目的:探索早发性前列腺癌的临床病理特征;建立并验证年轻男性前列腺穿刺结果的预测模型.方法:回顾性分析2011-2021年在四川大学华西医院行前列腺穿刺活检术的213例≤50岁男性的临床影像资料,以探索与早发性前列腺癌发生相关的预测因素.美国SEER数据库中4 608例同时期早发性前列腺癌患者的数据被获取,并用于比较东西方早发性前列腺癌患者的临床病理差异.129例具有基于多参数磁共振成像(multi-parametric MRI,mpMRI)的前列腺影像报告和数据系统(prostate imaging reporting and data system,PIRADS)评分及表观弥散系数(apparent diffusion coefficient,ADC)值的患者被随机分配至训练组(91例)和验证组(38例),分别用于早发性前列腺癌预测模型的构建和验证.单因素和多因素logistic回归分析被用于模型建立.模型通过列线图进行可视化展示并通过受试者工作特征(receiver operating characteristic,ROC)曲线、校准曲线以及决策曲线分析进行验证.结果:本研究所纳入患者的中位年龄为46岁.最终46例(21.6％,46/213)检出前列腺癌.相较SEER数据库人群,本中心的早发性前列腺癌患者具有显著更高的中位前列腺特异性抗原(prostate-specific antigen,PSA)(14.2 ng/mL vs 5.7 ng/mL,P＜0.001)、初诊骨转移占比(34.1％vs 4.6％,P＜0.001)以及国际泌尿病理学会(International Society of Urological Pathology,ISUP)4～5 级占比(46％vs 8％,P＜0.001).多因素logistic回归分析结果显示,年龄(OR=1.323,95％CI:1.111～1.649,P=0.005)、前列腺特异性抗原密度(PSAD)(OR=1.718,95％CI:1.058～3.108,P=0.038)、PIRADS 4 分(OR=26.632,95％CI:3.572～584.050,P=0.006)、PIRADS 5 分(OR=46.789,95％CI:5.548～1 107.801,P=0.002)以及 ADC 值(OR=0.995,95％CI:0.991～0.999,P=0.021)是早发性前列腺癌的独立预测因素.上述临床影像参数被用于预测模型的构建.在内部验证中,模型的ROC曲线AUC值为0.842.在校准曲线以及决策曲线分析中,模型亦具有良好表现.结论:本研究构建了预测年轻男性罹患早发性前列腺癌的模型,若能在未来大样本的外部队列中得到有效验证,这一模型将有望成为辅助前列腺穿刺抉择的工具.
Clinical and pathological characteristics of early-onset prostate cancer and construction of a predictive model
Objective:To explore the clinical and pathological characteristics of early-onset prostate cancer(PCa)and establish a predictive model for prostate biopsy results in men ≤50 years.Methods:Clinical and multi-parametric MRI(mpMRI)data of 213 men aged ≤50 years who underwent prostate biopsy in our hospital were retrospectively analyzed to explore predictive factors associated with early-onset PCa.Data of 4 608 early-onset PCa patients from the U.S.SEER database were obtained for comparing clinical and pathological differences be-tween Western and Eastern populations.One hundred and twenty-nine patients with prostate imaging reporting and data system(PIRADS)scores and apparent diffusion coefficient(ADC)values based on mpMRI were randomly assigned to the training group(n=91)and validation group(n=38)for the construction and validation of the ear-ly-onset PCa prediction model,respectively.Univariate and multivariate logistic regression analyses were conduc-ted for model development.The model was visually presented using nomograms and validated using receiver oper-ating characteristic(ROC)curves,calibration curves,and decision curve analysis.Results:The median age of the patients included in this study was 46 years.Among 213 cases,46(21.6％)were diagnosed with PCa.Compared to the SEER database population,early-onset PCa patients in our center had significantly higher median PSA lev-els(14.2 ng/mL vs 5.7 ng/mL,P＜0.001),initial bone metastasis rate(34.1％vs 4.6％,P＜0.001),and ISUP Grade 4-5 proportion(46％vs 8％,P＜0.001).Multivariate logistic regression analysis revealed that age(OR=1.323,95％CI:1.111-1.649,P=0.005),PSA density(PSAD)(OR=1.718,95％CI:1.058-3.108,P=0.038),PIRADS 4(OR=26.632,95％CI:3.572-584.050,P=0.006),PIRADS 5(OR=46.789,95％CI:5.548-1107.801,P=0.002),and ADC(OR=0.995,95％CI:0.991-0.999,P=0.021)were inde-pendent predictors of early-onset PCa.These clinical and imaging parameters were used for model construction.The model demonstrated good performance with an area under the ROC curve(AUC)of 0.842 in internal valida-tion.The model also exhibited good calibration and decision curve analysis performance.Conclusion:This study developed a predictive model for early-onset PCa in young men.If effectively validated in large external cohorts in the future,this model could serve as a valuable tool to assist in decision-making for prostate biopsy.
万方数据
维普
