Publication
Article
Author(s):
In late March, FDA converted its accelerated approval for pembrolizumab (Keytruda; Merck & Co) for certain patients with advanced microsatellite instability–high or mismatch repair–deficient solid tumors to a full approval, marking the latest advance for tumor-agnostic therapies.1
FDA broke new ground in 2017 when pembrolizumab, an immune checkpoint inhibitor, received the accelerated tumor-agnostic approval,2 and it seemed the time had come for regulators to evaluate treatments based on their ability to target a given mutation, not just a specific tumor type. Since then, the number of tumor-agnostic approvals has increased (see Table).3 The first targeted therapy to receive a tumor-agnostic approval from FDA was larotrectinib (Vitrakvi; Bayer), in 2018, to treat patients with solid tumors with NTRK gene fusions.4
Nonetheless, the path to approving targeted therapies in cancer based on changes they cause to a certain gene or protein has proved to be less than straightforward, and developing pathways and evidence to support this new concept remains a work in progress. In December 2018, FDA released its broad framework for the use of real-world evidence,5 but nearly 4 years passed between that guidance and FDA’s October 2022 draft guidance, “Tissue Agnostic Drug Development in Oncology.”6
Regulatory approval is only one step—payer reimbursement is a different story. In a white paper published shortly before FDA’s tissue-agnostic guidance,7 the research and data analytics organization IQVIA discussed the “assessment challenges” this new paradigm presents. “The factors that cause uncertainty in evidence packages for decision makers in these cases, primarily small heterogeneous populations, are not typically an issue with tumor-specific drugs,” the authors state. “The situation requires a different framework for generating and evaluating evidence.”
A case study involving the targeted therapy entrectinib (Rozlytrek; Genentech), which, like larotrectinib, also disables proteins of mutated NTRK genes, served as a case study for the IQVIA authors, who examined challenges the drug faced with the European Medicines Agency. Entrectinib received FDA approval in 2019.8 Although the IQVIA paper was formally directed to a European audience, many of the same issues apply in the United States. The authors review lessons learned, which could inform regulators in the United States and Europe as they evaluate therapies for increasingly small populations.7
Use of real-world data (RWD) and development of real-world evidence (RWE), they argue, could significantly reduce the time and cost of such approvals, based on recommendations drawn from a panel IQVIA convened of oncologists, epidemiologists, RWD leaders, statisticians, and representatives from industry. The group identified several issues and challenges, including those that affect reimbursement7:
The IQVIA authors foresee a need for flexibility from all stakeholders: from those designing real-world studies to reimbursement decision-makers evaluating evidence. A “paradigm shift” will be required, they write, that calls for academia, pharmaceutical companies, payers, and others to “collaborate more closely to develop a common framework for evidence generation.”7
To better understand these challenges, Evidence-Based Oncology (EBO) spoke with Emma Brinkley, innovation director of scientific services, IQVIA, and Laura Perez, principal real-world data scientist, Roche, who were among the coauthors of the IQVIA white paper. Brinkley, based in Raleigh, North Carolina, has worked on developing external comparators, using claims data, electronic health record data, and data generated by individuals. Perez, based in Basel, Switzerland, trained as an epidemiologist in the public health arena before coming to Roche. She, too, works on issues involving external controls, with a recent focus on targeted therapies.
This interview has been lightly edited for clarity.
EBO: Can each of you discuss the rise of tumor-agnostic therapies and why the use of RWE will be important in making these treatments available to diverse patient populations?
Brinkley: As a starting point, RWE can help you better understand the natural history of disease, which is really critical to understanding who your true underlying population is. That’s where historically clinical trials have been focused on these large academic medical centers, and you’re often missing a large segment of the population. Once you understand what the true underlying population is, that’s where RWE can help you. You’re able to better come up with a plan to increase access to those patients—whether it’s going to different sites in your clinical trials, whether it’s using real world data—just to study them in an observational way.
That’s where, especially in clinical trials, there’s been a big push by IQVIA, Roche, and others to increase diversity and representativeness in clinical trials. But I think that’s where, starting at that top level, if you can understand who the population is,…the key [lies to] increasing access moving forward.
Perez: I agree with everything Emma said. But just to add [to] that, now that we are in the sphere of more targeted therapies, what we see is that the population available for trials gets smaller and smaller and smaller. And so, what we are facing when we design a trial, [and look at] the timelines and the reality of being able to enroll the patients, if you want to go into the trial in different tumors, [is that] it becomes almost unrealistic. Real-world data is needed to reach representativity in the trial…. It is linked absolutely with the natural history [of the disease], because if there are underlying characteristics of this population, you cannot grasp them anymore. You need to rely on these external sources.
EBO: As we understand who the population is, we know that the FDA came out several years ago with its framework for the use of RWE. There has been a lot of discussion about that framework and subsequent guidance from FDA. I’d like to hear from each of you how you see the term “framework” and what that involves. Is the use of the word framework around RWE broader than what was described by FDA?
Perez: The FDA framework [involves] everything. I will almost say that when we talk about the tumor-agnostic [therapies], it’s a bit of a subset; broadly, we talk about the FDA framework—of how it can help early design, it can help in postmarketing, with safety events. But in the tumor-agnostic area, what I see is the framework accompanying the whole trajectory of the drug, from very early [on] in the process to the discussions with the health authorities.
Brinkley: I completely agree. That was actually my perspective as well. The FDA guidance––and a lot of the guidance we’re seeing coming out––is, by design, quite broad; but they needed to apply [it] across these different-use cases, across therapeutic areas. And so I think what this paper has done, and what the working group did, was take the high-level guidelines, the high-level framework we’re seeing from FDA, and say, “OK, how do we apply this in the context of tumor-agnostic therapeutics?” How do you do that? There are specific considerations, right? If you’re going to group patients based on line of therapy, given the small sample sizes we see here, how do you do that? So I think this [paper] takes the more general guidance we’ve seen from FDA and others and focuses it and makes it specific to this area.
EBO: Who are the stakeholders that are most important to have at the table when we are thinking of RWE and developing it? Emma, would you like to start with this one?
Brinkley: Sure. In this context, certainly the sponsors and the payers are really key. Other stakeholders include academia; [for the IQVIA white paper] we spoke with different therapeutic area experts, other pharmaceutical companies, and the stakeholders themselves, and then specialties within each area. There’s epidemiology, biostatistics, and having clinicians at the table; that’s really important so you get those diverse perspectives. And that’s what I thought was so wonderful about this working group, [that] we brought a lot of those different voices together. There was a really rigorous debate, and I think that’s important.
The other piece that’s important is bringing in the voice of the patient. It’s something I’m really passionate about. I advocate with both my IQVIA hat and in my outside involvement with different rare disease organizations, and I think we are seeing increasing acceptance but also a push to incorporate patient-reported outcomes into research. What is the patient experience living with these conditions? What outcomes would be meaningful to them? How do they perceive benefit vs risk? I think that’s something that benefits the whole process.
Perez: From the methodological pathway, I would say “the usual suspects” that Emma mentioned, but the focus is different. We must include the different data owners and a way to consistently collect what is necessary—not each one collecting what they feel like but collecting data for the purpose of one’s own research or insurance. So there must be a common understanding that unless we collect in the same way, using the same variables, and that we need to answer the same questions, we will never be able to put this together. And so we cannot reach a level of representativeness in any source to really be meaningful. The data owners need to generate ICT [information and communication technology]…. The payers need to participate because we should aim for changes in how we evaluate these drugs to get reimbursed.
As an example, it was very difficult with entrectinib to get reimbursement because there was no mechanism where we could say, “Let’s understand the uncertainties now. And let’s plan some sort of data collection together with a trial. And in one year, let’s reevaluate if…this uncertainty [has diminished].” There was no pathway into that, to do pay for performance. So the payer is the other key stakeholder for me.
And in between, linking that, are the clinicians; they have knowledge of how these patients are treated. What are the important variables to look at? What is important for patients that can guide both the data collection path [and] also [that] of the reimbursement? And when we were working on our paper, the clinicians in the group were absolutely essential to bring us together, to define that it’s line of therapy and tumor types that are essential to know when we want to know something about patients.
EBO: What are the emerging issues in the tumor-agnostic therapy discussion, where the data being discussed will be especially important in speaking to payers—similar to the example mentioned that is highlighted in the paper?
Perez: The starting point is to know…when we start with a drug, are we starting [something] completely new or do we already know something on some tumors? That should define you differentially on your pathway. For entrectinib, it was a very new alteration. Nobody knew anything about the drug. So the bar must be really high, which is totally understandable. But when you talk maybe 2 years from now, when you have not 3 tumors but…7 in your data set, you will know more.
If drugs have already worked in 1 or 2 data sets, the path toward a tumor-agnostic approval should be a bit easier. And here what we face is, there was always a fear—from both the health authorities and the industry—for us to deviate from saying we want to have a tumor-agnostic level, instead of thinking stage by stage. So it can be, we want 1 tumor, we clear 1 tumor; and then we clear another, and then we clear another one. And when you have 3 you can start thinking about a tumor-agnostic approval.
EBO: Emma, did you want to add anything?
Brinkley: I’ll just reiterate what Laura said: You mentioned this balance between deciding whether to try to go after the tumor-agnostic approval upfront or go tumor by tumor. Then, when there’s this larger totality of evidence, it might tip you into a tumor-agnostic label or approval.
EBO: Are there other unresolved regulatory questions, either at FDA or elsewhere, that you would like to discuss?
Brinkley: There is one piece related to this, which Laura spoke to, that involves the data source and the people who own the data. I think data governance and interoperability is a really key area. Building off what Laura mentioned, it’s one thing to have the data in the same format, but how do you get access to the data? How do you maintain patient privacy and data security while also maximizing the usability of the data? And that’s something we saw in the United States with the evidence accelerator, which was led by the Reagan-Udall Foundation that was focused more on COVID-19.8 One of the phrases they used was, “How do you connect the pipes?” In addition to thinking about how do we spend our standardized data at the source, how do we better connect those pipes in a way that maintains patient privacy, maintains really high standards and data security, but also allows us to use these data in really meaningful ways that impact patient lives?
The other piece would be, there [are] certainly no perfect real-world data. There’s no perfect world evidence. And so I think the more examples we can get from FDA or from payers on what they will accept, the more it will give industry confidence in terms of how to move forward with these guidelines that they put in place.
EBO: Laura, do you have anything to add?
Perez: Yes, the interoperability of data sources in this area for safety will be super important. My understanding is that it’s probably in a way easier to do in the United States than in Europe, where we are different countries, with different regulations, compliance, et cetera. This is going to be very challenging in Europe. The other thing is, in working together from very early [on] with FDA or with other health authorities, to define a pathway early [on], instead of being reactive to what we are asked. We can’t bring in a single-arm trial, and have them say, “Oh, do you have all the data?” and then we look for the data.
In this field, there are so few patients around, I don’t think it can work like that. It has to be planned beforehand. And all the efficacy measures—all the measures that will make the drug be OK or not OK—…should be predefined. My wish is more new thinking about how we are going to come to that place in this field, to preemptively prepare everything, and not being reactive to what we see in a single-arm trial.
References