guide salvage therapy. Patients receiving androgen deprivation therapy between radical prostatectomy and prostate specific membrane antigen positron emission tomography/computerized tomography were excluded from the study. We used multivariable logistic regression to assess predictors for the detection of prostate cancer recurrence outside the prostatic fossa on prostate specific membrane antigen positron emission tomography/computerized tomography. We minimized overfitting of the model and used decision curve analysis to determine clinical utility.
Median prostate specific antigen at scanning was 0.40 ng/ml (IQR 0.30-0.70). Overall 174 (42%) patients had prostate cancer recurrence outside the prostatic fossa. Prostate specific antigen at time of scanning, and grade group, N stage and surgical margin status at radical prostatectomy specimen were significant predictors for detecting prostate cancer recurrence outside the prostatic fossa. The bootstrapped AUC of this model was 0.75 (IQR 0.73-0.77). The decision curve analysis showed a net benefit by a model based probability from 16%. Limitations include the retrospective design and the missing histological correlation of positive lesions.
Next to the prostate specific antigen at time of scanning, grade group, N stage and surgical margin status at radical prostatectomy specimen are significant predictors for detecting prostate cancer recurrence outside the prostatic fossa on prostate specific membrane antigen positron emission tomography/computerized tomography. The presented model is implemented in a dashboard to assist clinicians in determining the optimal time to perform Ga-prostate specific membrane antigen-11 positron emission tomography/computerized tomography in patients with biochemical recurrence after radical prostatectomy.
我们开发了一种模型，用于预测前列腺特异性膜抗原正电子发射断层扫描/计算机断层扫描在根治性前列腺切除术后生化复发患者中探测前列腺窝外前列腺癌复发的概率。
我们回顾性纳入了 419 例根治性前列腺切除术后生化复发（前列腺特异性抗原<2.0ng/ml）的连续患者，他们接受 Ga-前列腺特异性膜抗原-11 正电子发射断层扫描/计算机断层扫描指导挽救性治疗。在根治性前列腺切除术和前列腺特异性膜抗原正电子发射断层扫描/计算机断层扫描之间接受雄激素剥夺治疗的患者被排除在研究之外。我们使用多变量逻辑回归来评估前列腺特异性膜抗原正电子发射断层扫描/计算机断层扫描中探测前列腺窝外前列腺癌复发的预测因素。我们最小化模型的过度拟合，并使用决策曲线分析来确定临床实用性。
扫描时中位前列腺特异性抗原为 0.40ng/ml（IQR 0.30-0.70）。总体上有 174 例（42%）患者出现前列腺窝外前列腺癌复发。扫描时的前列腺特异性抗原水平、Gleason 分级分组、N 分期和根治性前列腺切除标本的手术切缘状态是探测前列腺窝外前列腺癌复发的显著预测因素。该模型的 bootstrap AUC 为 0.75（IQR 0.73-0.77）。决策曲线分析显示，基于概率的模型具有 16%的净获益。局限性包括回顾性设计以及阳性病变的组织学相关性缺失。
除了扫描时的前列腺特异性抗原水平外，Gleason 分级分组、N 分期和根治性前列腺切除标本的手术切缘状态是探测前列腺特异性膜抗原正电子发射断层扫描/计算机断层扫描中前列腺窝外前列腺癌复发的显著预测因素。所提出的模型已在仪表板中实现，以帮助临床医生确定在根治性前列腺切除术后生化复发患者中进行 Ga-前列腺特异性膜抗原-11 正电子发射断层扫描/计算机断层扫描的最佳时间。
