For the past few weeks, I’ve been discussing and debating with friends and colleagues a recent study on the genetics of training. The study in question was published in the journal Biology of Sport last month, by researchers affiliated with a British company called DNAFit. Its results are… intriguing, to say the least.
Before I get into the details, I should start with something that everyone agrees on: Your genes influence how you respond to training. Starting with Claude Bouchard’s Heritage study several decades ago, evidence has piled up that if you give identical training programs to different people, they’ll respond differently—and those responses tend to run in families.
From there, it’s a short leap to conclude that if you (and your genes) don’t respond well to one type of training, you might respond well to a different type of training. That’s one of the big takeaway messages I got from David Epstein’s The Sports Gene a few years ago. As Epstein himself told RW:
… I do think we should pay more attention to the kind of training that we see benefits from. This made a huge difference for me. I started running track in high school. I was going to run cross-country for the first time my last year and I tried to make up for lost time by running a load of kilometres. I got up to 128 kilometres s a week, more mileage than probably 99.99 per cent of high schoolers—and I ran even worse. It totally killed my speed. And I remember then getting back into speedwork, and even my distance times would get better. When I ran at university, when I focused on 200s and 400s, my kilometre time would improve. Even my cross-country times would improve.
The next leap is to say, okay, if my genes dictate which type of training is best for me, let’s test my genes and figure out what genes I’ve got and what type of training I should do.
That is, in fact, what at least 39 different companies offer to do now, according to a consensus statement published last year in the British Journal of Sports Medicine, which concluded that such testing is, at this point, basically useless for prescribing training. That’s because so many genes influence training response that knowing one (or 15) of them doesn’t tell you much.
One of those companies is DNAFit, and that’s where their new study comes in. They assigned 123 young athletes (of whom 67 completed the study) to do eight weeks of strength training one to two times a week.
Both groups did the same six strength-training exercises; the only difference was that one group did “power” workouts with 10 sets of two reps for each exercise, while the other group did “endurance” workouts with three sets of 10 to 20 reps for each exercise.
Before starting, each volunteer gave a spit sample for the DNAFit test, which assesses the presence of 15 gene variants that previous studies have linked to power or endurance performance, then combines them in a proprietary (and confidential) algorithm to classify each subject as better suited to power or endurance training. The DNA results were “double-blinded,” hidden both from the athletes and from the researcher supervising the training.
Half of the power athletes were assigned to power workouts, while the other half did endurance workouts; the endurance athletes were similarly divided. After eight weeks, those whose training matched their genes had improved their countermovement jump (a measure of explosive power) and performance in a three-minute cycling test (a measure of endurance) by 6 to 7 per cent. In contrast, those whose training and genes were mismatched improved on both measures by 2 to 3 per cent.
This is a remarkable result. But I spoke to half-a-dozen scientists about the study, and every single one of them was skeptical about the findings. And there are, indeed, some methodological issues. One is the number of drop-outs.
Another is the fact that the DNAFit test produces a 0-to-100 scale of power vs. endurance, but the results are presented as just two categories. How do people who score 49 or 51 per cent power respond to the different types of training? I asked to see the individual results, but the authors were unwilling to send them.
In the end, the biggest red flag to me wasn’t any particular methodological flaw—it was that the results simply seemed too good to be true. And that, perhaps paradoxically, is why I ended up deciding to write about the study despite my doubts. Because if I choose to ignore studies simply because the results are unexpected, then what I’m really saying is that I don’t believe in studies!
Of course, it’s perfectly possible for good studies to give bad results, either because of some unwitting error or simply because of random chance. So we need to try to give appropriate weight to new studies: Be skeptical of small studies, in-house studies, studies that run counter to expectations (all of which are true in this case). I’m not convinced the new study is right, but I’m not convinced that it’s wrong, either. So I’m trying to present the results with appropriate context.
As an aside, one of the questions that occurred to me when I read the study is: How much of this could the subjects have guessed in advance? Personally, if I had to guess what my genes say, it would be that I respond best to endurance training. But when I asked Craig Pickering, one of the authors of the study, about that, he noted that he had tested at 65 per cent endurance—and he was a sprinter on the British 4 x 100-metre relay team at the 2008 Olympics, with a best of 10.14 seconds!
On that note, it’s worth pointing out that the genetics of training response aren’t necessarily the same as the genetics of what events you’re good at. Pickering may be born to sprint, but nonetheless—according to this test, at least—may respond well to training with more reps and a lighter load compared to fewer reps with a heavier load. That’s not necessarily a contradiction.
The last point I want to add here is that we should keep separate the debate about (a) whether a genetic test can tell you what sort of training you respond best to, and (b) whether, if the science is eventually validated, you should buy a genetic test to find out what sort of training you respond best to.
On the second question, I remain a skeptic, though I don’t see much harm in it if you’re interested. But I suspect discomfort with the second question is coloring the response to the first question—observers are skeptical of the DNAFit study not just because of its methodology and results, but because those results are being used to hawk a product to consumers. I understand that skepticism, but ultimately the science should be judged on its merits, regardless of whether it comes from academia or industry.
That, in the end, is why I decided to write about the study. As you can probably tell, I remain conflicted—but I’m certainly looking forward to seeing whether the results will be replicated by future studies.