摘要
目的基于血尿生化指标和临床参数探寻血清前列腺特异性抗原(prostate specific antigen,PSA)位于灰区时(4~10ng/m1)前列腺癌的危险因素,筛选具有临床价值的指标,并根据相关指标构建前列腺癌的预测模型。方法选取2023年3月至2025年3月南京中医药大学附属医院(江苏省中医院)泌尿外科收治的238例血清PSA为4~10ng/ml且接受经会阴前列腺穿刺活检的患者作为研究对象,根据病理结果分为良性前列腺增生组(n=135)和前列腺癌组(n=103)。分析患者年龄、体质量指数(body mass index,BMI)、总前列腺特异性抗原(total prostate specific antigen,tPSA)、游离前列腺特异性抗原(free prostatespecific antigen,fPSA)、fPSA/tPSA、前列腺体积(prostate volume,PV)、前列腺特异性抗原密度(prostate specific antigen density,PSAD)、血尿常规、血生化、高血压、糖尿病、脑梗死、高血脂症、前列腺影像报告和数据系统2.1版(prostate imaging reporting and datasystem version2.1,PI-RADSv2.1)等指标,使用二元logistic回归分析评估对前列腺癌有预测价值的指标,进而建立诊断前列腺癌的预测模型Logit(P)并绘制ROC曲线,比较各独立变量和预测模型Logit(P)的曲线下面积,选取各因素和预测模型Logit(P)的最佳截断值,以此评估其诊断效能。结果两组患者年龄、fPSA、fPSA/tPSA、PV、PSAD、脑梗死病史、PI-RADSV2.1等因素比较,差异均有统计学意义(P<O.05),其中,年龄、fPSA/tPSA、PV、PSAD、PI-RADSv2.1是预测前列腺癌的相关因素,其用于诊断前列腺癌的曲线下面积分别为0.586、0.343、0.217、0.735、0.714。基于临床因素构建预测模型为:Logit(P)=年龄×0.074-PV×0.02+PI-RADSV2.1×O.643-fPSA/tPSA×0.062-4.531,预测模型Logit(P)的ROC曲线下面积为O.831。结论基于年龄、fPSA/tPSA、PI-RADSV2.1以及PV建立的联合预测模型的诊断效能高于单一指标,在有限的检查项目中可有效降低前列腺癌的漏诊率,减少不必要的前列腺穿刺。
Objective To identify the risk factors of prostate cancer when serum prostate-specific antigen(PSA)is in the gray zone(4-10 ng/ml)based on hematuria biochemical indicators and clinical parameters,screen out indicators with clinical value,and construct a prostate cancer prediction model using the relevant indicators.Methods The clinical data of 238 patients with serum PSA levels ranging from 4 to 10ng/ml who underwent transperineal prostate biopsy in the Department of Urology,Jiangsu Provincial Hospital of Traditional Chinese Medicine from March 2023 to March 2025 were collected.According to the pathological results,the patients were divided into the benign prostatic hyperplasia group(135 cases)and the prostate cancer group(103 cases).Indicators including age,BMI,total PSA(tPSA),free PSA(fPSA),fPSA/tPSA ratio,prostate volume(PV),PSA density(PSAD),blood routine,urine routine,blood biochemistry,hypertension,diabetes,cerebral infarction,hyperlipidemia,and PI-RADS score were analyzed.Binary logistic regression analysis was employed to evaluate the indicators with predictive value for prostate cancer.Subsequently,a prediction model Logit(P)for the diagnosis of prostate cancer was established,and receiver operating characteristic(ROC)curves of each independent variable and the prediction model Logit(P)were plotted.The area under the curve(AUC)of each independent variable and the prediction model was compared,and the optimal cut-off values of each factor and the prediction model were selected to assess their diagnostic efficacy.Results There were statistically significant differences in age,fPSA,fPSA/tPSA ratio,PV,PSAD,history of cerebral infarction,and PI-RADS score between the two groups(P<O.05).Age,fPSA/tPSA ratio,PV,PSAD,and PI-RADS score were identified as independent predictive indicators for prostate cancer.Their AUCs for the diagnosis of prostate cancer were 0.586,0.343,0.217,0.735,and 0.714,respectively.The prediction model constructed based on clinical factors was Logit(P)=0.074×age-0.02×PV+0.643×PI-RADS score-0.062×fPSA/tPSA−4.531.The AUC of the Logit(P)model was 0.831.Conclusion The diagnostic efficacy of the combined prediction model based on age,fPSA/tPSA ratio,PI-RADS score,and PV is higher than that of any single indicator.When examination items are limited,this model can effectively reduce the missed diagnosis rate of prostate cancer and decrease unnecessary prostate punctures.
作者
徐渭
张犁
Xu Wei;Zhang Li(Department of Urology,Affiliated Hospital of Nanjing University of Chinese Medicine,Nanjing,Jiangsu 210000,China)
出处
《泌尿外科杂志(电子版)》
2025年第4期31-37,共7页
Journal of Urology for Clinicians(Electronic Version)
关键词
前列腺癌
前列腺特异性抗原
临床因素分析
Prostate cancer
Prostate-specific antigen
Analysis of clinical factors
