The Role of Real-World Evidence in Regulatory Decision Making
Flatiron’s former CMO & CSO, Amy Abernethy, and Dr. Rick Pazdur of the FDA's Oncology Center of Excellence discuss the path forward for RWE at the 2018 Flatiron Research Summit.
Of note, the conversation touched on one of most exciting applications of RWE: the generation of a real-world external control arm. While Dr. Pazdur emphasized the importance of randomization as the gold-standard approach to eliminating bias, it was clear that there are a number of promising applications of real-world external controls.
Amy Abernethy: So before we get going, I just wanted to stop and say thank you.
Rick Pazdur: Pleasure.
Amy Abernethy: On behalf of the entire oncology community. I don't know if you can tell the commitment from the FDA, Dr. Pazdur and their entire team, they have been here all day. But they have actually been a part of this whole real world evidence conversation for as long as it's been going on. So thank you very much.
Rick Pazdur: Pleasure. And I've also been excited. You have me on the hot seat, right?
Amy Abernethy: Yes.
Rick Pazdur: You could ask me anything you want.
Amy Abernethy: Oh, this is going to be fun. But I want to start. Let's go back. Why did you join the FDA in the first place?
Rick Pazdur: It's very interesting because I'm the only person at the FDA that actually was on the end of getting a non-approval letter from the FDA. So I know what it feels like, okay? I was working at MD Anderson. I'm originally from Chicago and did my training there at University of Chicago and rushed and left Chicago to get my first faculty job at Wayne State which had a big oncology program at that time. And then I went to MD Anderson, where I went through the academic ranks of being a full professor. And I was doing a lot of clinical trials, especially phase one drug studies. And I had gone with several drug companies to the FDA to present data. And I was just impressed about how the interaction was with these two entities, the drug companies and the FDA, and what wasn't being said at many of these meetings and some of the misunderstanding that existed. And I asked myself, who are these people at the FDA? I had never heard of them, okay? They were kind of in this cocoon separated from the oncology world. And when I saw a job that was opening at the FDA to be a director there of the oncology division, I said I want to go there because I have a vision about what this place should be. Okay, and I would be happy to mention what that vision is.
Amy Abernethy: I would love to hear it.
Rick Pazdur: Well at that time the FDA was, as I said, in its own cocoon. And basically I think many of the reviewers got some of their jollies at sticking it to the drug companies rather than working with them. And I really felt that we needed to get A, new people there. Okay? People that had more of an academic orientation. And also to have a much more fluid interaction with the community, the stakeholders. Not only the drug companies, but with the academic world so to speak. You have to remember, no one goes to medical school to work at the FDA. I don't think anyone that has wrote on their application to medical school, I'm going to medical school to work at the FDA. So how do you get and attract people there? And you have to build an infrastructure there that really presents an alternative to academics, basically. So throughout the tenure, almost 20 years that I have been there, the oncology unit has changed tremendously. When I first went there it had about 10 people. 10 medical oncologists. And now we have 100, okay? So a huge increase over a 20 year period of time. But this also echoed what was going on in the pharmaceutical world. Many of you who might be aware of this, 40% of all drug development now of the pharma world is in the area of oncology. We do 50% of all breakthrough therapies. So not only are we pushing out more drugs, but these drugs have significant impacts on people. When I began oncology in my fellowship in 1979 in Chicago, we had about 35 oncology drugs. Every year we now approve, for the past several years, about 15 to 20 new molecular entities. So the field is just going up, up, up, and up, and up.
Amy Abernethy: Just as one small example. Can you comment on your NCI FDA mechanism to bring in new clinicians and oncologists?
Rick Pazdur: Well many of our people are from the NCI Clinical Center. A lot of people like to stay in the Washington area. They have family commitments and we have a big draw from the NCI. But we also have a draw from Johns Hopkins also. We have fellowship programs and encourage people to really float between the two institutions. We have some individuals, one individual at this time, that has a joint appointment between the clinical center and the FDA. But we really promote that interaction between the National Cancer Institute and ourselves, but also with other academic programs. Some of the issues during this 20 year period of time. When I first began, reviewers did every disease. They would do myeloma and melanoma, okay? On the same day. But it soon became readily apparent that we needed to establish disease-specific teams to develop really academic interests of these people and also really to encourage them to interact with the outside world. So we have a series of meetings, small symposiums that the reviewers really work on to work with disease experts. They're called mini-symposium where we invite people in. And my emphasis is really to make sure that our review staff is really developing their careers in this. And really I, at this point in my career, and for the past several years obviously, my major emphasis has been on the development of the staff there at the FDA. Because at the bottom, at the end of the day, really what we're interesting in is really our staff. That is our product, so to speak. We cannot be any better or any worse than the people that are working there. And it makes a tremendous difference if people are actually communicating and working on projects with people and demonstrating the correct degree of flexibility. Sean has mentioned this in several of his talks echoing a statement I always make. There are sins of commission and sins of omission. And when we don't approve a drug that is potentially effective, that probably is a much more serious sin. When we approve a drug that doesn't work, people simply won't use it. If the drug is not approved, people don't have access to that and will never obviously witness the effect of the drug on the American public.
Amy Abernethy: So kind of following along those lines, 20 years. What have you seen change in oncology besides just the number of drugs being approved?
Rick Pazdur: I remember a time when the only drug for the treatment of renal cell cancer was what?
Amy Abernethy: IL-2?
Rick Pazdur: Before that. Megace. Yep. That was the only drug. I remember a time when the only drug for the treatment of CLL was chlorambucil. I remember a time when the only treatment for multiple myeloma was melphalan and prednisone. So things have dramatically changed not only in the treatment of patients for their neoplastic disease, but I think also for supportive care. I don't know when you did your training, but when I was doing my training there were wards full of people with spinal cord compressions, hypercalcemia, you name it, bone complications. And that doesn't exist now because of the introduction of bisphosphonates. Also the antiemetics. When I was beginning my training and through much of the early years, the only treatment that we had for antemetics was Compazine and Phenergan. Tremendously different area. I think one of the important things, and for bringing this up, is one of the points that I really emphasize and one of the things that can come from real world data, is what is the impact that all of these drugs are making? And I always emphasize, we use a large variety of endpoints in the approvals of drugs. Response rates, time to progression, profession free survival, different surrogates or evidence of biological activity many times. But when we have multiple drugs that are approved in a disease, what is the actual impact on that disease? And that's sometimes hard to quantitate. But that's really what the American public wants. I don't think the American public is really questioning and like having a, especially somebody that has far advanced cancer, really cares whether the drug was approved on a response rate or overall survival. They want a drug that is active in their disease. And ultimately they want progress in the disease where when you have a diagnosis, they have a longer life expectancy. And generally that can be done by using multiple drugs. It's not going to be one drug. They're going to be used in different combinations. They are going to be used sequentially. I sometimes ask our staff, okay we've approved in multiple myeloma all of these drugs, okay, you know? I remember when we first approved one of the first drugs in this disease. And now we have approved 10 plus drugs of various classes. And the life expectancy has changed dramatically for this population. So I was asking the staff, well what is now the life expectancy? And they just looked at me blankly. I don't know. And I said well, you know you're focusing on each application and this is again getting lost in the trees and not seeing the forest here. What's actually going on with the disease? And I think that is important. And that is one of my concerns, that we really have to emphasize in the oncology community that we have to demonstrate the appropriate degree of flexibility to get drugs out there that will be used in combination. We have been highly criticized by many people that we should be demanding overall survival for every single drug. And obviously that is not a possible situation in this world for many reasons, which I will be happy to discuss. But if we demanded that, we would really put the brakes on drug development entirely, have unnecessary clinical trials that are done. For example with our CAR T cells approvals with complete response rates in very refractory patient populations, no one in their right mind would be asking for a randomized study for demonstrating overall survival in that population. Likewise for small disease populations such as ROS1 lung cancer, one can't do a randomized study. In situations where you have lack of equipoise because we already know the drug is highly effective, to try to compare it to a known ineffective comparator just to demonstrate a survival endpoint probably is not in the best interest of patients. And really the society has really, and even pharmaceutical companies generally, have had a push and pull in this direction. And a lot of that has to do, and we haven't really gone into any detail in this conference, of we're not developing drugs only for the United States. We're developing for the world. And therefore many of the interactions that we have between Flatiron and the FDA, we really have to be cognizant that we need to bring in our European colleagues. Although we here in the United States have been very flexible in accepting single arm trials. I will tell you the EU has heartburn over this. And a lot has to do with their reimbursement issues with these drugs once they are approved. But we really have to be cognizant of that. One of the things that we have done in the Center of Excellence, in the Oncology Center of Excellence, is really ramp up our international communications with the EU. And we have a monthly telecom with the EU, Swiss Medica, Australia, China. Or not China, but Japan. And really go over all of our applications and discuss common problems that we have. The United States almost always gets these applications before the rest of the world. And we really kind of have the privilege of doing the first look at this. And obviously these other regulatory agencies are very interested in hearing what we have, since our staff is much more greater in number and intensity as far as their review practices.
Amy Abernethy: So kind of thinking about some of the different topics that you just brought on. I think what I might do is take a few things in sequence. Because this is really important. Indeed we've got lots of new molecular entities. And that part's exciting. I think the other thing that you've hit upon is we've got a developing landscape in terms of what do we even think of as evidence generation? Can you comment how you think about totality of the evidence and what you look at as you're thinking about it in 2018 versus what you did in 2005, for example?
Rick Pazdur: Yeah, when I first came to the FDA, I'll just say the word on the street, okay, and when I was developing this drug, was you need two randomized trials and both of them have to show survival advantages. That doesn't happen, forget it. Go eat cake, do whatever you want. The drug ain't going to get approved, okay? That obviously is an impossible situation. So we have to look at, and really what I tell the staff at the end of the day, the ultimate question that you have to ask yourself, is the American public going to be better with this drug than without it? That's the ultimate question. Is there a positive risk benefit here in terms of the population? What is the risk that we're taking with approving this drug? And here again this is a situation where companies have to address risk but we also have to address this concept of regulatory risk. And what I mean by that is when we have an end stage population of patients that do not have any other options available to them, then that degree of risk that we take may be much greater than if we were dealing with a first line indication that has many available therapies that have shown survival advantages. So when one takes a look at the totality of data, we generally have a study, okay, but what are the supporting studies? Okay, and they could be in other disease areas. What is the magnitude of benefit that one is seeing? And what types of clinical trials are we going to be asking for that? For example, if one is using a drug in combination and one sees a spectacularly higher response rate, a response rate of 60, 70%, do we really need a factorial design randomized trial to isolate the effect of that study? Or could we use real world data, for example, of what is the response rate of drug X? What is the response rate of drug Y? If they're like 20 and 10%, then it's obvious what we're doing here. So we have to demonstrate that degree of flexibility. And I think that's only reasonable to do. Unfortunately other therapeutic areas don't face the same situations that we do. And I'm sure many sponsors that are in the room have different interactions with different divisions at the FDA. Obviously one has to be realistic that when one is developing an antihistamine, there should be different expectations than a cancer drug for an end stage population. And it's very interesting. We were writing this guidance. It's called the Effectiveness Guidance, okay? And it's a general guidance of what we look at as terms of effectiveness. Two randomized trials, da da da da. This idea of clinical benefit. And I was saying you know, maybe we need to get a separate guidance on serious and life-threatening diseases and oncological diseases. Because there are different issues here that people face. And I think most people would realize that, that for a benign disease, one is going with a normal life expectancy or a relatively normal life expectancy. One would have different expectations about not only the amount of evidence but the risk benefit, the degree of toxicity. We in oncology have a much different toxicity profile that we accept, including drug deaths. This would drive the people in cardio, renal bonkers if they saw this, the toxicities that we accept. And I generally ask myself, well why do we accept this degree of toxicity? And the standard FDA line is well these are serious and life-threatening diseases, oncology diseases. That isn't really the true answer.
Amy Abernethy: What is the true answer?
Rick Pazdur: The true answer is, well there are many other serious and life-threatening diseases. End stage heart failure is a fatal disease. Let's face it, okay? Why do we accept this in oncology? It's our history. These drugs started out as poisons in the 1950s. Basically, nitrogen mustards. Our patients have grown to accept, rightly or wrongly, that when given chemotherapy drugs, your hair is going to fall out, you're going to get myelosuppression, you're going to have diarrhea, you're going to have da da da da da da. It's almost an expectation. So we have built this in and we have educated the public on this, but that has worked to our advantage and disadvantage. Luckily now we are getting better at control of symptomatic, the adverse effects, as well as developing drugs that have less of these effects, obviously, as we have moved to more targeted therapies.
Amy Abernethy: But I would assume in the same way that you are thinking oncology is different than other therapeutic areas, you highlighted that within oncology it's different, for example, for early stage breast cancer versus a person with late stage disease. Do you want to kind of comment on that in terms of effectiveness?
Rick Pazdur: Well here again, I think we're willing to take a different endpoint. For example, response rates. We're willing to take smaller databases depending on what the trial is. We realize that there is an urgency. We are able to use accelerated approval and ask for more confirmatory studies after the approval on these. So there are fundamental differences that exist between that risk, okay?
Amy Abernethy: So let's talk about potential applications for real world evidence in the near term. Sometime in the next, let's say one to three years. What do you see coming on the horizon of where you think real world evidence is going to be as either primary or supplemental? And how have you been considering preparing for that internally?
Rick Pazdur: Okay, well let me just give you an idea. Because I had this discussion with a drug company and they didn't take me up on this, okay? And they developed their own little clinical trial that they wanted to do here. Because they can't lose control. And I said okay, you have MSI high, this tissue-agnostic indication that we gave a drug company. And I said to several drug companies, why don't you do the following study? And it's really not a study. It's really kind of like a gemisch of real world data, so to speak. I said okay, these patients that have MSI high generally don't have any other therapy available to them. We're talking about patients, colon cancer patients that have progressed, gallbladder, carcinoid patients where there was no therapy, cholangiocarcinoma where there is no therapy. I said, why don't you just go out, make an announcement that you'll give free drug. Then just collect response rate data. Just ask the doctors, did the patients respond? And did they get any toxicities that weren't in the product label, okay? And why don't you just collect the X-rays on the people that responded and have them reviewed? You don't need to collect every single X-ray. Just collect the X-rays that they responded. Because nobody like doesn't call, it's usually, we're always like, response rates go down when we review. They never go up way after the data has been presented at the FDA. It isn't that the people are like hiding responses from us, okay? So like, all I'm interested in is, we have a population of patients that don't have any other therapies. So could we develop systems such as that? And we could use real world data. Some of the information you've already presented here where we're interested in single arm trials, where we want to know is this tumor marker, for example, prognostic or predictive? Are these populations going to be different? What is the adverse events of, augmenting adverse events? What is the activity of a single arm in a combination? One of the drugs in a combination, if we're looking at response rates. Where I think we are going to get into difficulty is with this synthetic control arm. And I do want to talk about that and if it's come.
Amy Abernethy: Tell me more.
Rick Pazdur: Is that the R word has not been mentioned here throughout the day.
Amy Abernethy: Oh yes, what's that word?
Rick Pazdur: The R word is randomization. And one has to remember. Why do we randomize? Randomization takes into account things that we don't know about. That is the whole kit and caboodle of randomization. We don't know everything about a disease. And even though we could do very matched controls, et cetera, and the propensity scores of the control arm, that may not represent an evolving picture of a disease. Let me just give you an example. Two examples. Say we were in the 1960s and were looking at lymphomas. And we just said okay, we're going to treat all lymphomas as it was one disease, basically. Well we know that there is a large cell lymphoma, Hodgkin's disease, da da da da da da da da da da, all with different prognosis. And when you use a synthetic arm, you may very well know and have it very well characterized. But what about the experimental arm? You may have 10 million patients in this arm. But that experimental arm may come from a few institutions that may have very different populations of patients. Where I think this may work out better is where you have an attempt to control the homogeneity of the populations. When you have a targeted disease, for example, ALK positive lung cancer, that may be better than, or BRAF positive melanoma, I'm just using these as examples, where it's a kind of a molecular defined disease, you might have a better handle to control the homogeneity and really mitigate against the heterogeneity of the potential differences in these populations. And we've seen this, particularly in diseases that are poorly characterized such as all of sarcomas, all of triple negative breast cancer, which is not a diagnosis, but is a diagnosis of exclusion of what it isn't, so to speak. So once you get into these very fuzzy diseases, you have a lot of problems. Now that being said, we did a very interesting project just last week.
Amy Abernethy: Go, oh yeah, I want to hear.
Rick Pazdur: I won't give you the data. But it's called, I developed the title of this, okay, Project Switch. And what we did, we took all of the, I think they were PD-1 drugs probably, all of the trials or most of the trials were. But all of the second line lung cancer drugs that had Taxotere as the control, okay? And there were like five or six or whatever it was. And then I told our statistician to switch the arms around. So arm A had the control arm of arm B. Arm B or trial two had arm C and they went all over the place. And then we took a look at the results. They weren't that different.
Amy Abernethy: The hazard ratios don't change?
Rick Pazdur: Not that much. There were some changes. And especially when we combined all of the data together and excluded, obviously, the arm that was in the trial, so we had a much bigger database, the actual estimation of the hazard ratio was closer. But there were a few exceptions where there would be an erroneous decision. But on the most part, it looked like the hazard ratios were all centering around one thing. Which is interesting, but that's not real world data. Remember, these are control arms of a study. And what's interesting, we didn't do any propensity scores on these. No, no, no, no. This was just rapid and dirty, basically. We just switched the arms around. Now they probably had very similar eligibility criteria. But it is of interest. Because really what we're talking about here is not only real world data, but what we really call external data to the clinical trial.
Amy Abernethy: So can I play what I've just heard back to you and make sure that I'm following? So as we think about this idea of external controls, places where you think probably there is greater viability. I heard you say for sure when we can see a response rate of 80%, then basically describing the context of it, it's usually 20%. The second thing that I heard was when we can really in a detailed, refined way describe the population in terms of histology and biomarkers. So for example, ALK positive lung cancer. But on the other hand, really heterogeneous populations such as sarcoma, we probably should be a bit more dubious. And then the last thing that I heard there was that there are some interesting opportunities popping up, such as Project Switch (I like that name), that highlight that we've got a lot to learn and maybe scenarios such as second line Taxotere, which is fairly specific may be an opportunity.
Rick Pazdur: Now it's going to be very interesting if we do that same experiment where we have a lot of studies in multiple myeloma, okay? Where the inclusion criteria of studies is greater than two lines of therapy and you have people that have two lines, three lines, six lines, four lines, five lines, six lines, okay? That's a different population. So it's going to be very interesting to see how, if we do this trial in different diseases, if we're able to get similar results.
Amy Abernethy: So we call that Project Switch Pentalign. Is that right?
Rick Pazdur: Yeah.
Amy Abernethy: So kind of taking all of that, the other thing that I feel like I'm hearing you say is we've got a lot of experimentation to do to figure out how we're going to get this right.
Rick Pazdur: Yeah, and one of the things that we have asked people to do is look at a randomized trial and then also look at the same time, data from your place, so to speak, and see what would we get if we did it contemporaneously, doing the analysis in a randomized trial and we did data from Flatiron? Would we get a similar hazard ratio? I think we have to have much more proof in that type of scenario, in a disease-specific scenario and an aligned-specific scenario before we would just say okay, we're just going to flop in, we don't need to do randomization at this time. Because here again, if you could remember one thing from this talk, okay, that you could carry away is what is the importance of randomization? And that is really the control of variables that you don't know, okay? And that is a very important principle that is missing from many people's lexicon. The other very important thing that we take a look, from a regulatory point of view, is the control of bias. Okay? Sean was mentioning when we started to take a look at X-rays for time to progression and there was a 30% discordant rate. Well when the first time that happened, one of our senior reviewers said oh my god, this is terrible! Look at, there's 30% discordance between the reviewer, or the sites, the doctors that reviewed these and the IRC. Oh, this data is just junk, okay? It's just junk. You know, we can't, somebody did something wrong here. Well then we started taking a look that it wasn't only in this trial, it was dun dun dun dun dun dun dun dun dun dun dun dun. Every single trial had a 30% discordance rate basically. And that's just the business
Amy Abernethy: Right.
Rick Pazdur: What we're doing with these X-rays are basically reading abstract art, so to speak. And it's like well, I think this is this and this is this. When you take a look at the data that you get from these X-ray reports, there is a suggestion of, there is a possible progression here, there may be a new lesion here, a clinical correlation is indicated. It's like what the hell is going on here? So the issue that we decided, it's not about the discrepancy, it's about whether there is a bias present. And that is a systematic bias and that's what we're interested in. Not so much whether there is a discrepancy in reading. Because remember, sloppiness in a clinical trial just obscures the result. End of discussion. It makes your life or the company's life more difficult in demonstrating efficacy.
Amy Abernethy: So kind of following along on those lines, one of the things that you've probably noticed, or at least I felt I was seeing across the course of the day, is this issue around small cohorts. We all thought it was going to be large populations. It turns out we're constantly looking at small groups of patients. How do you think about small cohorts? What do you think we need to be doing to try and both improve confidence in what we're finding and reduce bias?
Rick Pazdur: Well I think we have been very lucky in oncology because we're redefining diseases into these smaller cohorts. Here again, when I started, lung cancer was basically small cell or non-small cell. That was it. Pick your poison, basically. Now we're getting into much, much smaller subsets of patients. And fortunately for us in oncology, which other therapeutic areas don't have, are response rates. Why is a response rate different than any other endpoint? The answer is that it's directly attributed to the drug. Okay, that tumor doesn't shrink because of the natural history, with the exception of a few diseases. And those are far and few between. But that is a direct effect of the drug. When you tell me that a population in a single arm trial has a survival of 10 months, well you might have favored that with different cohorts of patients that may have been prognostically better. For example, if they came from MD Anderson, they would have had to be able to have the financial resources, the ability to get on a plane, the ability to take a taxi, the ability to stand the 95 degree Texas weather, so to speak, with 100% humidity. It does a wonder for your hair, but maybe not for your body.
Amy Abernethy: And you've got to be able to do it.
Rick Pazdur: Yeah. And so there is a lot there. So that is a unique endpoint that we have. And the reason why I'm bringing this up, small cohorts actually are not that difficult for us because we're able to say that okay, the drug has an effect. At the end of the day, what the FDA cannot do is put out a placebo. That is going to distress confidence. Okay, whether the, especially in a life-threatening disease, is there positive risk benefit? That's up to you. Different people have a different opinion of what's benefit, what's risk. We constantly have this discussion, of what is clinical benefit? And you know, it's defined as Sean put it on the slides, is somebody feeling better and overall survival and their functioning, so to speak. But I have taken a much different approach to this. And perhaps this has really helped the field move along. And we are accepting, for example, tumor shrinkage as clinical benefit. Because really, as you know in clinical practice, if you've got a patient and their tumor, they've got this big humongous mass in their chest and it went down, most doctors would say you're benefiting. They're not going to say oh, we've got to like, I don't know about this or this is a surrogate, okay? No, you're benefiting. Likewise if you have a rapidly progressive disease and the tumor is going dun dunk dun dunk dun dunk dun dunk, and you give them a drug and their tumor is not growing anymore, most people would say logically that's benefit to you. And I think we've gotten into this concept of clinical benefit versus benefit. And here again, before I worked at the FDA, I never heard the word clinical benefit. That's not in the lexicon of doctors. The patient benefited. They don't say the patient had clinical benefit. It's they benefited. But we've got ourselves into this kind of square box here. So we've had critics then that say oh no, these drugs should demonstrate a survival advantage. And what people can't understand is that with small populations, you're never going to do a randomized study. It's just impossible to do. But here again, this is an are where real world data can come up with actually augmenting additional numbers of patients, following these patients after approval for giving us further information. Many of these, even though they can't do randomized trials, we ask for post-approval studies to give us more information with greater numbers of patients. And that will give us more confidence in the regulatory decision that we're making.
Amy Abernethy: So let me follow that line of thought for just a second. You've kind of mentioned post-approval studies. You haven't necessarily called them requirements or commitments, but that entire space, how do you see that space evolving, especially with the availability of real world data? And one of those conversations, what do you anticipate that they're going to look like?
Rick Pazdur: They could have many, many forms. One of the things that I feel that we don't do is really study minorities in clinical trials. We have been under great pressure to do that. And one of the reasons why we don't, and this is not the FDA, it's the drug companies, is that these trials are being done internationally. And hence, we simply have increasing numbers of patients that are coming from outside of the United States, particularly from the former Soviet Union, Eastern Europe, et cetera. And hence we do not have representation of Hispanics and African Americans. So I think I would really love to see how these drugs, particularly in diseases that may have significant impacts in racial minorities, such as prostate cancer in African Americans and multiple myeloma, how these drugs actually work in those populations. Likewise in the elderly, okay? I don't think we can commit and delay approvals because people were not studied. That's not going to be realistic to say well we can't approve a drug because we don't have enough patients over the age of X or we don't have enough African Americans or we don't have enough Hispanics or enough Asians in it. But that's information that we could get and I think that's important information for the American public to have. I think what I really would like to see, and I think this will be the future if I had a crystal ball, here again I don't think we could just say we're going to obviate randomization. What I'd like to see is could we have randomized studies that real world data is fed into to promote and escalate the timely completion of these randomized trials. That would be one of my expectations.
Amy Abernethy: Can you give me a couple of examples of pragmatic trials or simple trials that are fed by real world data? Some examples of things that you think need to be addressed in that space early on.
Rick Pazdur: Well as I mentioned, some of these have to do with subpopulations of patients. I think that's an important one. The end organ dysfunction. Dosing. Dosing, especially with the PD-1 drugs, is a nightmare now. And we're getting increasing applications coming in with requests for dosing changes and we're unsure what implications that has. People are not interested in doing, academic people are simply not interested in doing dosing studies. Nobody is going to get tenure and promotion to discover whether the dose of some PD-1 drug is every two weeks versus every four weeks. Let's face it. It's a New England Journal of Medicine paper. But yet it's very important for people to understand. So I think that's important. I think one of the other things is, for us as FDA people, is that how are these drugs actually being used? As you implicated in your other statements. We know that physicians play around with dosing. I did it. Grandma is 80 years old. I'm not going to give her 100 milligrams of Taxotere, okay, and knock her off, okay. So you have to play around with the dose.
Even though that's a part of the label of metastatic breast cancer. You've got to use judgment, so to speak. I think at the end of the day that people have got to realize that what the FDA is about, and that is to get these drugs out, especially with a life-threatening disease, and in an expeditious fashion. And hence we have the data that we have and then build these post commitments to really get more information. I always say that the life history of a drug basically starts with the drug approval. It doesn't end with the drug approval.
Amy Abernethy: So if I share with you a vision, let me see if this aligns with what you're saying.
Rick Pazdur: Is it a vision or a nightmare?
Amy Abernethy: It's a vision. So I'm going to share with you a vision. So a vision that approvals happen earlier. I think we've seen that more and more. But then our responsibility becomes to then monitor how drugs are being used and update that. Is that consistent with how you see things? How would you read it?
Rick Pazdur: Yeah, and I think that is important for the American public to understand, is if we are going to approve drugs earlier on less data, that sometimes there is going to be a mistake. It's kind of like the story of removing somebody's appendix. If you're going to be correct, you're going to have some people that you took their appendix out and it's normal, so to speak. But if we actually demonstrate the correct degree of regulatory flexibility, you're going to have some drugs that are going to have to come off the market. And that's just the reality of the world. Now is the American public ready for that? That's an education process. But they have to understand that there are risk judgments here, as I mentioned before. We have to demonstrate regulatory risk. And that's part of the regulatory risks, that some drugs may need to come off the market. And we've been fairly successful with the accelerated approval program. And I think we will continue to use it. I'd like to see greater flexibility written in where we could not only use it for surrogates, but also in situations, small clinical trials where a survival endpoint has been observed. That may be interesting. But the trials were very small and we have concerns about whether that population when one is studying less than 100 patients, if that's representative of the population being studied.
Amy Abernethy: And so then imagine with me how do you think about real world evidence and the label? Like how is that story going to evolve?
Rick Pazdur: I think the evolution will be first safety. That's a no brainer because we kind of take that already, okay? But then I think we will look at rare populations. Here again, it's a risk base. So think of where the risk is. It's highly unlikely that we're going to give you in breast cancer a first line claim based on real world data on a single arm study with multiple drugs.
Amy Abernethy: Fair.
Rick Pazdur: But it will go in a risk oriented approach. I'll share with you, and I've been thinking about this a long time, is that really the clinical trial is really the litmus test. It's just a test of how the drug will work in the real world. What the FDA, what I am interested in, and what the FDA, if I could speak for the FDA as the box, so to speak, what we're really interested in and should be interested in is how it does work in the real world. Not how it works in the clinical trial. As I said, the clinical trial itself is an artificial situation designed because we didn't have real world data.
Amy Abernethy: Right.
Rick Pazdur: The whole promulgation of the Food and Drug Cosmetic Act occurred in 1962. Long before most of you, probably everyone in this room except me wasn’t born. And I was 10 years old then. And basically clinical trials were new at that time. They were brand, randomized clinical trials, hey, this is high tech stuff here, baby. This is randomizing people. This is like unheard of almost. They did it for polio. But it wasn't a big thing that companies were doing here. And what happened is we took this clinical trial as a test case, a test of how the drug may work in the real world. But really what we're interested in is how the drug works in the real world because who really gives a damn about this clinical trial in an artificial population?
Amy Abernethy: Right.
Rick Pazdur: So I think in the eventual outcome what we will be doing actually, as we get all the pieces together and the technology together, is actually be doing randomized studies in the context of clinical practice. Because that's what we're really interested in. The clinical trial is an artificial construct and a litmus of how we would think the trial might work in the best situation. But it isn't the real world.
Amy Abernethy: It's not the real world. So let me, in the last couple of minutes, imagine with you the world that you want to see in the future. What do you think is going to happen in oncology drug development and what excites you there in terms of both new drugs but also new evidence generation?
Rick Pazdur: I always state that one of my greatest regrets in my life is that I will not be here for the next 40 years. Because I think that this is going to be a tremendous time period of growth and knowledge generation in the field of oncology. My only trepidation is the following. Who's running the show? You know? It's not certainly the National Cancer Institute. The companies have redundant development programs that really are sometimes waste, well not sometimes, many times wasteful. I've been a public critic of the number of PD-1 drugs that are available. One doesn't need umpteen of the same drugs being developed in smaller and smaller indications. But who's really taking a look at the big picture of how we should really develop these drugs and develop and use the national resources not only of patients but also of the dollars that we're spending. Billions and billions. If you take a look at the total of pharma investment in oncology, it's billions and billions and billions of dollars. And how to avoid redundancy and how really to make true real advances in the diseases that we are treating. We had the Moonshot program, for example. And if one took a look at how we got to the Moon, for example, we didn't just say well, let any company like devise a spaceship here and we'll think of some way of getting to the Moon. There was a concentrated and organized approach. Now here again there are fundamental differences with the space program where you're dealing with technology versus basic science and cancer and clinical science and cancer. So there are some differences here. But that does concern me. Like who's running all of these billions of dollars that are going and how can we avoid redundancy?
Amy Abernethy: And so who do you think should?
Rick Pazdur: I don't know. Maybe Amy Abernethy. Queen for the day.
Amy Abernethy: So then kind of as we think about that, one of the reasons that I was even asking this is that I think the other part of this is going back to what you said a few minutes ago. The place that we hold right now in the US is twofold. We have a responsibility to the US as our constituency and our population. And I think that you said something really important about our responsibility worldwide. Kind of expand for me how you see that and where you think we should go with that.
Rick Pazdur: We have to realize that we are not the sole player, okay? And that's hard for many Americans. Especially with the slogan of our present president of Make America Great Again, so to speak. I have much more of an international approach to this. I lived in, I mentioned I lived in Detroit during the 1980s when the car industry was going woomf. Nobody would buy a US car. And they were all made in Detroit. Now the US companies have realized that this is a global market, okay? And these cars are made in multiple areas of the world and assembled and all the parts are assembled. And we have to be realistic that we are living in an international world. The companies that we're seeing are international in scope, okay? And they have to answer not only to the US. They have to answer to the EMA, they have to answer to Swiss Medica, Canada, and most importantly, that's coming up strong, and I spent two weeks there, China. China is a growing power and a power that is going to be a big player in the pharmaceutical world, I firmly believe. When I was there and I spent a significant period of time there interacting not only with the regulatory authorities in China, but also with the investigators. The major question that was asked to me, will the FDA take only Chinese data? Not once, 10 times.
Amy Abernethy: What was the answer?
Rick Pazdur: Yes. We will take data from wherever it is. It has to be applicable to the US population. But you could see there is a billion plus people there. The use and the design of clinical trials and the use of medications can be unprecedented in what they can bring to us. We also have to be realistic. We are, and I emphasize this to our staff when we are having these debates about whether we should take data from Pacific Rim countries and how applicable it is, I said you know, we are a nation of immigrants. We have people directly from these countries that are US citizens. So it's not that we are excluded, that we should exclude these. They are part of the American fabric, so to speak. So I think internationalization is not to hide ourselves away from this. It is very wrong to do. It's interesting. When I began, and here again I don't want to go back that far in history. Yeah, okay, it's the 70s, okay. Okay. Oncology in the rest of the world was like nonexistent. Okay? When I would go to ASCO meetings, there would be several posters from some countries in Europe and it would be kind of small, poorly done trials. Now the quality of data, the thought processes that go on in the clinical investigation community there are as equal to the United States if not better. And to think that we are not, that we have a monopoly on this is ridiculous in my point of view. It is really a worldwide problem. And we're going to be far better working together with all of the components. And that's why I spent, for example in the OCE, so much time in really developing international contacts. And again, I have a great deal of interest in China. In fact we've been mentioning to the AACR to actually do an Asian track to talk about what's going on in China. Because what we really have now is many of the Chinese nationals that left China to take educational opportunities here in the US have gone back to China and are establishing companies and developing their own PD1 drugs, et cetera. And here again they may be bringing these into the United States for regulatory submissions. And they already have a development plan that have been laid out by the formal approvals of these drugs.
Amy Abernethy: So interesting. Well with that I want to say thank you from the bottom of my heart. And thank you from the entire oncology community for everything.
Rick Pazdur: Thank you.
Flatiron Research Summit
November 7, 2018
- Amy Abernethy, MD, PhD — Former Chief Medical Officer, Chief Scientific Officer & SVP Oncology at Flatiron Health
- Rick Pazdur, MD — Director, Oncology Center of Excellence at FDA