Abstract C025: Effect of Chemotherapy on Spatial Relationships in the Colorectal Cancer Immune Microenvironment

Microsatellite stable (MSS) stage IV colorectal cancer (CRC) is treated with chemotherapy (CT) and targeted therapy, but resistance is common, with only a 13% 5-year overall survival, and more effective, durable treatments such as immunotherapy are needed; however, the immunologic effects of CT – including anti- and pro-tumor immune effects on cancer cells, effects on the tumor immune microenvironment (TIME), and the temporal impact of different agents – remain unknown. Evaluation of spatial relationships in the intact TIME can guide immune response augmentation and rational combination sequence design. The CXCL9:Spp1 macrophage polarity axis, defined by both gene expression and spatial organization, demonstrated Spp1+ macrophages (SM) emerge in diseased tissue, are associated with hypoxia and tumor necrosis, and predict poor outcomes in patients with CRC by suppressing anti-tumor immunity in liver metastases through SM sequestration of CD8+ cytotoxic T cell (CTL). CXCL9+ macrophages (CM) associated with CTL infiltration and better outcomes. Archival pre- and post-CT liver (L) (n=24) and select primary (P) FFPE samples (n=5) of 12 patients with stage IV MSS CRC (7 responders [response >1yr], 5 non-responders (NR) [relapsed <6mo]) with liver metastasis at diagnosis (2016-2024) were obtained. The Phenocycler (CODEX) platform, a multiplexed single-cell imaging technology enabling imaging of 50+ protein targets including immune, epithelial, tumor, endothelial, and stromal markers was implemented. This tool enables identification of tertiary lymphoid structures and cellular neighborhood analysis, providing in situ mechanistic insights into temporal TIME changes and cellular interactions. Preliminary analysis was performed on a post-CT matched P and L sample from a NR using surrogate markers for CM (MHCII+CD68+, previously verified as CXCL9+) and SM (CD31+Vim+CD68+, previously verified as Spp1+). We observed greater viable tumor and marginally greater CTLs in the P site, while the L had more CD15+CD66+ tumor associated neutrophils. The P site had 10x more dendritic cells (DCs) and natural killer cells, and despite the modest increase in CTL, also had more T regulatory cells (Tregs). Both sites had similar percentages of CMs. Strikingly, a 10x greater population of SMs was present in the P site compared to the L. These macrophages closely associated with tumor in the P site, while Tregs clustered with CTLs and DCs, consistent with a compensatory immunosuppressive mechanism in the post-CT, NR setting. Conversely, SMs, CMs, Tregs, CTLs and DCs were further from tumor in the L site, reflecting an immune-deserted phenotype. Analysis of remaining samples using validated CXCL9 and Spp1 is underway and will be overlaid on current surrogate markers. In this preliminary analysis, we demonstrate that distinct immune cell populations can be identified and spatially analyzed to better understand spatial tumor-immune arrangements in relation to CT response in CRC with liver metastasis, a challenging disease with an unmet need for immunotherapeutic biomarkers.

Duke Scholars

Author Nicholas Christian DeVito Medicine, Medical Oncology