By Mark McQuain
Almost exactly one year ago, this blog asked rhetorically whether your polygenic risk score was a good thing. Jon Holmlund raised this issue again last week, mentioning a company called Genomic Prediction. This company’s claim about the merits of their technology deserves close ethical scrutiny and is my reason for mentioning them yet again.
Genomic Prediction is increasingly calling for IVF clinics to use their version of expanded pre-implantation genetic diagnosis (EPGD), that is using big data analysis to select which embryo to implant – literally choosing the embryo you want based not upon the presence of single point genetic mutations known to be harmful but polygenic sequences (multiple genes, and even multiple upon multiple genes) statistically related to increased risk for complex diseases like heart disease and diabetes. Jon discussed the ethics of selecting “the children we want”, hinting that selecting for eye color or sex was “a step” further down the road.
Genomic Prediction is sprinting down that road.
Just last month, Genomic Prediction entered into discussion with some IVF clinics in the U.S. to screen embryos for “mental disability”. Per the firm’s co-founder Stephen Hsu:
“[EPGD] isn’t accurate enough to predict IQ for each embryo, but it can indicate which ones are genetic outliers, giving prospective parents the option of avoiding embryos with a high chance of an IQ 25 points below average”
This claim is entirely different from a claim that this particular embryo has this particular single point mutation that will cause this particular disease. Rather, the present claim of EPGD is better stated like this: When we saw this same polygenic pattern in some large number of prior embryos, a majority of them had an IQ 25 points below normal. Some questions naturally follow: Did any have above average IQs (and how far above average)? How big is the majority (75%, 85% or 97%)? How many embryos were studied to come up with this statistic?
Gamblers like Las Vegas odds makers are beginning to use big data analysis to better predict the outcome of sports contests. The result is that the Las Vegas odds makers can tell us that when number one ranked University of Alabama plays unranked Citadel in 1000 football games, Alabama will win 999 times. That also means that Citadel will win one of those 1000 games, and, prior to the start of the game, you never know which of those 1000 games you are watching. Last weekend, at halftime, Alabama and Citadel were tied 10-10. At that point, it looked like Citadel was going to win that one game out of 1000 and defy the odds makers.
And that is the point. Many have already decided that it is good to terminate the life of an embryo if he or she has a point mutation for a serious disease. Is it just as good to terminate the life of an embryo if he or she has the mere risk of some non-disease trait we find undesirable like the wrong eye color, sex or intelligence? Do we really have sufficient data to make this decision? How much risk is too much or too little? How do we know if the decision is a good one? Will the decision get us the “Children We Want”? Is it really our decision to make?
It is necessary we answer questions like these if we wish to take on the responsibility to decide who lives and who dies, a decision that we previously relegated, depending on your worldview, to God or Nature.