预测根治性前列腺切除术后Gleason评分升级列线图的构建

Construction of a nomogram for predicting upgraded Gleason scores after radical prostatectomy

目的构建和验证一个预测根治性前列腺切除术(RP)术后Gleason评分升级(GSU)发生风险的列线图。方法回顾性收集2018年1月至2022年12月于武汉市中心医院接受RP治疗的179例患者的临床资料。以分层随机分组法按6∶4比例随机划分为训练集和验证集。使用LASSO-Logistic回归筛选出GSU的独立预测因子并以此构建列线图。采用受试者工作特征(ROC)曲线、霍斯默-莱梅肖(H-L)检验和决策曲线分析(DCA)对列线图进行验证。在对两组数据进行组间比较时, 符合正态分布的计量资料采用Student’st检验, 非正态分布的计量资料采用Mann-WhitneyU检验, 计数资料则使用卡方检验。结果 GSU的独立预测因子为总前列腺特异性抗原[比值比(OR):4.56, 95%可信区间(CI):1.79~11.64, P<0.01)、前列腺影像报告和数据系统(PI-RADS)评分(OR:3.47, 95%CI:1.63~7.39, P<0.01)、游离前列腺特异性抗原密度百分比(OR:4.83, 95%CI:1.07~21.74, P<0.05)、穿刺入路(OR:4.60, 95%CI:1.61~13.15, P<0.05)和活检Gleason评分(OR:4.38, 95%CI:1.29~14.90, P<0.05)。训练集和验证集的曲线下面积(AUC)分别为0.827(95%CI:0.750~0.905)和0.799(95%CI:0.688~0.910)。H-L检验显示出了列线图良好的校准度(训练集:χ^(2)=5.501, P>0.05;验证集:χ^(2)=11.986, P>0.05)。DCA结果显示, 该列线图具有良好的临床使用价值。结论本研究构建一个新的列线图预测GSU风险。经内部验证, 该列线图显示出良好的预测性能。

Objective To construct and validate a nomogram to predict the risk of Gleason score upgrading(GSU)after radical prostatectomy(RP).Methods The clinical data of 179 patients who un-derwent RP at Wuhan Central Hospital from January 2018 to December 2022 were analyzed retrospectively.The data was randomly split into a training set and a validation set at a 6:4 ratio using stratified random sampling method.Student's t-test was employed for normal distribution parametric data,Mann-Whitney U test for non-normal data,and Chi-square test for count data between groups,with P<0.05 considered sta-tistically significant.Using LASSO-Logistic regression,key GSU predictors were identified for nomogram development.The model's efficacy was assessed via receiver operating characteristic curve,Hosmer-Leme-show(H-L)test,and decision curve analysis(DCA).Results The independent predictive factors for CSU included total prostate-specific antigen[odds ratio(OR):4.56,95%confidence interval(CI):1.79-11.64,P<0.01],prostate imaging reporting and data system(PI-RADS)score(OR:3.47,95%Cl:1.63-7.39,P<0.01),percent free prostate-specific antigen density(OR:4.83,95%Cl:1.07-21.74,P<0.05),biopsy approach(OR:4.60,95%Cl:1.61-13.15,P<0.05),and biopsy Gleason score(OR:4.38,95%CI:1.29-14.90,P<0.05).The area under the curve for the training and valida-tion sets wase 0.827(95%Cl:0.750-0.905)and 0.799(95%Cl:0.688-0.910),respectively.The H-L test indicated good calibration for the nomogram(training set:χ^(2)=5.501,P>0.05;validation set:χ^(2)=11.986,P>0.05)and DCA showed its usefulness.Conclusion This study built a nomogram for GSU risk prediction,demonstrating strong performance in internal tests.