Biocom, the regional life-science association representing more than 550 member companies in Southern California, recently invited GEN's Editor-in-Chief, John Sterling, to moderate a breakfast panel discussion in San Diego called The Promise of Personalized Medicine—What Challenges Lie Ahead?
The panelists included Mark Erlander, Ph.D., CSO, bioTheranostics; David Nelson, Ph.D., president & CEO, Epic Sciences and director of the Clearity Foundation; Darlene Solomon, Ph.D., CTO, Agilent Technologies; and Ashley Van Zeeland, Ph.D., cofounder, Cypher Genomics, and director of Strategic Partnerships at the Scripps Translational Science Institute.
The following Q&A highlights a number of the main points made by the panelists.
John Sterling: What are some specific challenges personalized medicine needs to overcome before it can move a step closer to the clinic on a broad scale?
Dr. Erlander: I'll answer from the perspective of our test for the classification of metastatic cancers. It's a 92-gene RT-PCR assay. The key issue is how do you take a test for classifying tumors via gene expression, which was described in a seminal paper in Science 10 years ago, to an assay that has now been run on over 10,000 patients serviced by more than 2,500 oncologists and pathologists? I see the need to address three challenges.
Number one is to demonstrate clinical utility to the oncologist and pathologist. It is no longer enough for a diagnostic to just show great accuracy or a strong positive or negative predictive value. It must be clear that patients benefit from the test.
Our second challenge is reimbursement. This is another hurdle because the clinical utility studies that you carry out to convince a pathologist or an oncologist of the usefulness of your test may not necessarily be the same studies you need to do to qualify for reimbursement. We have Medicare coverage for our test, but it involved a two-year effort to get that coverage. Different studies were needed to satisfy Medicare's requirements.
Our third challenge is the FDA and many of you know what the landscape there is at this point.
Dr. Nelson: Mark hit on something very, very central. Historically, the pharmaceutical industry has been consumed by the problem of technical risk. Most drugs fail and that has profound implications for how we have approached the business models. Our industry has been living in this world of technical risk and there are all kinds of biological reasons why things are going to fail.
The beautiful thing about personalized medicine is we're going to mitigate the technical risk dramatically. We're going to start identifying the patients who will benefit from each individual drug. We're almost completely certain we can develop diagnostic tests that will measure things appropriately. So the technical risk around doing what we're doing is really low.
Thus, the main challenge in personalized medicine is not technical risk but the business risk. How do you find a financial model that appropriately incentivizes the diagnostic companies to actually develop the tests in the first place? For example, we're looking at some of the new cancer therapeutics such as crizotinib as targets for personalized medicine. This is a lung cancer drug developed by Pfizer. There is also Zelboraf, a melanoma drug from Genentech.
These are fantastic drugs. They can bring in hundreds of thousands of dollars for the pharmaceutical companies, but how much are the diagnostic companies getting? Hundreds of dollars.
Crizotinib brings this point home. You can't have the drug without the test. So the issue centers on offering financial encouragement to the diagnostic companies so that they create the necessary tests. We still need to jump through all the same regulatory hurdles. And there is the reimbursement issue and the FDA. But until we solve the business model question, we have a significant challenge.
Dr. Erlander: Zelboraf, the BRAF inhibitor for melanoma that you talked about, is a good example of the business model issue. The BRAF test reimbursement is not large. But diagnostic companies can also offer their own proprietary tests. With proprietary tests that show clinical utility, you can seek a larger reimbursement.
Then there is the question of intellectual property which, in my opinion, is the one place where diagnostic companies can potentially leverage their technology to gain a stronger financial foothold in the personalized medicine arena.
Dr. Solomon: In terms of personalized medicine challenges, demonstrating clinical utility is huge. Even after you have identified a patient cohort, so much is required to take a drug through clinical trials and regulatory approval with a statistically demonstrated benefit to patients.
Another point for discussion is the absolute requirement to educate physicians and clinicians about personalized medicine. We often think about education as a challenge that will come into play once we have these promising new companion diagnostics. But I've come to appreciate the magnitude of this gap and the need to act now. I also wonder whether better education across the entire medical system would enable us to be more effective in recruiting patients. It is important to have the right set of people participating in personalized medicine trials so that we can further accelerate genotype/phenotype correlations. Education is big and it is something we can begin to address now.
Dr. Van Zeeland: A real need is not only educating the physicians but also the consumers and patients about what personalized medicine is and also what it is not. We have to manage expectations about true outcomes. Some people think personalized medicine is a magic bullet.
John Sterling: Although you said earlier that the big problem is coming up with an effective business model, are there areas in personalized medicine that still need a technological breakthrough to move this science more quickly into the clinic?
Dr. Nelson: Many companies are working hard to bring technologies to bear on personalized medicine. I think we're getting close to being able to sequence more efficiently, even looking at things like RNA and DNA in circulating blood. Again, technically, we're making good strides.
If I had to pick one technical challenge that we face as an industry it would be interoperability, which is much more important than what everyone in the industry seems to be fixated on, which is standards. Many groups are using completely different methods, and the assumption is that we should all get exactly the same result or answer. But the challenge with that is that people have spent decades trying to solve the standards problems, often to no avail.
The much more thoughtful solution is to say, look, what we're really trying to do is solve a particular clinical question using a specific technology. Each test should be judged on its fitness for use around a certain clinical question.
Clinical questions are driven by one question and one answer. Does the analyte you're looking at have clinical utility for that particular application? We may all be measuring different things that may all have inherent value. And so one of the challenges we face is how do we take all these disparate methods and gather all that useful information to make it interoperable inside of the healthcare system.
For example, rather than trying to force everybody to come up with the same sequencing results or the same protein results, the real question should be does your test work to measure what it needs to measure to provide clinical utility? And then how do we provide a solution for interoperability to take that result and feed it into the medical system in a way that's meaningful?
Dr. Solomon: Yes, some technological challenges are more tactical while others are more strategic. As we consider what it's going to take to advance personalized medicine, we can acknowledge that the diagnostic side and our ability to measure genomic parameters have come a long way. Going forward, I think technology breakthroughs in front-end sample preparation and back-end analysis can really change the game. Many tumors and organs that aren't readily accessible for sampling, as well as the cost of taking and preserving those samples, require major technical advances. On the back-end, we need improved data analysis tools to generate better information, higher success rates out of trials, and lower trial costs. To me, these are the tactical opportunities we need to address to boost the workflow today.
On a more strategic level, we need to better understand disease and how to target therapeutic developments. The therapy issue is by no means trivial. It's one more layer of personalized medicine. And so, while we've come a long way in genomics, we need to focus our strategy on a more holistic view of understanding and treating disease. The integration of DNA with RNA, proteins, and metabolites is going to be key to more effective and efficient discovery and development of next-generation therapeutics and biomarkers.
Toxicity is also an important challenge for drug development and personalized medicine. The 'omics' technologies are on the verge of enabling new paradigms for in vitro systems toxicology. Because in vitro methods based on cell lines have the potential to be more predictive of human response and more cost-effective compared to traditional animal testing, they will further accelerate translation of research into clinical practice.
Dr. Van Zeeland: For me, three primary advances need to take place. The first speaks more to electronic health records (EHR). The EHRs have to be reconfigured in a way where you can start adding personalized medicine-type of information in a reasonable way and access it. But this is not just for the benefit of a single patient. If you push this information into the cloud, and you can start drawing inferences between patients with interesting phenotypes or profiles, you start moving from just a genomics understanding to a more holistic patient view and outcomes. Then you can start creating personalized medicine in a distributed system.
Sample prep does not require a huge technological breakthrough. It's just a paradigm shift involving tumors. We need fresh, frozen tumors. No more FFPE. This type of situation would truly start changing things and indicate what types of information we can draw out of the tumor. Circulating tumor cells and being able to monitor them in real time is going to be key.
The third advance relates to the "big data" question. Again, how do you create the algorithms that provide the clinical decision support for a patient in light of the fact that there is simply too much data for physicians to handle on their own. You need these advanced algorithms to help guide somebody through this process because there is often a ton of data.
Dr. Erlander: Harkening back on some of the points other people on the panel have made, I agree that information technology is really where the problem is now. We are generating lots of data, and it's really about trying to understanding what are the key pieces of data and how to integrate all this. This is a massive problem because it's not only an information problem, but it's also a biology problem that requires drawing inferences that actually make sense. You may find something important in one cohort of patients but then you have to go validate that information in another cohort. These things all take time and then you need to convince people like oncologists and pathologists that the findings you indicated were important were indeed so.