A fibroblast-associated signature predicts prognosis and immunotherapy in esophageal squamous cell cancer.

Background: Current paradigms of anti-tumor therapies are not qualified to evacuate the malignancy ascribing to cancer stroma's functions in accelerating tumor relapse and therapeutic resistance. Cancer-associated fibroblasts (CAFs) has been identified significantly correlated with tumor progression and therapy resistance. Thus, we aimed to probe into the CAFs characteristics in esophageal squamous cancer (ESCC) and construct a risk signature based on CAFs to predict the prognosis of ESCC patients. Methods: The GEO database provided the single-cell RNA sequencing (scRNA-seq) data. The GEO and TCGA databases were used to obtain bulk RNA-seq data and microarray data of ESCC, respectively. CAF clusters were identified from the scRNA-seq data using the Seurat R package. CAF-related prognostic genes were subsequently identified using univariate Cox regression analysis. A risk signature based on CAF-related prognostic genes was constructed using Lasso regression. Then, a nomogram model based on clinicopathological characteristics and the risk signature was developed. Consensus clustering was conducted to explore the heterogeneity of ESCC. Finally, PCR was utilized to validate the functions that hub genes play on ESCC. Results: Six CAF clusters were identified in ESCC based on scRNA-seq data, three of which had prognostic associations. A total of 642 genes were found to be significantly correlated with CAF clusters from a pool of 17080 DEGs, and 9 genes were selected to generate a risk signature, which were mainly involved in 10 pathways such as NRF1, MYC, and TGF-Beta. The risk signature was significantly correlated with stromal and immune scores, as well as some immune cells. Multivariate analysis demonstrated that the risk signature was an independent prognostic factor for ESCC, and its potential in predicting immunotherapeutic outcomes was confirmed. A novel nomogram integrating the CAF-based risk signature and clinical stage was developed, which exhibited favorable predictability and reliability for ESCC prognosis prediction. The consensus clustering analysis further confirmed the heterogeneity of ESCC. Conclusion: The prognosis of ESCC can be effectively predicted by CAF-based risk signatures, and a comprehensive characterization of the CAF signature of ESCC may aid in interpreting the response of ESCC to immunotherapy and offer new strategies for cancer treatment.