根治性前列腺切除术后患者生化复发的贝叶斯预测模型的构建及临床价值  

CONSTRUCTION AND CLINICAL VALUE OF BAYESIAN PREDICTION MODEL FOR BIOCHEMICAL RECURRENCE AFTER RADICAL PROSTATECTOMY

作　　者：贺大千 王昕凝 王琨翔 闫伟华[2] 张弛[3] 张铭鑫 杨学成[1] HE Daqian;WANG Xinning;WANG Kunxiang;YAN Weihua;ZHANG Chi;ZHANG Mingxin;YANG Xuecheng(Department of Urology,TheAffiliated Hospital of Qingdao University,Qingdao 266003,China)

机构地区：[1]青岛大学附属医院泌尿外科,山东青岛266003 [2]青岛大学附属医院病理科 [3]西北工业大学机电工程学院工业工程系

出　　处：《精准医学杂志》2022年第3期252-256,共5页Journal of Precision Medicine

基　　金：国家自然科学基金资助项目(82071750)。

摘　　要：目的基于贝叶斯网络构建前列腺癌根治术患者术后1年内生化复发预测模型,并探讨其临床预测的价值。方法回顾性分析2018年1月—2021年3月于青岛大学附属医院行腹腔镜下或机器人辅助下根治性前列腺癌切除术的209例患者的相关临床资料并进行单因素分析。将单因素分析中的危险因素应用bayesiaLab软件构建生化复发预测模型,采用受试者工作特征(ROC)曲线和曲线下面积(AUC)评价模型预测效果的优劣。结果209例患者中共43例患者术后1年内出现生化复发,生化复发的发生比例为20.57%。单因素分析结果显示,是否辅助治疗、临床及病理分期、切缘性质、包膜侵犯、精囊侵犯、淋巴血管及周围神经侵犯、根治病理Gleason评分、PI-RADS评分及手术方式是前列腺癌根治术后患者1年内生化复发的危险因素(χ^(2)=2.026~26.306,P<0.05),基于上述10个因素建立贝叶斯模型。ROC曲线显示,模型预测患者生化复发的AUC为81.43%,准确度为80.95%,灵敏度为88.37%,特异度为96.55%,阳性预测值为85.71%,阴性预测值为80.00%。结论基于贝叶斯网络成功构建了前列腺癌根治术患者术后1年内生化复发预测模型,该模型可用于前列腺癌患者术后生化复发的预测。Objective To construct a Bayesian network-based prediction model for biochemical recurrence within one year after radical prostatectomy and explore its value for clinical prediction.Methods We retrospectively analyzed the clinical data of 209 patients who had undergone laparoscopic or robot-assisted radical prostatectomy in The Affiliated Hospital of Qingdao University from January 2018 to March 2021.Univariate analysis was conducted to determine risk factors for biochemical recurrence within one year after radical prostatectomy,and then these factors were included to construct a model for predicting bioche-mical recurrence with BayesiaLab software.The performance of the predictive model was assessed using a receiver operating characteristic curve and the area under the curve(AUC).Results Among the 209 patients,43(20.57%)had biochemical recurrence within one year after operation.Based on the univariate analysis,risk factors for biochemical recurrence within one year after radical prostatectomy included adjuvant therapy,clinical and pathological staging,resection margin status,extracapsular extension,seminal vesicle invasion,lymphovascular and perineural invasion,Gleason score,PI-RADS score,and the method of surgery(χ^(2)=2.026-26.306,P<0.05).The Bayesian model for predicting biochemical recurrence based on the above 10 factors showed:AUC,81.43%;accuracy,80.95%;sensitivity,88.37%;specificity,96.55%;positive predictive value,85.71%;and negative predictive value,80.00%.Conclusion The Bayesian network-based model established in this study was successful in predicting biochemical recurrence within one year after radical prostatectomy.

关 键 词：前列腺切除术 前列腺肿瘤 生化复发 贝叶斯定理 危险因素 预测模型 

分 类 号：R737.25[医药卫生—肿瘤]