Spatial transcriptomics and scRNA-seq: decoding tumor complexity and constructing prognostic models in colorectal cancer.

Introduction: Recent advancements in transcriptomic analysis, combined with single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics, have deepened our understanding of the tumor microenvironment and cellular heterogeneity, laying the groundwork for personalized therapies. The aim of this research is to explore the heterogeneity of tumor cells in colorectal cancer (CRC) and evaluate their prognostic value in different therapeutic contexts, emphasizing the impact of tumor cell heterogeneity on disease progression.

Methods: scRNA-seq alongside spatial transcriptomics was employed to analyze the heterogeneity of tumor cells in CRC, the spatial distribution of tumor cells, and their interactions with the microenvironment.

Results: We identified nine distinct tumor cell subtypes, with MLXIPL + neoplasm prevalent in advanced CRC, while ADH1C + and MUC2 + neoplasms were more common in early-stage CRC. MLXIPL + neoplasm was significantly associated with chemotherapy and targeted therapy efficacy. Analysis of spatial transcriptomics indicated that MLXIPL + neoplasm is located in the core area of the tumor cells. We developed a 13-gene prognostic signature (PS) using machine learning algorithm (StepCox backward), which predicts the prognosis of CRC patients. Furthermore, the patients with low PS score demonstrated higher immune cell infiltration and immune regulatory factors, suggesting enhanced immune surveillance and treatment response.

Conclusions: The findings highlight the critical role of tumor cell heterogeneity in CRC progression and treatment response, suggesting the need for personalized therapeutic strategies targeting different subpopulations. The constructed PS demonstrates significant clinical application value in predicting patient prognosis.