Assessing the role of immune aggregates as potential biomarkers of immunotherapy response in appendiceal adenocarcinoma (AA) using spatial proteomics

Appendiceal adenocarcinoma is a rare tumor representing less than 1% of all gastrointestinal malignancies. Unfortunately, over 50% of AA patients present with stage IV disease with many patients being unresectable and with few systemic therapy options. Data from a recent clinical trial suggests improved overall survival on atezolizumab and bevacizumab combination therapy for patients with unresectable metastatic AA (NCT03074513). We previously studied general immune effects in a subset of these patients using spatial transcriptomics, suggesting immune aggregates like tertiary lymphoid structures (TLS) could be important biomarkers of response in AA. Here, we aim to characterize the landscape of immune aggregates using single-cell spatial proteomics to further understand potential biomarkers of immunotherapy response in AA. Single-cell spatial proteomics was performed on pre-treatment FFPE biopsies from eight trial participants using nanoString's CosMx Spatial Molecular Imager (SMI) with the 64-plex human IO protein panel. Data processing was done using nanoString's AtoMx platform and exported to the R package Seurat for analysis. Cell type annotations were obtained using CELESTA. Immune aggregates were manually identified and labeled using CosMx staining images and Napari. The profiled trial cohort (n=8) was grouped by clinical response, measured as whether a patient was alive (n=4) or deceased (n=4) at trial follow-up. Peritumoral immune aggregates were manually annotated across all patients using the CD45 staining images from AtoMx and Napari to add labels to cells. Overall, we detected 75 unique immune aggregates across 7 of the 8 patient samples profiled. At the tissue level, we identified higher Treg density (p=0.085, t-test) in deceased patients compared to living patients. Among cells in immune aggregates, there was a trend towards higher B cell (p=0.14) and fibroblast (p=0.16) density in aggregates from living patients. Neighborhood analysis performed within these structures uncovered 9 cellular neighborhoods, representing many components of tertiary lymphoid structures. Unsupervised clustering of immune aggregates according to cellular neighborhood composition resulted in the separation of aggregates with more germinal center B cells from others with more CD4+ and CD8+ T cells. With challenges in tissue collection and sequencing of AA surgical specimens, spatial proteomics provides a meaningful framework for studying the tumor microenvironment and treatment response in such tissues. The presence of mature immune aggregates in AA tumors suggests a potential role in driving response to immunotherapy, and work remains ongoing to further characterize these aggregates to best predict response to immunotherapy in future AA patients.