Macarrón Víctor, Losantos-García Itsaso, Peláez-García Alberto, Yébenes Laura, Berjón Alberto, Frías Laura, Martí Covadonga, Zamora Pilar, Sánchez-Méndez José Ignacio, Hardisson David
Department of Pathology, Hospital Universitario La Paz, 28046 Madrid, Spain.
Biostatistics Department, Hospital Universitario La Paz, 28046 Madrid, Spain.
Cancers (Basel). 2025 Jan 16;17(2):273. doi: 10.3390/cancers17020273.
: The EndoPredict assay has been widely used in recent years to estimate the risk of distant recurrence and the absolute chemotherapy benefit for patients with estrogen (ER)-positive/human epidermal growth factor receptor 2 (HER2)-negative breast cancer. However, there are no well-defined criteria for selecting patients who may benefit from the test. The aim of this study was to develop a novel nomogram to estimate the probability of obtaining a high-risk EndoPredict result in clinical practice. : The study cohort comprised 348 cases of T1-3/N0-1a/M0 ER-positive/HER2-negative breast carcinoma. A multivariate analysis was conducted using a training cohort (n = 270) based on clinicopathological features that demonstrated a statistically significant correlation with the EndoPredict result in a univariate analysis. The predictive model was subsequently represented as a nomogram to estimate the probability of obtaining a high-risk result in the EndoPredict assay. The predictive model was then validated using a separate validation cohort (n = 78). : The clinicopathological features incorporated into the nomogram included tumor size, tumor grade, sentinel lymph node status, pN stage, and Ki67. The internal validation of the model yielded an area under the curve (AUC) of 0.803 (95% CI = 0.751, 0.855) in the receiver operating characteristic (ROC) curve for the training cohort, with an optimal sensitivity and specificity at a threshold of 0.536. The external validation yielded an AUC of 0.789 (95% CI = 0.689, 0.890) in its ROC curve, with optimal sensitivity and specificity achieved at a threshold of 0.393. : This study presents, for the first time, the development of a clinically accessible nomogram designed to estimate the probability of obtaining a high-risk result in the EndoPredict assay. The use of easily available clinicopathological features allows for the optimization of patient selection for the EndoPredict assay, ensuring that those who would most benefit from undergoing the test are identified.
近年来，EndoPredict检测已被广泛用于评估雌激素（ER）阳性/人表皮生长因子受体2（HER2）阴性乳腺癌患者远处复发的风险以及化疗的绝对获益。然而，对于选择可能从该检测中获益的患者，尚无明确的标准。本研究的目的是开发一种新型列线图，以估计在临床实践中获得高风险EndoPredict结果的概率。
研究队列包括348例T1-3/N0-1a/M0 ER阳性/HER2阴性乳腺癌病例。基于单变量分析中与EndoPredict结果具有统计学显著相关性的临床病理特征，使用训练队列（n = 270）进行多变量分析。随后将预测模型表示为列线图，以估计在EndoPredict检测中获得高风险结果的概率。然后使用单独的验证队列（n = 78）对预测模型进行验证。
纳入列线图的临床病理特征包括肿瘤大小、肿瘤分级、前哨淋巴结状态、pN分期和Ki67。模型的内部验证在训练队列的受试者工作特征（ROC）曲线中得到的曲线下面积（AUC）为0.803（95%CI = 0.751, 0.855），在阈值为0.536时具有最佳敏感性和特异性。外部验证在其ROC曲线中得到的AUC为0.789（95%CI = 0.689, 0.890），在阈值为0.393时实现了最佳敏感性和特异性。
本研究首次展示了一种临床可用列线图的开发，该列线图旨在估计在EndoPredict检测中获得高风险结果的概率。使用易于获得的临床病理特征可优化EndoPredict检测的患者选择，确保识别出最能从该检测中获益的患者。
