初诊前列腺癌患者骨转移预测模型的建立及验证

摘要

目的 根据初诊前列腺癌患者的临床病理资料,建立并验证用于预测前列腺癌骨转移风险的列线图.方法 采用随机方法把入组的501例患者分为建模样本(300例)及验证样本(201例).在建模样本中,采用Logistic回归分析与骨转移相关的临床病理资料并建模,再根据回归系数画出相应的列线图模型.最后在验证样本中比较列线图模型的分辨度及符合度.结果 前列腺特异抗原值、cT3、cT4及Gleason评分≥8分是前列腺癌骨转移的独立预后因素(P＜0.05),OR值分别为5.65、2.89、9.07和2.87.列线图模型在建模样本中的一致系数为0.830,而在验证样本中的一致系数为0.799,并表现出较好的符合度.结论 用列线图根据初诊前列腺癌患者的临床病理资料可预测出患者的骨转移风险,以此决定患者是否有接受骨扫描的必要.%Objective To develop and validate a nomogram used to predict the bone metastasis risks according to the clinicopathological factors of patients with newly diagnosed prostate cancer.Methods The 501 cases were randomly assigned into development sample (300 cases) and validation sample (201 cases).In the development sample,Logistic regression analysis was used to explore the predictors of bone metastases,and then a nomogram was built based on regression coefficients and validated in the validation sample.Results Prostate specific antigen,cT3,cT4 and Gleason score ≥ 8 were the independent prognostic factors (P ＜ 0.05),and the OR values were 5.65,2.89,9.07 and 2.87respectively.The concordance index was 0.830 in the model sample and 0.799 in the validation sample.Conclusion A nomogram,buih based on the clinicopathological factors,could be used to predict the risk of bone metastases and then could be helpful for the rational use of bone scan.