OR WAIT null SECS
What works when and for whom in the era of comparative effectiveness research.
The 2010 US Patient Protection and Affordable Care Act authorized the creation of the Patient-centered Outcomes Research Institute (PCORI) and requires that the PCORI Research Project Agenda "be designed as appropriate to take into account the potential for differences in the effectiveness of healthcare treatments, services, and items used with various subpopulations, such as racial and ethnic minorities, women, age, and groups of individuals with different comorbidities, genetic and molecular subtypes, and quality of life preferences, and include members of such populations as subjects in research as feasible and appropriate."1
This requirement in the statute presents the potential to discover and disseminate crucial comparative effectiveness research (CER) findings that individual patients and their clinicians need for making informed healthcare decisions, especially when there can be known harmful consequences to some patients. Achieving this enormous goal will move our nation closer to knowing what works for whom, under what conditions, and realizing the promise of what's often been termed personalized medicine. Indeed, results from genetic and molecular subtype research have advanced the ability to recommend treatments with the greatest likelihood of success, given that we know a patient's genotype. This important and direct use of genomic data is the view that's normally associated with personal medicine.
However, treatment decisions—and the tools for making many treatment decisions—also utilize non-genomic information on patient characteristics. That is, healthcare providers and systems know race, ethnicity, sex, age, comorbidity status, and other relevant health information for patients. Therefore, a more holistic definition of precision or personalized healthcare can now be considered, one that seeks the promise of knowing which interventions are most effective for which patients and under what conditions. Personalized or precision healthcare also incorporates the personal needs, preferences, healthcare access, and adherence attributes that each person brings to the healthcare encounter and decision-making process, in addition to genomic information, if available. Although PCORI is a US initiative, the need for increased precision in healthcare is important worldwide.
Recently, Amy Abernethy, Felix Frueh, and Jens Grueger presented their viewpoints on the challenges both facing PCORI and for achieving the hope for personalized healthcare from the perspective of a medical oncologist, a pharmacy benefit management (PBM) research leader, and a pharmaceutical developer, respectively. Four relevant questions incorporating each of these professional's viewpoints were answered and discussed, thus ensuring that the order of response for each perspective continually changed. The questions are as follows:
Will healthcare product development embrace the identification of subgroup effectiveness early and throughout the development process?
Abernethy: I'm a medical oncologist caring for melanoma patients. Undoubtedly, subgroup effectiveness is making its way into my clinic, and I see this on two main fronts. The first is in terms of the clinical trials where I try to match my patients by biomarker, by site of disease, or by line of therapy. These trials are increasingly more honed, and this means that I've got more patient characteristics to remember as I try to get the patient into the appropriate trial. Subgroup effectiveness doesn't feel new to me, but the number of things I need to remember about each patient for specific trials or interventions is increasing rapidly.
Second, in the post-marketing setting, I'm seeing subgroup comparative effectiveness and subgroup effectiveness in terms of how I take care of my patients. For example, consider Vectibix® (panitumumab), which was approved in September 2006 by the FDA for patients with epidermal growth factor receptor expressing colorectal cancer with disease progression on or following fluoropyrimidine, oxaliplatin, and irinotecan-containing chemotherapy regimens. To consider prescribing panitumumab, I obviously have to remember a number of things about my patient inherent in the product's approval. Moreover, the number of characteristics that oncologists try to coordinate in tailoring treatments to their patients increases every day. Given that drugs don't get to patients without going through us, information derived from specific subgroup analyses is directly impacting patient care by changing our therapeutic choices.
Frueh: I'll make just a couple of observations about the question itself, which I found quite interesting. First, you ask about healthcare product development, and I think it's important that we don't equate that only with drugs. There are many healthcare products, and therefore many venues that can actually look at tailoring something new to individual patients and identify subgroups. Second, you mentioned effectiveness rather than efficacy. I think that's another critically important distinction. When you're developing a package to submit to the FDA for regulatory approval, you're demonstrating clinical utility, but not necessarily clinical effectiveness. Often, information on effectiveness comes much later, when we actually realize how a new medical product performs in a real-world setting. Third, the question focuses on the development process. If you look at the pipeline, I believe that about two-thirds or so of all drugs that are currently being developed, be that small molecules as well as biologics, have a biomarker associated with them in one form or another. So in that sense, I think that the advent of personalization, or the more precise delivery of care, is currently part of most ongoing development processes.
The next step is the follow-through. Take Effient® (prasugrel) for example, where a lot of subgroup work was been done pre-market, but the follow-on post-market effort hasn't really happened. Opportunities are being identified, but they're not being translated into clinical practice. I think that's an area where we're probably going to see much change over the next few years. What is important in all of this is that we look at medical product development as a continuum that doesn't stop at the point the product comes to the market. We need to identify subgroup effectiveness opportunities as early as possible, even if this sometimes will be in the post-market setting.
Grueger: First, just as a reflection here, I work in the Primary Care Business Unit at Pfizer. In the 1990s and into the 2000s, Pfizer perfected the blockbuster model, yet it is the antipode of what we are speaking about here. Even a company that has worked this way in the past is realizing now that the future will be in better targeting our medicines to the patients with unmet needs. This is not just in highly specialized oncologic areas, either, where we have lots of examples of personalized medicine. I expect this is also coming into the primary care space because we have to face the fact that generics have become the standard of care for the diseases we are looking at. As part of the future of product development in this area, we need to identify the patients who are not effectively treated by the plethora of generic medicines. So my response here is "yes," and we have both risk factor data and genetic data in all of our clinical development programs. We are trying to understand the heterogeneity of treatment response to our medicines at an early stage. That way, we can include these subpopulations in the design of the Phase IIIb trials.
Regarding subpopulations, we have to make choices. It is important to understand the subgroups or strata where there is a response difference. I can confirm Felix's point there. I'm not sure whether two-thirds of the Pfizer portfolio of drugs in development are personalized medicines, but two of our late-stage products have significant genetic subgroup differences. First, critzotinib is an oral anaplastic lymphoma kinase (ALK) inhibitor approved by the US Food and Drug Administration (FDA) for the treatment of patients with locally advanced or metastatic non-small cell lung cancer that is ALK-positive. Second, we have an Alzheimer's disease drug where we know from the Phase II data that the safety profile is influenced by the genetic status of patients; for ApoE4-positive carriers, the safety profile is different from the ApoE4-negative patients. I think the personalized medicine hasn't come as fast as people have thought it would 10 years ago, but I think we're at the point where quite a substantial number of products are coming to market with this personalized information.
What roles do payers and manufacturers have in realizing the hope for personalized healthcare?
Frueh: Before payers pay a bill, they want to know what they're paying the bill for. I think the role of the payer in this context could be as a very rigorous evaluator of how the treatment, intervention, and therapy impacts outcomes. I think that the outcomes-based focus will significantly increase once we have more choices, and can make those choices based on an individual basis. Therefore, I think the role of payers over the next few years will be to discriminate between what works and what doesn't work, and to put incentives in place to make sure that this is being followed, but with a broader view in the payer community. In particular from the PBM perspective, the logistics and infrastructure that we have access to are very interesting. There really is no other place with a hub or central place to access physicians, patients, laboratories, and payers. If we're talking about personalized medicine, all of these groups need to converge in one place to mediate the interaction between all of these stakeholders and provide encouragement for the use of these novel products. For example, we can inform patients and physicians about the availability of new genetic tests and make them accessible. Hence, we have the logistics, infrastructure, and operational backbone that creates a very interesting place to enable precision healthcare and personalized medicine.
It is important to note that this also requires education and decision support. I remember when I was working at the FDA, we developed two online courses for physicians for personalized medicine—one in collaboration with the American Medical Association, and the other one with the American College of Clinical Pharmacology. Despite offering continuing medical education credits, they just flopped because there was no motivational rationale for the physicians to go online and take these courses. What has been missing is the introduction of a teachable moment. When you think about the PBM infrastructure that I was just mentioning, we are creating this teachable moment at the time when somebody comes to the pharmacy to fill a prescription for a drug where a personalized medicine tool, like a genetic test, is available to understand: the best dose, whether the drug is going to work, and/or the patient's risk for an adverse event. Therefore, we can take that opportunity at the pharmacy to inform the physician and the patient about the availability of such a test, and that's the teachable moment that otherwise doesn't exist. Moreover, we can do that on a very large scale. And since payers are responsible for paying the bills and ensuring that medication is used appropriately, you create that incentive at the needed point in time. I think it is this educational and decision support that's what is clicking with physicians.
The last point I want to make is about metrics, because payers, patients, and physicians are interested in long-term beneficial outcomes, as well as monitoring these outcomes. Therefore, I think it's critical that we're not just looking at making a decision at a particular time point and ignoring what happens afterwards, but instead following patients over time and evaluating them at appropriate intervals to see whether or not a particular intervention makes a difference. With this follow up, we can then adjust and optimize the delivery of care.
Grueger: I interpreted the question in a slightly different way than Felix. I've placed my emphasis on the "and" that is between "payers" and "manufacturers." I think we have to collaborate in this area because personalized healthcare treatment strategies will be complicated. In order to be implemented, we need the data that includes the evidence that we have created in the development program, together with incentives and processes that will then be helping us first to implement these strategies in the real world, and then to validate those strategies. This is when we will know the difference between efficacy and effectiveness to determine if these treatments work, and whether the genetic information can determine at all times whether something works or not. In many cases, the problem will be that we don't have 100% sensitivity and specificity for these strategies. We will need to collaborate together in those gray zones where there is some benefit for some patients, but it's not the extreme benefit that we are hoping for from a personalized strategy.
Personalized or precision treatments based on heterogeneity of treatment effect will also show us that the value of what we as an innovator deliver to the marketplace varies across different patient groups. Therefore, it will show how we should deal with a situation where the value may be very high in some patients and lower in other patients. The treatment may still be a good value, and better than what existing treatment strategies offer. This reminds me of the discussions that we have outside of the United States related to value-based pricing, and wondering if we need to solve the problem here in the United States. That is, do we need to say that depending on how the products are going to be used, the reimbursement from payers to the manufacturer varies depending on whether it is a real breakthrough situation? Should there be a smaller premium if it's just a small improvement on an existing therapy? These are complex situations, but I think it is critical that we can solve this in order to then provide the right incentives for implementing and further developing these strategies.
Abernethy: I agree with Felix and Jens that the generation of effectiveness information in the outcome setting is critical to going forward, and that we must try to continuously understand how to hone therapies to determine what works for whom and when. It is also crucial to transfer that information to clinical practice in a way that supports clinical decision-making at the time of the teachable moment. The days of continuing medical education, where I would sit in a room and listen to a lecture, and then two years later I'm supposed to somehow remember what I learned and integrate it to this particular patient on this particular day with this particular targeted agent, are pretty much gone. Therefore, my recommendations are for payers and manufacturers together. So I really focus on generating more of the information in partnership with the clinical community—and with government—about how to understand what works in the outcome setting. We need more pragmatic CER that continuously and progressively defines what works. Specifically, I always think about treatment sequencing when I decide which drug to recommend from my toolbox for this particular patient, especially when we have multiple drugs that target multiple pathways or the same pathway. Demonstration projects supported by payers and manufacturers are needed that help us understand: what it looks like to have personalized medicine in action; where it's not just the genetic test but also patient values; how to communicate these choices effectively; and what are all the core components of healthcare necessary to turn personalized medicine into something that can be done in the clinic.
We also need reimbursement mechanisms, especially to support participation in clinical trials, efforts to facilitate evidence development, and ways to best implement lessons learned in the clinic. As Felix said, it's not just about drugs but targeted strategies in order to achieve personalized medicine; "strategies" requires the testing of how all of healthcare and the process of personalization works in concert, not just one particular drug for one particular target.
On the manufacturer's side, we need to support mechanisms that help clinicians sort through a myriad of signals to be able to match treatments to patients; I think manufacturers need to understand all of the signals that are coming in. We need to support patient values and adherence. We need to conduct continuous post-marketing research. And we need to be ready to constantly update knowledge. Everything in my practice changes around June 5 each year, when the American Society of Clinical Oncology (ASCO) Annual Meeting happens. As soon as ASCO hits, my practice changes, and then it continuously evolves from that point over the course of the year.
For you, at what point does your company engage with payers in product development and look for input to perhaps design a later-stage clinical trial to generate information that is also relevant for the reimbursement part? (Frueh's follow-up question to Grueger)
Grueger: First, even before the products come into Phase I, we develop guidance documents internally to focus the research in areas that would be considered of value from a regulator, payer, and patient perspective, because that's what our researchers are looking for these days. They know that just having some new mode of action is not sufficient. Most important, of course, is when the first proof-of-concept data are available in Phase IIa. I think many companies are now asking for scientific advice from payers at that point in the very same way that we ask for that advice from the regulators; that includes the national payers outside of the United States and the private payers within the United States. It's an interesting experience because this requires new skills, like those at the Medco Research Institute. Both the innovator and the payer will have to predict what information will be relevant in five years' time that will drive reimbursement strategies for personalized medicine.
Are healthcare databases designed to address the challenges of achieving personalized healthcare?
Grueger: No. We have a choice between breadth and depth. Medco can offer data for more than 50 million lives covered in their databases, in some way or the other. Within those databases, there will be some pockets where we have very deep data, such as lab data from genetic labs or from other investigations. I think Felix in his introduction already said the art is to combine these databases, and I think that's where a lot of investment is currently going from healthcare reform. There was even more money invested into health information technology than there was in CER, so hopefully some of that money is really going in the direction to create databases that allow us to collect the information to drive personal healthcare.
Abernethy: Databases are critical, and this is a pretty big issue near and dear to my heart. However, we need more than the ability to look into databases and mine them; we need to be able to use the databases for achieving personalized medicine and personalized healthcare. So my answer is no, because our databases don't talk to each other very well. We lack the standards, the governance, the trust, and essentially the mechanisms to have free and liquid data flow across databases, both to facilitate stratified analyses and understand different interventions for specific patient populations. Moreover, when caring for the individual sitting in front of me, how do I take specific characteristics about this person and have that information be automatically matched into multivariable, prognostic, predictive, and other kinds of models that likely sit outside of my electronic health record and flow back to me, coupled with clinical decision support and other mechanisms? This is going to require a process of free-flowing liquid data that then also coordinates into our data systems and allows for continuous CER. We see this happening in some pockets and in some discreet systems where we understand what the walls of those systems look like, and how information can flow and work together. But thinking about it in a much larger framework has been difficult to date.
Frueh: I have a slightly different opinion. I think current databases can be used for personalized healthcare in certain circumstances. They're certainly not designed for this particular use, but the data that's contained might in fact be very useful to at least create hypotheses that you can then take forward and either confirm or disprove. For example, when the reports came out that Plavix® (clopidogrel) was being metabolized predominately through an enzyme called cytochrome P450 2C19, we didn't really have a lot of genetic information to figure out whether or not the presence of this enzyme translates into different patient outcomes. What we did have, however, was drug-drug interaction information, where we could look at drugs that interact with this pathway and mimic the genetics that we're interested in. So by pheno-copying genetics and looking into longitudinal pharmacy databases—and then outcomes databases to see whether or not patients who were on both drugs have worse outcomes than patients who are only on Plavix—we actually had a proxy for the impact genetics could have. If we consider that interaction as a hypothesis that we could then take forward and prove valid or not valid in clinical practice, I think it's that alternative use that we have to respect when looking at the current databases. That doesn't mean that we have the perfect system. I completely agree that there are new types of databases, more complex databases, and databases in particular that address molecular features of individual patients that we need to look at, but they're being created as we're moving along. I don't think that you're going to go out there and just design one database that becomes your personalized healthcare database for an individual patient. There are going to be various different iterations.
What I'm advocating for is that we look at what we have, to see whether or not there is a possibility to combine some of these datasets and explore opportunities where large genetic data pools exist to see whether or not we can combine them with pharmacy records, for example, and look for correlations so that we don't have to reinvent the wheel. The data is already there, and I don't think we have put a lot of thought into making smarter use of the data we already have.
How can PCORI research related to personalized healthcare be adopted into US policies?
Frueh: I view PCORI research findings as an opportunity to look at novel mechanisms and tools that we can deploy in evaluating whether something is valuable, and then take that information forward into clinical practice knowing whether something makes a difference. This has implications for the larger use of coverage with evidence development. I think what the Centers for Medicare and Medicaid Services have started in certain cases is something that I could see being taken up in the private payer environment as well. It's a good system to get your feet wet and evaluate whether something actually has the potential to work.
I personally have a problem with PCORI leaving economics completely out of the equation if we're talking about comparative effectiveness. You can't just look at one bucket versus the other bucket, and not look at whether or not it's economically feasible to use any one of these buckets in the real world. I think there needs to be a reality check that is based on the economic environment we're working in. And I just find it peculiar that, while we're talking about healthcare reform focused predominately on cost, we're creating a PCORI that leaves the cost part completely out of the equation.
Abernethy: I see the explosion of information and the huge, on-the-ground difficulty of cognitive overload being one of the biggest challenges for personalized medicine at the front line of clinical medicine. In order for PCORI research related to personalized healthcare to be adopted into US policies, PCORI needs to handshake with other parts of the government that are responsible for creating the kind of information systems, informatics standards, governance, etc., to ultimately facilitate the information that's going to support personalized medicine and the information flow. In particular, the Office of the National Coordinator for Health Information Technology, the Agency for Healthcare Research and Quality, and the National Institutes of Health need to be informed by PCORI and vice versa, with frequent conversations around common data elements and CER. I think that the inclusion of industry, as well as payers, is critical to not only this conversation but also the one around the information that's coming out of PCORI.
Second, I think PCORI needs to ensure that the conversation around personalized medicine does not get limited to genetics and genomics, but really continues to be personalized healthcare. That way, it is about all the contexts of patient care, including this person's values, how they're experiencing healthcare, and their own social situations. The message from PCORI should also continue to include important outcomes measures, like patient-reported outcomes.
Third, I think PCORI should work together with payers to develop better processes for coverage with evidence development, pragmatic clinical trials, and other evidence development schema. This is going to be how we learn the best methods to make this happen from an evidence standpoint. And finally, PCORI needs to work within the context of demonstration projects to understand what personalized medicine means within the microcosms of healthcare. It's not just BRAF to BRAF-positive melanoma; it's really BRAF inhibitors in the context of melanoma that's got an appropriate BRAF mutation for a particular person where we've got the right line of therapy, where we understand other issues around this person's tumor, and we understand issues in terms of toxicity over time, as well as how far they live from a place that can give them that particular drug, etc. There are a lot of things we have to do in order to figure out how to take care of this particular person within this particular context, and that requires demonstration projects.
Grueger: Personalized healthcare is central to PCORI; it is patient-centered. I think that's another word in that direction. The starting point is to strive for appropriate databases, and for appropriate standards for these kinds of analyses and research. The next step is that PCORI is funding and stimulating research in these areas, with valid research on the basis of better data than we are currently generating. I think these are the core missions; I wouldn't overload it. If PCORI achieves these two or three things, they will be extremely successful.
Editor's Note: This discussion took place on May 24, 2011, at the International Society for Pharmacoeconomics and Outcomes Research Annual Meeting in Baltimore, MN. Kathleen Wyrwich, PhD, moderated the discussion.
Kathleen W. Wyrwich,* PhD, is Senior Research Leader at United BioSource Corporation, 7101 Wisconsin Avenue, Suite 600 Bethesda, MD, e-mail: firstname.lastname@example.org. Felix Frueh, PhD, is President of Medco Research Institute. Amy P. Abernethy, MD, is Associate Professor of Medicine at Duke University School of Medicine. Jens Grueger, PhD, is Vice President and Head of Global Health Economics & Pricing, F. Hoffmann-LaRoche AG.
1. Subtitle D—Patient-Centered Outcomes Research, Public Law, Federal Register, 111–148 (2010), http://www.pcori.org/images/PCORI_EstablishingLeg.pdf.
*To whom all correspondence should be addressed.