Science Math Biology: New Engines of Growth

23andMe: A Letter to our Readers

If you could take a simple test to find out what might ail you, would you take it?

Anne Wojcicki is betting that you will. The Yale-educated biologist is co-founder and CEO of 23andMe, a genetic analysis company that can tell you whether you’re at risk of passing on some 50 inherited conditions, contracting 120 serious diseases or having an adverse reaction to more than 20 drugs.

Sign up online for just $99, and 23andMe will send you a test kit. Spit in the test tube, mail it back, and in six to eight weeks, you’ll receive a report on your genetic makeup. Your personal genotype will summarize the probability you might someday contract any of those ailments, as well as outline your ties to certain genetic families, such as Neanderthals, who went extinct about 34,000 years ago.

Along with your genotype, you’ll get a ticket on the Brin family’s voyage into the future of human biology—because 23andMe is no ordinary biotech start-up. Co-founder Wojcicki is married to Google co-founder Sergey Brin, and Brin and Google Ventures hold sizable stakes in 23andMe. Wojcicki’s company says its mission is Googlesque: “to be the world’s trusted source of personal genetic information.” But earning that trust will be a tall order, and dominating the industry will be challenging, even for a venture backed by Google’s financial resources and computer know-how.

A high-profile company in the direct-to-consumer personal genomics industry, 23andMe is an amalgam of science and personal health services that wraps basic DNA information in a Silicon Valley model of online community sharing and crowdsourced research. It presents personal genomics as a way to predict health problems while also sharing neat stuff about your ancestors and, by the way, contributing your genetic information to the greater cause of disease research. While that sounds simple, the business of providing genetic information to retail customers raises nettlesome privacy, regulatory and policy issues. The issues—some mind-bendingly technical, some intuitively troubling—are being hashed out in medical schools, law schools, Food and Drug Administration hearings and the genomics industry itself.

Wojcicki is betting that, eventually, millions of customers will take 23andMe’s test, creating a database that the company’s legion of scientists can analyze with statistical software to find gene irregularities that point to potentially serious conditions. It’s a burgeoning field called bioinformatics; 23andMe itself has identified previously unknown links, called “associations,” between damaged genes and serious conditions.

Sergey Brin took the test and found he’s predisposed to Parkinson’s disease. But don’t mistake 23andMe for a billionaire’s vanity play. The driving force of progress in genomics is Big Data—the computational biology and bioinformatics tools that researchers use to create and analyze databases of genetic information that are growing exponentially. While privacy concerns about gene databases remain in flux, researchers have used large-scale genetic studies to revolutionize understanding of some of the deadliest cancers, and personal genomics is being integrated into medical practice by blue-chip providers, including the Scripps Institute and the Cleveland Clinic. And with China moving into the U.S. genomics market, no company is better suited for a Big Data battle than Google.

The direct-to-consumer (DTC) genomics industry, which arose as genetic research technology made consumer testing feasible, has been volatile since its inception about 10 years ago. Regulators have questioned genomic companies’ claims that consumers can use the test results to make health and medical decisions. Privacy and ethical concerns abound. The past year saw major upheaval, as pharma companies paid top dollar to buy DTC players, hoping to use the technology to leapfrog competitors in disease research and drug discovery. Reflecting the high stakes, investors have challenged one recent deal, claiming insiders enriched themselves at the expense of public shareholders. Another DTC competitor recently opted out of the consumer market and now requires patients to buy its tests through a doctor.

What’s more, the cost of full DNA sequencing is falling fast, and some hospitals use Chinese systems to provide full genome scans. Wide availability of full genomes could render obsolete the basic genotyping—which only shows general tendencies—that 23andMe provides. Driving the cost reduction is BGI Shenzen, a sort of Foxconn of bioscience, with large facilities and low wages. BGI owns more than 150 gene sequencers and could be producing 10 to 20 percent of all genetic data globally. Last year BGI bought Complete Genomics, a U.S. company that was struggling financially despite having invented a complex sequencing method. The declining cost of genomics technology has attracted a new DTC competitor, which avoids regulatory scrutiny by offering “information only” reports to genomic do-it-yourselfers.

It doesn’t sound like an industry anyone would want to dominate, especially entrepreneurs as relentlessly successful as Google’s founding family. But as the dust settles after Round 1 in the DTC genomics arena, Anne Wojcicki and Sergey Brin are the last pair standing. And this summer, Wojcicki doubled down on her DTC bet, launching a national television advertising campaign to raise awareness of personal genetic testing.

Family Ties

Perhaps it’s fitting that the Brin family is leading the consumer genomics expedition, because in no other sector of American life is Big Data having such a widespread, yet often unnoticed, impact. Computational biologists are the Jedi Knights of genomics: scientists and medical researchers who generate petabytes of experimental data at an astonishing rate with low-cost, high-throughput instruments; concoct algorithms that tease insights out of titanic data sets; and invent computing infrastructures capable of routing data and calculations to geographically dispersed supercomputers and research teams.

23andMe, seeking to build a database of genetic samples by providing basic tests directly to consumers, sits at the center of a paradigm that it’s helping to create, combining data creation and analysis with a Web-based approach to disease research. Five years since the birth of the consumer genetics industry, 23andMe is the sole pure-play company in the sector. The company entered 2013 with new capital from a $50 million financing in late 2012, which funded a major Web site revamp and an aggressive communications plan to close the nation’s newfound “genetic literacy gap.” And while competitors have receded in the face of regulatory scrutiny, 23andMe has not only persevered, it’s the only DTC genetics company to request FDA approval for some of its health-related tests. One more thing: It was awarded a patent for a method of identifying exposure to Parkinson’s disease based on variations at two genetic sites that previously hadn’t been known to have a role in the condition. The patent research also validated 23andMe’s research approach by confirming known associations, and demonstrated how Web contact could accelerate research by allowing people far from medical centers to participate. “23andMe assembled 3,426 cases and 29,624 controls to track down two new genes that contribute to Parkinson’s disease, much faster than would have been possible in academic medicine,” says Ricki Lewis, Ph.D., a geneticist and a genetic ethics professor who writes the DNA Science Blog for PLOS, an open-access publisher of scientific and medical research. “That’s hardly frivolous. It’s crowdsourcing science.” Most importantly, she says, “it works.”

The software and IT techniques for creating and managing enormous life science databases are the basis for a booming new field, bioinformatics. In the past few years, the genomic Jedi have focused on developing industry standards to make it easier for research scientists to study genetic information. One such initiative is the Genome Analysis Tool Kit (GATK), developed by a team from the Broad Institute of M.I.T. and Harvard and Massachusetts General Hospital. The GATK uses a programming technique called MapReduce, which separates analytic calculations from the computer infrastructure management tasks that bring data back and forth from storage centers. Key researchers quickly adopted the tool kit, and the GATK now runs critical analysis programs for the 1000 Genomes Project and the Cancer Genome Atlas.

Perhaps not surprisingly, MapReduce was invented at Google, which holds a patent for the technique. Google executive chairman Eric Schmidt says the opportunity to transform and accelerate cancer research fits Google’s mission of making the world a better place. He oversees Google’s effort to provide computational infrastructure to the Broad Institute team, whom he calls the smartest people he’s ever known, working with databases so large that even he was surprised at the magnitude of the computing challenges.

The Big Data payoff has been powerful. Last April, researchers involved with one of the largest studies of DNA mutations in common cancers announced findings that fundamentally changed doctors’ understanding of the disease. Researchers discovered that the genetic structure of the most dangerous cancer of the uterine lining closely resembles that of the worst ovarian and breast cancers—indicating that cancer will increasingly be seen as a disease defined primarily by its genetic profile rather than by the organ or tissue where it originates. The findings resulted from genetic analysis of hundreds of tumors as part of a National Institutes of Health project that had more than 100 researchers study DNA anomalies in common cancers. Doctors are revising testing and treatment protocols based on the study. But there’s a catch. The ability to extract insight from genetic databases also makes it possible to pick out an individual from the metaphorical genetic haystack—in effect removing the privacy and anonymity that have been central to medical research, and shaking the trust that research institutions, patients and study participants rely on. It’s the ultimate disruptive technology.

The Big Data Paradox

The decision by most companies to stop marketing directly to consumers, or provide consumers only raw data, highlights the privacy and ethical issues that the DTC genomic movement has surfaced. The use of genetic samples in research has generated considerable controversy, according to Megan Allyse, a professor at the Stanford Center for Biomedical Ethics. “Although the format may have changed with 23andMe’s foray into genetic research, the issues have not,” she says. The key question is whether 23andMe is adequately disclosing what consumers are signing up for when they use a 23andMe test. The company is open about the existence of its research unit, she says, but questions remain about whether consumers have a proper opportunity to give informed consent to have their genetic information used in 23andMe research aimed at developing commercial products or services. “Companies like 23andMe are engaging in a trade,” says Allyse. “Information about your genetic makeup in return for the use of your genetic material for research, publication and patenting.” This is similar to the trade that academic researchers sometimes offer, “except that in the case of 23andMe the research participant pays for the privilege.” That’s a marked departure from the voluntary research participation that serves as the “cornerstone” of modern bioethics, says Allyse, and “commercial genetics companies will need to devote more attention to ensuring that their customers are fully aware if the company intends to retain and to conduct research on the data from customer samples and claim intellectual property of the results.”

Genomics law recognizes that consumers have a right to know what researchers and institutions intend to do with their data. But Big Data raises new issues—the computer systems that are helping researchers find new treatments are also making it virtually impossible to guarantee privacy to consumers or patients who contribute their genetic information.

Genomics databases “pose several substantial legal and ethical problems,” according to Henry T. Greely, a Stanford Law School professor and director of the Center for Law and the Biosciences. “Neither the usual methods for protecting subject confidentiality, nor even anonymity, are likely to protect subjects’ identities in richly detailed databases.”

“In these settings,” says Greely, “anonymity is itself ethically suspect.” In fact, some genomic Jedi warn that patient privacy is a thing of the past. “It is no longer clear that we can promise anonymity to participants,” according to a research team at the University of Washington in Seattle. The team was drawn from three departments at the school—the Institute of Translational Health Sciences, the Department of Bioethics and Humanities and the Department of Biomedical Informatics and Medical Education. The researchers warned that despite the “best efforts of researchers to meet the demands of anonymization, a number of features of biorepository research have combined to make it increasingly hard to achieve in practice.” These include the primary benefits of Big Data science: the creation of detailed genotypic databases, linkage to detailed clinical data, and the use of bioinformatics tools for analysis.

Results of broad genome studies were available until recently on public Web sites because they were thought to reveal little about individuals participating in a study. But statisticians can infer from genetic data whether a specific individual or a close relative participated in a broad genomic study. In a paper in Nature Genetics, 13 researchers from leading U.S. genomics, cancer and biotech institutes published their approach and said, “This method could be used to determine if specific individuals participated in a clinical study.” The co-authors predict researchers’ ability to pick out individuals in a genetic crowd is only going to increase: “Our results should be considered as a lower bound for the power to detect membership and phenotype in an aggregate genotype dataset, as more efficient methods may yet exist.” Going forward, the team emphasized the need for researchers to update policies and practices guiding genomic data sharing in order to “merit the ongoing trust of the research subjects who consent to participate in scientific studies.” In what might be an indication of the future scope of consent disclosure, 23andMe’s privacy policy reflects researchers’ decreasing ability to hide the identity of individual subjects. In defining “Aggregated Genetic and Self-Reported Information,” 23andMe notes that its effort to ensure anonymity means only that an individual’s information will be “combined with data from a number of other users sufficient to minimize the possibility of exposing individual-level information while still providing scientific evidence.”

Regulatory Lines

GENOMIC scientists look for mutations in genes that might cause diseases or inherited conditions. In contrast to the full sequencing used in experiments like the NIH cancer study, which focus in detail on specific genes, consumer genomics services typically rely only on the basic technique of genotyping. Genotyping tells customers how their genetic profile differs from a standard model based on a limited number of comparisons between their DNA sample and well-known genetic markers on a baseline human genome. Full sequencing assesses the 3 billion letters in a person’s DNA, while genotyping only looks at several hundred thousand specific sites. As a result, genotypes “can only be used to infer slight increases in disease incidence, often described as ‘predisposition,’ ‘risk’ or ‘susceptibility’ factors,” according to Population Diagnostics, a New York company that makes a technology platform for discovery and validation of causative genetic biomarkers. On its Web page explaining genomic reports, Illumina Inc., which manufactures gene-sequencing instruments, notes that “medical and research doctors’ ability to analyze and interpret genome information is still quite limited.”

In light of concerns that the limited predictive value of genotyping tests could be construed as medical advice, the U.S. Food and Drug Administration and some state regulators have been reining in what DTC genetics companies claim their tests can do for consumers, sometimes barring sales to consumers. In 2010, the FDA sent 23andMe, as well as other DTC genomics services, a letter asserting jurisdiction over the 23andMe Personal Genome Service as a device intended for the diagnosis of disease: “23andMe has never submitted information on the analytical or clinical validity of its tests to FDA for clearance or approval. However, your Web site states that the 23andMe Personal Genome Service is intended to tell patients in advance how they will respond to certain medications...” The clincher: “Consumers may make medical decisions in reliance on this information.”

The Genomics Law Report notes that “direct-to-consumer” means a test or service “that an individual can order, receive, review and share with others without being required, at any stage in that process, to engage a healthcare professional as an intermediary.” The 23andMe Web site conveys a sort of personal genetic rights manifesto, endorsing individual control of genetic data and extolling the virtues of gaining insight into one’s DNA makeup in order to make better health care decisions. That said, 23andMe does make clear the limitations of the reports it provides, and provides customers with links to find nearby genetic counselors, a fast-growing health care profession.

While 23andMe’s pending FDA approval may resolve these issues, states are enforcing laws on the books. In Maryland, for example, state law prohibits any entity from offering lab testing except to medical professionals. The state Department of Health’s 2012 Annual Report and Staffing Analysis says, “Direct to consumer testing is dangerous because it occurs without physical examination or medical assistance. It can also lead to inaccurate diagnoses and a higher cost for the consumer for irrelevant testing.” Maryland ratcheted up enforcement in 2012, directing more than 50 Web sites offering DTC testing to add a disclaimer prohibiting ordering by Maryland residents. 23andMe restricts sales to Maryland residents as well as consumers in New York, which requires such tests to be performed by a company with a New York State lab license. The company won’t say whether it’s seeking legislative or regulatory changes in those states.

Climbing The Helix

DESPITE the regulatory focus at the consumer level, there’s no doubt that unlocking secrets encoded in the double helix of DNA can lead to medical breakthroughs, and genetics experts are in demand. Merger and acquisition activity in the DTC genomics industry reached warp speed beginning in 2012. Navigenics, founded in 2006, by 2010 had opted out of the consumer market, choosing to sell through doctors. In 2012, Life Technologies acquired Navigenics, only to be acquired itself by Thermo Fisher Scientific for $13.6 billion in April 2013. A week after the deal was announced, a shareholder challenged the sale, claiming the price was too low.

One-time DTC player Pathway Genomics now requires consumers to order its tests through a doctor; its Web site tells users “we will reveal your genetic factors relevant to many common conditions so that you and your doctor can use this information to make better health choices.” deCODE Genetics, a strong competitor in computational genetics, had difficulty attracting consumers and wound up in bankruptcy in 2009. deCODE had identified new targets for Alzheimer’s disease and built a whole-genome database on 70,000 Icelanders, attractive assets for a big company seeking genomics expertise. Amgen acquired deCODE for $415 million in cash in 2012; its 2013 strategy highlighted a new “pick the winners” approach to research and development, “which takes advantage of our industry-leading position in human genetics that has resulted from our acquisition of deCODE.”

Having competitors taken off the market might look like a boon for 23andMe. But in late 2012, Gene by Gene Ltd. launched a unit called DNA DTC, using technology similar to 23andMe’s to offer full genomes without any interpretation or analysis, the first time a company had done so, according to the Genomics Law Report. While providing only raw data appears to avoid the need for regulatory approval, a new entrant signals that Round 2 in the consumer genetic testing battle might be getting under way.

Doctor, Doctor, Give Me The News

WHILE Big Data tools have highlighted—and perhaps amplified—the trust and privacy issues in medical and genetic research, 23andMe projects a groundswell of consumer demand. In a study released on DNA Day, the anniversary of the publication of the paper confirming the structure of DNA, 23andMe said 73 percent of Americans who hadn’t had a genetic test would like to do so in the future. But only 2 percent of Americans have had such tests, and most don’t understand the fundamentals of genetics or how DNA operates. 23andMe plans to fix that. “Our goal is to close the genetics literacy gap and help educate people about the basics of genetic inheritance,” said Joanna Mountain, Ph.D., senior director of research at 23andMe. The company plans to show consumers “the connections between DNA and health, and the potential benefits of genetic testing.” It’s a big opportunity—more than 70 percent of those surveyed said they’d like to find out which health conditions they’re most at risk of developing, and more than 55 percent said they’d consider making diet, exercise and other lifestyle changes based on the results. Almost as soon as DTC genetic tests hit the market, concerns arose that consumers might become depressed or panic after learning they were predisposed to serious diseases. “Genome scans give people considerable information about their DNA and risk of disease, yet questions have been raised if these tests are ready for widespread public use,” says Dr. Eric J. Topol, director of Scripps Translational Science Institute. 23andMe cautions customers that their reports could contain troubling information. But those fears appear to be unfounded. In 2008 Scripps began a 20-year study of up to 10,000 participants, the Scripps Genomic Health Initiative, to assess how people respond to genetic testing. The initial results: Genetic screening didn’t induce anxiety in study participants. In fact, many who showed a high risk for developing a disease expressed strong intent to undergo the corresponding health-screening test. “Early detection is a critical factor in preventing most diseases, yet a lot of us don’t get our health screenings as recommended,” says Topol.

Genetic testing made headlines after film star Angelina Jolie’s decision to undergo a double mastectomy after learning she carried a faulty BRCA1 gene. The BRCA1 gene repairs damaged DNA. A faulty BRCA1 gene can allow defective cells to grow uncontrollably to form a tumor. In its summary of cancer risks, 23andMe calls BRCA1 problems a “rare but serious mutation.” Jolie’s mother died of ovarian cancer at age 56, and Jolie wrote in a New York Times op-ed piece that she herself had an 87 percent risk of developing breast cancer and a 50 percent risk of developing ovarian cancer.

23andMe’s genotyping test looks for smaller variations in genes that might someday form the basis of new tests that could indicate a propensity to develop cancer among women who don’t have the major risk factors for breast cancer. For now, the company points out that its tests have limited ability to tell women whether they’re predisposed to breast cancer, and suggests women heed family history. A highlighted text box on the page explaining breast cancer advises that “if you have a strong family history of breast cancer you should consider talking to your doctor, who may suggest getting a clinical genetic test. The effects of rare but serious mutations in genes such as BRCA1 and BRCA2 far outweigh those of the [small variations] reported here.” While not specifically referring to any genomics company, Jolie wrote that many women “do not know that they might be living under the shadow of cancer. It is my hope that they, too, will be able to get gene-tested, and that if they have a high risk they, too, will know that they have strong options.”

Jolie’s case highlights the power of genomics to improve health and save lives, and will almost certainly boost interest in clinical and DTC genetic testing. Despite regulatory challenges, fierce industry competition and its reporting of such whimsical factors as the consistency of a consumer’s earwax, industry observers say 23andMe has established DTC genetics as a platform for serious science. Blogging from the DTC panel at the 2012 annual meeting of the American Society of Human Genetics, Ricki Lewis predicted a bright future for the industry. “With consumers on board, scientists seeming to have accepted DTC testing, and doctors having to keep up with their patients who come in with test results, I think DTC genetic testing is here to stay”. And as Big Data expands its impact, Lewis wrote, the practice of genomic medicine is “poised to explode.” 

Authors

  • Christopher R. O'Dea

    Contributor, Korn Ferry Institute