Integrating single-cell and bulk transcriptomes identifies B cell features associated with neoadjuvant chemoradiotherapy sensitivity in rectal cancer

Neoadjuvant chemoradiotherapy (nCRT) is the main treatment for Locally Advanced Rectal Cancer (LARC). The response to nCRT varies from a complete response to no response. The impact of the B cells in this process is poorly understood. This study aimed to characterize the B cells types associated with response or resistance to nCRT. We applied the "Scissor" algorithm to integrate single-cell RNA-seq data with bulk transcriptome data from colorectal cancer samples, thereby identifying B cell subpopulations associated with nCRT response and exploring the clinical significance of B cell-related characteristic genes in nCRT for rectal cancer. At the single-cell level, we identified a B cell subpopulation characterized by the expression of HLA-DRB5, HLA-DQA2, HLA-DQB1, CD74, and ACTG1, which was associated with nCRT response in rectal cancer. Using subpopulation-specific trait genes, rectal cancer patients were classified into three distinct subtypes with unique features. Subtype A shows higher PD-L1 expression suggesting that patients in this subgroup are more likely to achieve favorable responses to immunotherapy. Subtype C shows lower hypoxia scores and a higher proportion of patients deriving clinical benefit from nCRT, suggesting that this subgroup may be more sensitive to neoadjuvant treatment. We developed a machine learning-based predictive model for pathological complete response (pCR) to nCRT in rectal cancer, achieving an area under the curve (AUC) of 0.911 in the training set and 0.819 in the 64-sample validation cohort. This study reveals that a B cell subpopulation characterized by the co-expression of HLA-DRB5, HLA-DQA2, HLA-DQB1, CD74, and ACTG1 is significantly associated with nCRT response in rectal cancer. These findings offer actionable insights for optimizing clinical treatment strategies, including patient stratification and personalized therapy selection.