A distinct dimension of immunotherapeutic biomarker: Beyond immune infiltration and tumor antigenicity.

person Anlin Li Department of Medical Oncology, Sun Yat-sen University Cancer Center, Guangzhou, China info_outline Anlin Li, Yunpeng Yang, Hongxi Chen, Guangyang Tang, Ruihan Hu, Kangqiao Xiong, KEHUI Chen, Wei Du, Linfeng Luo, Huaqiang Zhou, Yuwen Yang, Zhixin Yu, Shen Zhao, Wenfeng Fang, Yuanyuan Zhao, Li Zhang, Shaodong Hong

Authors person Anlin Li Department of Medical Oncology, Sun Yat-sen University Cancer Center, Guangzhou, China info_outline Anlin Li, Yunpeng Yang, Hongxi Chen, Guangyang Tang, Ruihan Hu, Kangqiao Xiong, KEHUI Chen, Wei Du, Linfeng Luo, Huaqiang Zhou, Yuwen Yang, Zhixin Yu, Shen Zhao, Wenfeng Fang, Yuanyuan Zhao, Li Zhang, Shaodong Hong Organizations Department of Medical Oncology, Sun Yat-sen University Cancer Center, Guangzhou, China, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China Abstract Disclosures Research Funding National Natural Science Foundation of China Guangdong Basic and Applied Basic Research Foundation, Fundamental Research Funds for the Central Universities, Sun Yat-sen University Background: Biomarkers for immune checkpoint blockade (ICB) are classified into two categories: those reflecting inflammation (e.g., PD-L1 expression and CD8+ T cell infiltration), and those indicating antigenicity (e.g., tumor mutation burden [TMB] and neoantigen burden [NAB]). Most newly discovered ICB biomarkers are found to be associated with these two types of information. Consequently, they might simply represent alternative assessments of previously uncovered biology. Considering that ICB resistance largely occurred in 'hot' and antigenic tumors, the exploration of the third dimension of biomarker is warranted. Methods: A total of 2646 patients with RNA-seq data were collected from in-house and public randomized controlled trials (RCTs) (n=3) and cohorts (n=21) across multiple cancer types, including lung (n=1381), urinary tract (n=557), skin (n=345), kidney (n=214), stomach (n=76), liver (n=41), and esophagus (n=32). All patients received ICB mono- or combined therapy, except in the RCTs, where patients received ICB or chemotherapy (Chemo). We first identified genes that are independent of the two categories of biomarkers (i.e., PD-L1 expression, T-cell inflamed profile, CD8+ T cells, tertiary lymphoid structure, TMB, NAB, and microsatellite instability). Independence was defined as a spearman level within -0.2 to 0.2. These genes were then subjected to gene co-expression analysis to identify potential modules associated with ICB outcomes. Results: We revealed a novel functional module that shows weak correlation with previously established biomarkers. The bulk and single-cell RNA-seq confirmed its enrichment in subsets of stromal cells, thus it was named the Third Dimensional Stromal (TDS) score. High TDS consistently predicted less ICB efficacy in each cohort and jointly in meta-analysis (OS: HR=1.56, P<0.0001; PFS: HR=1.65, P<0.0001). The advantages of ICB over Chemo were lost in high-TDS tumors (meta-analysis: OS: HR=1.03, P=0.23; PFS: HR=1.14, P=0.25), demonstrating its predictive rather than prognostic role. The negative predictiveness of TDS further increased in 'hot' tumors with high CD8+ T cell infiltration (meta-analysis: OS: HR=1.96, P<0.0001; PFS: HR=1.69, P<0.0001), but was far less predictive in 'cold' tumors. Importantly, even in 'hot' tumors, high-TDS defined a population where ICB could not improve survival compared to Chemo (meta-analysis: OS: HR=0.83, P=0.23; PFS: HR=0.96, P=0.77). Conclusions: The TDS framework represents a unique classification of pan-cancer patients, rather than a repetition of existing systems, thus can be integrated into the current multipart biomarker panel. Additionally, since we found a distinct biomarker nature between 'hot' and 'cold' tumors, this study emphasizes the importance of exploring biomarkers and resistance mechanisms specifically among patients who harbor inflamed tumors.