exercises. So the researcher… profile in relation to physical exercise capacity, focusing specifically on a gene known as CREB1. People with a high-risk CREB1 genotype, rather than a protective CREB1 genotype, typically have poorer aerobic capacity and struggle to benefit from physical exercises. So the researchers split participants into groups according to these genotypes.
Erman Misirlisoy, PhD
A very nice article and I agree with the main thrust. Certainly people’s beliefs about their relative risks for certain bad outcomes will impact their behavior which can of course feedback into impacts on their health in various ways. This is why it is so critical that we define “risk” precisely and avoid the types of statements like the above I highlighted.
It is the kind of statement that drives me nuts, and is one of the reasons people get so bent out of shape about the results of these genetic tests. How exactly is a “high risk CREB1 genotype” defined? And what exactly is “high risk?” The first question can be answered the second is subjective and depends on how one defines both the term “risk” and most especially the term “high.” If you actually look at the paper in which CREB1 supposedly was shown to have this “strong association” with exercise heart rate response to regular exercise you quickly see that the claims are not all they are cracked up to be. Turns out there were (at least) 9 SNPs associated with this effect and together they only accounted for 20% of the variance in exercise response rate. Of this paltry 20% only 1 SNPs was actually in the CREB1 gene accounting for a measly 5.4% of the variance. Their were at least 5 other mutations required in other genes in the locus before the original linkage signal could be fully accounted for. And, lest we forget, that is only a 20% variance association. Now I do not have any idea what genetic test was run to determine who was in the so called “high risk” group and who was not, but even if it looked for all 9 SNPs and found them all, this high risk group does not seem all that risky to me. Therefore to say that people with a “high risk” genotype “typically” have poorer aerobic capacity is simply incorrect, unless you consider something with a 20% variance association as “typical.” I would suggest that most people would not consider this typical but would in fact consider it atypical. And I would wager the genetic test did not look for all 9 SNPs and possibly only looked for the one in CREB1 in which case the “typical” is now only 5.4%. This analysis of course also glosses over the details of what 20% and 5.4% actually mean in the context of regression modeling of variance. Suffice to say it does not mean 1 in 5 people or 1 in 20 people. Moreover, the study was based on an analysis of data from the Heritage family study which has a number of well known biases and flaws that I am going to guess were not explained to the test population of so called “high risk” individuals. More likely it was mentioned in passing or in a confusing statement included in the subjects study waiver package.
Now this paper is from way back in 2010 and I have no doubt follow up studies have “confirmed” the association. Though I would also note that other studies have questioned these links or called into question the basis of the entire project of linking single point mutations to various unhappy health outcomes. It turns out that CREB1 has seen a lot of activity and shown “associations” with all sorts of interesting things in the past decade. In fact, as recently as 2017 a single CREB1 SNP was found to be “associated” with prospective memory. My unfounded claims alert radar went into danger mode with this one as to suggest that a single point mutation in a single gene could have measurable and repeatable effects on a phenomenon as complex as memory beggars the imagination. Moreover, my confidence is not bolstered when I read in the abstract “…. After multiple testing adjustments, a significant association was found…” Maybe I am an old school scientist or something, but adjusting the test parameters either during or after the experiment is generally frowned upon as a solid application of the scientific method. Specifically when using statistical models to “prove” such associations “test adjustments” could be read as “statistical manipulation to achieve the desired result.”
How long until this is a reality?