Machine learning-Based integration develops a metastasis-Related rhythmic gene signature for improving outcomes in pancreatic cancer

Circadian dysregulation has been shown to be associated with the progression of pancreatic cancer (PC). However, the circadian machinery comprises numerous circadian rhythm-related genes (CRRGs), whose roles in tumor biology remain unclear. A metastasis-related rhythmic gene signature (MRRGS) was established based on differentially expressed genes (DEGs) identified from public PC datasets. A risk model was developed, and Gene Set Enrichment Analysis (GSEA) was employed to explore functional differences between low-risk and high-risk groups. The impact of MRRGS was further assessed through mutation and copy number variation (CNV) analysis. The correlation between modeling genes and immune infiltration scores was examined using the ESTIMATE, xCell, and ssGSEA algorithms. Finally, potential therapeutic drugs were tested in various PC patient populations. Five CRRGs, including S100A10, CD47, UBXN1, ERBB3, and ST3GAL1, were identified and validated for the development of the risk model. GSEA analysis revealed that high-risk PC patients were associated with ribosome-related pathways and MHC class II pathways. Mutation and CNV analyses indicated that high-risk patients exhibited SMAD4 mutations and a higher frequency of CNVs, while low-risk patients displayed significant alterations in genes such as GSTM1, CFHR3, and RNF43. Furthermore, results from the ESTIMATE, xCell, and ssGSEA analyses demonstrated a notable negative correlation between the ERBB3 and S100A10 genes and immune infiltration-related genes. Based on drug screening outcomes, Sepatronium bromide is recommended for drug trials in high-risk populations, while SB505124 is suggested for low-risk populations. The MRRGS constructed from S100A10, CD47, UBXN1, ERBB3, and ST3GAL1 may serve as predictors of prognosis and drug sensitivity for PC patients.