Jing Song, Xiaomei Song, Yue Xie, Hong Guo, Yupeng Cun
Background: Despite the use of immunotherapy in esophageal squamous cell carcinoma (ESCC), treatment failure occurs occasionally in patients, yet the underlying mechanisms remain poorly understood.Methods: We conducted large-scale single-cell RNA sequencing (scRNA-seq) data analysis, which integrated seven independent datasets from 192 ESCC patients to yield over 440,000 high-quality single cells, to systematically characterize the tumor microenvironment (TME) landscape during ESCC progression and immunotherapy response.Additionally, we performed high-resolution spatial transcriptomics (stRNA-seq) using the 10x Visium HD platform on paired pre-and post-treatment tissues from two patients (one immunotherapy responder and one non-responder), which enhanced the findings from the scRNA-seq data and mapped therapy-induced TME at the spatial level.Multiplex immunohistochemistry was employed based on seven patients to confirm distinct patterns of intercellular crosstalk underlying differential therapeutic outcomes.Results: In scRNA-seq data, we found that B lineage cells were reduced during ESCC progression but were enriched in immunotherapy-resistant patients.Further analysis of malignant ESCC cells suggested that immunotherapy resistance might be associated with a subpopulation of tumor cells exhibiting aberrantly elevated cholesterol biosynthesis.Cell communication analysis of scRNA-seq and stRNA-seq data collectively revealed that immunotherapy resistance was linked to cellular crosstalk between cholesterol-biosynthetic tumor cells and germinal center (GC) B cells within tertiary lymphoid structures.Notably, single-cell, spatial data, and multiplex immunohistochemistry demonstrated that cholesterol biosynthesis-associated ESCC cells express elevated levels of MIF.This disrupts GC reactions by competing with the CXCL12-CXCR4 signaling axis via MIF-CXCR4 interactions, thereby impairing B cell-mediated immunity.Conclusions: MIF + tumor cells in GCs may be a biomarker for predicting immunotherapy resistance in ESCC.