前列腺癌根治术后生化复发的危险因素分析

Risk factors for biochemical recurrence of prostate cancer after radical prostatectomy

目的分析前列腺癌根治术(radical prostatectomy,RP)后生化复发的危险因素,并建立风险预测模型。方法本研究采用回顾性队列研究方法,分析2007年9月至2022年3月在本院接受前列腺癌根治术的患者279例,收集并分析临床指标共21项。纳入时间因素,应用Lasso回归筛选指标,Kaplan-Meier法绘制生存曲线,采用单因素、多因素Cox比例风险回归模型分析危险因素并建立预测模型,受试者工作特征(receiver operating characteristic curve,ROC)曲线评价模型的判别能力。并基于危险因素建立生化复发的风险预测列线图,应用校正曲线评价预测的准确性,决策曲线分析其临床效益。结果中位随访时间37.10(17.60,61.30)月,中位年龄68.00(65.00,72.00)岁,其中86例(30.8%)生化复发。单因素分析显示共12个指标与RP术后生化复发密切相关(P<0.05),并应用Lasso回归筛选得到7个具有代表性的指标集:营养风险指数(nutritional risk index,NRI)、临床T分期、病理Gleason评分、病理T分期、切缘阳性、淋巴结侵犯、精囊侵犯。再进行多因素分析显示,病理高分期(≥pT3期)(HR=1.895,95%CI:1.045~3.435,P=0.035)、切缘阳性(HR=1.808,95%CI:1.006~3.250,P=0.048)、淋巴结侵犯(HR=2.161,95%CI:1.118~4.175,P=0.022)、NRI≤106.9(HR=0.598,95%CI:0.378~0.946,P=0.028)是生化复发的独立危险因素。建立预测模型,其AUC为0.743(95%CI:0.677~0.810,P<0.001),敏感度为0.826,特异度为0.534。并建立预测列线图,C-index:0.741(95%CI:0.677~0.805;P<0.001),校正曲线表明一致性较好。临床决策曲线表明在阈值概率为15%~85%的区间可临床获益,临床效用较好。结论病理高分期(≥pT3期)、切缘阳性、淋巴结侵犯、NRI≤106.9是前列腺癌根治术后生化复发的独立危险因素,建立的预测模型能进行有效判断,为临床决策提供参考。

Objective To analyze the risk factors for biochemical recurrence in prostate cancer patients after radical prostatectomy(RP)and establish a risk prediction model for the recurrence.Methods A retrospective cohort study was conducted on 279 patients who underwent RP surgery in our hospital from September 2007 to March 2022.There were 21 clinical indicators collected and analyzed.After time factors were included,Lasso regression was applied to screen the indicators,Kaplan-Meier method was used to draw survival curve,univariate and multivariate Cox proportional risk regression models were employed to analyze the risk factors and establish a prediction model.Then receiver operating characteristic(ROC)curve were plotted to evaluate the discriminative performance of the model.Then a nomogram was draw based on the obtained risk factors.Calibration curve was used to evaluate the accuracy of prediction,and decision curve was employed to assess the clinical benefit.Results During a median follow-up time of 37.10(17.60,61.30)months,86 cases(30.8%),at a median age of 68.00(65.00,72.00)years,experienced biochemical recurrence.Univariate analysis showed that 12 indicators were strongly associated with biochemical recurrence after RP(P<0.05),and Lasso regression screening indicated 7 representative indicator sets,including nutritional risk index(NRI),clinical T-stage,pathological Gleason score,pathological T-stage,positive surgical margin,lymph node invasion,and seminal vesicle invasion.Further multifactorial analysis revealed that pathological high stage(≥pT3 stage)(HR=1.895,95%CI:1.045~3.435,P=0.035),positive surgical margin(HR=1.808,95%CI:1.006~3.25,P=0.048),lymph node invasion(HR=2.161,95%CI:1.118~4.175,P=0.022)and NRI≤106.9(HR=0.598,95%CI:0.378~0.946,P=0.028)were independent risk factors for biochemical recurrence.The predictive equation was established with an AUC of 0.743(95%CI:0.677~0.810,P<0.001),a sensitivity of 0.826 and a specificity of 0.534.And the established nomogram has a C-index of 0.741(95%CI:0.677~0.805;P<0.001),with good agreement indicated by calibration curve.Clinical decision curve displayed that there was a net benefit when the threshold probability was 15%~85%.Conclusion High pathological stage(≥pT3 stage),positive surgical margin,lymph node invasion,and NRI≤106.9 are independent risk factors for biochemical recurrence of prostate cancer after RP.Our prediction model can make effective judgments and provide references for clinical decision making.