外周血CD8+ TEMRA细胞对胃癌新辅助治疗 病理响应的预测价值：一项回顾性研究

Predictive Value of Peripheral Blood CD8+ TEMRA Cells for Pathological Response to Neoadjuvant Therapy in Gastric Cancer: A Retrospective Study

摘要: 目的：探讨胃癌的新辅助治疗(NAT)下外周血免疫图谱的动态演变及其与病理退缩分级(TRG)的关联，并筛选独立预测因子，构建术前疗效评估模型。方法：回顾性纳入2023年12月至2025年3月于重庆医科大学附属第一医院胃肠外科接受新辅助治疗并完成手术切除的48例胃癌患者。按照治疗方式分为新辅助经动脉化疗栓塞联合免疫治疗组(TACE组，33例)与新辅助全身化疗联合免疫治疗组(NACT组，15例)。主要终点为病理退缩分级(TRG)，TRG 0~1级界定为病理响应良好。采用错误发现率(FDR)校正进行横断面差异分析；对18例配对样本(治疗前后)进行纵向动态比较；结合最小绝对收缩和选择算子(LASSO)回归与多因素Logistic回归筛选独立预测变量，绘制受试者工作特征(ROC)曲线，并通过留一法交叉验证(LOOCV)评估模型内部判别效能。结果：全队列中17例(35.4%)实现病理响应良好。横断面多重检验校正提示，术前外周血CD8+ TEMRA细胞比例在响应良好组中显著富集，且为唯一经校正后保持统计学意义的亚群(FDR < 0.05)。纵向配对分析显示，新辅助干预后效应/活化表型细胞(NK、CD8+ HLA-DR+、CD8+ CD38+)比例显著上调，而初始/耗竭表型细胞(CD19+、CD8+ Naive、CD8+ PD-1+)比例显著回落，其中CD8+ PD-1+ T细胞下调幅度最为显著(FDR < 0.001)。组间比较显示，TACE组NK细胞比例显著高于NACT组(FDR = 0.014)，但CD8+ TEMRA分布无方案依赖性差异。LASSO降维及多因素Logistic回归确证，CD8+ TEMRA是预测良好病理响应的独立相关因素(OR = 1.12, 95% CI: 1.040~1.200, P = 0.003)。联合预测模型全队列曲线下面积(AUC)为0.852，经LOOCV内部惩罚后校正AUC为0.824。结论：术前外周血CD8+ TEMRA细胞富集程度与胃癌新辅助治疗后的病理良好响应呈独立正相关。基于该指标构建的联合预测模型在内部验证中展现出稳定的判别效能，可作为无创性外周标志物辅助胃癌围手术期疗效评估，但其临床应用仍需在更大样本中、结合标准化流式细胞术检测流程及明确的临床决策阈值后进一步验证。

Abstract: Objective: To investigate the dynamic evolution of peripheral blood immune profiles during neoadjuvant therapy (NAT) in gastric cancer and its association with tumor regression grade (TRG), and to identify independent predictors for constructing a preoperative efficacy assessment model. Methods: A retrospective cohort of 48 gastric cancer patients who received neoadjuvant therapy and underwent surgical resection at the Department of Gastrointestinal Surgery, The First Affiliated Hospital of Chongqing Medical University between December 2023 and March 2025 was enrolled. Patients were stratified by treatment modality into neoadjuvant transarterial chemoembolization plus immunotherapy group (TACE group, n = 33) and neoadjuvant systemic chemotherapy plus immunotherapy group (NACT group, n = 15). The primary endpoint was tumor regression grade (TRG), with TRG 0~1 defined as favorable pathological response. Cross-sectional differential analysis was performed with false discovery rate (FDR) correction; longitudinal dynamic comparison was conducted in 18 paired samples (pre- and post-treatment). Independent predictive variables were identified through least absolute shrinkage and selection operator (LASSO) regression combined with multivariate logistic regression. Receiver operating characteristic (ROC) curves were generated, and internal discriminative performance was evaluated via leave-one-out cross-validation (LOOCV). Results: Favorable pathological response was achieved in 17 patients (35.4%) in the entire cohort. Cross-sectional analysis with multiple testing correction revealed that preoperative peripheral blood CD8+ TEMRA cell proportion was significantly enriched in the favorable response group and remained the only subset retaining statistical significance after correction (FDR < 0.05). Longitudinal paired analysis demonstrated that effector/activated phenotype cells (NK, CD8+ HLA-DR+, CD8+ CD38+) were significantly upregulated following neoadjuvant intervention, while naive/exhausted phenotype cells (CD19+, CD8+ Naive, CD8+ PD-1+) were significantly downregulated, with CD8+ PD-1+ T cells showing the most pronounced reduction (FDR < 0.001). Inter-group comparison showed that NK cell proportion was significantly higher in the TACE group than in the NACT group (FDR = 0.014), whereas CD8+ TEMRA distribution exhibited no regimen-dependent difference. LASSO dimensionality reduction and multivariate logistic regression confirmed CD8+ TEMRA as an independent correlation of favorable pathological response (OR = 1.12, 95% CI: 1.040~1.200, P = 0.003). The combined predictive model achieved an area under the curve (AUC) of 0.852 in the entire cohort, with a corrected AUC of 0.824 after internal penalization via LOOCV. Conclusion: Preoperative peripheral blood CD8+ TEMRA cell enrichment demonstrates an independent positive correlation with favorable pathological response following neoadjuvant therapy in gastric cancer. The combined predictive model constructed based on this indicator exhibits stable discriminative performance in internal validation and may serve as a non-invasive peripheral biomarker to assist perioperative efficacy assessment in gastric cancer, but its clinical application requires further validation in larger cohorts using standardized flow cytometry workflows and an explicit clinical decision threshold.