Incoming Assistant Professor Duke University Palo Alto, California, United States
Introduction:: Chronic inflammation due to recurrent tissue injury can perturb tissue homeostasis and thereby contribute to tumorigenesis. Frequently this occurs to exposed epithelium of organs such as the esophagus, lung, stomach, and intestine. Yet only some patients with inflammation-induced metaplasia of the epithelium progress to cancer. Consequently, understanding the link between inflammation (immune cells) and stroma (mesenchymal cells) that creates a unique environment leading to epithelial metaplasia is critical for prevention and treatment.
Materials and Methods:: We leveraged the Phenocycler technology to characterize 405 tissue samples from 26 Barrett’s esophagus (BE) (metaplasia) who have not progressed to adenocarcinoma with 6 BE patients that have progressed to adenocarcinoma as well as 23 samples from patients from all disease states: normal, metaplasia, dysplasia, and cancer. The Phenocycler Fusion enabled fast spatial phenotyping of whole slide FFPE samples with both a custom-created 56-antibody panel and 100+ RNA molecules in 1.1 million cells leading to identification of 57 cell types spanning major immune cells and various metaplastic states of the epithelium. We also use a computational technique to identify multicellular neighborhoods that are consistent organizations of cells that make up the functional structures within the tissue.
Results, Conclusions, and Discussions:: Our CODEX multiplexed imaging results (Fig. 1A) indicated that there is a misdifferentiation of the epithelial to an intestinal-like epithelial and then dedifferentiated state within cancer (Fig. 1B). Additionally, we were able to identify multicellular neighborhoods associated with pathological tissue (Fig. 1C). One of the key takeaways from this analysis was that there was a mirroring change or dedifferentiation of the cellular microenvironment to first a more intestinal-like stromal environment (Fig. 1C) and then leading to an environment with increasing diversity with disease progression (Fig. 1C). We compared the environments between 3 different types of esophageal cancer on 3 tissue microarrays (Fig. 1F). Largely cell type proportions were similar across the different cancers taken except for presence of foveolar cells within adenocarcinoma associated with BE (Fig. 1G). All samples had an increase within CD4+ Tregs within carcinoma (Fig. 1H). We observed metaplastic-like multicellular structures associated with adenocarcinoma with BE, that were consistent with increased risk within previous samples consistent with what we observed previously (Fig. D, I-J).
Finally, we analyzed a cohort of patients who had BE but did or did not progress to adenocarcinoma. From the CODEX data we observed that progressors had increased CD4+ Treg cells, Neutrophils, and NK cells (data not shown) and plasma cell multicellular neighborhoods decreased in progressors compared to non-progressors (data not shown). To further understand these cellular interactions, we collected a single-cell transcriptomic (100 RNA) dataset (Fig. 1K-L). Paired in situ transcriptomics revealed transitions in fibroblast populations with progression of disease, involved in immune cell recruitment and maintenance of inflamed-epithelial cellular neighborhoods.
Consequently, the power of capturing both multiplexed protein and RNA levels from single-cells across a large cohort of FFPE human esophageal samples enabled description of cell and cell-cell interactions potentially driving disease progression. This will help to think about new therapies targeting the stroma could reverse the cellular rearrangements governing dedifferentiation or provide risk stratification. In the future, we will use this data with our scRNAseq and ECM proteomics collected to further connect our cellular rearrangements to observed cross-talk between fibroblasts, immune, and epithelial cells.