程序性细胞死亡配体-1 (PD-L1) 是乳腺癌免疫检查点阻断治疗的一个有前景的靶点。然而，乳腺癌中PD-L1表达的术前评估很少被探索。为了确定基于术前动态对比增强（DCE）MRI的放射组学特征评估乳腺癌中PD-L1表达的能力。回顾性。196个原发性乳腺癌症患者术前MRI和术后病理评估PD-L1表达，分为训练组（n = 137，28例PD-L1阳性）和测试队列（n = 59，12例PD-L1阳性）。3.0T；乳腺评估DCE序列的体积成像。利用最小冗余最大相关性方法和最小绝对收缩和选择算子算法从第一阶段DCE-MRI中提取放射组学特征。基于瘤内、瘤周以及瘤内和瘤周组合区域构建了三个放射组学特征。使用受试者工作特征 (ROC) 曲线 (AUC) 下面积、敏感性、特异性和准确性来评估特征的性能。单变量和多变量逻辑回归分析、t 检验、卡方检验、Fisher 精确检验或 Yates校正、ROC 分析和单向方差分析。 P < 0.05被认为是显着的。在测试队列中，与瘤内（AUC，0.816；P = 0.528）和瘤周放射组学特征（AUC，0.846；P = 0.905）相比，组合放射组学特征（AUC，0.853）表现出优越的性能在 PD-L1 状态评估中，尽管差异未达到统计学显着性。基于术前乳腺 MRI 的瘤内和瘤周放射组学特征显示出对乳腺癌 PD-L1 状态的无创评估具有一定的潜在准确性。4 技术效果：分期2.© 2023 国际医学磁共振学会。

Programmed cell death ligand-1 (PD-L1) is a promising target for immune checkpoint blockade therapy in breast cancer. However, the preoperative evaluation of PD-L1 expression in breast cancer is rarely explored.To determine the ability of radiomics signatures based on preoperative dynamic contrast-enhanced (DCE) MRI to evaluate PD-L1 expression in breast cancer.Retrospective.196 primary breast cancer patients with preoperative MRI and postoperative pathological evaluation of PD-L1 expression, divided into training (n = 137, 28 PD-L1-positive) and test cohorts (n = 59, 12 PD-L1-positive).3.0T; volume imaging for breast assessment DCE sequence.Radiomics features were extracted from the first phase of DCE-MRI by using the minimum redundancy maximum relevance method and least absolute shrinkage and selection operator algorithm. Three radiomics signatures were constructed based on the intratumoral, peritumoral, and combined intra- and peritumoral regions. The performance of the signatures was assessed using area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, specificity, and accuracy.Univariable and multivariable logistic regression analysis, t-tests, chi-square tests, Fisher exact test or Yates correction, ROC analysis, and one-way analysis of variance. P < 0.05 was considered significant.In the test cohort, the combined radiomics signature (AUC, 0.853) exhibited superior performance compared to the intratumoral (AUC, 0.816; P = 0.528) and peritumoral radiomics signatures (AUC, 0.846; P = 0.905) in PD-L1 status evaluation, although the differences did not reach statistical significance.Intratumoral and peritumoral radiomics signatures based on preoperative breast MRI showed some potential accuracy for the non-invasive evaluation of PD-L1 status in breast cancer.4 TECHNICAL EFFICACY: Stage 2.© 2023 International Society for Magnetic Resonance in Medicine.