23andme stages a comeback
The Dna-testing company’s journey back to health is a lesson in learning to adapt.
Anne Wojcicki, CEO of the groundbreaking Dna-testing company, got crossways with the Fda—and had to rethink everything. It was a blessing in disguise.
WEARING A BLUE FLEECE JACKET, RUNNING SHORTS, AND SNEAKERS, HER HAIR IN A PONYTAIL, ANNE WOJCICKI IS RECOUNTING HER EVENING OUT THE NIGHT BEFORE. SHE AND A COUPLE OF FRIENDS HAD GONE TO SEE U2 AT LEVI’S STADIUM IN SANTA CLARA (SHE’S FRIENDLY WITH THE BAND), AND WOJCICKI HAD FORGOTTEN NOT ONLY HER FRIENDS’ TICKETS BUT HER OWN ACCESS BADGE, TOO. SO, SHE SAYS, “WE JUST BUSTED OUR WAY IN.”
Wojcicki, sitting in a glass conference room at the Mountain View, California, office of 23andme, the company she cofounded in 2006, seems not to appreciate the irony of her story. Just four years earlier she paid a heavy price for trying to bulldoze her way through obstacles: In November 2013, the FDA gave 23andme a forceful slap-down, barring the company from selling its revolutionary DNA spit test, which had stirred passionate interest and generated headlines for promising to democratize genetic testing and yield data-driven cures for disease. (Fast Company had even run a cover story calling her “The Most Daring CEO in America.”) The FDA declared her company’s service an “unapproved medical device,” a crippling blow.
While 23andme could still sell its ancestry product, Wojcicki and her investors—including Google cofounder Sergey Brin (Wojcicki’s then-husband, who has personally invested in the company), Google itself (where Anne’s sister Susan was employee No. 18 and is now CEO of Youtube), and Russian billionaire Yuri Milner— hadn’t stuck their necks out and put up $118 million to build a genealogy service. 23andme’s unique value proposition, and its founders’ mission, was to empower individuals with actionable genetic health information. Suddenly all that potential was in doubt.
How 23andme turned that uncertainty around—and, in fact, expanded its brand—is a tale of perseverance, adaptability, and what in days gone by might have been called gumption. Less than a year after the FDA’S rebuke, for instance, Wojcicki signed a lease on a new, larger office space, even though it wasn’t clear at that time if the company would survive. A glass hive with a rooftop garden and free lunch and snacks for the nearly 300 employees, the space sports reminders of what the company has already overcome. A green couch sits in a common seating area, one of a crop of hand-me-downs from Wojcicki’s first apartment in New York. “We’ve been really frugal,” she says. “We used to pride ourselves on these $19 Ikea chairs that would break if you weighed over 220 pounds. We would
go visit competitors in the early days and say, ‘Oh, they’ve got fancy chairs—they’re burning cash.’ ”
Today, 23andme is back on track and, as Wojcicki puts it, “I still want more.” In April, the company received FDA consent to restore some of its health reports, notably those that assess a customer’s genetic risk of developing Parkinson’s and late-onset Alzheimer’s. Beyond this “major accomplishment,” as Wojcicki describes it, the company has also forged large-scale partnerships with businesses such as Genentech, Procter & Gamble, and Target; launched an in-house drug-development program; and raised $155 million in a Series E round, elevating its valuation to an estimated $1.1 billion.
Still, the current version of 23andme’s product offers only a fraction of the 254 reports it featured in 2013. And while the company now boasts what Wojcicki calls “a great, very supportive relationship” with the FDA, getting the next set of approvals—for reports on cancer risk and drug response—is hardly assured. Meanwhile, competitors such as Ancestry and Helix have suggested they may start offering their own health reports (Ancestry has filed for an IPO), and uncertainty over genetic privacy issues continues to loom over the sector.
Yet considering all the drama and setbacks, Wojcicki might be forgiven a bit of selective enthusiasm. Building a cult brand within the staid, highly regulated health industry is unusual; sustaining users’ and investors’ interest through years of controversy ratchets up the challenge further. How Wojcicki has managed 23andme’s comeback, and what she has planned next, is a playbook for disruption. As she recalls, when she first heard about the FDA clampdown, “I was like, ‘Hey, what can we do in six weeks [to fix this]?’ I learned pretty quickly, like, nothing. I’ve learned that things take longer than you expect.”
Wojcicki received a BS in biology from Yale
before going to work as a biotech analyst with a New York hedge fund in the late ’90s, experiences that helped her appreciate not only the enormous public-health opportunities that would be created by the Human Genome Project, but also the inefficiencies of a health care system where it seemed everybody but the consumer got to call the shots. This is what motivated her to start 23andme. When the company began offering its Personal Genome Service (PGS), in 2007, there was no precedent. Genetic screening for a panel of inheritable conditions was routinely offered by doctors to would-be parents, to assess their likelihood of passing on harmful mutations to their offspring. And Myriad Genetics had a doctoradministered test with a list price of more than $3,000 that detected genetic mutations associated with increased risk of breast and ovarian cancer, leading some women (famously, actress Angelina Jolie) to seek mastectomies and/or ovary removal. These tests had to be prescribed by doctors and were often interpreted by genetic counselors.
23andme was doing something quite different, leveraging technical advances in genomic
sequencing, and drastically falling prices, to offer extensive DNA analysis directly to consumers. Analysts and reporters covering the company frequently described 23andme’s business model of bypassing the medical establishment as hubris, a case of brash Silicon Valley confronting sluggish convention. Wojcicki pushes back at any suggestion that the company was inclined to sidestep regulators. When 23andme launched its first health report to consumers, she notes, it had CLIA (Clinical Laboratory Improvement Amendments) approval, the usual standard for diagnostic tests that are handled through a single-lab facility. Wojcicki says she didn’t consider 23andme’s test to fall under the FDA’S purview at first, in part because they weren’t diagnosing diseases, just reporting risks. “An APOE mutation isn’t diagnosing you with Alzheimer’s,” she says. “It’s a risk factor.” But as doctors, geneticists, and state regulators started to question the reliability and appropriateness of 23andme’s tests, the company saw where things were headed and in 2008 started meeting regularly with the FDA about a possible path to approval by the agency. “We had some humility,” says Wojcicki. “We did show up.”
There seemed to be little sense of urgency on the FDA’S part, until June 2010—shortly after 23andme started selling its kit through Amazon—when the FDA sent warning letters to the company, along with direct-to-consumer genomics competitors Navigenics, decode, Knome, and Illumina. (Another company, Pathway, had received a letter in May, four days before it planned to start selling its test in Walgreens.) “You should take prompt action to respond to this letter,” wrote Alberto Gutierrez, the director of the FDA’S office of in vitro diagnostic device evaluation and safety. According to Gutierrez, 23andme’s Personal Genome Service didn’t meet the criteria for a single-lab test, and because consumers might make important medical decisions based on the information it contained, it was a diagnostic “device” under the Federal Food, Drug, and Cosmetic Act.
The company found itself in new territory. While it was allowed to keep selling the test, it faced fresh scrutiny. For the next two and a half years, representatives from 23andme and the FDA had “14 face-to-face and teleconference meetings, hundreds of email exchanges, and dozens of written communications,” according to the FDA. 23andme submitted its first application for FDA clearance in July 2012—unsuccessfully—and followed it with another submission at the beginning of September. Into 2013, the record shows extensive back and forth as 23andme attempted to respond to FDA feedback and requests for additional evidence.
Then, in May 2013, 23andme abruptly went quiet, which didn’t sit well with the FDA. A company spokesperson explains that “the delays on our part really were related to miscommunication and lack of understanding the process.” This quiet stretch with regulators coincided with a noisier public posture, punctuated by the launch of new 23andme TV ads. This expanded branding effort seemed to trip some sort of invisible wire at the FDA. After hearing nothing official from 23andme for six months, the FDA sent a heated shut-down letter scolding 23andme for failing to submit evidence that its reports could be trusted. “We have been diligently working to help you comply with regulatory requirements,” this new letter from Gutierrez read. “FDA has not received any communication from 23andme since May. Instead, we have become aware that you have initiated new marketing campaigns, including television commercials that, together with an increasing list of indications, show that you plan to expand the PGS’S uses and consumer base without obtaining marketing authorization from FDA.”
Wojcicki turns reflective when recalling the letter. “Without a doubt, we didn’t have the right team in place or the experience in how to get regulated,” says Wojcicki. “More and more, I’ve learned how far off we were.” 23andme was forced to stop selling its health reports immediately.
“Change what you can; manage what you
can’t,” reads the tagline for a 23andme ad campaign from around this time, advice that Wojcicki and Andy Page, a former president of Gilt Groupe who came on as 23andme president just months before the 2013 letter arrived, took to heart. “It was a huge wind out of our sails,” says a former employee who was there during the crisis.
In the weeks after receiving the letter, 23andme issued press releases declaring its intention to continue marketing parts of its service that the FDA didn’t explicitly forbid—information about ancestry and “raw genetic data without interpretation”—and pledging to keep working with the FDA to restore the full range of health tests.
To that end, Wojcicki hired a regulatory affairs expert, Kathy Hibbs, an attorney who had been working at Genomic Health, a cancer genomics company. Hibbs had been introduced to Wojcicki by a mutual friend and only agreed to meet as a favor. “It didn’t occur to me that they’d recruit me,” she says, “and it didn’t occur to me that I’d be interested.” But she felt that the company’s interaction with the FDA “had the potential to establish the standards for similar products in the future. That’s a powerful thing, and it’s an actual advantage.” The FDA’S warning letter “was pretty provocative,” she admits, but “I don’t think they were trying to kill the company.”
But it would take time—and luck—as well as better science to repair 23andme’s trust with the FDA. Hibbs has tackled the process in chunks, identifying those areas with the highest likelihood of approval first, and then moving to the more touchy ones, including the tests for BRCA gene mutations and drug response (such as sensitivity to the blood thinner warfarin).
In the meantime, the company had no choice but to shift its marketing to focus on its ancestry product. The FDA’S warning letter gave the
“WITHOUTADOUBT,WEDIDN’T HAVETHERIGHTTEAMINPLACEOR THEEXPERIENCEINHOWTOGET REGULATED,”SAYSWOJCICKI. “MOREANDMORE,I‘VELEARNED HOWFAROFFWEWERE.”
company impetus to explore the more whimsical side of its business, such as finding relatives you never knew about. This relatively low-cost consumer DNA test turned out to be a shockingly effective way to create a genetic database the likes of which no other public or private entity has ever seen. Selling the $99 ancestry reports alone, the company added more than half a million new customers between December 2013 and June 2015, when it announced its 1 millionth genotyped customer. Today, the company has collected DNA from more than 2 million people.
Thanks to this unexpected appeal, 23andme now controls the world’s largest and most diverse collection of human genotypes (600,000-plus unique genetic markers per individual) paired with self-reported phenotypic information (hundreds of data points per average individual on traits ranging from lactose intolerance to male pattern baldness to a family history of cancer). More than 80% of 23andme customers—even those who just purchase the ancestry product— opt in to its research program, allowing the company to share its anonymized data with academic and industry partners. The information, in turn, has allowed scientists to run gene-association studies—trawling through huge amounts of data to find mutations that are strongly linked to particular traits or health conditions. Plus, because 23andme does an exceptional job at keeping customers engaged well beyond getting their test results—the company pushes frequent updates on genetic research, a combination of serious medical research as well as cocktailparty-friendly studies on the genetics of cilantro aversion or the tendency to be a “morning person”—it can prompt them to fill out an endless string of surveys. “The core element of our company is this partnership with the customer,” says Wojcicki.
If a drug company wants to study Parkinson’s, for example, it can quickly identify some 15,000 people with the condition through 23andme—or nearly 300,000 with depression or high blood pressure—and ask them to complete additional surveys, recruit them to participate in clinical trials or additional kinds of tests, or perhaps, one day, contact them about a new treatment.
Researchers working with 23andme’s data have already started to publish breakthrough findings. A project with the Michael J. Fox Foundation has identified novel biomarkers for Parkinson’s disease. One with Pfizer detected 15 new regions of the human genome linked to a higher risk of serious depression. Genentech paid a reported $10 million up front (with a potential $50 million more for hitting key milestones) to access genomic data from 23andme’s Parkinson’s community. These sorts of successes have attracted other partners. Procter & Gamble is using genetic insights gleaned from 23andme to help inform its marketing of products ranging from over-the-counter sleep aids to skin creams, steering people whose genetic profile suggests they’d be unresponsive to retinol, for example, to alternative antiaging products. “We can go in the direction of personalized skin care,” says Frauke Neuser, a principal scientist at P&G’S Olay brand.
23andme’s head of business development, Emily Drabant Conley, says that “in the early days it was all uphill. It was hard. Now there’s this realization that human genetic data is valuable. It feels like we’re in a different time period.”
If selling test kits to consumers represents 23andme’s central business, and forging
data-sharing partnerships with companies is its primary line extension, then the company’s in-house therapeutics-development division could be considered its moonshot. Creating a drug may be the utmost embodiment of Wojcicki’s mission to transmute data into real improvements in public health. “It’s our responsibility to help people figure out how to prevent developing a disease, or to find a cure for it,” she says. “In science, the best way to try to drive change is to lead by example. If I really believe—and I do—that starting with human genetic data will make the drug discovery process faster and more effective, then the best way to prove that is to do it.”
Richard Scheller always saw the potential for drug development in 23andme’s database. The award-winning biochemist, who had spent 14 years leading research strategy, drug discovery, business development, and early-stage drug-development activities at Genentech, had played a key role in Genentech’s investment in 23andme’s 2007 Series A. Personally, he says, “up until I started, the database wasn’t large enough to be really exciting for me. But it got there, and now with the FDA approvals and more sophisticated marketing, it’s growing incredibly fast.”
The walls are blindingly white at 23andme’s research lab in South San Francisco, and the centrifuges, mass spectrometers, and laboratory refrigerators still gleam like new. Scheller, dressed in a checkered flannel shirt and drooping gray jeans, leads a tour of the facility, sipping Diet Coke from a cup. For the past two years he has been overseeing a team that has grown to more than 30 scientists, hoping to use 23andme’s database to alter the economics of bringing new drugs successfully to market. On average, a new drug costs more than a billion dollars to develop. This is because for every one that works, 10 don’t. Drugs with a basis in human genetics, though, have a success rate one-and-a-half-fold to twofold higher, says Scheller. “If we can be even twice as good as the industry average, we would be in pretty good shape.”
Scheller is currently focusing on several conditions, including cardiovascular and liver disease, cancer, skin diseases, and immunological disorders. “We want to bias ourselves to something that’s going to work,” he says. That means leaving the development of neurological drugs—scheller’s specialty—to partners. “The most important thing for a young company is to be successful, and diseases of the brain are hard. That’s not a place to start.”
THEFDA’SWARNINGLETTER TO23ANDME“WASPRETTY PROVOCATIVE,”SAYSREGULATORY AFFAIRSHEADHIBBS,BUT“I DON’TTHINKTHEYWERETRYING TOKILLTHECOMPANY.”
If the company happens upon a drug compound that works, 23andme would likely sell or license the marketing rights to a partner. “What pharma companies do really well is running clinical trials and marketing,” says Wojcicki. “Those are really complicated and expensive things to do. Where we can potentially be really good is on the discovery side.” Her goal is to get a new drug to the point where it’s proven effective in humans.
Even in the rosiest scenario, it will be years
before Scheller’s team sees any big commercial reward for its efforts. In the meantime, 23andme expects to continue working with the FDA to expand its menu of direct-to-consumer health tests. The company is on much better terms now with the agency regulating it. “I think they’re really proud of the work they did,” says Hibbs of the FDA. “It gives them the ability to embrace innovation.” FDA spokesperson Tara Goodin says via email that “since 2013, 23andme has diligently collaborated with the FDA to create a model for offering a genetic health risk test directly to consumers that mitigates risks associated with how a consumer interprets and uses the test results.” Of course, the fact that everyone’s playing nicely now doesn’t mean that future approvals the company is seeking— for cancer risk and drug response, potentially the most actionable information for consumers—will be any easier to win.
One of the reports that Wojcicki badly wants back is for mutations in the BRCA1 and BRCA2 genes, which are highly predictive for breast and ovarian cancer. The tests are not only potentially lifesaving but lucrative:
The No. 1 provider of BRCA testing,
Myriad Genetics, earns some $632 million annually from its services. But Wojcicki will face headwinds—in part because of her own decisions. 23andme has chosen to forgo using so-called next-generation sequencing (NGS) technology for now, since it is expensive and results can be hard to interpret. Unlike Myriad’s NGS test, 23andme’s is designed to screen only for the three most common BRCA variants found among Ashkenazi Jews, among whom the incidence of BRCA mutations is about 1 in 40, versus about 1 in 400 in the general population.
“When 23andme customers got a positive BRCA result, it was a positive,” says Jill Hagen- kord, who led a next-generation sequencing effort at 23andme from 2014 to 2016 and is now chief medical officer at Color Genomics—whose $249 doctor-ordered test uses NGS to screen for several hereditary cancer risk factors. But when it was negative? “There’s a concern about people who think they got a BRCA test but really didn’t,” Hagenkord says.
Opting not to use NGS will also limit the types of drug discovery that the company will be able to do, says Scott Hebbring, a research scientist at the Center for Human Genetics at Marshfield Clinic Research Institute in Wisconsin. “Having just the common variants from genotyping,” he says, “can make it harder to find variations with large effects that could be rare but potentially the most interesting for understanding disease.”
23andme also faces privacy fears, which have only increased since Wojcicki launched her company a decade ago. “We’re living in a time when we can no longer be completely sanguine about the safety of any of the data we share with the world,” says Misha Angrist, an associate professor in the Social Science Research Institute at Duke University. “We’ve seen too many reports of Fortune 500 companies, to say nothing of hospitals, getting hacked.”
The Genetic Information Nondiscrimination Act of 2008 (GINA), which protects individuals from genetic discrimination in health insurance and employment, was “a good start” on regulating the legitimate uses of genetic data, says Hebbring. “But it’s by no means comprehensive.” It doesn’t apply to military personnel or employees of companies with fewer than 15 workers, and GINA will not protect consumers with known genetic conditions from being discriminated against for disability insurance, long-term-care insurance, and life insurance. (In those industries, carriers are expressing new concerns about consumers who game the system by buying long-term-care coverage only when they discover—but don’t disclose—an increased genetic risk for Alzheimer’s.) If the Affordable Care Act is repealed, its explicit prohibition against discrimination for preexisting conditions, including ones revealed through genetic testing, could be weakened in significant ways. The reality, or even just the perception, that your DNA could be used against you could have a chilling effect on people’s willingness to get genetic testing for any purpose.
All of this makes it crucial for 23andme to position itself as an approachable and accessible brand. “I think there’s a whole world of people who don’t get any health care because it’s too cumbersome and expensive,” Wojcicki says. That’s why she “loves” Walmart’s $59 nurse practitioner visit and CVS’S Minuteclinic. The desire to connect with consumers in the real world, outside the health care bubble, is what pushed her to get mainstream retailers—including CVS, Sam’s Club, and Target—to start carrying 23andme’s health-and-ancestry test over the past year and persuade Amazon to start selling it again. It also helped motivate her to capture the popular imagination through partnerships that wouldn’t have been imaginable when the company started out, like the recent marketing campaign with Universal Pictures around the release of Despicable Me 3, featuring an animated short in which the character Gru discovers a longlost brother using 23andme’s DNA Relatives tool.
“We’re definitely a mission-driven company,” Wojcicki says. “There’s a lot of skepticism about that, but we really want to drive positive change in health care.” Many would-be disrupters in Silicon Valley talk about changing the world. But doing so takes both skill and stamina. “In some ways, what happened with the FDA has forced me to learn the lesson of patience,” says Wojcicki. “Everything is in steps.”
“INSCIENCE,THEBESTWAY TODRIVECHANGEISTOLEADBY EXAMPLE,”WOJCICKISAYS.
Emily Drabant Conley, the company’s head of business development, says that she has seen a huge uptick in partnership interest.
Kathy Hibbs, who leads regulatory affairs for 23andme, has worked strategically to gain FDA approval.