Abstract PR-08: Spatial transcriptomics to profile the non-small cell lung cancer tumor microenvironment and identify novel predictive biomarkers for checkpoint blockade

Christina Cho; Francesc Lopez; Gavitt Woodard; Baozhu Huang; Jianwei He; Tayf Badri; Michal Kidacki; Matthew Vesely; Evelyn Ng; Guilin Wang; Lieping Chen

doi:10.1158/2326-6074.TUMIMM24-PR-08

Immune checkpoint inhibitors (ICIs) have dramatically changed the way we treat advanced non-small cell lung cancer (NSCLC). Compared to chemotherapy alone, ICIs significantly improve overall survival, and these survival benefits are durable. Unfortunately, only a minority of patients benefit from ICI therapy. Most advanced NSCLC patients are non-responders. Several factors may contribute to poor response rates. One, we may not be correctly identifying the patients who could benefit from ICIs. Currently, PD-L1 is the only predictive biomarker; yet not all tumors that express PD-L1 are responsive. Two, blocking checkpoints to restore cytotoxic T lymphocyte (CTL) function may be insufficient to cure disease. The tumor microenvironment (TME) likely remains hostile to CTLs even after ICI therapy. The TME may continue inhibiting infiltrated CTLs or may prevent CTL infiltration in the first place. To effectively re- invigorate CTL-mediated anti-tumor responses, we must remodel the TME into one that is hospitable for CTLs; however, our understanding of the TME is limited. Here, we demonstrate the use of spatial transcriptomics to (1) identify novel predictive biomarkers for ICI therapy, and (2) generate a comprehensive profile of the NSCLC TME. Using Nanostring’s GeoMX Digital Spatial Profiler (DSP), we performed spatial analysis on 56 human NSCLC tumors. Using the DSP’s state-of-the-art UV masking technology, we generated transcriptomic profiles of CD8+ CTLs, the vimentin+ stroma, and the vimentin- tumor. We grouped the profiles of each compartment using their inflammatory status—hot/inflamed, or cold/not-inflamed—to identify differentially expressed genes that correspond to the immune phenotype of the tumors. We identified 112 DEGs in CD8+T cells and found that T cells in cold tumors were functional but poorly migratory. Strikingly, T cells in hot tumors were dysfunctional but motile and able to infiltrate tumors. 41 DEGs were identified in the Vimentin+ stroma. The stroma of hot tumors was enriched in CEACAM1, HMGB1, and LGALS9—all ligands of TIM3, a known immune checkpoint. The stroma of cold tumors expressed higher levels of JAG2, LTB4R, and SEMA4C. Bioinformatic analysis of TCGA data indicated that all three molecules are negatively correlated with CD8A levels in NSCLC tumors, which suggests that these molecules could promote T cell exclusion. 29 DEGs were identified in the tumor, and hot tumors were enriched in pathways associated with cellular migration and inhibition of immune effectors. In depth analysis of molecules upregulated in cold T cells revealed five new potential predictive biomarkers. CD2, CD52, KLRK1, NKG7, and PTPRC are upregulated in responders, and higher expression of these molecules corresponded to improved response to anti-PD1 therapy. In sum, our innovative studies highlight the utility of spatial transcriptomics in profiling the TME and identify new biomarkers that could be used to predict response to ICI therapy. Citation Format: Christina Cho, Francesc Lopez, Gavitt Woodard, Baozhu Huang, Jianwei He, Tayf Badri, Michal Kidacki, Matthew Vesely, Evelyn Ng, Guilin Wang, Lieping Chen. Spatial transcriptomics to profile the non-small cell lung cancer tumor microenvironment and identify novel predictive biomarkers for checkpoint blockade [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Tumor Immunology and Immunotherapy; 2024 Oct 18-21; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2024;12(10 Suppl):Abstract nr PR-08.

Abstract PR-08: Spatial transcriptomics to profile the non-small cell lung cancer tumor microenvironment and identify novel predictive biomarkers for checkpoint blockade

View Online

Abstract

Details

Metrics