Abstract
ESCC presents a significant global health challenge due to its high mortality rates and varying responses to treatment. This underscores the critical need for novel diagnostic and predictive biomarkers to improve treatment outcomes. Initially, we conducted single-cell transcriptome sequencing on a total of 128,688 cells obtained from 10 patients as part of our research. Utilizing machine learning and cross-validation techniques, we developed a model incorporating 12 genes that distinguish malignant cells from non-malignant ones. In vitro, we explored the effects of IGFBP2 knockdown on the proliferation, invasion, and migration of ESCC cells. The clinical relevance of IGFBP2 was confirmed through IHC and Kaplan-Meier survival analyses. Furthermore, using bioinformatics tools such as GSVA and xCell on public databases, we discovered that high expression of IGFBP2 is associated with an immunosuppressive tumor microenvironment in ESCC, characterized by reduced CD8+ T cell infiltration. This was validated then through IHC. In summary, our study integrates single-cell sequencing and sophisticated computational techniques to highlight IGFBP2 as a promising biomarker and therapeutic target in ESCC.