De-risking drugs using genetics
Using Mendelian Randomization to de-risk EPO-increasing drugs
Happy Friday! Did you check the Feb roundup post? Bookmark it for your weekend read. Meanwhile here is a tiny post on an example of using genetics to de-risk drugs (ensuring a drug is safe and free of any suspected adverse effects) that I posted on Twitter two years ago. This is similar to the TYK2 story that I posted two weeks ago.
From the Twitter archives
A great paper with a study design that elegantly combines molecular experiments with statistical genetics to study drug adverse effects. Some interesting learning points here.
The main aim here is to identify genetic variants that alter EPO (erythropoietin) levels and use them as MR instruments to evaluate the cardiovascular safety of EPO-increasing drugs.
The authors performed a GWAS of circulating EPO levels and found three GWAS hits. Surprisingly, none of them were cis variants i.e., located near the EPO gene ( located in Chr 7)
We always expect the cis variants to show up first in a GWAS, followed by trans. This is because cis effects are stronger than trans effects. But here the phenotype is a hormone that is under feedback regulation. Some of the factors that stimulate EPO production such as hemoglobin levels are themselves strongly genetically influenced.
So here the genetic effects of the regulatory factors that influence EPO levels are much stronger than the genetic effects that directly influence EPO levels via gene/protein expression. Hence trans effects surface before the cis.
But note that trans variants are not good MR instruments as they can affect the outcome (cardiovascular disease) in ways other than through exposure (altering EPO levels), which violates one of the MR assumptions.
So, the authors searched for cis variants with sub-threshold significance and found one, which is an eQTL for EPO.
Then, they experimentally validated the effect of this cis variant on EPO levels using CRISPR techniques.
Having validated the instrument, the authors next performed Mendelian Randomization analysis and evaluated if genetically increasing EPO levels increased the risk for cardiovascular diseases. They found no evidence that increasing EPO will cause cardiovascular adverse events.
Not often we come across papers that combine the power of molecular biology and statistical genetics techniques to answer clinically important questions. Cool work by Harlow et al.
