Associations of plasma protein levels with risk of colorectal cancer: a proteome-wide Mendelian randomization study.

Abstract

Background: The treatment of advanced or metastatic colorectal cancer (CRC) poses a global challenge. Mendelian Randomization (MR) has been primarily applied for repurposing licensed drugs and uncovering new therapeutic targets.

Objective: This study aims to systematically identify potential plasma protein targets for CRC using proteome-wide Mendelian randomization and evaluate their potential side effects through phenome-wide association studies (Phe-WAS).

Methods: We conducted a comprehensive proteome-wide MR study to assess the causal relationships between plasma proteins and the risk of CRC and evaluate their potential side effects through Phe-WAS. The plasma proteins were sourced from the Finland and Iceland decode database, encompassing GWAS data for plasma proteins (Olink-619 samples across 2925 proteins, SomaScan -828 samples across 7596 proteins and Iceland decode database across 4907 proteins). Additionally, GWAS data for CRC were extracted from the UK Biobank-SAIGE database, including 3051 cases and 382,756 controls. Subsequently, colocalization analysis was performed to identify shared causal variants between plasma proteins and CRC. Finally, a phenome-wide association study (Phe-WAS) was conducted to examine the potential adverse effects of druggable proteins for CRC, utilizing the extensive UK Biobank-SAIGE database, encompassing 783 phenotypes.

Results: The MR analysis identified GREM1, DKKL1, and CHRDL2 as plasma proteins whose genetically predicted levels were positively associated with CRC risk, whereas TMEM132A was inversely associated with CRC risk (P_fdr < 0.05). The colocalization analysis identified these four proteins as shared variation with CRC (PPH3 + PPH4 > 0.7), suggesting that these proteins represent potential direct targets for CRC intervention. Further phenotype-wide association studies showed no significant potential side effects of these targets (P_fdr > 0.05).

Conclusion: This proteome-wide Mendelian randomization study offers a comprehensive molecular landscape of CRC, identifying GREM1, DKKL1, CHRDL2, and TMEM132A as potential therapeutic targets. Our research provides a critical foundation for future experimental validation and therapeutic development in colorectal cancer management.