TMIC-35. Cancer cell – fibroblast communication in the leptomeninges

Abstract Leptomeningeal metastasis (LM), the spread of cancer into cerebrospinal fluid (CSF)-filled coverings of the brain and spinal cord, is a dismal complication of cancer leading to rapid neurologic disability and death. How cancer within this unique microenvironment grows remains incompletely understood. To address this, we carried out differential untargeted proteomics of breast cancer patient spinal fluid and uncovered multiple inflammatory species present at significantly elevated concentrations in the setting of LM. Subsequent targeted proteomics of these specimens revealed a cluster of inflammatory mediators reminiscent of those generated by cancer-associated-fibroblasts (CAFs), including IL-6 and IL-8. We hypothesized that leptomeningeal fibroblasts, while not canonical fibroblasts, might play the role of a CAF in LM. Indeed, we found that leptomeningeal fibroblasts support cancer cell growth in vitro, accelerate cancer cell migration, and promote cancer cell growth in vivo. To identify the key cancer cell-generated mediators that provoke this process, we carried out orthogonal proteomic and genomic screens, and identified DKK1. To understand how these leptomeningeal fibroblasts might respond compared to classical CAFs, we collected cancer cells and fibroblasts from models of breast and lung cancer LM as well as pancreatic cancer and subjected these to single cell RNA seq on 10X. We find that the leptomeningeal fibroblasts and CNS fibroblasts more generally are transcriptionally distinct from classical pancreatic, lung, and mammary fat pad fibroblasts. Further, we find transcriptionally distinct populations of leptomeningeal fibroblasts in the setting of LM. The first, a DKK1 responsive population that upregulates AP-1 signaling through JNK and STAT3 pathways. The second, a cell population upregulating NFKB and TGFB gene signatures. We hypothesize these two populations promote cancer separately through either tumor growth or migration supporting pathways. Together, our work extends the CAF phenotype beyond classical fibroblasts to include pial cells, suggesting novel therapeutic opportunities within CNS malignancies.