DNA Day Hype

Happy DNA Day everyone. On this date in 1953, Nature published four articles on the structure of DNA, including the 800-word, data-free masterpiece by Watson and Crick—but also the work of Rosalind Franklin, Raymond Gosling, and Maurice Wilkins that did actually have data, and without which the first Watson and Crick paper would have been handwaving fluff. The Watson-Crick paper is a rightful classic of the scientific literature, but it’s too easy to forget those who provided the evidence to back them up.

MRC Biophysics Unit from PaulingBlog.
The MRC Biophysics Unit in 1951, from Paulingblog. Wilkins is scrunched up at the far left. Gosling is on his feet straining his lower back at the right.

To celebrate, the genetic testing company 23andMe posted a DNA Day infographic that is a marvelous inadvertent evidence of genetic oversell. That’s the best kind, because it unself-consciously undermines its own claims.

Rosalind Franklin
An unusual image of Franklin at the microscope, and the familiar portrait, from fantagabriele.blogspot.com.

These claims are about health. Last year, the company was ordered to stop marketing their genomic testing service as a health service and it agreed to stop selling it altogether. It would henceforth focus on the genealogy side of their service. They are evidently sneaking back in, though, with ads—sorry, “infographics”; so much more documentary-like than “advertisements”— like this one.

Ninety-one percent of Americans, it trumpets in giant type at the top of the ad, “correctly believe that knowing their genetic information can be helpful in managing their health.” On one level, Well, duh. Everyone knows that some diseases run in families: you don’t have to have a high level of genetic literacy to be aware that knowing whether your mother, grandmother, great-grandmother, and aunts have had breast cancer is a pretty useful little medical tidbit. The statement is worded so vaguely as to be meaningless. The remaining nine percent probably have some strongly hereditary learning disability that keeps them from correctly knowing how to feed themselves.

On a second level, though, I’d like to know what percentage of Americans incorrectly believe that knowing their genetic information can be helpful in managing their health. What percentage, for example, think that having one of the BRCA risk-factor alleles means they are going to get breast cancer unless they have a mastectomy? What percentage believe that a 300% increase in risk for an extremely rare disease—from one in 3 million, say, to one in a million—is cause for alarm? What percentage think that the association of a single nucleotide polymorphism with a genetic disease means that biomedicine has the cause—let alone a cure—for that disease? What percentage of Americans, in short, have no understanding of probability, pleiotropy, penetrance, or gene–environment interaction, and yet read ads from companies such as 23andMe and think, “Yee-haw! I can learn what diseases I’m going to get, and which ones not, just by spitting in a cup!”

Watson and Crick
The dynamic duo. From The Sandwalk.

The infomercial continues downward, with more statistics: smaller numbers in smaller type. Thirty-one percent know that genetic testing can “show their body’s ability” to metabolize caffeine, etc. At the bottom, though, the numbers get large again. “People still need a refresher on the basics of genetics,” they say. Forty-nine percent of women “believe their sex chromosome is XY.” Their sex chromosome is XY? What percentage of genetic testing companies employ staffers who can write simply and accurately about genetics? Another statistic: forty-one percent don’t realize DNA is organized into chromosomes.

Finally, in tiny print at the very bottom, they tell us that the survey was conducted on 1000 “nationally representative Americans” by an “independent research firm, Kelton.” Kelton Global is a marketing firm that specializes in repositioning companies that have lost market share or want to break into new markets. Their motto is “helping brands navigate change.” They take surveys, track metrics, re-brand companies, and so forth. Their niche is using numbers to persuade and making statistics say what their clients want them to say.

Let’s make a few postulates for the sake of argument. Let’s say that this is a real sample, designed seriously by people who understand statistics. Let’s say the questions were worded better than this and that those questioned understood what they were being asked. Let’s assume the ad was just badly written. It may be that these are totally unjustified, but we’ll give them the benefit of the doubt for just a moment.

If their numbers are in fact meaningful, what they show is that people are buying the hype about genetic testing without understanding it. How happy should we be that people who don’t know what a chromosome is nevertheless believe that genetic testing can tell them about their health? We’re not talking about informed decision making about subtle and complex data; we’re talking drinking the Kool-Aid. What this ad says, most of all, is that even though officially 23andMe is out of the health-claim game, they are still very interested.

Why aren’t genome profiles free? A cynic’s view

Charles Seife’s piece over at sciamblogs the other day gave me one of those forehead-slapping, “if-it-was-a-double-helix-it-would-have-bit-me” moments. For 23andMe, the “test” or genome profile is small potatoes. The real product of 23andMe isn’t the saliva test.

It’s the database, stupid.

Think about gmail or Facebook. When you sign up for a “free” account you agree to allow them to send you targeted promotions, in the form of ads that appear in the margins. It’s hilarious how bad their algorithms are. When I rant about some bonehead right-wing politician, I start getting suggestions to follow Mitt Romney. When my wife posted wedding pictures, she got ads for wedding registries. My google profile is amazingly bad–mixed in with a few genuine interests, it lists dozens of things I have no interest in whatsoever (fishing, dolls and accessories, apartments and residential rentals) as well as things so general they say nothing meaningful about me (consumer resources, search engines). I’ve often found that reassuring. Though I know the software is bound to get better, it’s comforting at least for the moment to know that they don’t actually know me that well. I’m not naive enough to think that will last forever.

What 23andMe is really about, says Seife, is doing that same kind of profiling with your genome. Historians and anthropologists of science have long been interested in “biologization” and “medicalization.” Ugly words, useful concepts. Biomedicine tends to shift our gaze from labor to biology (and especially health). Is violence a crime or a disease? Do you identify more strongly as a carpenter, soldier, or professor, vs. as celiac, PTSD, or a breast-cancer survivor? Conceiving some corner of our world in biological terms can have profound implications. If violence is a crime, we treat it with fines, incarceration, or death. If it is an organic disease, we treat it with drugs and counseling. Conceivably, we may someday treat it with gene therapy. As Foucault pointed out, both criminalization and medicalization involve behavior modification–just in different formats. Medicalization can be more humane, but it can also strip away one’s autonomy and subtly and dangerously shift power relationships.

Every time someone sends in their little vial of spit to 23andMe, the company adds to a large-and-growing database of genomic data linked to a broad range of personal tastes and behaviors. Like Google and Facebook, they make you agree to let them send you ads based on the data they collect, which is augmented by their social-media site. It is a genomic version of Google. Welcome to the all-volunteer biological surveillance state.

Seife points out that Anne Wojcicki was married to Google co-founder Sergei Brin, that Google is a heavy investor in 23andMe, and that the price of the saliva test has been dropping steadily and is now below $100.

My cynical view is that the company has an easy end-around available for the FDA letter demanding that they halt marketing their test: give it away. It’s not marketing if they’re not selling, right? Investopedia defines marketing as “The activities of a company associated with buying and selling a product or service.” It does go on to list advertising as one key aspect of marketing, but I wonder whether they could successfully argue that promoting a free product isn’t marketing per se.

Legal definitions aside, if Seife is right, my guess is that before long genome profiling will be like web browsers and email: something almost everyone does, and that except for a few willing to pay for a premium private service, something that provides so many benefits that we tolerate its ads as a necessary trade-off of modern life.

Is this any more insidious than gmail? If we say yes, we risk running headlong into the genetic determinism this blog rails against. We have to be careful about privileging biological information over social information. If Facebook’s suggestions are laughable, they’d likely seem prescient in comparison with what they’d predict based on my genomic profile. The “genes for” most things explain tiny amounts of variance and tend to have low penetrance. Other than a few strongly Mendelian diseases, a genome profile currently says very little about you, simply because it’s based on small probabilities of uncertain precision.

But like the algorithms analyzing your social profile, those combing your genomic profile will improve, and probably at a rate faster than any of us expect. Most importantly, your genomic profile will merge with your social profile, which will greatly enhance the accuracy of both. Your social profile will become biologized–rooted in and interwoven with your DNA.

The gradual way in which 23andMe is heading toward an open-source business model may simply reflect the high cost of getting the biotech version of Google Plus off the ground. As profits increase, they can afford to drop the price. When it hits zero–when they start giving away the test–rest assured that ad revenues will then be enough to keep the shareholders happy.

23andMe, FDA, and the history of hype

Yesterday I and seemingly everyone else interested in genomes posted about the FDA letter ordering the genome diagnostics company 23andMe to stop marketing their saliva test. FDA treats the test as a “medical device, because “it is intended for use in the diagnosis of disease or other conditions or in the cure, mitigation, treatment, or prevention of disease, or is intended to affect the structure or function of the body.” The company first issued a bland, terse statement acknowledging the letter and then company president Anne Wojcicki signed a short post affirming the company’s commitment to providing reliable data, promising cooperation with FDA, and reasserting her faith that “genetic information can lead to better decisions and healthier lives.” (I say she “signed” it because of course we have no way of knowing whether she composed it and she’s no fool: surely the text was vetted by Legal.) In other words, the company followed up with a bland, less-terse response, carefully worded to reassure customers of the company’s ethical stance and core mission. Reactions to the FDA letter range from critics of the company singing “Hallelujah!” to defenders and happy customers are attacking FDA for denying the public the right to their own data. The 23andMe blog is abuzz and, hearteningly, a few sane souls there are trying to dispel misinformation.

I am doing history on the fly here. If journalism is the first draft of history, let’s take a moment to revise that first draft—to use the historian’s tools to clear up misconceptions and set the debate in context as best we can. The history of the present carries its own risks. My and other historians’ views on this will undoubtedly evolve, but I think it’s worth injecting historical perspective into debates such as these as soon as possible.

*

We must be clear that the FDA letter does not prohibit 23andMe from selling their test. It demands they stop marketing it. The difference may not amount to much in practice—how much can you sell if you don’t market your product?—but the distinction does help clarify what is actually at stake here. FDA is not attempting to instigate a referendum on the public’s access to their own DNA information. They are challenging the promises 23andMe seems to make. This is, in short, not a dispute about access, but about hype.

The company seems to promise self-knowledge. The ad copy for 23andMe promises to tell you what your genome “says about you.” “The more you know about your DNA,” they trumpet, “the more you know about yourself.” On one level, that’s perfectly, trivially true: your genome does have a lot to do with your metabolism, body structure, how you respond to disease agents, and so forth. The problem is, we as yet know very little about how it all works. The 23andMe marketing exploits a crucial slippage in the concept of “knowledge,” which FDA correctly finds misleading. In short, the marketing implies a colloquial notion of knowledge as a fixed and true fact, while the science behind the test is anything but.

Historians and other scholars of science have thought a lot about the concept of scientific knowledge. In 1934, Ludwik Fleck wrote about the “genesis and development of a scientific fact,” namely the Wasserman test for syphilis. It is a pioneering classic in a now-huge (and still growing) literature on how scientific facts are created. Science claims to gather facts about nature and integrate them into explanations of natural mechanisms. A moment’s reflection reveals that very few scientific facts last forever. Most, perhaps all, undergo revision and many are discarded, overthrown, or reversed. They are historical things, not universal truths. A surprisingly small amount of what I learned in science courses 20 and 30 years ago is still true. As that great philosopher of science John McPhee wrote, “science erases what was previously true” (Oranges, p. 75). Because scientists search for universal, timeless mechanisms, they easily slip into language suggesting that they discover universal, timeless truth. But there is uncertainty, contingency, malleability built into every scientific fact.

This goes double for genome information. The 23andMe product, like every genome test, provides probabilities of risk, not mechanisms. Probabilities are messy and hard to understand. They carry an almost irresistible tendency to be converted into hard facts. If you flip a coin 9 times and it comes up heads every time, you expect the next flip to come up tails. And if you get heads 49 times in a row, the next one has got to be tails, right? Even if you know intellectually that the odds are still 50:50, just like on every previous flip. You can know you have a particular gene variant, but in most cases, neither you nor anyone else knows exactly what that means. Despite the language of probability that dots the 23andMe literature, their overall message—and the one clearly picked up by many of their clientele—is one of knowledge in the colloquial sense. And that is oversell.

Human genetics has always been characterized by overstatement and hype. In the early 1900s, the rediscovery of Mendel’s laws persuaded many that they now understood how heredity works. Although every scientist acknowledged there was still much to learn, prominent students of human heredity believed they knew enough to begin eliminating human defects through marriage and sterilization laws. We now view such eugenic legislation as almost unbelievably naive. Combine that naivete with race, gender, and class prejudice and you obtain a tragically cruel and oppressive eugenics movement that resulted in the coerced sterilization of many thousands, in the US and abroad—including, of course, the Nazi sterilization law of 1933, based on the American “model sterilization law,” which culminated not only in racist forced sterilization but euthanasia.

Human-genetic hype hardly ended with the eugenics movement. In 1960s, as human diseases were finally being mapped to chromosomes, it seemed transparent that if a chromosomal error that produces an individual with an XXY constitution feminizes that individual (which it does), then an extra Y chromosome (XYY) must masculinize. Such “super-males,” data seemed to suggest, were not only taller and hairier than average, but also more aggressive and violent. It was, for a while, a fact that XYY males were prone to violent crime.

The molecular revolution in genetics produced even more hype. When recombinant DNA and gene cloning techniques made it possible to try replacing or augmenting disease genes with healthy ones, DNA cowboys hyped gene therapy far beyond existing knowledge, promising the end of genetic disease. The 1995 Orkin-Motulsky report acknowledged the promise of gene therapy but noted,

Overselling of the results of laboratory and clinical studies by investigators and their sponsors…has led to the mistaken and widespread perception that gene therapy is further developed and more successful than it actually is.[1]

Soon after this report was published, Jesse Gelsinger died unexpectedly in a gene-therapy trial, patients in a French gene-therapy trial for adenosine deaminase (ADA) deficiency unexpectedly developed leukemia, and the gene-therapy pioneer W. French Anderson was arrested, tried, and convicted on charges of child molesting—in other words, abusing and overestimating his power over the children whose health was entrusted to him. The risks of failing to heed warnings about genetic oversell are high.

*

Like gene therapy, genome profiling has great promise, but the FDA letter to 23andMe is a stern reprimand to an industry that, like gene therapy and the entire history of human genetics, blurs the line between promise and genuine results.

The current controversy over commercial genome profiling has two qualities that distinguish it as particularly serious. First, unlike previous examples of overselling human genetics, it is profit-driven. The “oversell” is more literal than it has ever been. Although 23andMe presents as a concerned company dedicated to the health of their clientele, they are also—and arguably primarily—dedicated to their stockholders. In a for-profit industry, oversell is a huge temptation and that risk needs to be made transparent to consumers.

Second, the 23andMe test is being sold directly to individuals who may not have any knowledge of genetics. The tendency to convert risks into certainty is higher than ever. The knowledge they sell is a set of probabilities, and further, those probabilities are not stable. The consumer may not—indeed probably doesn’t—appreciate how much we know, how much we don’t know, and how much we don’t even know we don’t know. The company claims to be selling knowledge but in fact they are selling uncertainty.

In a characteristically insightful and clarifying post, the geneticist (and 23andMe board member) Michael Eisen doubts whether the 23andMe test will ever meet FDA’s definition of a “medical device.” It is not an MRI machine or a Wasserman test. It’s something new. Adequate regulation of products such as the 23andMe genome profile will require rethinking of what exactly the company is marketing.

Putting this controversy in context, then, illustrates another critical risk: the risk of failing to acknowledge the uncertainty underlying the science. In some sense, the more we learn, the less we know.

 


[1] Orkin, S. H., and A. Motulsky. Report and Recommendations of the Panel to Assess the NIH Investment in Research on Gene Therapy.  Bethesda, MD: National Institutes of Health, 1995.

A blow for personal genome testing

Hey honey–remember when I accidentally left the chicken coop open and they all flew away? Well I think they’ve come back home to roost!

Last summer, we did an analysis of the 23andMe commercial promoting their genetic testing service and the egotistical identity politics it both taps into and contributes to. The ad was all about how your genes were “You” and knowing about them would enable you to predict your genetic future. Genetic profiling can in some cases give robust statistical estimates of likelihood of certain genetic conditions, but it is safe to say that we rarely know what that means. And it’s presented as though we do.

Now we find that FDA is ordering 23andMe to stop marketing their tests.

The 23andMe saliva sample kit, says FDA, is a “medical device,” “intended for use in the diagnosis of disease or other conditions or in the cure, mitigation, treatment, or prevention of disease, or is intended to affect the structure or function of the body.

They cite the company’s claims to allow patients’ genome profiles to help them assess “health risks,” and “drug response,” and specifically as a “first step in prevention” that enables users to “take steps toward mitigating serious diseases” such as diabetes, coronary heart disease, and breast cancer.

23andmescreenshot

This is not a shot over the bow–it’s the last straw. FDA has warned 23andMe repeatedly, going back to July, 2012, that they were making health claims about their product that they couldn’t back up.

The company offers two types of products: a genealogical “panel” or profile, and a health panel. The genealogical panel is popular but is apparently considered a harmless hobby, or at least outside the purview of the Public Health Service. It is not clear whether FDA (which, like the National Institutes of Health and the Centers for Disease Control) falls under the sprawling PHS will have any concerns about genealogical applications of the saliva test, but that would seem unlikely. The problem for 23andMe is that, as shown by the ad we analyzed earlier, they have been pushing the health panel very hard. Family trees are a hobby; health is where the real money is.

Direct-to-consumer medicine trails an appealing democratic, anti-authoritarian perfume that seems to make people slightly drunk. Mild intoxication can be pleasant, need not be dangerous, and sometimes can be a spur to creativity. But it can also impair your judgment. When you’ve gotta drive the kids home, you may need a couple cups of good strong regulatory coffee and a couple hours to sober up before getting behind the wheel.

A good deal of “preventive, participatory, personalized” medicine is profit-driven, and stockholders don’t necessarily have the public’s health foremost in mind. The FDA warning is a good illustration of why it’s important to balance the goal of stimulating innovation and economic growth with the goal of maximizing health. For the former, the free market can be a powerful tool. But for the latter, sometimes you need a little good old-fashioned meritocratic oversight.

h/t Robert Resta, Mark Largent

 

23andMe, myself, and I

Here is the new ad from 23andMe that will begin airing shortly on cable TV*:

Genomics is going mainstream and the best news is first that it’s real simple and second that it’s all about me.

Let’s take the most obvious first: the “me” meme. Of course this relates to the company name, but the ad takes me to a new level. It makes “you” your DNA. I give them points for a couple of qualifiers — it “helps” make me who I am, one character says. But the overall message is that you are your genes.

It also exploits the meme of egocentrism. Nearly everything today seems to be all about me. Memoirs are the hottest genre of nonfiction. We have a magazine called “Self.” One of the most common themes on commercial websites is to have a “My [company name]” area, which usually just means they have your personal information to use to sell you more stuff. There’s even a “.me” internet domain, which they advertise “is all about you.” Who isn’t curious about himself? I’m the most interesting topic in the world! And 23andMe will tell me about my true inner nature for just $99.

One element of personalized medicine, then, is narcissism. Another, more noble, element is individuality. No one is more committed to his individuality than I am—but I’m also wary of its dark side: selfishness. I am struck by the single reference to future generations (“what I will pass on to my kids”). Again, this is a two-sided coin. In the Progressive era, the literature on genetic medicine emphasized family and community. There isn’t a hint of that here. On the one hand, then, the ad is free of the eugenic message of controlling human evolution. On the other, it’s relentlessly selfish. Most likely, the reason for staying away from issues such as family, community, and responsibility is that it enables them to steer way wide of abortion. This ad is about me, not my kids and not the future. That’s actually a new and rather radical development in genetics. 

A persistent theme in popular literature from the 19th century to the 21st, is that hereditary information provides certainty. This despite the fact that one of the signal insights from genomics is how uncertain its results are. Genetic medicine today is all about probabilities, and to make informed decisions based on our genetics we have to understand how probability works. The ad works against this principle, promising certainty where there is only chance. “Now, I know” says one woman. No, you don’t. Now, you have a sense of risk—not certainty. This is a dangerous over-simplification.

double helix
Simplified double helix from Watson and Crick’s 1953 paper.

This sense of simplicity is also carried in the graphics. Note how there’s hardly a double helix in it. “Your” DNA is reduced to circles, dots, and lines. They move and whirl entertainingly and there’s just enough suggestion of complexity to carry the message that you can’t understand “you” without them‚ 23andMe. If DNA becomes as central to identity as companies such as 23andMe want to make it, this ad suggests that its iconic image may fade. Even the stripped-down ribbons and bars version is simply too complex for TV.

karyotypedowns
An early karyogram (of Down syndrome) from the 1960s.
A comprehensive chromosome map from UCSF.
Screen shot from 23andMe commercial. Her “DNA” is those two colorful cylinders by her ear.

Most of the genetic “knowledge” promised is simple enough to be carried in the one- and two-syllable words that dominate mass-market media. Genetic medicine, stuffed as it is with Latinate and Greek words, is a tough sell in that market, but the ad pulls it off. At 0:21 we hear the longest word in the ad: “hemochromatosis.” The speaker pauses after the second syllable, to suggest empathy with viewers who get hung up on such terms. According to the Mayo Clinic website, hemochromatosis is indeed usually inherited, is rarely serious, is most common in men, and is the most common genetic disease in Caucasians. The ad script gives this word to a black man. Thus, one of the ad’s subtle messages is to erase racial differences—even differences supported by scientific evidence. It’s a commonplace in TV ads nowadays to feature men and women of many hues, but the 23andMe ad takes it a step further.

Another theme of the commercial is the way it suggests communities based around biological identities of health and disease. Once, our primary identities were with those who lived near us, or shared our work or hobbies or politics. But politics has become personal, our communities are digital, and our identities center around health. The sociologist Nikolas Rose calls this “biological citizenship.” The 23andMe website features forums where members who share particular mutations or risks can discuss diets, lifestyle habits, child-bearing decisions–or their pets, if they wish. They are communities based around health. The ad sends the message that race, class, and gender are no longer our defining social themes: what matters now is health and disability.

We hear so much about the importance of educating the public about their biology as a key component of contemporary personalized medicine, but in this ad that biology is reduced to bumper-sticker-like phrases about this circle “saying” I will have blue eyes and that line segment “saying” I have a risk of this or that disease. Learning about me will be fun, easy, and inexpensive. Thank goodness I can mail off a C-note, spit in a cup, and in a few weeks get a report that simplifies it all in language I can understand. The ad ends with a rainbow of people chanting “Me. Me. Me.” It’s the “Om” of the 21st century.

 

*h/t to Bob Resta for sending the link to the ad, and to Shirley Wu (@shwu) for a tweet that showed me that the hemochromatosis passage was too terse in yesterday’s version. I’d been wanting to add something about biological citizenship and Shirley’s comment suggested a way to do it.