20: Single-cell transcriptomic analysis identifies divergent tumor epithelial cell states in colorectal metastasis

Tumor heterogeneity in metastatic colorectal cancer (mCRC) is a key determinant of treatment outcomes and survival. However, site-specific heterogeneity of tumor epithelial cells in mCRC remains unknown. Using single-cell RNA sequencing (scRNA-seq), we offer novel insights into tumor epithelial cell diversity in mCRC.

 


Methods:

Utilizing scRNA-seq, we analyzed 22 samples from 14 mCRC patients, encompassing primary tumors, and liver, lung, and peritoneal metastases. Our results were compared with data from premalignant polyps and primary CRC. We utilized the scanpy.tl.score gene function to compute signature scores for consensus molecular subtypes (CMS), epithelial intrinsic CMS (iCMS), stem-cell, and metaplasia signatures.

 


Results:

Using the iCMS classification, we discovered a distinct dichotomy in tumor epithelial cells across mCRC. Liver and lung metastases predominantly associated with the iCMS2 subtype, while peritoneal metastases were primarily identified as the iCMS3 subtype (Figure 1a). This distinction was further corroborated in our data using the original CMS classification, which categorized liver metastases as the CMS2 or 3 subtypes, whereas peritoneal metastases were classified as CMS4 subtype. Notably, we discovered mutually exclusive stem cell and metaplasia gene signature in liver and peritoneal metastases (Figure 1b). Liver metastases were enriched in stem cell signatures, including the canonical stem cell genes ASCL2, AXIN2, OLFM4, and CDX2. Conversely, peritoneal metastases, were enriched in the metaplasia gene signature characterized by ANXA1, S100A4, and TROP2. We validated our epithelial signatures in independent datasets which recapitulated the stem cell and metaplastic molecular programs in mCRC. When comparing mCRC to pre-malignant polyps, we identified a significant overlap: Stem-cell signatures dominant in liver metastasis were also enriched in conventional adenomas. In contrast, the metaplasia signature associated with peritoneal metastases was prevalent in serrated polyps, suggesting distinct cellular origins of liver and peritoneal metastases. Additionally, metaplasia signature scores were higher in treated than in chemo-naïve mCRC epithelial cells.

 


Conclusions:

Our single-cell analysis reveals two dichotomous gene signatures within tumor epithelial cells that distinguish liver and peritoneal metastases. These insights suggest divergent epithelial cell origins for liver and peritoneal metastases, stemming from distinct pre-cancerous lesions, conventional adenomas vs serrated polyps.