前列腺癌穿刺活组织检查阳性率影响因素的分析及前列腺癌预测模型的建立

目的：建立前列腺癌logistic回归预测模型，为决策前列腺穿刺提供更充分的指征。方法：回顾性分析117例行前列腺穿刺活组织检查术的病例资料，按照病例收集的时间先后顺序分为两组，前2/3(78例)分入建模组，后1/3(39例)分入验证组，建立logistic回归预测模型，并通过受试者工作特征(receiver operating characteristic，ROC)曲线下面积来评估该模型的预测价值。结果：直肠指诊(digital rectal examination ，DRE)、经直肠超声(transrectal ultrasound，TRUS)、MRI、前列腺特异性抗原密度(prostate-specifi c antigen density，PSAD)以及前列腺特异性抗原游离比值(free PSA/total PSA，fPSA/tPSA)是前列腺穿刺的影响因素(P<0.01)。根据各影响因素的回归系数值，建立前列腺癌logistic回归预测模型：logit P=−2.362+2.561×DRE+1.747×TRUS+2.901×MRI+1.126×PSAD−2.569×fPSA/tPSA。ROC曲线下面积为0.907，当P的临界值取值0.12时，其敏感度、特异度分别为81.80%，89.30%。结论：利用logistic回归分析建立预测模型可以为前列腺穿刺提供更充分的指征。当该模型预测概率P>0.12时建议行前列腺穿刺活组织检查。

Analysis of influential factors for prostate biopsy and establishment of logistic regression model for prostate cancer

Objective: To establish logistic regression model for prostate cancer and provide basis for prostatebiopsy.Methods: A total of 117 cases of prostate biopsy were retrospectively analyzed in chronologicalsequence. All cases were assigned into a model group (n=78) and a validation group (n=39).Logistic regression model was established and its value was estimated by receiver operatingcharacteristic (ROC) curve.Results: Digital rectal examination(DRE), transrectal ultrasound(TRUS), MRI, prostate-specificantigen density (PSAD), and free PSA/total PSA (fPSA/tPSA) were the influential factors forprostate biopsy (P<0.01). The established logistic regression model for prostate cancer by regressioncoefficient was: logit P=−2.362+2.561×DRE+1.747×TRUS+2.901×MRI+1.126×PSAD−2.569×fPSA/tPSA and area under curve was 0.907. When the cutoff aimed at 0.12, the sensitivityand specificity were 81.80% and 89.30%, respectively.Conclusion: Logistic regression model for prostate cancer can provide sufficient basis for prostatebiopsy. Prostate biopsy should be performed when P value is more than 0.12.