基于^(18)F-PSMA-1007 PET/CT及临床病理因素预测前列腺癌转移的诊断价值

The diagnostic value of predicting prostate cancer metastasis based on ^(18)F-PSMA-1007 PET/CT and clinical and pathological factors

目的:探讨^(18)F标记前列腺特异性膜抗原(PSMA-1007)PET/CT最大标准化摄取值(SUVmax)、平均标准化摄取值(SUVmean)结合血清前列腺特异抗原(TPSA)、国际泌尿病理协会(ISUP)分组预测前列腺癌(PCa)转移的价值。方法:回顾性分析本院2020年1月-2022年2月70例经穿刺活检病检证实为PCa且有完整^(18)F-PSMA-1007 PET/CT检查资料的初诊患者,根据^(18)F-PSMA-1007 PET/CT和随访结果分为未转移组和转移组,比较两组患者临床病理因素和SUVmax、SUVmean的差异。将单因素分析有意义(P<0.05)的因素纳入二元Logistic回归分析,建立Logistic回归模型,用受试者操作特征(ROC)曲线评价其鉴别转移的诊断效能及预测价值。应用MedCalc比较不同曲线下面积(AUC)间的差异。以P<0.05认为差异有统计学意义。结果:70例PCa患者,发生转移45例,未发生转移25例,转移组和非转移组ISUP分组(χ^(2)=14.737,P<0.000)和TPSA分组(χ^(2)=15.240,P<0.001)差异有统计学意义,而两组间年龄差异无统计学意义。转移组和未转移组前列腺原发病灶SUVmax(t=5.972,P<0.001)、SUVmean(t=4.131,P<0.001)差异有统计学意义。根据单因素分析有统计学意义ISUP、TPSA、SUVmean、SUVmax进行二元Logistic回归分析,结果显示ISUP、TPSA、SUVmean、SUVmax均进入预测模型,提示ISUP、TPSA、SUVmean、SUVmax都是PCa转移的独立预测因子,最有预测价值的特征参数是SUVmax,OR=3.008(95%CI 1.584~5.710),P=0.001;其次是SUVmean OR=0.273(95%CI 0.116~0.645),P=0.003。TPSA OR=0.148(95%CI 0.026~0.787),P=0.025。ISUP OR=0.156(95%CI 0.026~0.918),P=0.040。ROC曲线分析单独SUVmax、SUVmean、TPSA、ISUP诊断PCa转移的AUC分别为0.848、0.811、0.806、0.749,敏感度分别为80.0%、86.7%、82.2%、84.4%,特异度分别为76.0%,64.0%,72.0%、56.0%。而预测模型对PCa转移阳性预测诊断的AUC为0.942,敏感度和特异度为80.0%和92.0%。MedCalc对单独SUVmax、SUVmean、TPSA、ISUP和预测模型AUC分别进行比较,差异均具有统计学意义,Z值分别为2.014(95%CI 0.003~0.187)、2.368(95%CI 0.023~0.240)、2.733(95%CI 0.039~0.234)、3.389(95%CI 0.082~0.305),P值分别为0.044、0.018、0.006、0.001。结论:基于^(18)F-PSMA-1007 PET/CT SUVmax、SUVmean、TPSA和ISUP的预测模型对PCa转移有较好的阳性预测价值,可为临床预测PCa转移发生概率提供理论依据。

Objective:To explore the value of ^(18)F labeled prostate specific membrane antigen(PSMA-1007)PET/CT standardized uptake maximum(SUVmax),standardized mean uptake(SUVmean)combined with serum prostate specific antigen(TPSA)and International Association of Urological Pathology(ISUP)group in predicting prostate cancer(PCa)metastasis.Methods:70 newly diagnosed PCa patients with complete ^(18)F-PSMA-1007 PET/CT data confirmed by biopsy in our hospital from January 2020 to February 2022 were divided into non-metastasis group and metastasis group according to ^(18)F-PSMA-1007 PET/CT and follow-up results.The clinical pathological factors and the differences in SUVmax and SUVmean between the two groups were compared.The significant clinical pathological factors of univariate analysis were included in binary logistic regression analysis,and the logistic regression model was established.The diagnostic efficacy and predictive value of differential metastasis were evaluated by receiver operating characteristic(ROC)curve.MedCalc was used to compare the differences in the areas under curves(AUCs).The difference was considered statistically significant(P<0.05).Results:Among the 70 patients with PCa,45 cases had metastasis and 25 cases had no metastasis.There were significant differences in the score of the ISUP group(χ^(2)=14.737,P<0.000)and TPSA(χ^(2)=15.240,P<0.000)between metastasis group and non-metastasis group,but no significant difference in age.There were significant differences in SUVmax(t=5.972,P<0.000)and SUVmean(t=4.131,P<0.000)between metastatic group and non-metastatic group.ISUP,TPSA,SUVmean and SUVmax were analyzed by binary logistic regression analysis.The results showed that ISUP,TPSA,SUVmean and SUVmax entered the prediction model,suggesting that TPSA,ISUP,SUVmean and SUVmax were independent predictors of PCa metastasis,and the most valuable characteristic parameter was SUVmax,OR=3.008(95%CI:1.584~5.710),P=0.001.The second valuable parameter was SUVmean,OR=0.273(95%CI:0.116~0.645),P=0.003.The remaining two parameters had OR values of 0.148(95%CI:0.026~0.787,P=0.025)and OR=0.156(95%CI:0.026~0.918,P=0.040)respectively.According to ROC curve analysis,the AUCs of SUVmax,SUVmean,TPSA and ISUP for diagnosing PCa metastasis were 0.848,0.811,0.806 and 0.749,respectively,the sensitivities were 80.0%,86.7%,82.2%and 84.4%,and the specificities were 76.0%,64.0%,72.0%and 56.0%,respectively.The AUC value of the regression model for positive predictive diagnosis of PCa metastasis was 0.942,and the sensitivity and specificity were 80.0%and 92.0%,respectively.MedCalc compared the differences in AUC values of individual SUVmax,SUVmean,TPSA,ISUP and prediction model.The AUC of the prediction models differed significantly from individual SUVmax,SUVmean,TPSAL,and ISUP,with Z values of 2.014(95%CI:0.003~0.187),2.368(95%CI:0.023~0.240),2.733(95%CI:0.039~0.234),3.389(95%CI:0.082~0.305)with P values of 0.044,0.018,0.006 and 0.001.Conclusion:The predictive model based on ^(18)F-PSMA-1007 PET/CT SUVmax,SUVmean,TPSA and ISUP had good positive predictive value for PCa metastasis,and can provide theoretical basis for clinical prediction of PCa metastasis probability.