免疫检查点抑制剂在多种肿瘤治疗中展现出良好的疗效。生物标志物是用于选择接受全身性抗癌治疗的生物学指标，但只有少数一些临床上有用的生物标志物，如PD-L1表达和肿瘤突变负荷，可用于预测免疫治疗的反应。本研究建立了一个包含基因表达和临床数据的数据库，以识别对抗PD-1、抗PD-L1和抗CTLA-4免疫治疗有响应的生物标志物。对包含同时具有临床反应和转录组数据的数据集进行GEO筛选，不考虑癌症类型。筛选仅限于涉及抗PD-1（nivolumab、pembrolizumab）、抗PD-L1（atezolizumab、durvalumab）或抗CTLA-4（ipilimumab）药物的研究。执行接收器操作特征（ROC）分析和曼-惠特尼检验，以识别与治疗反应相关的特征。该数据库由19个数据集和1434个肿瘤组织样本组成，涉及食管、胃、头颈、肺、尿路和黑色素瘤，包括基因表达和临床数据。免疫治疗期间与抗PD-1耐药性相关的最强基因候选者为SPIN1 (AUC=0.682，P=9.1E-12)、SRC (AUC=0.667，P=5.9E-10)、SETD7 (AUC=0.663，P=1.0E-09)、FGFR3 (AUC=0.657，P=3.7E-09)、YAP1 (AUC=0.655，P=6.0E-09)、TEAD3 (AUC=0.649，P=4.1E-08)和BCL2 (AUC=0.634，P=9.7E-08)。在抗CTLA-4治疗队列中，BLCAP (AUC=0.735，P=2.1E-06)是最有前途的基因候选者。在抗PD-L1队列中未发现具有治疗相关性的靶标预测。在抗PD-1组中，我们能够证实错配修复基因MLH1和MSH6与生存的显著相关性。我们建立了一个Web平台，用于进一步分析和验证新的生物标志物候选者。总之，我们建立了一个数据库和Web平台，以调查大型固体肿瘤样本中的免疫治疗反应的生物标志物。我们的研究结果有助于确定新的患者群体，使其适合接受免疫治疗。© 2023 The Author(s).

Immune-checkpoint inhibitors show promising effects in the treatment of multiple tumor types. Biomarkers are biological indicators used to select patients for a systemic anticancer treatment, but there are only a few clinically useful biomarkers such as PD-L1 expression and tumor mutational burden, which can be used to predict immunotherapy response. In this study, we established a database consisting of both gene expression and clinical data to identify biomarkers of response to anti-PD-1, anti-PD-L1, and anti-CTLA-4 immunotherapies. A GEO screening was executed to identify datasets with simultaneously available clinical response and transcriptomic data regardless of cancer type. The screening was restricted to the studies involving administration of anti-PD-1 (nivolumab, pembrolizumab), anti-PD-L1 (atezolizumab, durvalumab) or anti-CTLA-4 (ipilimumab) agents. Receiver operating characteristic (ROC) analysis and Mann-Whitney test were executed across all genes to identify features related to therapy response. The database consisted of 1434 tumor tissue samples from 19 datasets with esophageal, gastric, head and neck, lung, and urothelial cancers, plus melanoma. The strongest druggable gene candidates linked to anti-PD-1 resistance were SPIN1 (AUC = 0.682, P = 9.1E-12), SRC (AUC = 0.667, P = 5.9E-10), SETD7 (AUC = 0.663, P = 1.0E-09), FGFR3 (AUC = 0.657, P = 3.7E-09), YAP1 (AUC = 0.655, P = 6.0E-09), TEAD3 (AUC = 0.649, P = 4.1E-08) and BCL2 (AUC = 0.634, P = 9.7E-08). In the anti-CTLA-4 treatment cohort, BLCAP (AUC = 0.735, P = 2.1E-06) was the most promising gene candidate. No therapeutically relevant target was found to be predictive in the anti-PD-L1 cohort. In the anti-PD-1 group, we were able to confirm the significant correlation with survival for the mismatch-repair genes MLH1 and MSH6. A web platform for further analysis and validation of new biomarker candidates was set up and available at https://www.rocplot.com/immune . In summary, a database and a web platform were established to investigate biomarkers of immunotherapy response in a large cohort of solid tumor samples. Our results could help to identify new patient cohorts eligible for immunotherapy.© 2023. The Author(s).