前列腺癌预测新模型的建立及其评估

Establishment and evaluation of a new model for predicting prostate cancer

目的：联合应用血清总前列腺特异性抗原（tPSA）,游离前列腺特异性抗原（fPSA）及结合前列腺特异性抗原（cPSA）,构建出新的前列腺癌（PCa）预测模型：tPSA/fPSA×cPSA1/2,即前列腺癌预测因子（PCP）,进而评估PCP诊断PCa的临床应用价值。方法：从我院2011年12月1日～2014年12月1日期间的病例中,筛选出拥有明确病理报告且术前血清前列腺特异性抗原（PSA）检查结果满足2≤tPSA〈10ng/ml的PCa患者54例,BPH患者579例;满足10≤tPSA〈20ng/ml的PCa患者48例,BPH患者147例。应用Logestic regression及ROC曲线对比分析tPSA、fPSA、fPSA与tPSA的比值（f/tPSA）、cPSA及PCP诊断PCa的价值,应用列线图分析PCP预测PCa的能力。结果：在PCa及BPH两组间,tPSA、f/tPSA、cPSA与PCP的中位数（9.2,5.5;0.13,0.19;8.0,4.4;22.1,11.0）均存在明显差异（P〈0.000）。当2≤tPSA〈10ng/ml时：PCP的ROC曲线下面积（AUC）为0.680,明显高于tPSA的0.588、fPSA的0.571、f/tPSA的0.675、cPSA的0.613;在诊断PCa的敏感性均为90.7%左右的前提下,PCP诊断PCa的特异性为22.8%,明显高于tPSA的11.1%、fPSA的11.2%、f/tPSA的17.4%、cPSA的15.5%。当10≤tPSA〈20ng/ml时：PCP的ROC曲线下面积为0.686,明显高于tPSA的0.603、fPSA的0.643、f/tPSA的0.679、cPSA的0.647;在诊断PCa的敏感性均为91%左右的前提下,PCP诊断PCa的特异性为29.3%,明显高于tPSA的10.9%、fPSA的10.2%、f/tPSA的23.1%、cPSA的18.4%。结论：当2≤tPSA〈10ng/ml或10≤tPSA〈20ng/ml时,PCP明显较PSA、fPSA、f/tPSA、cPSA对PCa更具预测价值。

Objective：To build a new prostate cancer predictor（PCP）model for predicting prostate cancer（PCa）by using the combination of total prostate specific antigen（tPSA）,free PSA（fPSA）and complexed PSA（cPSA）.Method：The diagnoses of all the included patients were confirmed by pathology in our hospital from December 1st,2011 to December 1st,2014.There were 54 cases of PCa and 579 cases of BPH whose tPSA were among 2-10ng/ml,while there were 48 cases of PCa and 147 cases of BPH whose tPSA were among 10-20ng/ml.Logistic regression and receiver operating characteristic curve（ROC）were employed to copmpare the value of PCP with tPSA,fPSA,the ratio of fPSA to tPSA（f/tPSA）and cPSA in the differential diagnosis of PCa and BPH.Nomogram was employed to analyse capability of PCP for predicting PCa.Result：When tPSA was 2-10ng/ml,the area under the curve（AUC）of PCP（0.680）was significantly higher than tPSA（0.588）,fPSA（0.571）,f/tPSA（0.675）and cPSA（0.613）.When the sensitivity for diagnosis of PCa was about 90.7%,the specificity of PCP（22.8%）was apparently higher than that of tPSA（11.1%）,fPSA（11.2%）,f/tPSA（17.4%）and cPSA（15.5%）.When tPSA was 10-20ng/ml,the AUC of PCP（0.686）was significantly higher than that of tPSA（0.603）,fPSA（0.643）,f/tPSA（0.679）and cPSA（0.647）.When the sensitivity for diagnosis of PCa was about 91%,the specificity of PCP（29.3%）was apparently higher than that of tPSA（10.9%）,fPSA（10.2%）,f/tPSA（23.1%）and cPSA（18.4%）.Conclusion：PCP is a novel model for predicting PCa and has more predictive value than tPSA,fPSA,f/tPSA and cPSA when tPSA is among 2-20ng/ml.