NEW YORK (GenomeWeb) – At a time when data sharing is becoming the standard in the genomics field, Ambry Genetics this week took the practice a step further by shouldering the cost of sequencing the exomes of 10,000 cancer patients and freely sharing aggregate, allele-frequency data.
Ambry spent around $20 million to launch AmbryShare, which so far contains deidentified information from individuals who had or have hereditary breast and ovarian cancer, received commercial testing from the company, and had consented to research. Ambry then performed research-level exome analysis on the same blood sample and released deidentified data in AmbryShare.
Aaron Elliott, Ambry's chief operating officer, explained to GenomeWeb that AmbryShare doesn't contain exome sequences from individual patients. "What we're releasing is basically the allele frequency," he said. "So, if you go into the database and you searched for a particular variant or particular mutation, what you're going to see is the frequency of that mutation in this cohort of samples."
In the past two decades, scientists have identified around 20 genes — including the most famous BRCA1 and BRCA2 genes — associated with hereditary breast and ovarian cancer. Undoubtedly, there are others, but these can only be identified by casting a wider net, which exome sequencing does by assessing all the expressed genes. "So, this data holds a wealth of information to pull out new genes that are implicated," Elliott said.
Ambry Genetics was among the first labs to share data on BRCA gene mutations within public variant repositories after the US Supreme Court in 2013 invalidated a number of Myriad Genetics' patent claims on isolated BRCA1/2 sequences. Following the court's decision, advocacy organization Genetic Alliance announced Free the Data!, a project to encourage patients to get their BRCA test report from healthcare providers and submit the information to an open database called ClinVar.
As of November last year, Ambry has made close to 10,000 submissions on more than 200 genes into ClinVar, a resource that labs can use to identify areas of agreement and disagreement about whether variants have pathogenic or benign associations to disease. Ambry ranks seventh among top ClinVar submitters.
Many of Ambry's competitors in the lab community have also contributed to ClinVar and some have even launched data-sharing projects largely focused on classifying BRCA1/2 variants. For example, Quest and France's national institute of health and medical research, Inserm, launched BRCA Share last year, hoping to enlist the help of the lab community to speed up the classification of rarely seen BRCA mutations in cancer. Laboratory Corporation of America has signed on to BRCA Share, which aligned the efforts of two of the largest reference labs in the US.
The University of Utah, along with its Huntsman Cancer Institute and its non-profit lab ARUP, also announced their own open-source BRCA1 and BRCA2 gene mutation database last year. This database contains two types of resources on BRCA variants — one that classifies variants from published literature and family studies, and another that provides in silico predictions of risk to advance understanding of variants of unknown significance (VUS). The first resource contains data on around 2,500 reported BRCA1/2 variants, while the in silico database contains predictions on approximately 1,900 BRCA1 and 3,400 BRCA2 amino acids.
However, since AmbryShare draws on data from exome sequencing that the company has paid for, Ambry has gone "above and beyond" others in the field, Ambry CEO and stage IV prostate cancer survivor Charles Dunlop told GenomeWeb. Although this kind of information exists at universities, private institutions and at some diagnostic companies, Dunlop said they weren't publicly sharing this information to advance the field's collective understanding of the science.
At a Scripps Translational Science Institute conference last week, David Haussler from the University of California, Santa Cruz made an enthusiastic plug for the life sciences community to share data. Every disease is a rare disease at the molecular level, he said, and "if you can only compare between the patients who have been treated at only your institution, you've just tied one hand behind your back and shot yourself in the foot."
Haussler is one of the leaders of the Global Alliance for Genomics & Health, within which 387 institutional members from 38 countries are developing a framework for voluntary and secure sharing of genomic and clinical data. The Global Alliance last year launched the BRCA Challenge to pool data on the 12,000 known BRCA1/2 gene variants from around the world. Haussler seemed encouraged that a number of commercial testing firms have joined the challenge and are willing to share data.
"There's only one company I know that absolutely refuses to share this information," he said. "And you can guess who that might be." Researchers, patient groups, and labs that challenged the patent eligibility of Myriad Genetics' claims on BRCA1/2 genes, have continued to criticize the company for maintaining its proprietary database as a trade secret.
This week, at the American College of Medical Genetics and Genomics, Myriad presented data showing the ability of Pheno — its proprietary family history weighting tool — to analyze variants of unknown significance (VUS) in high-risk cancer genes, such as BRCA1/2, MLH1, MSH2 and MSH6, as well as moderate cancer risk genes, such as ATM, CHEK2 and PALB2. According to Myriad's analysis, Pheno was more than 99.5 percent accurate in definitively determining the association between VUS and cancer.
While Myriad continues to grow and improve its proprietary database, the company has also highlighted the discrepancies in variant classification among public resources, such as ClinVar, and has cautioned labs and healthcare providers against using them to make clinical decisions. But as advanced sequencing tests are increasingly performed in a medical context, yielding more VUS with unclear links to disease, many labs and researchers are voluntarily joining data-sharing efforts to try to crowdsource their classifications.
Data-sharing proponents have even suggested that payors insist labs openly report this information in repositories like ClinVar and elsewhere as a condition for insurance coverage. Heidi Rehm, director of the Laboratory for Molecular Medicine at Partners Healthcare Personalized Medicine, has publicly discussed how accrediting organizations could require that labs share variant interpretations as a way to assess quality.
Amid this increasing push for genomic data sharing, Ambry has launched AmbryShare, hoping to advance the field's collective knowhow. AmbryShare will be useful for discovering new genes implicated in disease, Ambry's Elliott said, whereas with public variant databases like ClinVar, "you're just contributing variants of characterized genes."
Drugmakers might be particularly interested in the aggregate allele-frequency information in the databases, he suggested. "If there is a drug that's targeting a specific mutation," pharmaceutical companies may want to know "what's the frequency of that mutation in people that have hereditary breast cancer," Elliott said.
The database, however, isn't restricted to experts in the field. Anyone can register online and can access this information.
Since only data from patients who had or have hereditary breast or ovarian cancer are currently included in AmbryShare, they were "very motivated" to contribute to advancing science, Elliott said. Ambry didn't track why some patients may have declined to contribute to AmbryShare, but according to company officials, more than 90 percent of its customers consent to donate their data for research. For those who decline to do so, Ambry discards samples after conducting the test they've ordered.
Within genomic databases, even when the information is deidentified, there is a risk of privacy breaches, which in turn can place participants at risk for discrimination. However, researchers from Baylor College of Medicine recently published their experience in a pilot project, in which seven cancer patients agreed to openly share data from their sequenced tumors and matched normal samples. The researchers said the project showed patients are willing and able to provide "true informed consent" for sharing their genome sequencing data.
The initial data release in AmbryShare is conservative in scope, but according to Dunlop, the company is currently talking with a panel of experts to figure out how to grow AmbryShare so it can eventually include genomic information on other diseases, as well as more granular genomic and phenotypic data.
In mid-February, the company launched its Super Lab, a 65,000-square-feet facility that will enable Ambry to process three times the number of samples daily and increase its capacity ninefold. After the Super Lab went live, the company worked quickly to launch AmbryShare, which Dunlop said the company has been planning for a year.
Dunlop couldn't provide an estimate for how much funding the company would need to expand AmbryShare, but he said, "We're going to keep doing this."