Comprehensive single-cell atlas of colorectal neuroendocrine tumors with liver metastases: unraveling tumor microenvironment heterogeneity between primary lesions and metastases

Abstract

Colorectal neuroendocrine tumors with liver metastases (CRNELM) are associated with a poorer prognosis compared to their nonmetastatic counterparts. A comprehensive understanding of the tumor microenvironment (TME) heterogeneity between primary lesions (PL) and liver metastases (LM) could provide crucial insights for enhancing clinical management strategies for these patients. We utilized single-cell RNA sequencing to analyze fresh tissue samples from CRNELM patients, aiming to elucidate the variations in TME between PL and LM. Complementary multidimensional validation was achieved through spatial transcriptomics, bulk RNA sequencing, and multiplex immunohistochemistry/immunofluorescence. Our single-cell RNA sequencing analysis revealed that LM harboured a higher proportion of CD8 + T cells, CD4 + T cells, NK cells, NKT cells, and B cells exhibiting a stress-like phenotype compared to PL. RGS5 + pericytes may play a role in the stress-like phenotype observed in immune cells within LM. MCs in PL (PL_MCs) and LM (LM_MCs) exhibit distinct activation of tumor-associated signaling pathways. Notably, COLEC11 + matrix cancer-associated fibroblasts (COLEC11_mCAFs) were found to be significantly associated with LM_MCs. Cell communication analysis unveiled potential targetable receptor-ligand interactions between COLEC11_mCAFs and LM_MCs. Multidimensional validation confirmed the prominence of the characteristic stress-like phenotypes, including HSPA6_CD8_Tstr, HSPA6_NK, and COLEC11_mCAFs in LM. Moreover, a higher abundance of COLEC11_mCAFs correlated with poorer survival rates in the neuroendocrine tumor patient cohort. Overall, our study provides the first single-cell analysis of the cellular and molecular differences between PL and LM in CRNELM patients. We identified distinct cell subsets and receptor-ligand interactions that may drive TME discrepancies and support metastatic tumor growth. These insights highlight potential therapeutic targets and inform strategies for better managing CRNELM patients.