Unpacking the Clinical Genome
By Clinical Informatics News Staff
June 23, 2014 | The Clinical Genome Conference (TCGC), now in its third year, was held in San Francisco June 10-12. The program explored a wide range of issues facing genomics in the clinic including the payer process, genome interpretation software, clinical utility and integration, references and standards, and communicating findings to the patient. Here are some of the highlights from the event.
Genome Variation and Clinical Utility
Jonathan Keats, assistant professor at the Translational Genomics Research Institute, discussed the Multiple Myeloma Research Foundation’s CoMMpass clinical trial. Multiple myeloma is a good test case for cancer genomics models, Keats said. There’s an antibody that corresponds directly to tumor burden, so disease progression is tested easily with a blood test. The CoMMpass study revealed that multiple trajectories exist for the evolution of tumors, Keats said. Over the course of the disease, physicians can track “clonal tides”, he said, as tumor genomes gain and lose deletions over the course of several relapses. However, all mutations aren’t important. Most mutations aren’t expressed, Keats said, estimating that perhaps 50% of genes are expressed in any given tissues. The rate begs the question: are we treating mutations that are not actually impacting the patient’s disease?
Gail Jarvik, head of the division of medical genetics at the University of Washington Medical Center, presented lessons learned from the Clinical Sequencing Exploratory Research (CSER) Consortium (cheekily dubbed @hail_CSER on Twitter). Jarvik highlighted terminology: “mutations” are not always pathogenic; while “polymorphism” does not always signal benign. Since “variant” can mean any change and “polymorphism” signals a change that appears in at least 1% of the population, she favors single nucleotide variant (SNV) instead of single nucleotide polymorphism (SNP).
The CSER Consortium made its own list of actionable genes that could be returned to patients, and compared how they were classified by different groups. In the “bake-off” of six variants being categorized by six different groups—as pathogenic, likely pathogenic, a variant of uncertain significance, or benign—11 of the 36 responses varied. Jarvik called the variety “concerning”. There was also discrepancy in how long classification took. Most reviews were done by two to four people, and in many cases Variant Review Committees also considered the classification. The general time frame was 15 to 75 minutes to classify a variant. Interestingly, Jarvik noted that when reclassification happened, variants were almost always classified as less pathogenic.
But when asked if she thought the time and energy were necessary—should we just classify variants as “known pathogenic” and “everything else”?—Jarvik still came down on the side of including “likely pathogenic” and “variant of uncertain significance” classifications. Medical decision-making is different, she said, even for these classifications. We need the granularity.
Making a firm stand for whole genome sequencing, Elizabeth Worthey, assistant professor at the Medical College of Wisconsin, said that WGS covers more of the genome with better coverage than whole exome sequencing. 40x whole genome sequencing gives better exonic coverage than 100x whole exome sequencing. Everyone knows whole genome is best, she said. It’s just more expensive right now. WGS does not even take longer, she stressed. In fact, if time is short, WGS is a best first test!
Worthey has experience. She was on the team that sequenced Nicholas Volker in 2009, the young boy in Wisconsin who put a face and a name to clinical sequencing. At the time, though, the team opted for exome sequencing because whole genome sequencing would have cost about $1 million. Costs have plummeted since then.
Worthey also weighed the complexity of the genome and diagnoses. What if the patient has more than on condition, she asked? Without WGS, their diagnostic odyssey would stop at the first concerning gene, possibly missing an important comorbidity.
With just six weeks on the job, Jill Hagenkord, 23andMe’s new Chief Medical Officer, already has the vision. She proposed adding a fifth P to Lee Hood’s paradigm of P4 medicine: predictive, personalized, preventative, participatory, and now, philanthropic.
We’re in an era of genetic abundance, Hagenkord said, and it’s time to have a data-driven conversation about genomic data. There’s been no evidence that direct-to-consumer genetic testing provokes distress or leads to inappropriate treatment, Hagenkord said, citing a PeerJ paper from 2013* about consumer responses to DTC BRCA testing.
Though Hagenkord was mum on the timeline of the FDA and 23andMe conversations, she seemed confident that once the FDA approved the offering, all 23andMe patients could look at their health data results again. However she stressed that even now, the raw data from the Illumina SNP chips was available for download. The FDA does not object to us returning the results, she said.
Glen Weiss, director of clinical research at Cancer Treatment Centers of America, paid homage to Dr. George Sledge who proposed the idea of “stupid” and “smart” cancers at the 2011 ASCO meeting. Stupid cancers have a single dominant mutation, and targeting that dominant driver is usually effective. Resistance is rare, often occurs late, and can be reversed by attacking the dominant pathway. Weiss proposed chronic myelogenous leukemia as an example. Smart cancers, like breast or lung cancer, have multiple simultaneous drivers and a large mutational load. Treatment requires targeting multiple drivers and resistance is common and early.
Weiss does not believe that the clinic is ready for whole genome sequencing—listing concerns about timing of the testing and cost—but agreed that sequencing was good at identifying known, actionable mutations.
He argued for applying clinical sequencing to some patients, but not necessarily for all. For the “smart cancers” in particular, sequencing could unravel some of the various drivers. But even then Weiss does not believe that genomics will provide breakthroughs for most patients.
Atul Butte started his talk with packed disclosures slide listing—among other things—eight companies started by graduate students; he specifically mentioned NuMedii, Personalis and Carmenta Biocience during his talk. Now you can’t trust anything I have to say, he joked.
He then proceeded to issue his call to arms for open data and the entrepreneurial spirit. Data sharing is retroactive crowd-sourcing, he said. When you look up breast cancer studies in the NCBI database, you have access to 1,800 different laboratories that are helping you with your research—and they don’t even know it!
In fact, Butte said, almost half of his graduate students have started a company, all on the “best platform in the world”: Open Data. Butte believes that open data is the solution to skyrocketing drug discovery costs and the long lead times. He referenced the anti-depressant imipramine’s activity in non-small cell lung cancer. Fifteen months after building a computer model for the drug’s activity based on shared date, he had IRB approval for a clinical trial.
If you want to change the world, Butte said, you have got to take the science out of the lab. Start companies; quit writing papers!
But even accessing the data that is open can be a challenge. Malachi Griffith, at The Genome Institute at Washington University, spoke about mining the druggable genome, and the Drug-Genome Interaction database (DGIdb.org) that he and his twin brother Obi are developing. DGIdb hopes to answer questions about potential druggability, known drug-gene interactions, clinical actionability
John Quackenbush reminded the audience that as overwhelming as the data deluge is now, there’s more and more data all the time, and we never have everything. Histories are always incomplete. Lifestyle data is always hazy. We need to be careful how we use the data we have, he said.
The life sciences are still waiting for the true disruption that will make data accessible to everyone, Quackenbush said. Think about what the printing press did for the written word. That’s the kind of tool we’re looking for, he said.
But he also championed the idea of a precision medicine “ecoverse”—there will be multiple types of data and information, and we need to effectively deliver the right data to the right people. The $100,000 analysis is only for extremely rare diagnostic odysseys, with armies of people doing analysis and follow up experiments, he said. Practically, most data analysis will be more streamlined than that.
Interpretation and Translation to the Client
The final day of TCGC focused on clinical integrations. Elissa Levin, a certified genetic counselor and assistant professor of genetics and genomic sciences at the Icahn Institute for Genomics at Mount Sinai School of Medicine, talked about the challenges of interpreting and communicating the “healthy” genome.
At a class offered at Mount Sinai, genetic counseling students have the opportunity to sequence and interpret their own genomes. Of course most of these students won’t find anything shocking in their genomes, but their reactions and how they deal with their data is still telling.
80% of students thought it was “useful” to analyze their genomes, Levin reported, and found whole genome sequencing both over- and underwhelming. After class, students reported being more persistent and empathetic for their own patients.
Of all of the students that participated, at least one reported had test-related regret, and distress. This student had some extenuating life circumstances at the same time, but Levin stressed that that was not a reason to discount the reaction.
Everyone that gets tested will always have life circumstances going on, she said.
Mount Sinai and Sage Bio have launched the Resilience Project, which is looking for healthy individuals who may be carriers of mutations that we would normally associate with childhood illness or disease. Levin expects about 1 in 20,000 individuals to present some resilience, and says the project will need millions of volunteers to study. The project is currently working on an e-consent platform and other social platforms to leverage mobile health.
* Francke U, Dijamco C, Kiefer AK, Eriksson N, Moiseff B et al. (2013) Dealing with the unexpected: consumer responses to direct-access BRCA mutation testing. PeerJ 1:e8 http://dx.doi.org/10.7717/peerj.8
Text Only 2000 character limit
Page 1 of 1