血清生化指标对前列腺癌风险预测的比较研究

Comparative study on serum biochemical indicators for prostate cancer risk prediction

目的评估穿刺前生化指标对前列腺癌患病风险的预测价值并构建诺莫图模型。方法回顾性分析2021年9月至2022年12月于苏州大学附属第一医院行经会阴前列腺穿刺活检患者的临床资料,共纳入479例,其中340例用于建模分析,139例作为独立的外部验证集。建模数据集中,根据病理结果的不同将两组数据分为阳性组(前列腺癌)170例和阴性组(良性前列腺增生)170例。通过Logistic单因素和多因素分析,寻找前列腺癌的独立风险因素,并绘制诺莫图,通过受试者工作特征曲线检验模型的效能。结果阴性、阳性两组数据之间年龄、游离前列腺特异性抗原/总前列腺特异性抗原比值(f/t PSA)、前列腺体积、基于多参数磁共振的前列腺影像学和数据评分系统、血清磷、谷氨酰转肽酶为前列腺癌的独立风险因素(P<0.05),绘制Logistic多因素模型的诺莫图,用于前列腺癌患病的风险预测。内部验证提示模型拟合度较好(平均绝对误差=0.011,n=340),外部验证表明所得曲线与理想线相接近(平均绝对误差=0.045,n=139)。受试者工作特征曲线结果显示,全模型的曲线下面积最高(AUC=0.874),与基线模型及各单因素模型相比,具有较好的效益。进一步绘制决策曲线及临床影响曲线表明,全模型的临床获益均高于其他模型。结论本研究联合血清磷、谷氨酰转肽酶以及年龄、f/t PSA、前列腺体积、磁共振PI-RADS评分等构建前列腺癌风险预测模型,具有较好的前列腺癌诊断效能。

Objective To evaluate the risk predictive value of preoperative biochemical indicators for prostate cancer(PCa)and to construct a nomogram model.Methods The clinical data of patients undergoing perineal prostate biopsy in the First Affiliated Hospital of Soochow University from September 2021 to December 2022 were analyzed retrospectively.A total of 479 patients were included,among which 340 were used for model training and 139 were used as an independent external validation set.Based on different pathological results,the training patients were divided into two groups,i.e.,the positive group with 170 PCa cases and the negative group with 170 prostate hyperplasia cases.The independent risk factors in PCa were screened using Logistic univariate and multivariate analyses,and a nomogram was constructed.The reliability of the model was validated by receiver operating characteristic(ROC)curve analysis.Results Age,free prostate-specific antigen total Prostale-specific antigen(f/tPSA),prostate volume(PV),prostate imaging reporting and data system,serum phosphorus(P)andγ-glutamyl transpeptadase(GGT)between negative and positive groups were identified as independent risk factors in PCa(P<0.05).A nomogram was constructed from Logistic multi-factor analysis for PCa risk prediction.The internal validation indicated the high fitting of the model(mean absolute error=0.011,n=340),and the external verification showed that the curve was close to the ideal line(mean absolute error=0.045,n=139).ROC curves showed the highest overall area under the curve(AUC)of the whole model(AUC=0.874)and it had better benefits compared with the baseline model and each single factor model.Further decision curve and clinical impact curve showed that the clinical benefit of the whole model was higher than that of other models.Conclusions The model for PCa risk prediction constructed by integrating serum P,GGT,age,f/tPSA,PV and PI-RADS showed good efficacy for PCa detection.