胃肠肿瘤免疫微环境多靶点精准分型及临床治疗策略适配性研究
Multi-target Precision Classification of the Immune Microenvironment in Gastrointestinal Tumors and Its Clinical Adaptability to Treatment Strategies
摘要
背景与目的： 程序性死亡受体-1（PD-1）/程序性死亡配体-1（PD-L1）抑制剂在高度微卫星稳定（MSS）/错配修复正常（pMMR）型胃肠肿瘤中获益有限，传统单一标志物筛选体系难以满足精准治疗需求。本研究旨在构建包含调节性T细胞（Treg）标志物的多靶点免疫微环境分型体系，分析分型与患者临床特征、治疗响应及预后的关联性，为胃肠肿瘤个体化免疫治疗提供依据。方法： 纳入180例晚期胃腺癌/结直肠癌患者的存档组织样本，检测CD8、PD-L1、LAG-3、TIGIT及Foxp3（Treg标志物）五项指标，采用无监督聚类分析进行免疫微环境分型。收集患者人口学特征、肿瘤病理特征、治疗史、实验室检查结果及生存数据，分析不同分型与临床特征的关联性、治疗方案适配性及预后价值。构建列线图预测模型，通过ROC曲线和决策曲线分析（DCA）验证分型体系的预测效能。结果： 根据免疫组化分析结果，患者被分为5个免疫亚型：免疫激活型（16.7%，30/180）、共抑制型（30.0%，54/180）、免疫耗竭型（22.2%，40/180）、免疫失衡型（15.0%，27/180）和免疫荒漠型（16.1%，29/180）。各亚型与肿瘤分化程度、转移状态、ECOG评分及肿瘤标志物（CEA、CA19-9）水平均存在显著关联（P均<0.05）。免疫激活型患者中位无进展生存期（mPFS）为12.5个月，中位总生存期（mOS）为28.0个月；免疫荒漠型mPFS仅2.0个月，mOS仅7.0个月。多因素Cox回归分析显示，免疫分型是独立预后因素，荒漠型、失衡型、耗竭型相对于激活型的风险比（HR）分别为2.68（P<0.001）、2.15（P=0.002）和1.82（P=0.010）。免疫激活型患者PD-1抑制剂单药治疗疾病控制率（DCR）达64.7%；共抑制型患者PD-1抑制剂联合化疗DCR为40.0%，显著优于单纯化疗的31.5%（P<0.05）。与传统PD-L1 CPS相比，本研究分型体系预测疾病控制率的AUC显著更高（0.852 vs 0.644，ΔAUC=0.208，P<0.001），净重新分类指数（NRI）为0.42。在108例PD-L1阴性患者中，分型体系识别出28.3%（31/108）的共抑制型潜在获益人群。列线图模型整合免疫分型、ECOG评分和CEA水平，可个体化预测6个月PFS率。
结论： 包含Treg标志物的多靶点免疫分型体系可更精准地反映胃肠肿瘤免疫微环境特征，与患者临床特征、治疗适配性及预后密切相关，为胃肠肿瘤精准治疗策略制定提供了重要参考。
Abstract
Background & Objective: Programmed death receptor-1 (PD-1)/programmed death ligand-1 (PD-L1) inhibitors show limited clinical benefit in gastrointestinal tumors with microsatellite stable (MSS)/proficient mismatch repair (pMMR) status. Traditional single-biomarker screening systems fail to meet the requirements for precision therapy. This study aimed to establish a multi-target immune microenvironment classification system incorporating regulatory T cell (Treg) markers, analyze the correlations between the classification and patients' clinical characteristics, treatment response, and prognosis, and provide evidence for individualized immunotherapy of gastrointestinal tumors. Methods: Archived tissue samples from 180 patients with advanced gastric adenocarcinoma or colorectal cancer were enrolled. Five markers, including CD8, PD-L1, LAG-3, TIGIT, and Foxp3 (a Treg marker), were detected, and unsupervised clustering analysis was used to classify the immune microenvironment. Demographic characteristics, tumor pathological features, treatment history, laboratory test results, and survival data were collected to analyze the associations between different subtypes and clinical features, treatment suitability, and prognostic value. A nomogram prediction model was constructed, and the predictive performance of the classification system was validated by ROC curve and decision curve analysis (DCA). Results: Based on immunohistochemical analysis, patients were stratified into 5 immune subtypes: immune-activated type (16.7%, 30/180), co-inhibitory type (30.0%, 54/180), immune-exhausted type (22.2%, 40/180), immune-imbalanced type (15.0%, 27/180), and immune-desert type (16.1%, 29/180). Significant associations were observed between each subtype and tumor differentiation, metastatic status, ECOG score, and levels of tumor markers (CEA, CA19-9) (all P<0.05). The median progression-free survival (mPFS) for the immune-activated type was 12.5 months, and median overall survival (mOS) was 28.0 months; for the immune-desert type, mPFS was only 2.0 months and mOS was only 7.0 months. Multivariate Cox regression analysis showed that immune classification was an independent prognostic factor, with hazard ratios (HR) for desert, imbalanced, and exhausted types relative to the activated type of 2.68 (P<0.001), 2.15 (P=0.002), and 1.82 (P=0.010), respectively. Patients with the immune-activated type achieved a disease control rate (DCR) of 64.7% with PD-1 inhibitor monotherapy. Those with the co-inhibitory type showed a DCR of 40.0% with PD-1 inhibitor plus chemotherapy, significantly superior to 31.5% with chemotherapy alone (P<0.05). Compared with the traditional PD-L1 CPS, the classification system in this study showed significantly higher AUC for predicting disease control rate (0.852 vs 0.644, ΔAUC=0.208, P<0.001), with a net reclassification index (NRI) of 0.42. Among 108 PD-L1-negative patients, the classification system identified 28.3% (31/108) co-inhibitory type patients as potential beneficiaries. The nomogram model integrating immune classification, ECOG score, and CEA level could individually predict 6-month PFS rate. Conclusion: A multi-target immune classification system incorporating Treg markers can more accurately reflect the characteristics of the immune microenvironment in gastrointestinal tumors, and is closely correlated with patients' clinical features, treatment suitability, and prognosis. It provides an important reference for formulating precision treatment strategies for gastrointestinal tumors.关键词
胃肠肿瘤/免疫微环境分型/多靶点检测/Foxp3/治疗适配性/预后分层Key words
Gastrointestinal Neoplasms/Immune Microenvironment Classification/Multi-target Detection/Foxp3/Treatment Adaptability/Prognostic Stratification引用本文复制引用
张林,侯艳红,吴凯,李炳慧.胃肠肿瘤免疫微环境多靶点精准分型及临床治疗策略适配性研究[EB/OL].(2026-06-08)[2026-06-13].https://www.biomedrxiv.org.cn/article/doi/10.12201/bmr.202606.00026.学科分类
肿瘤学/临床医学
