As I reported earlier, 23andMe today announced its intention to pursue pharmaceutical R&D, in addition to its business selling direct-to-consumer genetic test kits. To lead its new drug research group, it has tapped Richard Scheller, the longtime head of research at Genentech and a distinguished neuroscientist.
Scheller wasn’t available to talk today, but after posting my earlier story, I had a conversation with 23andMe president Andy Page, who joined the company in 2013 after working for many years in Internet commerce and finance. Page filled in more details and gave updates on other facets of the company’s business.
A few quick bullet points:
—23andMe will continue its direct to consumer business.
—23andMe will raise capital soon to finance the new drug group.
—Scheller’s group could grow to roughly 15 people by year’s end.
—Most of the work this year will focus on data analysis, not “wet lab” work.
—Scheller will be in charge of pipeline and partnering decisions for the drug programs.
—23andMe does not gather health information through wearable devices, nor does it have near-term plans to do so.
I first asked Page when 23andMe decided to get into drug development. He said that for years, several people outside the company, including Art Levinson—the former CEO of Genentech, now CEO of Calico—have raised the question with CEO Anne Wojcicki of what to do with the 23andMe data once the database became large enough. “We always were and always will be a consumer company first, but doing research is a close second,” says Page.
When 23andMe partnered with Genentech to collaborate on Parkinson’s disease research, Scheller, then the head of research and early development at Genentech, “was the point on that deal. He saw our database, saw its potential, and suddenly it became the intersection of a lot of good things happening at the same time.”
“A couple days after he announced his retirement from Genentech, the agreement [to join 23andMe] came together at Anne’s house.” (Which means that Scheller had already agreed to his new job when the Genentech-23andMe deal was announced this January.)
So what was the threshold, I asked? How and when did the database become ‘large enough’ for 23andMe to do its own drug discovery work with it?
First, said Page, was the scale. It now contains the information of nearly 900,000 people. It’s not just their genomic information. 23andMe uses surveys to build other layers of information, too: age, gender, ethnicity, and so on. For people with specific diseases, the survey asks questions such as “what drugs are you taking” and “when did you first start to notice symptoms,” Page said.
“We have 250 million questions answered, and we’re constantly looking for ways to enrich the database,” he said.
Also driving the decision was a certain number of what 23andMe calls “escapers,” or people who have a mutation associated with a disease, yet they don’t have the disease. “They don’t show up in the medical system [because they don’t get sick], but they show up in our database.” For competitive reasons, Page declined to say how many of them, and with which mutations, are in the 23andMe database.
(Page is unrelated to Google CEO and cofounder Larry Page. Google was an early investor in 23andMe.)
With the launch this week of Apple’s ResearchKit program to let iPhone users contribute their health data into various disease research areas, I asked Page if 23andMe was also gathering data through wearable devices and the like—or if it planned to do so. “Whether we partner with a wearable company or not, we definitely think genetics should be married with all the other info being tracked. That will come in time. We’re looking forward to being able to do that,” he said.
But it’s not happening now. “Taking in wearable device data right now is not core to our near term strategic objectives,” Page said.
A big part of the company’s focus is “working with FDA in order to seek permission, or to have permission granted, to market the health portion of our product in North America,” Page said. In other words, to do what FDA banned it from doing in late 2013, when it ordered 23andMe to stop promoting its Personal Genome Service to consumers as a way to gain insight into medical conditions or to assess risk of disease. To do that, the kit would need to be approved as a medical device or diagnostic, with a certain body of evidence behind it. Page would only say there has been “progress” in discussions with the agency.
(The company received the FDA’s marketing clearance last month for a genetic test to determine whether a person is a carrier of the mutation that would put his or her children at risk of Bloom syndrome.)
Apple’s ResearchKit has also raised questions about selection bias: How to study health in populations that might not be representative of broader society. Is that a concern as 23andMe gets into drug R&D? After all, a majority of those in the database are there because they paid $99 to learn their genomic profiles. (23andMe has also distributed kits for free to people who agree to participate in research projects, such as those focusing on Parkinson’s, lupus, and African Americans.)
That cost—not to mention the curiosity spurring someone to pay for the service—would seem to skew the database toward those who have disposable income, a certain level of education, and are inclined to trust science.
It’s true, Page said, that “people first have to want to engage in their own genetics. They endorse our mission to access, understand, and benefit from the human genome. Naturally there will be populations that are underrepresented, we’re well aware of that, but there are lots of ways we try to make it as representative as possible.”
The free kits are one method. Another is replication studies, triangulating results from the 23andMe database with results from other databases.
The test kit business is doing fine, according to Page, with nearly 300,000 sold in 2014. (23andMe is privately held, so there’s no way to independently check financial claims.) So why get into the drug business? It can bring huge profit margins, but it also requires massive, long-term investment before reaching profits.
I asked if 23andMe’s low-margin consumer business needs a high-margin counterpart. Page said drug R&D “seemed to be a natural outgrowth of where we are” now that it has entered more than 20 collaborations with academic groups and pharma companies—Genentech, Pfizer (for irritable bowel disease), and Reset Therapeutics of San Francisco are the only three industry partners disclosed.
So what happens to those partnerships now that 23andMe is also mining the database for drug targets? Are the therapeutic areas covered by those partnerships now off-limits?
Page didn’t exactly say no. He stressed that the company doesn’t yet know what it will pursue. Then he noted that the exclusive work its partners do in the database is limited, and most of the work flows back into 23andMe. There will be a firewall: Scheller’s research group won’t know what 23andMe’s partners are doing, and vice versa.
It could be that Scheller’s team will feed new targets to partners, just as small biotechs with promising technology platforms do all the time. Or Scheller might decide to do work around a target and bring a drug into the clinic. The big budget decisions will be made by 23andMe’s executive team, Page said, not just by Scheller.
23andMe will raise capital this spring to fund Scheller’s group, Page said. Most of the work in 2015 will be bioinformatics and data analysis; “I’d be surprised if there are more than 15 people by the end of the year,” Page said, but the group will then “scale appropriately,” perhaps doubling in size in 2016.
And as 23andMe moves into new headquarters in Mountain View, CA—near the train station for those coming down from San Francisco, Page pointed out—Scheller will also start looking for lab space “likely in the Bay Area.”
“I’d be surprised if we didn’t have a lab in the next 18 months or sooner,” Page said.