Department of Radiology, The First Affiliated Hospital of Zhengzhou University, No. 1, East Jianshe Rd, Zhengzhou 450052, Henan Province, China.
Present address: Department of Radiology, Jiangxi Provincial People's Hospital, Nanchang, Jiangxi Province, China.
AJR Am J Roentgenol. 2021 Jun;216(6):1539-1548. doi: 10.2214/AJR.20.23528. Epub 2021 Apr 14.
The purpose of this study was to develop and evaluate a dual-energy CT (DECT)-based nomogram for noninvasive identification of the status of human epidermal growth factor receptor 2 (HER2; also known as ERBB2) expression in gastric cancer (GC). A total of 206 patients with histologically proven GC who underwent pretreatment DECT were retrospectively recruited and randomly allocated to a training cohort ( = 144) or a test cohort ( = 62). Information on clinical characteristics, qualitative imaging features, and quantitative DECT parameters was collected. Univariate analysis and multivariate logistic regression were implemented to screen independent predictors of HER2 status. An individualized nomogram was built, and its discrimination, calibration, and clinical usefulness were assessed. Tumor location, the iodine concentration of the tumor in the venous phase, and the normalized iodine concentration of the tumor in the venous phase were significant factors predictive of HER2 status (all < .05). After these three indicators were integrated, the proposed nomogram showed a favorable diagnostic performance, with AUCs of 0.807 (95% CI, 0.718-0.897) in the training cohort and 0.815 (95% CI, 0.661-0.968) in the test cohort. The nomogram showed a preferable fitting (all .05 by the Hosmer-Lemeshow test) and would offer more net benefits than simple default strategies within a wide range of threshold probabilities in both cohorts. The DECT-based nomogram has great application potential in terms of detecting HER2 status in GC, and can serve as a novel substitute for invasive testing.
本研究旨在开发和评估一种基于双能 CT（DECT）的列线图，用于无创识别胃癌（GC）中人类表皮生长因子受体 2（HER2；也称为 ERBB2）表达的状态。共回顾性招募了 206 名经组织学证实的 GC 患者，这些患者在预处理 DECT 前接受了治疗，并将其随机分配到训练队列（n = 144）或测试队列（n = 62）。收集了有关临床特征、定性成像特征和定量 DECT 参数的信息。进行了单变量分析和多变量逻辑回归，以筛选 HER2 状态的独立预测因子。建立了个体化列线图，并评估了其区分度、校准度和临床实用性。肿瘤位置、肿瘤静脉期碘浓度和肿瘤静脉期标准化碘浓度是预测 HER2 状态的重要因素（均<0.05）。在整合了这三个指标后，所提出的列线图显示出良好的诊断性能，在训练队列中的 AUC 为 0.807（95%CI，0.718-0.897），在测试队列中的 AUC 为 0.815（95%CI，0.661-0.968）。列线图的拟合效果较好（Hosmer-Lemeshow 检验均<0.05），并且在两个队列中，在广泛的阈值概率范围内，都比简单的默认策略提供了更多的净收益。基于 DECT 的列线图在检测 GC 中的 HER2 状态方面具有很大的应用潜力，可以作为一种新的替代有创检测的方法。
