Tag Archives: BGI

23andme: It’s all about the data

By Steve Dickman, CBT Advisors

There was a flood of news in late November about the stinging letter that Mountain View, California-based 23andme received from the U.S. Food and Drug Administration (FDA). Because it ignored FDA instead of continuing a years-long dialogue, 23andme was forced, over howls of protest, to stop selling its direct-to-consumer genetic testing panel.

Almost lost in the controversy was the company’s now derailed core strategy: to collect a million customers’ worth of genetic data, then mine the data for valuable insights that can give the company an insurmountable competitive advantage.

You could try to convince me that the strategy is moot now that 23andme has run into a brick wall at FDA. That aggregating data as a way both to derive medical benefit and to make money is now as dead as 23andme’s consumer genetics business.

23andme blimp


But I would push back. I think this regulatory battle, which 23andme has apparently lost in a rout, is just the first skirmish in what promises to be a game played over a much longer term and at much higher stakes. More about that below.

A year before the FDA’s letter, 23andme cut the price of its service to $99 and announced that it would attempt to reach one million customers by the end of 2013 after attracting only a reported 180,000 in its first six years on the market.

In my view, this change in business model explains much. The test used to cost $699, then $299 and, despite economies of scale, it is hard to imagine that 23andme was making much money selling it for $99.

What happened is this: when adoption was running way behind what it would likely have taken for 23andme to become a profitable testing company, there was a purposeful shift toward aggregating data. In the words of CEO Anne Wojcicki, “One million customers can be the tipping point that moves medicine into the molecular era.”

In my view, what stood behind this shift is the same widespread belief that informs much of the research being done on longer genome sequences: that the aggregation of enough “Big Data” will yield insights more valuable – and profitable – than anything that genomics has yielded until now.

This is why BGI in China, in its Million Human Genomes project, is attempting to sequence more genomes faster than has ever been done before.

It is also why Foundation Medicine has raised over $200 million in venture funding and IPO financing to be the first to market with a 200-gene test for cancer. Foundation does not simply want to be a first mover in massive sequencing of cancer genomes. As I have written before, I believe that it wants both the data that patients will provide as well as the high-margin revenue that will come from providing sequences of the relevant genetic segments at $5,000 or so per patient. It remains to be seen if it will get either the data or the revenue.

Journalist Ezra Klein, nailed it in his Dec. 5 column on Bloomberg View when he wrote “… the long-term play is [the] more interesting [one]: 23andMe wants to aggregate the genetic information of millions of individuals, then mine that data to make medical connections, find disease markers and discover treatments at a faster rate than would be possible using traditional techniques.”

In Klein’s view, the company “fumbled” its chance to work in concert with FDA to jointly develop regulatory guidelines under which it – and presumably its competitors – could live. This “fumbling” by 23andme, wrote Klein, has created “an opportunity for the political system to reassess an old law and determine whether it suits the newest technologies.”

I beg to differ. I do not think 23andme was that foolish. I think that by flagrantly waving its tests in the face of FDA, even going so far as to run national TV ads for them while spending six months not returning FDA’s calls, the company sought out the chance to challenge the very idea of its test being regulated as a medical device.

Indeed, Lauren Fifield, a senior health policy expert cited by MedCity News, predicted in late November that the company has purposely taken a stand. “My gut tells me,” Fifield is quoted as saying, “that the company isn’t challenging process but is instead challenging the very regulatory definition of what it is to be a device.” Fifield, the blog says, works closely with startups, the FDA, and other federal health agencies in her role at electronic medical records company Practice Fusion. “What remains to be seen,” she continues, “is whether the company and tech industry can convince the government that safety can be increased, or at least balanced, by innovation rather than set at odds.”

Look not just at the fact that 23andme lost. Look at how the company lost. The FDA letter stated that, after “more than 14 face-to-face and teleconference meetings, hundreds of email exchanges, and dozens of written communications, you have not worked with us toward de novo classification, did not provide the additional information we requested necessary to complete review of your 510(k)s, and FDA has not received any communication from 23andMe since May.”

You might try to persuade me that 23andme acted inattentively or naively when it gave FDA the cold shoulder. That is the argument made in The New Yorker blog on Nov. 27, 2013, by 23andme co-founder Linda Avey, who left the company in 2009. The FDA decision “…surprised me,” she told the New Yorker writer David Dobbs. “But she pointed out,” wrote Dobbs, “that 23andMe’s general counsel, whom she understands was leading the negotiations with F.D.A., left the company this summer; [so] perhaps it fell through the cracks. “The whole time I was there,” Avey told Dobbs, “we were in an outreach mode with the F.D.A. I can’t imagine there was that much of a cultural shift since then. It might be they weren’t paying close attention.” She admits this sounds strange, Dobbs wrote, but thinks that it is no stranger than any other explanation.

Look at what was at stake: the very future of the company, not to mention the option for consumers to have hundreds of thousands of their genes scanned for health-related variants. 23andme was the only remaining provider among the initial crop of consumer-focused companies to continue to offer these tests.[1]

With so much on the line, I have to believe that 23andme went into this battle with its eyes open. It initially conceded defeat – though even that took a week – in a press release put up on the company web site stating, “We have received the warning letter from the Food and Drug Administration. We recognize that we have not met the FDA’s expectations regarding timeline and communication regarding our submission. Our relationship with the FDA is extremely important to us and we are committed to fully engaging with them to address their concerns.”

Wojcicki was quoted in a New York Times blog saying that the company should have responded to FDA’s requests sooner rather than ignoring them for six months. “We completely recognize we’re behind schedule; we failed to communicate proactively,” she said. “They’re a very important partner, and everyone is focused on resolving it.”

But 23andme may also be borrowing a page from its investor Google in not necessarily attempting to resolve the tension with FDA but rather by trying to trump FDA’s factual and legal arguments with evidence of the utility of the data and widespread support of consumers who willingly share the data in order to see a bigger picture. How better to go into a regulatory or legal proceeding than to be armed with medical advances that were only made possible by data collection that, one could later argue, existed in a regulatory grey zone?

Now that the initial thrust by 23andme has been parried by FDA, the company will face a much tougher road to getting its tests back on the market, if it ever does.

But I would not underestimate the power behind the company, which might include the full force of Google, despite the public separation of Wojcicki and her husband, Google co-founder Sergei Brin. After all, Brin himself took an interest in the company when it revealed his increased risk for Parkinson’s disease, which he knew ran in his family. Furthermore, Anne Wojcicki’s sister, Susan Wojcicki, is Google’s senior vice president of ads and commerce. In addition to Facebook billionaire Yuri Milner and several venture capital firms, Google would appear to remain one of 23andme’s largest financial investors.

Aside from Google, enough consumers believe that they have been helped by 23andme’s tests that a court case or at least an impassioned appearance at Congressional hearings might start to turn things around.

The implications reach far beyond 23andme. In an interview published (paywall) in the Financial Times on Dec. 20, 2013, PayPal co-founder  and billionaire investor Peter Thiel lamented “how technological ambition has gone from the world, leaving what he calls an ‘age of diminished expectations that has slowly seeped into the culture.’ Predictably, given his libertarian bent, much of this is traced back to regulation.”

This is his explanation for why the computer industry (which inhabits “the world of bits”) has thrived while so many others (“the world of atoms”) have not: “The world of bits has not been regulated and that’s where we’ve seen a lot of progress in the past 40 years, and the world of atoms has been regulated, and that’s why it’s been hard to get progress in areas like biotechnology….”

The argument in favor of consumer genetics the way 23andme wants to practice it will be easier to make after there is overwhelming evidence in favor of its utility. I, for one, am not a customer. I have not been convinced that a 23andme test would do more for me than increase my anxiety about my genetic risks for a variety of ailments.

In that regard, FDA has a point beyond a merely procedural one. A clinical trial showing an advantage to a genetic test such as 23andme’s would go a long way toward that test achieving acceptance among both regulators and consumers.

23andme might go away as a provider of medical data. (The company still provides genealogical services.) But its skirmish has paved the way for a fight that could take the better part of the next decade and might result in either radical reform (no more FDA regulation of consumers’ own genes at all?) or in the offshoring of genetic analysis, with all its benefits and pitfalls, to more lenient regulatory environments, whether those turn out to be in China, in Iceland or somewhere in between.


Steve Dickman will be moderating a panel on Big Data in healthcare and drug discovery at Biotech Showcase in San Francisco on Jan. 14, 2014, at 8am Pacific time. He is CEO of consulting firm CBT Advisors, based in Cambridge, Massachusetts.

[1] Navigenics was acquired in 2012 by Life Technologies (now Thermo Fisher) and its consumer-facing business was shut down. DecodeME was discontinued before its parent, Iceland-based Decode, was acquired by Amgen in 2012. Pathway Genomics shied away from direct-to-consumer testing through Walgreens after a warning from FDA came in 2010.


Filed under Uncategorized

Genomes-R-Us: Is BGI now Complete?

By Steve Dickman, CEO, CBT Advisors

September 25, 2012 (originally published on Xconomy)

The sad endgame in the acquisition of Complete Genomics (CGI; NASDAQ: GNOM) came last week: having failed to create a sustainable business, CGI was put up for sale in June of this year, culminating in a takeover by sequencing powerhouse BGI for $117.6 million in cash plus $30 million in bridge financing.

Behind that headline is a fascinating story: a US company losing despite being right about its market; a Chinese company succeeding by vigorous price competition and then buying its rival; and a glimpse of the future of genomics-driven medicine.

On the surface, the sale of Complete Genomics looks like a case of overreach by the company’s investors and management coupled with poor execution. Complete, founded in 2005, had early on identified a superior business model for the coming era of cheap and frequent sequencing: take the sequencing activity and much of the interpretation out of the hands of hospitals and other healthcare providers and instead provide it on an outsource basis – both the sequence data itself as well as the all-important interpretation. For an apt analogy, think of Google’s core search business: why own a server farm when what you need are the search results?

I strongly remember meeting the late, visionary venture capitalist Alex Barkas of Prospect Ventures at the JP Morgan Healthcare Conference in early 2008 and hearing him forecast a glorious future for Complete Genomics. Even though the market was at that time buzzing about the next high-speed sequencing technology play, Pacific Biosciences (PacBio; NASDAQ: PACB), Barkas was supremely confident that CGI’s innovative business model would rule the day. That vision, driven by Complete Genomics CEO Cliff Reid as well as by Barkas and other investors, brought in VC and public investment of more than $250 million. The company went public at $9 a share and sold for as much as $17 a share before plummeting into the $2 a share range, where PacBio also now languishes. BGI’s purchase price correlates to $3.15 a share.

There were some momentary triumphs along the way, including technical breakthroughs, such as increasing the accuracy of sequencing. But, as Technology Review put it, “Though a 2011 paper published in Nature Biotechnology found that Complete Genomics produced more accurate DNA data than competitors, superior accuracy never translated into financial success.” CGI scored some small commercial successes along the way, such as landing the Mayo Clinic as a client in February of this year. Along the way, CGI was able to drop the price of a full human genome sequence to $4,200 in 2011, down from $12,000 in 2010.

But CGI’s revenues and, presumably, its margins dropped along with the price and the company never made up the difference on volume. Even worse, the company lagged in processing the genomes it had promised to sequence. The backlog in the end numbered in the hundreds of genomes. And even if CGI had been able to keep up with the influx of genomes it had, the customers did not come in sufficient numbers to create growth. A CGI business development executive told me in February that the only thing that would drive a higher stock price would be when the company proved its value and thereby showed big revenue gains. That executive has since left the company and, far from being able to build a sustainable business around this model, CGI first had to downsize and then had to be sold. And the buyer, in what must seem to insiders like a bitter irony, is a former competitor, the US subsidiary of Shenzhen, China-based BGI.

BGI is not just any competitor. In fact, BGI had arguably represented the biggest obstacle standing in the way of Complete’s success. As CGI tried to increase its market share by cutting prices, BGI responded by cutting them still further. BGI, using sequencers from Illumina, had a lower cost of capital due to the patience and strategic orientation of its investors. Like Amazon.com, the company and its investors focused not so much on quarterly earnings statements, but rather on BGI’s market share. They chose to operate BGI at what must have been a loss for several years and succeeded at driving CGI to the auction block. (BGI was founded in 1999, and in 2010 it received $1.5 billion in funding from the China Development Bank to expand its operations, according to Isaac Ro of Goldman Sachs.) BGI apparently succeeded in a big way. In January, 2011, Nature estimated that of the 30,000 human genomes that would be sequenced that year, BGI would be responsible for 10,000 to 20,000 of them. The lower prices were good for customers but bad for competitors (bye, bye, CGI).

A transition waiting to happen

So if BGI emerged triumphant from the bruising price war, why did it buy its former rival? Several reasons, all of them interesting. Like every other player in the commercial world of genomic sequencing and analysis, China-based BGI is on a journey from research to clinical applications. BGI hopes that the market finally (finally) expands once sequencing becomes a routine clinical activity ordered by physicians and reimbursed by insurance companies. In other words, like CGI was, BGI is eagerly preparing for sequencing to become part of routine disease diagnosis and determination of therapy.

The transition to clinical adoption of sequencing has been “just about to happen” for the last five or six years. If and when it does (and I am still betting that it will), BGI needed to be prepared. It was facing several obstacles, all of which can be overcome or at least reduced with the pickup of CGI:

  • Reduce or eliminate dependence on Illumina: Illumina is increasing and speeding up its service offerings. BGI had become dependent on sequencers from Illumina, the market leader in sequencer sales with over 60 per cent share, which had provided most of its 100-plus machines. (According to a research note published by Wall Street analyst Peter Lawson of Mizuho on Monday, Sep. 24, Illumina’s market share has actually reached 66 per cent.) Now that Illumina is moving into sequencing-as-a-service in a much bigger way, it will be more of a competitor to BGI. Thus, owning CGI and its proprietary sequencing technology (and different reagent suppliers) will give BGI an advantage.
  • Improve turnaround time: Shipping samples across the Pacific was not an efficient way for BGI to deliver data to customers in the U.S. market. Research institutes might have put up with it but clinicians will not. By buying Complete Genomics and its California-based sequencing “factory,” BGI is moving closer to its customers.
  • Add customers and capacity. BGI picks up not just CGI’s customers (like the Mayo Clinic) but also its 25 or so sequencers. Isaac Ro of Goldman Sachs last week told GenomeWeb that the deal accelerates BGI’s expansion into the US and gives it “an immediate infrastructure and service offering that will complement the facilities in China.” Former CGI developer Zhanzhi Hu told me in a phone interview that “If CGI has a healthy factory, it could crank out 1000 genomes a month – a not insignificant number.” BGI will need that capacity and more. A reliable industry source told me that the Mayo Clinic deal is expected to require sequencing of 200,000 human genomes over the next five years. Too bad for CGI that they could not hang on long enough to do all that sequencing!
  • Become a clinical laboratory. This is perhaps the most important reason. CGI applied in July to the US government to attain status as a CLIA lab. The decision, expected to be positive, should come in late 2012 or early 2013. The decision to buy CGI echoes the recent $50 million acquisition of former personal genomics company Navigenics by the second-largest sequencing manufacturer Life Technologies (LifeTech). In its acquisition announcement, LifeTech declared that it will shut down the Navigenics consumer business while maintaining its CLIA lab. (Illumina has had a CLIA lab since 2009).

Growing up and being clinical

If sequencing goes clinical, BGI will be able to play sooner and better based on its pickup of CGI. Although BGI already has a US sales presence, it has no way of serving clinical customers in the United States. If the CLIA lab designation comes through, then BGI will be able to sell clinical sequencing right away. One of the immediate drivers of the deal may have been Illumina’s predicted hesitancy (according to my industry sources) to sell clinical-rated instruments to BGI rather than research-only instruments once Illumina receives its expected 510(k) clearance from FDA.

There is also a cultural aspect. BGI has built a stellar reputation as a provider of genome sequence data. But it is not a US company. By keeping CGI up and running as a US subsidiary, BGI can – assuming that the deal goes through – sell its services more easily as it competes with US players like Illumina and LifeTech.

The race for improved sequencing hardware will not slow down. But as this acquisition shows, the more interesting battlefield, at least for the healthcare field, is in the interpretation of clinically obtained genomic data. The same week that CGI was acquired, Foundation Medicine secured a $42.5 million financing (funded in part by major clinical diagnostics players Roche and Laboratory Corporation of America) to pursue forward-looking genomic medicine in oncology; and the University of Texas M.D. Anderson Cancer Center announced an up to $3 billion “Moon Shots” initiative to significantly improve cancer care outcomes, in part by paying closer attention to genomic data.

Clinical sequencing is coming, first in diagnosing especially pediatric diseases of unknown origin and in oncology, then later in gastrointestinal disease (gut microbes…) and perhaps even, much later, in population screening. It just (barely) did not arrive in time to make a success of Complete Genomics. I suspect that BGI and its patient investors will have a better chance.


1 Comment

Filed under Uncategorized