February 2013, Vol 2, No 1
Interview with the Innovators
An Interview with Michael J. Pellini, MD, of Foundation Medicine
As the molecular drivers of cancer are better understood and targeted therapies are developed against those drivers, there is a growing need to perform comprehensive molecular analysis of tumors to determine the optimal treatment strategy for each patient. To this end, Foundation Medicine was founded in April 2010 by leaders in genome technology, cancer biology, and medical oncology from the Broad Institute, the Dana-Farber Cancer Institute, Harvard Medical School, and the Massachusetts Institute of Technology. Their goal: to develop a comprehensive cancer diagnostic test that helps physicians recommend treatment options for each patient based on the molecular subtype of their cancer.
The company’s first assay, FoundationOne, became available last year and is a genomic profile that identifies a patient’s individual molecular alterations by sequencing 236 genes known to be somatically altered in human cancer. Genomic alterations are matched to relevant targeted therapies, either approved or in clinical trials, that could be a rational choice for the patient based on the genomic profile of their cancer. This information is reported to the patient’s physician via the FoundationOne Interactive Cancer Explorer, the company’s online reporting platform. This assay utilizes next-generation sequencing technology on small amounts of routine cancer tissue with a reported >99% sensitivity and specificity for alterations within relevant cancer genes.
PMO recently had the great pleasure of meeting with Michael J. Pellini, MD, to discuss the unique approach of Foundation Medicine to personalizing cancer treatment. To view the interview in its entirety, please visit www.personalizedmedonc.com/
Dr Pellini joined Foundation Medicine as President and Chief Executive Officer in May 2011. Previously, he was President and Chief Operating Officer of Clarient, a GE Healthcare Company. Dr Pellini received a BA from Boston College, an MBA from Drexel University, and an MD from Jefferson Medical College of Thomas Jefferson University. Dr Pellini serves as a member of the Boston College Technology Council and the Board of Trustees for the Coriell Institute for Medical Research.
PMO Thank you so much for taking the time to speak with us, Dr Pellini. To start, how do you define personalized medicine in oncology?
Dr Pellini Personalized medicine in oncology is a concept that is relatively easy to define but difficult to implement.
Let’s start with the physician’s perspective. If we implement personalized medicine in oncology well, we enable each oncologist to select the appropriate therapy or therapies to target the drivers of each patient’s cancer.
If we end with a patient in mind, then that patient is confident that he or she is getting the most effective therapy to prolong life. That’s how it has to come together. We select the right therapy, attack the molecular appropriate drivers, and hopefully prolong the life of the patient.
PMO Personalized medicine is largely practiced in academic settings. In your opinion, what changes are needed to ensure that personalized medicine can be made broadly available to patients managed by community physicians?
Dr Pellini That is a key question for all of us. How do we drive the practice of personalized medicine in oncology from the academic centers to the community setting? It’s certainly one of the motivating factors in establishing Foundation Medicine.
Eighty percent to 85% of patients with cancer are treated in community practices around the United States; however, the key novel therapeutics tend to be used first within the academic medical centers and slowly roll out to community practices. But what we have to do as a company, what we have to do as a community, is drive this information to the community setting so those 80% to 85% of patients who are being treated in this country, and even outside the United States, are getting best-in-class testing and the most appropriate therapy for their disease.
Foundation Medicine launched a test called FoundationOne last year. This is a test that takes a comprehensive snapshot of each patient’s cancer to help the oncologist identify the molecular drivers to treat. Now that sounds like something that should occur in an academic medical setting, and it is occurring in the academic medical setting, but there’s no reason why that can’t be driven into the communities around the United States and beyond. In fact, we’re already seeing it.
There are additional issues to consider. One is that of reimbursement. How do we rethink the reimbursement challenge in the United States? Reimbursement is something that we have to tackle over the coming years because of the rapid technological changes in our space. And we also have to think about the regulatory side, the FDA. What role are they going to play in this transformation?
So there are several different factors that we have to take into consideration, but again, if we do this well, we’ll start to take diagnostic tests and therapeutics that are often only utilized in the key academic medical centers and gradually, or perhaps rapidly, drive them into communities.
PMO As you mentioned, Foundation Medicine recently launched FoundationOne to deliver genomic profiles to oncologists to help them guide personalized therapy. Can you talk a little bit about FoundationOne and how it’s going to work for community oncologists?
Dr Pellini FoundationOne is a comprehensive molecular profile that oncologists can use to help understand the key molecular drivers of their patients’ cancer. It’s unique in that we sequence the coding regions of 236 genes known to be somatically altered in human cancer. In addition, we capture all 4 major classes of alterations: insertions/deletions, base pair substitutions, translocations, and copy number alterations. There’s nothing else even close to it on the market today.
When we initially launched the test, the first uptake came from the academic medical centers. What we’re starting to see, though, is this migration outside of the academic medical centers into the community setting. Currently, about 40% of the tests are ordered by oncologists in the academic medical centers, 40% by oncologists practicing in the community setting, and 20% from the international community.
One of the reasons that this migration has occurred so rapidly is that in FoundationOne we have designed a test that is straightforward to use.
Second, it’s impactful. In more than 70% of the cases that are reported out, oncologists are receiving information that is clinically actionable for their patients. So it’s not only the amount of information they are receiving; rational options open up when you have this information. It helps them think about treating a patient perhaps a little bit differently than they otherwise would have.
We spent a lot of time on the development of this program, not just in terms of optimizing the test and making sure that the test has the robust sensitivity and specificity that is needed for the clinic, but we spent a lot of time making sure that it integrates easily into clinical practice, both in terms of the way the test is ordered and the way the test is reported back to the oncologist.
It’s about taking very complex information that’s generated on each patient’s cancer sample and finding a way to translate that into information that is easily understood and then acted upon.
PMO Confidence in the results of a program such as FoundationOne is extremely critical for a community uptake. What steps has Foundation Medicine taken to ensure that FoundationOne will provide the accuracy that patients and physicians are looking for?
Dr Pellini When you come out with a new molecular test, especially a complex new molecular test that should have an impact on the way patients are ultimately treated, one needs to make sure that it has robust sensitivity and specificity. With the new technology advancements in this field, we have to think about new ways to generate that type of analytic information. So if we’re generating the data, how do we know that the data are accurate?
A particular benefit, yet also challenge, of utilizing next-generation sequencing is that if you use this technology appropriately, it’s potentially better than the gold standard. For example, we routinely find an alteration in EGFR, or we’ll pick up a HER2 alteration or some type of translocation that might not have been identified by the standard molecular techniques that are used today. That presents us with a challenge, and 1 way to overcome that challenge is to go back and perform additional validation with specific molecular tests to confirm the next-generation sequencing results. But that can’t be done forever. We have developed new approaches to validate testing based on next-generation sequencing. We’ll present these data at AACR [American Association for Cancer Research], and we also recently submitted our validation approach to a prestigious journal.
We’ve taken a very simplistic approach to getting the message out about the sensitivity and specificity, or the accuracy, of this approach. It’s all about generating and presenting the data. In the last year alone, we’ve had articles in Nature Medicine and Journal of Thoracic Oncology. We had a total of 8 presentations at ASCO [American Society of Clinical Oncology] last year. We’re in a position to generate a significant amount of data with many of our collaborating academic medical center scientists and physicians. Our aim is simply to publish these data to demonstrate to the broader community that this is a new approach, but this is a new approach that’s potentially going to change the way that we think about molecular testing.
PMO Treating cancer patients has become an increasingly costly venture. How does incorporating new molecular profiling into that provide value? How do we justify those added costs to therapy?
Dr Pellini How does this approach of utilizing next-generation sequencing add value? How do we deal with the additional costs? It’s a question today that’s front and center, so I’ll try to hit that question from a couple of different perspectives.
Let’s first focus on the test itself.
In terms of the cost of this type of testing, we often hear that the cost of doing the human genome is now coming down to under $1000. But I think it’s really important to understand what that is and what it isn’t.
As we think about FoundationOne, consider the costs and the requirements that go into this test from 3 different perspectives. There’s an enormous amount of work that occurs before the sequencing. The sample preparation standard operating procedures that allow us to extract nucleic acid DNA and RNA from a very small piece of paraffin-embedded tissue is extremely challenging on its own. The library construction and hybrid capture procedures are the next presequencing components. These are procedures that our company has perfected over the past few years. Recognize there’s a real cost burden associated with this initial aspect of the testing.
The second part is the cost and effort associated with the sequencing itself. These costs are coming down, but we really need to differentiate running a germline sample, which is basically a blood sample, versus applying sequencing to human cancer tissue. Sequencing from cancer tissue is a much more costly and complex enterprise than running it on a germline sample.
Then there’s the final piece of the testing. Once we have all those data, what do we do with them? How do we take that information and translate it into something that’s actionable? These steps require complex informatics.
I want to make sure that we separate this informatics component of the $1000 genome from all the other costs that actually go into running this type of test for patients diagnosed with cancer. It’s critically important for the payer community to understand how important informatics has become, because this is a complex and new consideration, and it’s something the payers and industry must continue to work through.
The reason that we have to push in this direction and figure this out together is that the current paradigm for molecular testing in the United States must change. First, we are running out of tissue to continue down the current path. If we think we’re going to run 3, 5, or 10 individual molecular tests and continue to do that on small biopsies, think again. We’re running out of tissue, and we won’t get the answers that we ultimately need for patient care.
Another challenge is the additional cost incurred every time a new molecular test is ordered. As you know, it’s not $100 per molecular test; it’s roughly $500 per new test on average. So, if I’m a thoracic oncologist, and I’m treating a patient with non–small cell lung cancer and want to run 10 individual molecular tests, that’s at least a $5000 price tag. Further, the number of tests that are being ordered is only growing.
A final aspect of cost relates to the therapeutic decisions that are being made. This starts to get a little complicated, but today if patients are relatively late stage in their disease, chances are they’re getting a cytotoxic chemotherapy. Some of these chemotherapies have been around for a long time. The cost of the drug itself may not be that significant. The side effects, the emergency room visits, the impact on the patient, and even the broader cost to the healthcare system associated with those adverse events can become very significant. If we can use this approach to find a target therapy, or therapies, that are more effective and less toxic for the patient, the price of the diagnostic test becomes quite insignificant.
If we utilize this technology to work with smaller sample sizes, if we utilize this technology to avoid the increasing costs associated with the growing number of individual molecular tests that are going to be ordered, and if we utilize this technology to select the right therapeutic up front, it might change some of the fears that we have around targeted therapies and the diagnostic workup.
What I can tell you is that while we are in the early days of working with the payers, they have been receptive. Medicare has been receptive. They all understand that this is the next wave of molecular testing. They all understand the potential clinical value. We all know this approach is going to have an impact in the clinic, and they are with us, and now we’re simply starting the dialogue. Of course, that dialogue has to be backed up with clinical and pharmacoeconomic data to make sure that we all arrive at the appropriate end point.
The reason that I’m optimistic is because it is the right thing to do. It’s the right thing to do for the patient, it’s the right thing to do for the oncologist, and in this case we can say we ultimately believe it’s going to be the right thing to do for the payer as well.
PMO There are cancer patients who either don’t respond to therapy or have acquired resistance to therapy. How will FoundationOne specifically help those particular patients suffering from cancer?
Dr Pellini The question is how could FoundationOne now change the treatment options or help patients who have maybe failed first-, second-, or third-line therapy. They have metastatic disease. What comes next?
Often, those patients are relegated to broad-based cytotoxic chemotherapeutics. Sometimes there’s a benefit, but often there’s not, and there are significant side effects that typically go hand in hand with these drugs. It’s something the oncology community has been working through for decades.
Let me turn to the data to answer the question about how FoundationOne and this technology can be used. We have a data set from 95 patients diagnosed with non–small cell lung cancer. What’s interesting is that we found alterations in 33 genes in those patients. That’s kind of scientifically interesting, but what does it really mean? Well, 22 of those altered genes can be directly tied to a therapeutic that is either on the market or in clinical trials.
You may say there aren’t that many targeted therapeutics for non–small cell lung cancer, so what’s the benefit of this knowledge. Here’s the potential benefit: Today there are 3 tests that are routinely used for patients with non–small cell lung cancer in the community: EGFR, KRAS, and EML4-ALK. If you have patients who tested negative for those 3 biomarkers, they are relegated to broad-based chemotherapy. Second, if you have patients who have failed their first targeted therapy, and their next option is a chemotherapy with a marginal benefit, do you want to know that there might be an additional 20 drugs that are either in clinical trials or on the market that could potentially be helpful to your patients with that disease? That’s the way to think about FoundationOne; that’s the way to think about this next wave of molecular testing. It’s going to open up new doors for patients and for oncologists to select therapeutics and find clinical trials in a very rational setting.
PMO Regarding heterogeneity, there was an article published last March in the New England Journal of Medicine about the nuances of intertumoral heterogeneity. How will FoundationOne help researchers address the heterogeneity of tumors?
Dr Pellini While the heterogeneity of cancer became more of a mainstream topic thanks to the New England Journal of Medicine article and a New York Times follow-up article, it’s something that we’ve known for a long time in this community. It’s not surprising, especially if you have a pathology background, that tumors are, in fact, heterogeneous.
We have to rethink the way that we’re testing these samples. We spent an enormous amount of time making sure that FoundationOne had the appropriate level of sensitivity and specificity to capture the alterations that are found at a very low mutant allele frequency.
If you’re looking for an alteration that is found in 60% of the tumor’s cells, chances are you’re also going to capture it with a much simpler technology. But if the tumor is heterogeneous, as almost all are, the alterations may only be found in a very limited set of the tumor’s cells. If tissue is selected from 2 different areas, a technology is needed with an exquisite sensitivity and specificity to capture the alterations that are found at a low mutant allele frequency.
That’s how we need to think about capturing this information. It’s just not practical in today’s clinical age to think about taking multiple biopsies from a patient and then running multiple tests. The payers aren’t going to pay for the multiple biopsies and tests, and we might not have access to enough tissue. What we have to do is utilize this technology to deliver results with the appropriate level of accuracy while minimizing the risk that we face with heterogeneous tissue that we work with every single day.
PMO To achieve personalized medicine in oncology all healthcare professionals need to be educated. What efforts is Foundation Medicine making toward payers, physicians, patients, and other healthcare professionals to ensure that they’re up to speed on the newness of personalized medicine in oncology?
Dr Pellini Education is such a key element of enabling this approach to become mainstream. It’s not just the education of oncologists, it’s education of the payers, it’s education of the regulators. And a group that can play a vital role is the patient advocates.
We recently invited 10 of the most well-respected patient advocacy groups in the country to Foundation Medicine and spent a day and a half talking with them about this very issue. How do we educate patients? How do we educate oncologists? What role can they play? What role should we play?
I’ll tell you that I learned more in that 1 day than I think I’ve learned in any 1 day in a long time because it really is a different perspective. It’s the patient’s perspective. The only thing they care about is driving the advancement of cancer therapy forward. That’s it. How can we maximize the benefit of this new technology, of everything that we’re learning about cancer for patients? It starts with education. Education is perhaps the single most important underpinning of that revolution in cancer care. We have to think about the advocacy groups, because if you work closely with them, they’ll help get the message out. That’s critical.
Of course, we also have to think about oncologists. The oncologists are central to this migration of cancer care, especially out of the academic medical centers into the communities around the United States. Oncologists are data driven, and that’s why our company has spent so much time and effort on really developing the data and putting together the retrospective and prospective clinical studies. We’ve partnered with many of the major cancer centers, in the United States and overseas, and pharmaceutical companies to generate the data to demonstrate that this approach will ultimately have a dramatic impact in the way that we think about caring for patients with cancer.
Education also comes in the form of working with the FDA. We have met with certain groups at the FDA. The FDA has not completely sorted through how to work with next-generation sequencing. It’s a complex new technology. It’s going to take time, but they are receptive to it. There’s an understanding at the FDA that this technology will likely have a significant impact on the way pharmaceutical companies perform their clinical trials. It allows them to be much more targeted and ideally much more efficient in their approach to clinical trials. The FDA also understands that not only can this technology be impactful at the level of the clinical trial; it can be impactful at the level of the clinic, impacting how oncologists select therapies for each of their patients.
Payers also understand these concepts. They might not be as focused on the clinical trials because that’s the world of the FDA and pharma, but they are focused on making sure that they do get the right therapy to the patient to maximize the benefit that patient is going to receive, and also to minimize the negative, the side effects that the patient might have if he or she receives the wrong therapy.
So, we have a lot of different stakeholders when it comes to education. We have patient stakeholders, we have oncologist stakeholders, we have payer stakeholders, and we have regulators at the FDA.
Patient advocacy groups can play a role with each of them, but we as a company, we as an industry, we as a community have to continue to focus on education for all these groups. If we do it well, we will see oncology continue to evolve extremely quickly.
PMO Drug development is going to change in the future. What is your vision of the future of drug development as it relates to our pharmaceutical company colleagues utilizing next-generation sequencing technologies in drug development?
Dr Pellini How might next-generation sequencing impact drug development? I’ll tell you, one of the most significant surprises that I’ve had in my 18 or 19 months at Foundation Medicine is the adoption of this approach by pharmaceutical companies. I have to admit that when I came here I was a little skeptical that pharma would readily adopt what Foundation Medicine was doing because, by design, this approach is going to take a therapeutic that might be used for a broader population and narrow that population. To my pleasant surprise, it’s not only the development groups within pharma that understand it, but the commercial teams get it as well.
There have been a couple of very high-profile failed phase 3 trials, yet we know that there was a subset of patients who responded extremely well. How do we identify that subset? Well, if we have the molecular profile of those patients, we can look at the responders, match them up to the appropriate molecular alterations that were found, and determine whether there’s a match. If there is a match, think about how quickly that follow-up clinical trial can commence and how small that trial might need to be if it’s targeting a very specific population.
The second thing that we’re seeing as the therapeutics become more targeted is pharma looking for molecular information that continues to refine their patient populations. For example, we have at least 1 partnership where our test and our approach is being applied to most of their phase 1 and phase 2 patients in oncology on a global basis. That’s thousands of patients. That will add up to dozens of clinical trials, and a fully informative molecular profile is going to be run on each of those patients to help them continue to think about how to refine their therapeutic approach.
That should be promising to all of us, and again, it’s been a very pleasant surprise to see the pharmaceutical industry’s response to this approach.
Another path that I’ll go down touches on broadening their markets. We’re moving toward thousands of samples from patients with all different types of cancer. We’re running this information at a level of sensitivity and specificity that is helping us uncover new alterations, which means we have a treasure chest of new information that has emerged within Foundation Medicine. So why is that important?
If I’m a pharmaceutical company and have a therapeutic that is targeting an alteration found in 10% of patients with breast cancer, I might want to know if that alteration is found in patients with pancreatic cancer, with colorectal cancer, with rare sarcomas, and so on, so we can think about additional markets for this therapeutic.
We now receive regular requests from pharmaceutical companies regarding an alteration or the possibility of a combined trial with 2 therapeutics that are targeting 2 alterations, and they want to understand how often these alterations are found across various disease states. That’s information that we can provide in a timely fashion even with only several thousand patients in our database. What happens when our database goes to 10,000, 50,000, and 100,000 patients? You start to get a glimpse into the future of clinical trials and how pharmaceutical companies and companies like Foundation Medicine can work together and ultimately partner with oncologists and patients to refine clinical trials, thereby making them as efficient as possible.
PMO Can you provide us with a couple of clinical examples from your experiences?
Dr Pellini One of the things that drives us each and every day is that we get to see the impact that this approach can have in the clinic. It’s unusual that a day goes by when our head of clinical development or head of medical affairs or medical director doesn’t come to the senior team and says “guess what we learned today?” We’re learning a lot about this disease, and we’re learning a lot about how oncologists are treating their patients. Now we’ve been doing this long enough to understand the impact that it’s starting to have on some of these patients.
Let me give you a couple of examples that probably would not be intuitive to even the best oncologists in the United States.
Yesterday, I learned we received a specimen from a patient who was being treated for pancreatic cancer. The patient was not doing well. FoundationOne was run on this patient’s sample, and it identified a HER2 amplification. Not too many oncologists or scientists would think about testing for HER2 in a pancreatic cancer case. This finding was identified several months ago, and yesterday we learned that the patient has now been on Herceptin for a few months and was responding nicely.
These experiences are the things that keep us going every single day. It’s still anecdotal evidence, though we’re conducting the outcomes studies that we’ll need over the long term and are generating more publishable case studies all the time.
For example, last summer in the Journal of Thoracic Oncology we reported an extremely interesting case of a patient with non–small cell lung cancer who tested negative for EGFR, KRAS, and EML4-ALK. The patient sample was then sent to Foundation Medicine for testing. We identified a novel ALK fusion, which ultimately led to the treatment of the patient with crizotinib. We now have data that are going out 6 months showing the radiographic response, not only the clinical but the radiographic response to crizotinib.
I think it’s safe to say that most oncologists, if not all oncologists, would agree that without the type of approach that was applied with FoundationOne, there is no chance that the second patient would have received crizotinib, and there’s no chance that the patient with pancreatic cancer would have received Herceptin. We are seeing these cases almost every day now that we’ve been doing this long enough to receive patient follow-up.
It’s not the ultimate answer, but it really does give us hope. It gives the oncology community a sense that, collectively, we really are moving this in the right direction. Another patient who comes to mind was in hospice care, and there was another response based on a finding. FoundationOne identified an EGFR mutation in a woman with late-stage breast cancer. That’s counterintuitive. You typically do not see EGFR alterations in patients with breast cancer, yet it was identified, and the patient was started on Tarceva.
So we see these interesting connections. We see these seemingly counterintuitive connections, but it just tells us that we have a lot to learn about oncology and how we’re going to treat patients with cancer. We have a lot to learn about the approach of applying next-generation sequencing via FoundationOne. All of this is going to be an important part of cancer care in the future.
PMO Thank you so much for your time.
Dr Pellini My pleasure.
Molecular subtyping of early breast cancers using MammaPrint and BluePrint allows precise and accurate prediction of the molecular phenotype of the disease, which has the potential to guide selection of personalized therapy if the tests are used prospectively. A retrospective study of 208 tumor samples found that molecular subtyping with [ Read More ]
Dear Reader, Welcome to the second year of publication of Personalized Medicine in Oncology (PMO)! We launched this journal with the goal of providing the practicing oncologist with a translational guide to personalized medicine’s research advances and system improvements that bring targeted treatment into practice, thereby transforming personalized medicine from [ Read More ]