NEW YORK (GenomeWeb) – 23andMe's 11,000-patient-strong Parkinson's disease community, for which the consumer genomics firm already has genotyping and phenotypic data, offers Genentech an unprecedented opportunity to learn more about the incurable neurodegenerative disorder and hunt for molecular characteristics that can be targeted by drugs.
Under a collaboration announced this week, Genentech said it would enroll 3,000 Parkinson's patients from 23andMe's research community in a study, whole-genome sequence them, and analyze the data for molecular markers that can inform its drug R&D efforts.
"No one has ever done [whole-genome] sequencing in Parkinson's at this scale before, and these people have all been genotyped," Emily Drabant-Conley, the director of business development at 23andMe who led the deal with Genentech, told GenomeWeb. "When we choose the people that we sequence, they'll be chosen in a very smart way, which will likely include existing genetic factors, as well as symptom profiles and family history and things like that."
The partners haven't disclosed the specific Parkinson's-related genes that they might focus on. LRRK2, according to Drabant-Conley, will certainly be a focus of the research efforts given its importance to the disease. Of the approximately 1 million Parkinson's patients in the US, the Michael J. Fox Foundation (MJFF) estimates that 10 percent of cases have a genetic cause. Among this minority group, mutations in the LRRK2 gene are the most well known to be linked to increased disease risk in carriers. Genentech is in the early stages of studying an LRRK2 inhibitor and has published a paper describing the molecular features that contribute to overactivation of the kinase.
Within 23andMe's database, there are 1,200 individuals with and without Parkinson's who have LRRK2 mutations, the largest cohort of its kind. Sergey Brin – 23andMe CEO Anne Wojcicki's husband whose Google Ventures investment firm has financially backed the consumer genomics company – carries a risk mutation in LRRK2.
In 2012, 23andMe whole-genome sequenced 50 of its customers harboring the G2019S mutation in LRRK2 to try to find out why some developed Parkinson's and other didn't. While this study did not identify any variants that significantly impacted the effects of the G2019S mutation, researchers found a few variants "suggestive of PD-related function."
These resources, combined with the research that 23andMe has already done on Parkinson's, will give it a leg up on identifying the right Parkinson's patients for whole-genome sequencing, as well as their unaffected first-degree relatives who are customers and agree to participate in the project. "We'll be able to do some really cool linkage studies that wouldn't be possible if we were only looking at people with Parkinson's disease," Drabant-Conley said.
23andMe will seek consent first from those in its Parkinson's community, and then reach out to unaffected relatives. The company has said that more than 80 percent of its overall customer base agrees to participate in research. "The consent rate is much higher in the Parkinson's community because they are in [the community] for a specific disease," Drabant-Conley said. Based on this initial consent, 23andMe can share customers' deidentified and aggregated data with its research partners.
For the Genentech effort, 23andMe will re-consent the 3,000 study subjects because their individual level data will be shared with the drug company. "With sequencing it's really quite relevant to share that individual-level data, because we may be looking for mutations that occur in a handful of individuals," Drabant-Conley said.
Genentech and 23andMe hope to start sequencing study subjects in nine months. The partners haven't settled on which sequencing shop they will hire to do the job. Once the sequence data comes back, 23andMe and Genentech will collaboratively analyze the data. "Quite a bit of analysis will happen here at 23andMe," Drabant-Conley said. Genentech declined to reveal how much it will cost and how long it will take to sequence 3,000 people, but Drabant-Conley said much of this will depend on the sequencing firm and the technology used.
Under the terms of the deal, 23andMe will continue to hold rights over the genotype data and the phenotype-related survey answers on the Parkinson's community subjects that it has already collected. Genentech will have first rights to the WGS data that's generated through the collaboration and any internal analysis will be proprietary to the drugmaker. After the project is completed, 23andMe will be able to analyze the sequence data, too, and share de-identified results with other research partners.
Genentech and 23andMe wouldn't confirm the financial details of their arrangement. However, according to an article in Forbes, Genentech has made $10 million in an upfront payment to 23andMe, and the deal includes opportunities for 23andMe to bring in $50 million in milestone payments.
Due to its ongoing regulatory interactions with the US Food and Drug Administration, 23andMe is currently not marketing its health-related testing services to customers in the US. The company recently launched a UK version of its testing business, where it does offer health-related genetic testing.
This more restrictive environment has slowed the rate at which 23andMe is growing its customer base, but it hasn't stopped the company from pursuing its ambitions to transform the way healthcare research is done in traditional academic circles by engaging consumers through web-based communities.
Partnering with drugmakers is a big part of 23andMe's efforts to grow its research footprint. Drabant-Conley noted that 23andMe has inked a number of other deals with pharma companies and many of these will start to become public this year. In addition to its Parkinson's project with Genentech, 23andMe is working with Pfizer on a 10,000-patient inflammatory bowel disease study and has previously partnered with Genentech on a metastatic breast cancer project, called InVite.
Most of 23andMe's pharma research partners are using genotype and phenotype data it has already generated. "Many of our partners are using this data to validate targets that they have already discovered. They want to validate those targets in human data, and they want to do that earlier in the R&D process," Drabant-Conley said. Drug developers are also using 23andMe's data to identify new drug targets, she added, focusing on finding new genetic markers associated with diseases that may be druggable targets.
"Partners are also using 23andMe to be able to find exquisitely targeted research participants," she noted. "They may be studying a rare disease or condition and they're looking to enroll people in their clinical trial that have a certain genetic background. So, they're using 23andMe to find and engage with those individuals." As the cost of next-generation sequencing technologies continues to come down, Drabant-Conley expects that other drugmakers will want to cull 23andMe's customer base for sequencing projects.
As for the latest Genentech deal, given its scale and scope, 23andMe has high hopes for the project. Currently there are drugs that ameliorate for a time the tremors that Parkinson's disease patients suffer from, but there are no treatments that slow disease progression or prevent the illness. Drabant-Conley is hopeful that this study can make a meaningful contribution on that front.
If through this project Genentech can learn more about why people develop Parkinson's and is able to advance a treatment based on that knowledge, it would vindicate 23andMe's consumer-driven research ethos. It could provide a viable alternative to the randomized-controlled trial model that researchers are finding they cannot do for many personalized drugs intended for smaller and smaller subsets of people. If 23andMe's research model works for drugmakers, in the long term the consumer genomics firm will have played a key role in improving the lengthy, costly, and failure-ridden drug development process.
"One in 10 drugs that goes in to clinical trials fails because the vast majority of the time there was no human data until [drugmakers] put that drug into a human being. So, the idea is that if you can bring human data into the development process much earlier on, you'd be more successful," Drabant-Conley said. "If there is any way to do it, it will be to do it with human genetic data on thousands of people who have been chosen in a smart way, based on mutations they have and that we know about already… And this may be one of the studies that proves that because we're finally doing it in the right way."