Development and validation of a prediction model for identifying men with intermediate- or high-risk prostate cancer for whom bone imaging is unnecessary: a nation-wide population-based study

Objective: To develop and validate a nomogram that identifies men for whom bone scan is unnecessary. Material and methods: The study datasets were derived from the National Prostate Cancer Register (NPCR) of Sweden. All men in the NPCR ≤80 years of age who were diagnosed with intermediate- or high-risk prostate cancer and who had pretreatment bone imaging (99mTc MDP scintigraphy, plain x-ray, computed tomography, magnetic resonance imaging, and/or positron emission tomography fused with computed tomography) were included. Men diagnosed from 2015–2016 formed a development dataset and men diagnosed in 2017 formed a validation dataset. Outcome was metastasis on bone imaging as registered in NPCR. Multivariable logistic regression was used to develop a nomogram. Results: In the development dataset 482/5084 men (10%) had bone metastasis, the corresponding percentage in the validation dataset was 282/2554 (11%). Gleason grade group, clinical T stage, and prostate-specific antigen were included in the final model. Discrimination and calibration were satisfactory in both the development (AUC 0.80, 95% CI 0.78–0.82) and validation dataset (AUC 0.80, 95% CI, 0.77–0.82). Compared with using the EAU guidelines’ recommendation for selecting men for imaging, using the nomogram with a cut-off at 4% chance of bone metastasis, would have avoided imaging in 519/2068 men (25%) and miss bone metastasis in 10/519 (2%) men in the validation dataset. Conclusion: By use of our nomogram, bone scans of men with prostate cancer can be avoided in a large proportion of men.

研究目标：开发并验证一款列线图（nomogram），用于识别无需接受骨扫描的男性前列腺癌患者。
材料与方法：本研究数据集源自瑞典国家前列腺癌登记库（National Prostate Cancer Register, NPCR）。纳入对象为该登记库中年龄≤80岁、确诊为中危或高危前列腺癌，且已完成治疗前骨显像的男性患者，其中骨显像包括99mTc亚甲基二膦酸盐闪烁显像、X线平片、计算机断层扫描、磁共振成像，以及/或正电子发射断层显像-计算机断层扫描融合显像。2015-2016年确诊的患者构成建模数据集，2017年确诊的患者构成验证数据集。研究结局为登记库记录的骨显像提示骨转移。采用多因素logistic回归构建列线图。
结果：建模队列中共5084例患者，其中482例（10%）存在骨转移；验证队列中共2554例患者，对应骨转移比例为282例（11%）。最终模型纳入格里森分级组、临床T分期及前列腺特异性抗原水平。建模队列与验证队列的区分度与校准度均表现良好：建模队列的曲线下面积（Area Under the Curve, AUC）为0.80，95%置信区间（95% confidence interval, 95% CI）为0.78~0.82；验证队列的AUC为0.80，95%CI为0.77~0.82。与采用欧洲泌尿外科学会（European Association of Urology, EAU）指南推荐的骨显像患者筛选方案相比，以骨转移风险4%作为截断值应用本列线图，可在验证队列中减少25%（519/2068）的患者接受骨显像，且仅漏诊2%（10/519）的骨转移患者。
结论：应用本列线图，可使绝大多数前列腺癌男性患者免于接受骨扫描检查。