Name: Gregory Heath
Title: Senior Vice President and General Manager of Diagnostics, Illumina, since 2008
Experience and Education: Head of clinical genomics, senior vice president of global product marketing, senior vice president of global marketing and business development, and most recently, senior vice president of global business, Roche Molecular Systems, 2003 – 2008
Head of business development and licensing of the diagnostics division, F. Hoffman La Roche in Basel, 2000 – 2003
BS in psychology, Illinois State University
MA in experimental psychology, Hollins College
PhD in experimental psychology, Virginia Commonwealth University
As next-generation sequencing is increasingly considered for us in diagnostics and to guide treatment, physicians and researchers are faced with an array of challenges in incorporating the technology into the clinical setting.
At Cambridge Healthtech Institute's recent X-Gen Congress in San Diego, Gregory Heath, senior vice president and general manager of diagnostics at Illumina, spoke on a panel about the topic. Illumina has been offering whole-genome sequencing in its CLIA-certified and CAP-accredited laboratory since 2009. The service requires a physician's prescription and costs $9,500 for clinical cases where a medical need has been determined and $19,500 for anyone else.
Heath spoke with Clinical Sequencing News about the company's experience with sequencing patient genomes, the challenges it has faced, and what he sees as the future of clinical sequencing.
Can you describe the experience you've had with doing whole-genome sequencing of patients in your CLIA lab?
I can't disclose details around the specific cases, but the way we position it is: This is largely exploratory and largely experimental. So, we don't make any promises.
We have a five- to seven-day cooling-off period, so if someone wants to order the test, they have to really think about it for a while, along with their physician, and really discuss it.
Then we deliver the results to the physician and he uses it as an aid along with all his other diagnostic tools to really determine what the best course of therapy is for that patient.
What have been some of the challenges of sequencing patients' genomes, and what has the response been from both physicians and patients?
With regard to the CLIA lab, we return the results back to the physician, so they don't go to the patient, so there is this level of interpretation. We have a genetic counselor on staff that can help guide those decisions a little bit. The response has been positive. We've had some repeat business from the same accounts.
I think an important barrier is the physician's understanding of genetics and the ability to analyze the data in a meaningful way. The physicians aren't typically combing through the entire genome looking at every anomaly, but what they're doing is [using the genome to do] hypothesis testing from the broader clinical picture. It's a very efficient way to hypothesis-test.
Currently, if they have a suspicion that a gene is involved and they sequence that gene and are wrong, they have to order another sample and sequence it again, looking at a different gene. If they're wrong again, this can go on for years. If they're trying to diagnose a condition, it's much more efficient to be able to hypothesis-test in real time.
Do you include analysis along with the sequence data?
The analysis is relatively limited at this point. That's why we work with physicians that are knowledgeable and can interpret the results. They don't get a report like you would get from a cholesterol test. You don't get normal ranges. You don't get literature to interpret [the sequence data]. You do get support from the Illumina team, but these are fairly sophisticated users at this point.
Has it been challenging for physicians to interpret the data, and do you see that they're primarily looking at areas of the genome that they already understand?
I think that's where they go initially because you're … seeing at very high resolution that person's genome and there are going to be lots of anomalies that are maybe not described in the literature and not clinically meaningful.
No one comes in and says, 'My genome hurts.' [Patients] come in with a specific set of symptoms and the physician forms a hypothesis as to what's wrong in his mind and now he's looking in those areas [of the genome] first.
[ pagebreak ]
Are you sequencing solely cancer patients, or also patients with other types of diseases?
It's very broad. Whatever a physician deems is appropriate, he can order the test for that and it's been across a range of conditions — cancer, pediatric genetic disease, adult genetic disease, and what I call difficult-to-diagnose cases, which are not necessarily rare conditions, but cases where the symptoms are kind of vague. There's often a familial component, so the physician has a strong suspicion that it is genetic, but the current tools available don't give any good guidance. By doing whole-genome sequencing they can do this detective work in a much more efficient way.
At the X-Gen Congress meeting last month, you mentioned that whole-genome sequencing of cancer would likely be one of the first early applications of sequencing in a clinical setting. What other early applications do you foresee?
There are three that I would say are low-hanging fruits. I think the first one, which is not necessarily what we will do but the field will do, [is cancer]. I draw the analogy that PCR and infectious disease is like cancer and next-gen sequencing. PCR has this nice exquisite sensitivity that allows you to detect very low copy numbers of a virus. That was a great tool to apply to infectious disease where you could detect early infections and apply it in areas like blood screening. When I look at cancer and the heterogeneity you see in tumors, next-gen sequencing allows you to do very deep coverage and find minor mutations in those populations, and that's an important attribute of the technology that I think makes it very appropriate for cancer. It's the right technology applied to the right clinical question.
But, that said, obviously [next-gen sequencing] is a genetic tool, so any genetic disease is very appropriate. We're seeing [it be used] in complex cases and simple cases. Diseases like neurodegenerative diseases and cardiomyopathies. Some of these are relatively rare diseases and a physician may not have come across them often; that's why [sequencing] is a good tool for them to do that hypothesis testing I described earlier.
Probably the third area is viral resequencing for infectious disease. Resequencing HIV, hepatitis, to help adjust therapies based on viral load and how that virus has mutated.
Those are probably the three most immediate applications. Longer term, I can see doing things like metagenomics where you're looking at, let's say, the flora that exists in an individual's digestive track and trying to diagnose how it's absorbing nutrients. But, I think that's a little bit further off.
Will whole-genome sequencing be the way to go for clinical applications, or will some applications be more suited for targeted sequencing?
I think you're probably going to see both. I think there are definitely advantages to doing targeted sequencing when cost is an imperative, at this point in time. Whole-genome sequencing, as you know, has been getting a lot faster and a lot cheaper.
And I think what we're seeing, at least on the research side, is that, [for example,] what people used to consider a gene associated with a particular type of cancer might show up in another type of cancer, but people [hadn't previously] noticed it because they hadn't looked for it. So, if you develop a PCR assay looking at certain hotspots in certain genes, and it happens to be something like a tumor suppressor gene, you may miss mutations that knock out that gene, and so I think sequencing is a better way to go, because in that case, you're going to see any mutation that could compromise that patient and you won't miss mutations.
I think as more and more whole genomes get done we're going to learn more about the interplay between the non-coding regions and the complex geometries around how DNA folds, and [how] that interacts with genes that are in close proximity to other genes. I think we're going to find in the long run that whole genomes might be the best way to get the complete picture.
That said, I think in the short run we're going to see some targeted approaches that are associated specifically with therapies. If you know the mechanism of action of a drug and it's just targeting one specific gene or one specific hotspot of a gene, then you don't really need all that other information to prescribe that drug.
[ pagebreak ]
What do you see as the major regulatory hurdles, and how will those affect adoption of sequencing to the clinic?
Currently, the test is regulated as a lab-developed test under CLIA and CAP regulatory guidelines. And we're CLIA-approved and CAP-accredited. We also have a training program in the state of California to teach clinical lab scientists how to run sequencers, so we're building some of that expertise.
The groups that develop these guidelines are working actively to determine how they are going to regulate these [tests], and what standards can be applied more directly to the technology.
I think a lot of this is really being driven by the medical community, though, and groups like the American College of Medical Genetics are thinking about applying these tests in a lot of different spaces.
The challenge will be around how these things scale up as more information becomes available. [For example], in terms of [in vitro diagnostics], I think the main challenges are around intended use of the technology and how you manage risk. Specifically, what is this product going to be used for? Is this product going to be used for diagnosing KRAS mutations? Is it going to be used for carrier screening? What is the intended use of the product? Whenever you submit an IVD, the FDA requires you to specify an intended use.
So, will the challenge then be, if you're using sequencing to diagnose a specific disease, and you're doing whole-genome sequencing, what do you do if you find causative variants of another disease, or variants of unknown significance?
Sure, you're going to find variants of unknown significance, and I think in that setting if your intended use is limited to a specific application you'd probably do either a targeted approach or mask out the other data. The device side of the FDA has a tool designation for stuff like MRIs; they don't really have that today on the diagnostic side, but it would be great if they did.
What are the other major hurdles for implementing sequencing in the clinic?
I think there are going to be questions around reimbursement. I think there are going to be questions around who does the interpretation. Is this the responsibility of a pathologist? Is this the responsibility of a board-certified medical geneticist? Also, support from genetic counselors — there probably are not enough of them today to meet the need, let alone going forward when we have a lot more genetic information available. Those are some of the things. But I also don't believe too much in genetic exceptionalism. I think this is another tool that physicians can use. There are some unique characteristics of genetic information. It affects more than you; it affects your family -- your ancestors and descendants and your siblings, but there are just a few unique characteristics of it. I don't believe that we should treat it with such regard that it can't be used in very practical ways in medical practice. That said, I'm not a doctor.
Based on your experience in your CLIA lab and working with physicians, what about clinical sequencing is working well, and what do you want to see improved?
What's working well is when you see these approaches being applied and seeing them really impact intractable problems in health care. That's really very rewarding for the teams here. It's very rewarding for the guys who are on the front lines of patient care and of course for the patients themselves. They get answers to questions that might have plagued them for a long time.
What I think is still lagging a little bit is general education in the medical community and in the patients themselves around genetics and around probabilities. If you're looking at something like a carrier screen, that's going to affect the odds but it's not written in stone that this is absolutely going to happen. I think education around those two fronts will be a key driver that the community at large has to take on in the next several years.