Integrated transcriptomic and single cell analyses identify matrix stiffness-related candidate prognostic genes and epithelial dynamics in colorectal cancer

Abstract Colorectal cancer (CRC) is recognized as the third most common malignant tumor globally. Matrix stiffness (MS) has been associated with the progression of various cancer types and is often abnormally high in tumor tissues. Regardless, the precise molecular pathways through which MS influences CRC initiation and disease advancement remain underexplored. This research was designed to investigate MS-related prognostic genes and their potential biological relevance in CRC progression. Transcriptomic datasets alongside single-cell sequencing information were obtained from open-access repositories for the current investigation. Univariate Cox regression analysis, combined with other analytical methods, were used to identify prognosis-associated genes. Based on these findings, a risk prediction model was constructed and its validity was confirmed using separate training and validation cohorts. Computational biology techniques were then applied to delineate the characteristics of the tumor immune landscape and assess therapeutic agent responsiveness between different risk stratifications. Key cell populations were characterized using single-cell sequencing, and the expression patterns of prognostic genes were examined across distinct cell subtypes. Moreover, the expression levels of these prognostic markers in the experimental specimens were quantified using RT-qPCR. Four prognostic genes were identified— SIX4 , PCOLCE2 , TIMP1 , and TNNT1 —all of which functioned as risk factors (HR > 1, p  p  = 3.97e−42). A total of 95 drugs showed exploratory differences in computationally predicted sensitivity between risk groups, with Dasatinib being a representative example (adjusted p  = 2.38e−15). Epithelial cells were determined as the key cell population, and the expression dynamics of all prognostic genes exhibited broadly consistent trends throughout epithelial cell differentiation. Finally, RT-qPCR analysis preliminarily validated the bioinformatics-predicted expression patterns. SIX4 , PCOLCE2 , TIMP1 , and TNNT1 were identified as MSRG-informed prognostic genes, providing potential clues for future studies of CRC prognosis and MS-related tumor biology.