Page 163 - EJMO-9-2

Eurasian Journal of
Medicine and Oncology Genetic insights into CAD drug targets

The initial MR analysis was conducted using the p2 = 1E-4, p12 = 1E-5; where p1, p2, and p12 represent
“TwoSampleMR” package (version 0.5.8) in R software the predefined probabilities that an SNP is closely related
(version 4.3.1). 14,15 To ensure the reliability of the genetic to gene expression, CA risk, or both, respectively. The
instruments (IVs), a rigorous validation process was posterior probabilities (PP) derived from the colocalization
followed. The single nucleotide polymorphisms (SNPs) analysis correspond to one of five hypotheses: PPH : SNP
0
used as IVs were carefully selected based on their strong is not associated with either trait; PPH : SNP is associated
1
association with gene expression in the eQTL database with gene expression but not CA risk; PPH : SNP is
2
and their independence from confounding factors. For associated with CA risk but not gene expression; PPH :
3
each instrument, we assessed whether it satisfied the SNP is associated with both CA risk and gene expression
key assumptions of MR, which include: (i) relevance but is driven by different genetic variants; PPH : SNP is
4
assumption: each IV must be strongly associated associated with both CA risk and gene expression and
with the exposure (gene expression); (ii) exclusion is driven by the same genetic variant. Colocalization
restriction assumption: the IV must not influence significance was determined with PPH >0.70, indicating a
4
the outcome (CA) through pathways other than the colocalized relationship between the gene and CA, and was
exposure; (iii) independence assumption: the IV must be considered a potential drug target gene. 16
independent of confounders, which could otherwise bias
the causal inference. 2.6. The GWAS data for plasma metabolites
To assess and mitigate the effects of horizontal The GWAS data for plasma metabolites were derived from
pleiotropy, we applied MR Egger regression and used the the Canadian longitudinal study on aging, which included
Egger intercept test to detect any potential pleiotropic 8,299 participants, with 1,458 metabolites quantified using
effects. If the Egger intercept test revealed a significant the Metabolon HD4 platform while retaining samples
pleiotropic effect (p<0.05), we considered the results with <50% missing data. The study identified 309 pairs of
with caution and reported them accordingly. In addition, metabolites with shared enzymes or transporters through
Cochran’s Q test was employed to detect heterogeneity the Human Metabolome Database and calculated the
among the IVs. A significant p-value from Cochran’s Q ratios of these metabolite pairs for further analysis. In total,
test (p<0.05) indicated the presence of heterogeneity, 1,400 metabolites were evaluated and classified into eight
suggesting that the IVs may not be homogenous in their categories, including amino acids, carbohydrates, lipids,
effects. If heterogeneity was detected, we used the random- etc.
effects model in the IVW analysis to account for the 2.7. Mediation analysis
variability between IVs. Conversely, when no significant
heterogeneity was observed, a fixed-effects model was This study employed a two-step MR (two-step MR)
applied. method for mediation analysis to assess whether plasma
metabolites serve as mediators in the causal pathway
2.4. SMR analysis and HEIDI test between druggable genes and CA outcomes. In the first
SMR is a method used to determine the causal relationship step, the causal effect of druggable genes on the mediator
between gene expression and CA, effectively reducing the (Beta1) was calculated. Subsequently, in the second step, the
influence of confounders on causal inference. The HEIDI causal effect of the mediator on CA (Beta2) was evaluated.
test is employed to assess whether the association between The proportion of the mediation effect relative to the total
gene expression and CA is caused by the same genetic effect was calculated as R = (Beta1 × Beta2)/Beta0. The
variant, thereby determining if it represents a colocalized direct effect of exposure on the outcome was represented
signal. The SMR analysis and HEIDI test were conducted as Beta3 = Beta0 - Beta1 × Beta2.
using SMR software (version 1.3.1). If the p-value of the Beta0 represents the total causal effect of druggable
SMR test is <0.05 and the p-value of the HEIDI test is genes on CA outcomes, encompassing both direct and
>0.05, it indicates that the association is likely driven by a mediated effects.
shared genetic variant.
2.8. Phenome-wide association studies (PheWAS)
2.5. Colocalization analysis analysis
For important MR results identified during the discovery In this study, we conducted a PheWAS analysis to explore
and validation process, colocalization analysis of CA risk the pleiotropic effects and potential safety concerns of
was conducted using the R package “coloc.” The analysis candidate therapeutic targets. PheWAS analysis was based
17
used SNPs harmonized by the “TwoSampleMR” package, on publicly available data resources, including the PheWAS
with default prior probabilities set as follows: p1 = 1E-4, portal (https://azphewas.com/). The databases provide

Volume 9 Issue 2 (2025) 155 doi: 10.36922/ejmo.7387