Current gastric cancer staging alone cannot predict prognosis and adjuvant chemotherapy benefits in stage II and III gastric cancer. Tumor immune microenvironment biomarkers and tumor-cell chemosensitivity might add predictive value to staging. This study aimed to construct a predictive signature integrating tumor immune microenvironment and chemosensitivity-related features to improve the prediction of survival and adjuvant chemotherapy benefits in patients with stage II to III gastric cancer. We used IHC to assess 26 features related to tumor, stroma, and chemosensitivity in tumors from 223 patients and evaluated the association of the features with disease-free survival (DFS) and overall survival (OS). Support vector machine (SVM)-based methods were used to develop the predictive signature, which we call the SVM signature. Validation of the signature was performed in two independent cohorts of 445 patients. The diagnostic signature integrated seven features: CD3+ cells at the invasive margin (CD3 IM), CD8+ cells at the IM (CD8 IM), CD45RO+ cells in the center of tumors (CD45RO CT), CD66b+ cells at the IM (CD66b IM), CD34+ cells, periostin, and cyclooxygenase-2. Patients fell into low- and high-SVM groups with significant differences in 5-year DFS and OS in the training and validation cohorts (all P < 0.001). The signature was an independent prognosis indicator in multivariate analysis in each cohort. The signature had better prognostic value than various clinicopathologic risk factors and single features. High-SVM patients exhibited a favorable response to adjuvant chemotherapy. Thus, this SVM signature predicted survival and has the potential for identifying patients with stage II and III gastric cancer who could benefit from adjuvant chemotherapy.

中文翻译：

---

肿瘤免疫微环境和化学敏感性签名可预测胃癌对化学疗法的反应。

仅目前的胃癌分期不能预测II期和III期胃癌的预后和辅助化疗的益处。肿瘤免疫微环境生物标志物和肿瘤细胞化学敏感性可能为分期增加预测价值。这项研究的目的是构建一个结合肿瘤免疫微环境和化学敏感性相关特征的预测特征，以提高对II至III期胃癌患者生存率和辅助化疗获益的预测。我们使用IHC评估了来自223位患者的26种与肿瘤，间质和化学敏感性相关的特征，并评估了这些特征与无病生存期（DFS）和总生存期（OS）的关联。使用基于支持向量机（SVM）的方法来开发预测签名，我们将其称为SVM签名。在445名患者的两个独立队列中对签名进行了验证。诊断签名整合了七个功能：侵入性边缘的CD3 +细胞（CD3 IM），IM的CD8 +细胞（CD8 IM），肿瘤中心的CD45RO +细胞（CD45RO CT），IM的CD66b +细胞（CD66b IM）， CD34 +细胞，骨膜素和环氧合酶2。患者分为低和高SVM组，在训练和验证队列中的5年DFS和OS有显着差异（所有P <0.001）。在每个队列的多变量分析中，签名是一个独立的预后指标。该签名具有比各种临床病理危险因素和单一特征更好的预后价值。高SVM患者对辅助化疗表现出良好的反应。因此，