Illustration by Cei Willis
Last Friday, the FDA forced personal genomics company 23andMe to stop marketing its tests to the public in their current form. Before the order came in, customers would send a spit sample to the firm, who would sequence the DNA and look for genes indicating a risk of up to 254 diseases and conditions, providing a breakdown of any issues.
The FDA cited a lack of supporting evidence for some of the claims made and expressed particularly serious concern over their assessment of the BRCA gene, which is linked to breast cancer, suggesting 23andMe's tests might result in false positives that could lead to women undergoing traumatic and unnecessary surgery. The FDA’s actions have led to an explosion of opinion across the science blogosphere, but in all of that commentary a big question remains unanswered: What exactly is the point of personal genomics?
My first experience with the industry came about three years ago, when I was offered the chance to have a test done with Navigenics, a firm since taken over by a biotech firm called Life Technologies. Being a curious sort of guy, I jumped at the chance. A sample tube arrived via Fedex a few days later, which I duly spat into and sent back for analysis.
The results came back in the form of a sort of "wall of death"—a breakdown of all the things that might harm or kill me over the coming decades, detailing how likely I am to have each condition. Drilling into the figures, I can see that I have a higher risk of prostate cancer than 95 percent of the population and a 1 in 5 chance of developing Alzheimer’s—twice the average risk. So I’ll probably get cancer, but on the plus side I'll be too forgetful to care about it.
I love data, and I’m a huge fan of the "quantified self" concept. Every day, an army of gadgets monitors pretty much everything I do: the Runkeeper app to monitor all the exercise I do, MyFitnessPal to scan all the food I eat, and so on. So you’d expect me to geek out over a numerical portrait of my genome, but I found the experience underwhelming. Sure, there was the initial excitement at finding out what’s probably going to kill me if bad wine and clumsiness don’t get there first; but beyond that it simply didn’t engage me, because the more you drill down into the results, the more pointless they seem.
My "wall of death"—the results from my Navigenics test
To illustrate the problem, let’s look at the big risks. I have a 20 percent chance of developing Alzheimer’s, so what can I do about it? Their advice tells me to get more exercise and use my brain more. How about prostate cancer? I should exercise, stay in shape, and discuss screening in a couple of decades. Heart attack? Exercise, don’t smoke, check blood pressure. Atrial fibrillation? Keep in shape, eat less, don’t smoke or drink too much, check your blood pressure. Obesity? I think you can probably guess this one.
For all the data, there’s surprisingly little information. For every major risk, the preventative measures are the same: lead a healthy lifestyle. Well, duh. Whether my risk of a heart attack is 38 percent or 42 percent is basically meaningless. My risk of prostate cancer looks high—28 percent against a population average of 17 percent—until you realize that, like most conditions, prostate cancer can be caused by a mixture of genetic and environmental factors. In this instance, the page tells me that the causes are 58 percent environmental and 42 percent genetic, which makes the 28 percent headline figure essentially a load of bullshit. That calls into question their advice to seek early screening, given repeated findings that show it makes very little difference to mortality and may lead to lots of people having unnecessary treatment.
Then there’s the question of how accurate the tests are in the first place. This is one of the issues flagged by the FDA, and while I haven’t used 23andMe’s services myself, my friend and colleague at the Guardian, Suzi Gage, a scientist working in public health, expressed concerns to me about the results she received.
"There's info and links to the papers that they cite as evidence for each of the genetic traits they talk about," she said. "However, some of the papers are pretty shit. And a lot of them are based on a few specific genes, which have been implicated in LOADS of different traits by candidate gene studies, which are often not replicated (and not for want of trying either). The company 23andMe does make some attempt to rate the strength of evidence for the traits that they're talking about, and the ones with weaker evidence are toward the bottom of the screen, but I don't know if you might need a decent understanding of genetics to really clearly know how reliable the info presented is."
Regardless of your understanding, once you’ve had the test you’re basically on your own. Having bought the test, there’s no real need to visit the website again or spend more money—you’re done. It’s not like you can improve your genome over time. Nor can you, for example, integrate the results with Runkeeper and watch your risks change over time as you get more exercise. Having paid for this service once, there’s no need for me to spend another penny on personal genomics in the future. And if you were tempted to splash out, well, you’re better off getting a new pair of running shoes or a set of tiny plates.
That brings us to the key difference between quantified self and personal genomics companies. Runkeeper, MyFitnessPal, and the like are bent on improving the experience of a growing army of loyal users who engage with their services on a daily basis; 23andMe and its competitors will never have that kind of focus, because consumers aren’t the end game. The great irony at the heart of the personal genomics industry is that it isn’t really personal at all. Like all the richest Silicon Valley start-ups, it’s all about the data.
To understand this, consider the cost of the tests: 23andMe’s product is priced at $99, supposedly making the sub-$100 personal genome test a reality. However, that price point was only possible due to tens of millions of dollars' worth of investment sunk into the company. One recent funding round raised $50 million in financing from the likes of Google Ventures, Google co-founder Sergey Brin (whose wife, Anne Wojcicki, co-founded 23andMe) and various venture capital firms.
That level of investment makes sense when you realize that it isn’t about making cheap genome tests available to all, but money spent building the world’s largest database of genome data, happily handed over by 23andMe’s hundreds of thousands of users. In that respect, the company is following the path carved out by Google and Facebook previously: create a vast and unique lump of data—be it search data, the social graph, or genome data—then monetize it.
This was set out pretty explicitly by 23andMe board member Patrick Chung in a recent interview, "The long game here is not to make money selling kits, although the kits are essential to get the base level data. Once you have the data, [the company] does actually become the Google of personalized healthcare."
This means a couple of things. The first is that the company doesn’t really put consumers’ interests at the heart of its business. Their experiences and the advice they receive aren’t really that important in the long run—once you’ve spat in the cup, you’re of zero further interest to the firm. For that reason, it’s excellent to see the FDA—who do care about consumers—stepping in to question some of the results being issued.
The second is that there are huge implications for future research and reproducibility. On the one hand, without private funding this giant genome database wouldn’t exist; but on the other it’s now firmly in the grip of a group of venture capitalists. Will academics be able to afford the same kind of access as big biotech firms? And, if not, what does this mean for the reproducibility of any results gained from studying this data? Put simply, can we trust the claims of a company like 23andMe or its partners if we can’t see the evidence used to support them?
Regardless of the FDA’s actions, the personal genomics industry will continue; ultimately, I will be able to send my spit anywhere in the world. Meanwhile, it’s good for science journalists and bloggers to put aside the geek fandom and start asking some more serious questions about this technology. Who is it for, exactly? What would a consumer-friendly genomics company really look like, and what can they offer back to the public? And what does it mean for such an important set of data to be monopolized by a small group of people in a valley in California?