基于临床及TRUS特点的前列腺结节恶性风险预测模型研究  

Risk predictive model of prostate nodules based on the clinical characteristics and transrectal ultrasonography features

作　　者：方志伟[1] 许克新[1] 胡浩[1] 夏秋翔 霍飞[1] 陈京文[1] 高等会 王晓峰[1] 

机构地区：[1]北京大学人民医院泌尿外科,北京100044

出　　处：《临床泌尿外科杂志》2015年第4期306-308,311,共4页Journal of Clinical Urology

摘　　要：目的：通过分析经直肠前列腺超声影像学特点及部分临床特征建立前列腺结节恶性风险预测模型,提高穿刺阳性率,减少不必要的前列腺穿刺。方法：回顾2010年1月-2014年1月在我院行前列腺穿刺活检且经直肠超声（TRUS）发现前列腺结节的患者151例。收集总前列腺特异性抗原（tPSA）、游离前列腺特异性抗原（fPSA）、f/t比值（fPSA/tPSA）、前列腺体积（PV）、PSA密度（PSAD）与经直肠前列腺B超发现的前列腺结节的大小、结节回声、结节位置、边缘是否清晰等资料,采用Logistic单因素分析寻找前列腺癌独立预测因子,Logistic多因素分析制定前列腺结节恶性风险预测模型。结果：142例患者入选,共170个结节纳入最终分析,其中恶性结节58例。单因素分析中结节位置、回声、边界是否清晰、lgPV、lgtPSA、f/t比值在恶性结节及良性结节组的分布差异有统计学意义,多因素分析中边界、lgtPSA、lgPV进入最终的风险预测模型。logit（前列腺癌）=4.081-1.090×边界-2.941×lgPV＋1.836×lgtPSA。结论：本研究中受试者工作特征曲线下面积0.800,与国内外同类研究总体准确性相近,具有一定的准确性。Objective： To propose a prostate cancer risk predictive model by analyzing the clinical characteris tics and transrectal ultrasonography （TRUS） features of prostatic nodules so as to improve the positive rate of punctures and prevent the unnecessary punctures. Method： A total of 151 patients with prostate nodules found by TRUS-guided biopsy in our hospital from January 2010 to January 2014 were retrospectively examined. Single factor analysis of logistic was used to screen the independent risk factors of prostate cancer including total prostate specific antigen （tPSA）, free PSA （fPSA）, f/t ratio （fPSA/tPSA）, prostate volume （PV）, PSA density （PS- AD）, diameter of prostate nodule, location of prostate nodule, nodule boundary et al. The prediction model was developed by multivariate logistic regression analysis. Result.. Totally 142 patients and 170 nodules were finally recruited to this study, and 58 prostate nodules were diagnosed as prostate cancer. There were significant differ ences between cancer group and benign lesion group in terms of lgtPSA, f/t ratio, lgPV, location of prostate nodule, nodule boundary, and ultrasound echo of prostate nodule. Nodule boundary, lgPV, lgtPSA were brought into the final prostate cancer risk prediction model. Logit （prostate cancer） = 4. 081-1. 090 X boundary-2. 941 X lg- PV＋ 1. 836 ）〈 lgtPSA. Conclusion： The whole accuracy of this prediction model was 0. 800, which is similar to the outcome of previous domestic and overseas researches. Our study can help to predict the risk factors of prostate cancer in clinical practices.

关 键 词：经直肠前列腺超声 前列腺结节 风险预测模型 

分 类 号：R697.3[医药卫生—泌尿科学]