超声影像组学联合模型预测前列腺癌内分泌治疗无进展生存期的价值

Value of a combined ultrasound imaging radiomics model to predict progression-free survival in endocrine therapy for prostate cancer

目的探讨超声影像组学联合模型预测前列腺癌内分泌治疗无进展生存期的价值。方法回顾性收集2018年7月至2023年1月于内蒙古自治区人民医院与呼和浩特市第一医院接受内分泌治疗的前列腺癌患者283例,其中来自内蒙古自治区人民医院的198例患者按照7∶3的比例随机划分为训练集和验证集,来自呼和浩特市第一医院的85例患者作为独立的外部测试集。依据患者内分泌治疗开始18个月后是否进展为去势抵抗性前列腺癌分为进展组和未进展组。基于超声二维图像提取筛选影像组学特征并构建影像组学评分(Rad score),分析免疫病理等临床资料,采用逻辑回归分别构建三种前列腺癌内分泌治疗无进展生存期预测模型:临床模型、超声组学模型、超声组学-临床联合模型。通过ROC曲线和临床决策曲线评估模型对前列腺癌内分泌治疗无进展生存期的预测效能和临床实用性。结果5个超声影像组学特征被纳入超声组学模型;治疗过程中TPSA的最低值(PSA nadir)、治疗前中性粒细胞与淋巴细胞比值(NLR)、肿瘤增殖细胞核抗原67(Ki-67)表达水平被纳入临床模型;超声组学模型输出结果为筛选特征所计算出的Rad score与PSA nadir、NLR、Ki-67表达水平共同构建了超声组学-临床联合模型。联合模型在训练集、验证集和外部测试集中均获得了三种模型中最高的预测性能,曲线下面积(AUC)分别为0.85、0.84,且临床决策曲线显示其临床获益良好。结论本研究基于前列腺癌内分泌治疗前二维超声图像构建的超声组学-临床联合模型能够较准确地预测前列腺癌内分泌治疗无进展生存期,可为临床诊疗前列腺癌提供客观依据。

Objective To investigate the value of the combined ultrasound imaging radiomics model for predicting progression-free survival in endocrine therapy for prostate cancer.Methods A total of 283 prostate cancer patients who received endocrine treatment at the Inner Mongolia Autonomous Region People′s Hospital and the First Hospital of Hohhot from July 2018 to January 2023 were retrospectively collected,of which 198 patients from the Inner Mongolia Autonomous Region People′s Hospital were randomly divided into the training set and the validation set according to the ratio of 7∶3,and 85 patients from the First Hospital of Hohhot served as an independent external test set.They were classified into a progression group and a non-progression group based on whether the patients progressed to desmoplasia-resistant prostate cancer 18 months after the start of endocrine treatment.Based on the two-dimensional ultrasound images,the imaging radiomics features were extracted and the imaging radiomics score(Rad-score)were constructed,the immunopathology and other clinical data were analysed,and three prediction models were constructed using logistic regression:the clinical model,the ultrasonography model,and the ultrasonography-clinical combined model,respectively.The predictive efficacy and clinical utility of the models were assessed by the ROC curves and clinical decision curves.Results Five ultrasonographic features were included in the ultrasound model;the prostate-specific antigen nadir,the neutrophil-to-lymphocyte ratio before treatment,and the expression level of tumour proliferating cell nuclear antigen 67(Ki-67)were incorporated into the clinical model;and the Rad score computed from the output of the ultrasound model for the screening features,together with the prostate-specific antigen nadir(PSA nadir),the neutrophil to lymphocyte ratio(NLR)before treatment,and the expression level of Ki-67 were used to construct the ultrasound-clinical joint model.The joint model achieved the highest predictive performance in both the training and validation sets of the three groups of models,with the area under the curve of 0.85 and 0.84,and the clinical decision curve showed good clinical benefit.Conclusions The combined ultrasound-clinical model constructed in this study based on two-dimensional ultrasound images of prostate cancer before endocrine therapy can predict progression-free survival of endocrine therapy for prostate cancer more accurately.