23andMe Is Using Customers’ Genetic Data to Develop Drugs. Is This Brilliant or Dubious?
Leading direct-to-consumer (DTC) genetic testing companies are continuously unveiling novel ways to leverage their vast stores of genetic data.
"23andMe will tell you what diseases you have and then sell you the drugs to treat them."
As reported last week, 23andMe's latest concept is to develop and license new drugs using the data of consumers who have opted in to let their information be used for research. To date, over 10 million people have used the service and around 80 percent have opted in, making its database one of the largest in the world.
Culture researcher Dr. Julia Creet is one of the foremost experts on the DTC genetic testing industry, and in her forthcoming book, The Genealogical Sublime, she bluntly examines whether such companies' motives and interests are in sync with those of consumers.
Leapsmag caught up with Creet about the latest news and the wider industry's implications for health and privacy.
23andMe has just announced that it plans to license a newly developed anti-inflammatory drug, the first one created using its customers' genetic data, to Almirall, a pharma company in Spain. What's your take?
I think this development is the next step in the evolution of the company and its "double-sided" marketing model. In the past, as it enticed customers to give it their DNA, it sold the results and the medical information divulged by customers to other drug companies. Now it is positioning itself to reap the profits of a new model by developing treatments itself.
Given that there are many anti-inflammatory drugs on the market already, whatever Almirall produces might not have much of an impact. We might see this canny move as a "proof of concept," that 23andMe has learned how to "leverage" its genetic data without having to sell them to a third party. In a way, the privacy provisions will be much less complicated, and the company stands to attract investment as it turns itself into [a pseudo pharmaceutical company], a "pharma-psuedocal" company.
Emily Drabant Conley, the president of business development, has said that 23andMe is pursuing other drug compounds and may conduct their own clinical trials rather than licensing them out to their existing research partners. The end goal, it seems, is to make direct-to-consumer DNA testing to drug production and sales back to that same consumer base a seamless and lucrative circle. You have to admit it's a brilliant business model. 23andMe will tell you what diseases you have and then sell you the drugs to treat them.
In your new book, you describe how DTC genetic testing companies have capitalized on our innate human desire to connect with or ancestors and each other. I quote you: "This industry has taken that potent, spiritual, all-too-human need to belong... and monetized it in a particularly exploitative way." But others argue that DTC genetic testing companies are merely providing a service in exchange for fair-market compensation. So where does exploitation come into the picture?
Yes, the industry provides a fee for service, but that's only part of the story. The rest of the story reveals a pernicious industry that hides its business model behind the larger science project of health and heredity. All of the major testing companies play on the idea of "lack," that we can't know who we are unless we buy information about ourselves. When you really think about it, "Who do you think you are?" is a pernicious question that suggests that we don't or can't know who we or to whom we are related without advanced data searches and testing. This existential question used to be a philosophical question; now the answers are provided by databases that acquire more valuable information than they provide in the exchange.
"It's a brilliant business model that exploits consumer naiveté."
As you've said before, consumers are actually paying to be the product because the companies are likely to profit more from selling their genetic data. Could you elaborate?
The largest databases, AncestryDNA and 23andMe, have signed lucrative agreements with biotech companies that pay them for the de-identified data of their customers. What's so valuable is the DNA combined with the family relationships. Consumers provide the family relationships and the companies link and extrapolate the results to larger and larger family trees. Combined with the genetic markers for certain diseases, or increased susceptibility to certain diseases, these databases are very valuable for biotech research.
None of that value will ever be returned to consumers except in the form of for-profit drugs. Ancestry, in particular, has removed all information about its "research partners" from its website, making it very difficult to see how it is profiting from its third-party sales. 23andMe is more open about its "two-sided business model," but encourages consumers to donate their information to science. It's a brilliant business model that exploits consumer naiveté.
A WIRED journalist wrote that "23andMe has been sharing insights gleaned from consented customer data with GSK and at least six other pharmaceutical and biotechnology firms for the past three and a half years." Is this a consumer privacy risk?
I don't see that 23andMe did anything to which consumers didn't consent, albeit through arguably unreadable terms and conditions. The part that worries me more is the 300 phenotype data points that the company has collected on its consumers through longitudinal surveys designed, as Anne Wojcicki, CEO and Co-founder of 23andMe, put it, "to circumvent medical records and just self-report."
Everyone is focused on the DNA, but it's the combination of genetic samples, genealogical information and health records that is the most potent dataset, and 23andMe has figured out a way to extract all three from consumers.
A Single Blood Test May Soon Replace Your Annual Physical
For all the excitement over "personalized medicine" in the last two decades, its promise has not fully come to pass. Consider your standard annual physical.
Scientists have measured thousands of proteins from a single blood test to assess many individualized health conditions at once.
Your doctor still does a blood test to check your cholesterol and gauge your risk for heart disease by considering traditional risk factors (like smoking, diabetes, blood pressure) — an evaluation that has not changed in decades.
But a high-risk number alone is not enough to tell accurately whether you will suffer from heart disease. It just reflects your risk compared to population-level averages. In other words, not every person with elevated "bad" cholesterol will have a heart attack, so how can doctors determine who truly needs to give up the cheeseburgers and who doesn't?
Now, an emerging area of research may unlock some real-time answers. For the first time, as reported in the journal Nature Medicine last week, scientists have measured thousands of proteins from a single blood test to assess many individualized health conditions at once, including liver and kidney function, diabetes risk, body fat, cardiopulmonary fitness, and even smoking and alcohol consumption. Proteins can give a clear snapshot of how your body is faring at any given moment, as well as a sneak preview at what diseases may be lurking under the surface.
"Years from now," says study co-author Peter Ganz of UCSF, "we will probably be looking back on this paper as a milestone in personalized medicine."
We spoke to Ganz about the significance of this milestone. Our interview has been edited and condensed.
Is this the first study of its kind?
Yes, it is. This is a study where we measured 5,000 proteins at once to look for patterns that could either predict the risk of future diseases or inform the current state of health. Previous to this, people have measured typically one protein at a time, and some of these individual proteins have made it into clinical practice.
An example would be a protein called C-reactive protein, which is a measure of inflammation and is used sometimes in cardiology to predict the risk of future heart attacks. But what's really new is this scale. We wanted to get away from just focusing on one problem that the patient may have at a time, whether it's heart disease or kidney disease, and by measuring a much greater number of proteins, the hope is that we could inform the health of ultimately just about every organ in the body or every tissue. It's a step forward for what I would call "a one-stop shop."
"I'm very excited about personalized medicine through proteins as opposed to genes because you get both the nature and nurture."
Three things get me excited about this. One is the convenience for the patient of a single test to determine many different diseases. The second thing is the healthcare cost savings. We estimated what the cost would be to get these 11 healthcare measures that we reported on using traditional testing and the cost was upwards of 3,000 British pounds. And even though I don't know for sure what the cost of the protein tests would ultimately be, [it could come down to about $50 to $100].
The last thing is that the measurement of proteins is part of what people have called personalized medicine or precision medicine. If you look at risk factors across the population, it may not apply to individuals. In contrast, proteins are downstream of risk factors. So proteins actually tell us whether the traditional risk factors have set in motion the necessary machinery to cause disease. Proteins are the worker bees that regulate what the human body does, and so if you can find some anomalies in the proteins, that may inform us if a disease is likely to be ongoing even in its earliest stages.
Does protein testing have advantages over genetic testing for predicting future health risks?
The problem with genomics is that genes usually don't take care of the environment. It's a blueprint, but your blueprint has no idea what you will be exposed to during your lifetime in terms of the environment and lifestyle that you may choose and medications that you may be on. These are things that proteins can account for. I'm very excited about personalized medicine through proteins as opposed to genes because you get both the nature and nurture as opposed to genomics, which only gives you nature but doesn't account for anything else.
Proteins can also be tracked over time and that's not something you can do with genes. So if your behavior improves, your genes won't change, but your proteins will.
Could this new test become a regular feature of your annual physical?
That's the idea. This would be basically almost a standalone test that you could have done every year. And hopefully you wouldn't need other tests to complement this. This could be your yearly physical.
How much more does it need to be validated before it can enter the clinic and patients can trust the results?
This was a proof-of concept study. To really make this useful, we need to expand from 11 measures of health to a hundred or more health insights, to cover the whole body. And we need to expand this to all racial groups. Three of the five centers in the study were European – all Caucasian – so it's one of our high priorities to find groups of patients with better representation of minorities.
When do you expect doctors to be routinely giving this test to patients?
Much closer to five years than 20 years. We're scaling up from 11 disease states to 100, and many of those studies are underway. Results should be done within three to five years.
Do you think insurance will cover it?
Good question. I have been approached by an insurance company that wanted to understand the product better – a major insurer, with the possibility that this could actually be cost saving.
I have to ask you a curveball -- do you think that the downfall of Theranos will make consumers hesitant to trust a new technology that relies on using a single blood sample to screen for multiple health risks?
[Laughs] You're not the first person to ask me that today. I actually got a call from Elizabeth Holmes [in 2008 when I was at Harvard]. I met with her for an afternoon and met her team two more times. I gave them advice that they completely disregarded.
In many ways, what we do is diametrically opposite to Theranos. They had a culture of secrecy, and what we do is about openness. We publish, like this paper in Nature Medicine, to show the scientific details. Our supplement is much longer than the typical academic paper. We reveal everything we know. A lot of the research we do is funded by [the National Institutes of Health], and they have strict expectations about data sharing. So we agree to make the data available on a public website. If there is something we haven't done with the data, others can do it.
So you're saying that this is not another Theranos.
No, God forbid. We hope to be the opposite.
Kira Peikoff was the editor-in-chief of Leaps.org from 2017 to 2021. As a journalist, her work has appeared in The New York Times, Newsweek, Nautilus, Popular Mechanics, The New York Academy of Sciences, and other outlets. She is also the author of four suspense novels that explore controversial issues arising from scientific innovation: Living Proof, No Time to Die, Die Again Tomorrow, and Mother Knows Best. Peikoff holds a B.A. in Journalism from New York University and an M.S. in Bioethics from Columbia University. She lives in New Jersey with her husband and two young sons. Follow her on Twitter @KiraPeikoff.
Trying to get a handle on CRISPR news in 2019 can be daunting if you haven't been avidly reading up on it for the last five years.
CRISPR as a diagnostic tool would be a major game changer for medicine and agriculture.
On top of trying to grasp how the science works, and keeping track of its ever expanding applications, you may also have seen coverage of an ongoing legal battle about who owns the intellectual property behind the gene-editing technology CRISPR-Cas9. And then there's the infamous controversy surrounding a scientist who claimed to have used the tool to edit the genomes of two babies in China last year.
But gene editing is not the only application of CRISPR-based biotechnologies. In the future, it may also be used as a tool to diagnose infectious diseases, which could be a major game changer for medicine and agriculture.
How It Works
CRISPR is an acronym for a naturally occurring DNA sequence that normally protects microbes from viruses. It's been compared to a Swiss army knife that can recognize an invader's DNA and precisely destroy it. Repurposed for humans, CRISPR can be paired with a protein called Cas9 that can detect a person's own DNA sequence (usually a problematic one), cut it out, and replace it with a different sequence. Used this way, CRISPR-Cas9 has become a valuable gene-editing tool that is currently being tested to treat numerous genetic diseases, from cancer to blood disorders to blindness.
CRISPR can also be paired with other proteins, like Cas13, which target RNA, the single-stranded twin of DNA that viruses rely on to infect their hosts and cause disease. In a future clinical setting, CRISPR-Cas13 might be used to diagnose whether you have the flu by cutting a target RNA sequence from the virus. That spliced sequence could stick to a paper test strip, causing a band to show up, like on a pregnancy test strip. If the influenza virus and its RNA are not present, no band would show up.
To understand how close to reality this diagnostic scenario is right now, leapsmag chatted with CRISPR pioneer Dr. Feng Zhang, a molecular biologist at the Broad Institute of MIT and Harvard.
What do you think might be the first point of contact that a regular person or patient would have with a CRISPR diagnostic tool?
FZ: I think in the long run it will be great to see this for, say, at-home disease testing, for influenza and other sorts of important public health [concerns]. To be able to get a readout at home, people can potentially quarantine themselves rather than traveling to a hospital and then carrying the risk of spreading that disease to other people as they get to the clinic.
"You could conceivably get a readout during the same office visit, and then the doctor will be able to prescribe the right treatment right away."
Is this just something that people will use at home, or do you also foresee clinical labs at hospitals applying CRISPR-Cas13 to samples that come through?
FZ: I think we'll see applications in both settings, and I think there are advantages to both. One of the nice things about SHERLOCK [a playful acronym for CRISPR-Cas13's longer name, Specific High-sensitivity Enzymatic Reporter unLOCKing] is that it's rapid; you can get a readout fairly quickly. So, right now, what people do in hospitals is they will collect your sample and then they'll send it out to a clinical testing lab, so you wouldn't get a result back until many hours if not several days later. With SHERLOCK, you could conceivably get a readout during the same office visit, and then the doctor will be able to prescribe the right treatment right away.
I just want to clarify that when you say a doctor would take a sample, that's referring to urine, blood, or saliva, correct?
FZ: Right. Yeah, exactly.
Thinking more long term, are there any Holy Grail applications that you hope CRISPR reaches as a diagnostic tool?
FZ: I think in the developed world we'll hopefully see this being used for influenza testing, and many other viral and pathogen-based diseases—both at home and also in the hospital—but I think the even more exciting direction is that this could be used and deployed in parts of the developing world where there isn't a fancy laboratory with elaborate instrumentation. SHERLOCK is relatively inexpensive to develop, and you can turn it into a paper strip test.
Can you quantify what you mean by relatively inexpensive? What range of prices are we talking about here?
FZ: So without accounting for economies of scale, we estimate that it can cost less than a dollar per test. With economy of scale that cost can go even lower.
Is there value in developing what is actually quite an innovative tool in a way that visually doesn't seem innovative because it's reminiscent of a pregnancy test? And I don't mean that as an insult.
FZ: [Laughs] Ultimately, we want the technology to be as accessible as possible, and pregnancy test strips have such a convenient and easy-to-use form. I think modeling after something that people are already familiar with and just changing what's under the hood makes a lot of sense.
Feng Zhang
(Photo credit: Justin Knight, McGovern Institute)
It's probably one of the most accessible at-home diagnostic tools at this point that people are familiar with.
FZ: Yeah, so if people know how to use that, then using something that's very similar to it should make the option very easy.
You've been quite vocal in calling for some pauses in CRISPR-Cas9 research to make sure it doesn't outpace the ethics of establishing pregnancies with that version of the tool. Do you have any concerns about using CRISPR-Cas13 as a diagnostic tool?
I think overall, the reception for CRISPR-based diagnostics has been overwhelmingly positive. People are very excited about the prospect of using this—for human health and also in agriculture [for] detection of plant infections and plant pathogens, so that farmers will be able to react quickly to infection in the field. If we're looking at contamination of foods by certain bacteria, [food safety] would also be a really exciting application.
Do you feel like the controversies surrounding using CRISPR as a gene-editing tool have overshadowed its potential as a diagnostics tool?
FZ: I don't think so. I think the potential for using CRISPR-Cas9 or CRISPR-Cas12 for gene therapy, and treating disease, has captured people's imaginations, but at the same time, every time I talk with someone about the ability to use CRISPR-Cas13 as a diagnostic tool, people are equally excited. Especially when people see the very simple paper strip that we developed for detecting diseases.
Are CRISPR as a gene-editing tool and CRISPR as a diagnostics tool on different timelines, as far as when the general public might encounter them in their real lives?
FZ: I think they are all moving forward quite quickly. CRISPR as a gene-editing tool is already being deployed in human health and agriculture. We've already seen the approval for the development of growing genome-edited mushrooms, soybeans, and other crop species. So I think people will encounter those in their daily lives in that manner.
Then, of course, for disease treatment, that's progressing rapidly as well. For patients who are affected by sickle cell disease, and also by a degenerative eye disease, clinical trials are already starting in those two areas. Diagnostic tests are also developing quickly, and I think in the coming couple of years, we'll begin to see some of these reaching into the public realm.
"There are probably 7,000 genetic diseases identified today, and most of them don't have any way of being treated."
As far its limits, will it be hard to use CRISPR as a diagnostic tool in situations where we don't necessarily understand the biological underpinnings of a disease?
FZ: CRISPR-Cas13, as a diagnostic tool, at least in the current way that it's implemented, is a detection tool—it's not a discovery tool. So if we don't know what we're looking for, then it's going to be hard to develop Cas13 to detect it. But even in the case of a new infectious disease, if DNA sequencing or RNA sequencing information is available for that new virus, then we can very rapidly program a Cas13-based system to detect it, based on that sequence.
What's something you think the public misunderstands about CRISPR, either in general, or specifically as a diagnostic tool, that you wish were better understood?
FZ: That's a good question. CRISPR-Cas9 and CRISPR-Cas12 as gene editing tools, and also CRISPR-Cas13 as a diagnostic tool, are able to do some things, but there are still a lot of capabilities that need to be further developed. So I think the potential for the technology will unfold over the next decade or so, but it will take some time for the full impact of the technology to really get realized in real life.
What do you think that full impact is?
FZ: There are probably 7,000 genetic diseases identified today, and most of them don't have any way of being treated. It will take some time for CRISPR-Cas9 and Cas12 to be really developed for addressing a larger number of those diseases. And then for CRISPR-based diagnostics, I think you'll see the technology being applied in a couple of initial cases, and it will take some time to develop that more broadly for many other applications.