FDA last week issued a draft guidance that urges sponsors to conduct randomized controlled trials when they seek accelerated approval.
The document, titled Clinical Trial Considerations to Support Accelerated Approval of Oncology Therapeutics, is consistent with the agency’s previously stated advice cautioning against reliance on single-arm trials (The Cancer Letter, Sept. 23, Nov. 4, 2022).
In the draft guidance, FDA suggests one of two pathways sponsors can select as they design studies intended to support an accelerated approval:
- Two randomized controlled trials, one with an early endpoint, such as response rate, and a second powered for a longer-term endpoint such as progression-free survival or the gold standard, overall survival.
- One randomized controlled trial, dubbed the “one-trial” approach, which supports accelerated approval while simultaneously testing longer-term endpoints. “One-trial approaches” include follow-up to verify clinical benefit.
According to the draft guidance, single-arm trials have most commonly been used to support accelerated approval in oncology. These studies come with significant challenges, including small safety datasets, low-magnitude response rates that may not be reasonably likely to predict clinical benefit, and inability to establish differential contribution of effect for combination regimens.
“Most of the accelerated approvals in oncology have been based on single-arm trials,” Lola Fashoyin-Aje, deputy director of the Division of Oncology 3 in the Office of Oncologic Diseases at FDA, said to The Cancer Letter. “Largely reflecting the historical practice in oncology to evaluate new drugs in more treatment-refractory settings where standard of care may not have been available. However, the accelerated approval pathway is not synonymous with single-arm trials, and there should be a natural evolution with respect to study design toward randomization, as more drugs become standard of care.
“Our hope is that we will have fewer of these single-arm trials moving forward.”
In the document, FDA acknowledges that single-arm trials are in some cases appropriate. In some settings, a randomized control trial is not feasible. Here, examples include the case of dostarlimab in the curative-intent setting for patients with mismatch repair-deficient or MSI-H locally advanced rectal cancer (The Cancer Letter, Feb. 10, 2023).
The draft guidance specifies explicit expectations for the conduct of single-arm trials.
“What we outline in the guidance is that part of the consideration regarding whether a single-arm trial will be acceptable depends on many factors,” Fashoyin-Aje said, “The prevalence of the disease, the available therapy landscape, what is known about the natural history of the disease, etc. Therefore, the focus should always be patient-centeredness and determining the best approach to generating evidence in the specific context to reduce the risk to the patient, rather than expediency.”
The draft guidance comes on the heels of the December 2022 Food and Drug Omnibus Reform Act, which was enacted as part of the Consolidated Appropriations act of 2023. FDORA grants FDA “the authority to require post-approval confirmatory studies to be underway at the time of accelerated approval or within a specified period of time after approval, to report on the status of post-marketing studies, to update withdrawal procedures, and requires FDA to form an accelerated approval council,” Fashoyin-Aje said.
For the two-trial approach or the single-arm approach, FDA urges drug sponsors, not for the first time, to have fully enrolled their confirmatory trials at the time of accelerated approval.
“The sooner we initiate the trial,” Fashoyin-Aje said, “The quicker we can have assurance of the drug’s benefit or the quicker we can withdraw marketing authorization if needed.
“We have to remember that the patients who receive the drugs are the ones assuming the risk. And what we are emphasizing is that we need to shorten that window of uncertainty. We owe that to our patients. The new authority granted by FDORA will now necessitate agreement with FDA on the timelines for initiating these trials.”
Fashoyin-Aje spoke with Jacquelyn Cobb, a reporter with The Cancer Letter.
Jacquelyn Cobb: Why is this draft guidance needed right now?
Lola Fashoyin-Aje: Thank you for the opportunity to discuss this new draft guidance.
As you may be aware, the accelerated approval program has been under scrutiny for several years now. The 2020 approval of an Alzheimer’s drug brought additional visibility to many of the issues that critics of the program have highlighted for years.
Approval through the accelerated approval pathway is common for oncology drugs and this pathway has really accelerated access to new therapies for patients with cancer. So, we have gained a lot of experience in the use of this pathway, and this has informed our efforts to improve use of this pathway to better serve patients with cancer.
We have been very transparent about what we consider to be the pitfalls of the current approach to the use of the accelerated approval pathway for oncology drugs, including the over-reliance on single-arm trials, the almost knee-jerk approach to developing new drugs in the refractory setting, and the lack of a comprehensive approach that includes early initiation of the trial that will provide definitive evidence of clinical benefit rather than delaying it until after a marketing application for accelerated approval is submitted or approved.
So, the guidance provides recommendations that address some of these pitfalls, focusing on the operational aspects of these concepts. The timing of the guidance in relation to the legislative efforts around this issue was fortuitous and not at all by design, but we were glad to see alignment between what we have been emphasizing in oncology, and the new legislation which addresses the accelerated approval program more broadly.
Could you describe what the new legislation provides for?
LFA: Sure. The Food and Drug Omnibus Reform Act or FDORA, was enacted in December 2022 as part of the Consolidated Appropriations act of 2023, provides for amendments to the Federal Food, Drug, and Cosmetic Act, including provisions to modernize the accelerated approval program.
This includes granting the FDA the authority to require post-approval confirmatory studies to be underway at the time of accelerated approval or within a specified period of time after approval, to report on the status of post-marketing studies, to update withdrawal procedures, and requires FDA to form an accelerated approval council.
Going back to the guidance, what are the main takeaways for clinical investigators and drug companies?
LFA: There are a few takeaways. The first is that we consider randomized controlled trials to be the gold standard for being able to provide the data that is needed to make a benefit-risk determination, even in the context of an accelerated approval submission. This is because this design allows a within-trial comparative assessment of the new drug to a standard of care, and thus decreases the likelihood that factors other than the drug’s effect are responsible for the observed differences in outcomes.
The focus should always be patient-centeredness and determining the best approach to generating evidence in the specific context to reduce the risk to the patient, rather than expediency.
Most of the accelerated approvals in oncology have been based on single-arm trials, largely reflecting the historical practice in oncology to evaluate new drugs in more treatment-refractory settings where standard of care may not have been available. However, the accelerated approval pathway is not synonymous with single-arm trials, and there should be a natural evolution with respect to study design toward randomization, as more drugs become standard of care.
We also want to encourage the investigation of new drugs or approved drugs in new indications, earlier in the course of the disease. We have discussed this in the context of an OCE initiative called Project FrontRunner. The paradigm that has been followed for a long time is to introduce new drugs after patients have exhausted other lines of treatment. This approach does not really serve patients.
This is particularly the case when investigating drugs that may provide unprecedented treatment effects and have better safety profiles. Such drugs which reflect the advances in oncology, are essentially out of reach for patients until they have progressed on several therapies. By the third or fourth line in some diseases, most patients may have become ineligible to receive these new treatments either because of accumulated treatment-related toxicity over time, and/or because the disease itself has progressed such that they can no longer receive the treatment.
The guidance also highlights the importance of early discussion with FDA when the sponsor is developing the clinical developing program and at key milestones when the preliminary data may suggest that early access to the drug through the accelerated approval pathway would be reasonable.
There are important considerations about trial design, clinical endpoints, the approach to verifying benefit, etc., that should be discussed early on and throughout the clinical development program. Importantly, the guidance emphasizes early initiation of the confirmatory trial.
What does this draft guidance mean for oncology drug sponsors who are currently applying—or plan to apply—for accelerated approval using a single-arm trial?
LFA: We have been discussing the concepts outlined in the guidance with sponsors for quite some time and so, our hope is that sponsors are heeding our advice regarding the use of single-arm trials.
As we always do, we will consider the data package for those trials that have been completed and determine whether the data support accelerated approval. We recommend that sponsors initiate early discussions with the FDA regarding their development plans, particularly when their plans include an accelerated approval submission strategy.
When are single-arm trials most appropriate or acceptable? Is it only for rare cancers and orphan drugs? What about cross-trial comparisons to historical trials? What is an ideal scenario in which—or general rules to ensure that—historical data is justified and adequate for proving clinical benefit?
LFA: We are emphasizing a preference for randomized controlled trials with analysis of intermediate clinical endpoints like response rate as the basis for an accelerated approval submission.
However, we do recognize that single-arm trials may be appropriate in some settings, but this is not cancer-specific or treatment line-specific or based on orphan drug designation. When a sponsor proposes a single-arm trial for accelerated approval, implicitly, the comparison is to a historical control or to a placebo.
What we outline in the guidance is that part of the consideration regarding whether a single-arm trial will be acceptable depends on many factors including, the prevalence of the disease, the available therapy landscape, what is known about the natural history of the disease, etc. Therefore, the focus should always be patient-centeredness and determining the best approach to generating evidence in the specific context to reduce the risk to the patient, rather than expediency.
In the guidance, we describe the limitations of cross-trial comparisons including evolution in treatments that may render a population enrolled in a historical trial markedly different than the population enrolled in the new drug trial.
In some cases, we just do not have the data that characterizes the effects of standard of care therapy for the indicated population or the data could be very limited. Beyond standard of care for the treatment of given disease, there may also be differences in the concomitant medications that are administered to manage certain toxicities.
There is no ideal scenario, per se, where comparison to a historical control would be considered better than conducting a randomized controlled trial. However, there may be scenarios wherein it is acceptable to conduct a single-arm trial due to feasibility issues, for example, in the context of a very rare disease, a high response rate based on early data which raise concerns about equipoise, and other considerations.
At the February 9, 2023, meeting of the Oncologic Drugs Advisory Committee on the dostarlimab program for dMMR/MS-H locally advanced rectal cancer, we discussed a situation wherein the community felt it would not be possible to do a randomized controlled trial due to the high clinical complete response rate observed in the preliminary data. So, we may exercise regulatory flexibility for programs that need to deviate from the randomized investigation.
What is the current proportion of accelerated approvals applications that are green-lighted based on single-arm trials?
LFA: We have published on this. In a report we published in JAMA Oncology last December describing the drug and biological product approvals for malignant hematology and oncology indications reviewed by the Office of Oncologic Diseases over a 10-year span, 176 of the 563 approved indications, or 31%, were based on single-arm trials and of these, 116, or 66%, were for accelerated approval. Our hope is that we will have fewer of these single-arm trials moving forward.
In an analysis of all accelerated approvals granted in oncology over the past decade, FDA found that about 12% of oncology accelerated approval indications have been withdrawn from the market. Is this rate high or low, in your opinion? Or should it be higher?
LFA: When we grant AA, we do so based on our assessment that the effect on the surrogate or intermediate clinical endpoint is an improvement over available therapy that is reasonably likely to predict clinical benefit. Therefore, at the outset there is recognition of the risk that the effect on intermediate endpoint may not lead to a benefit based on a definitive endpoint of clinical benefit like overall survival. We make a determination based on the available data, but this is not a fail-proof assessment at the time of accelerated approval.
We would like to get it “right” 100% of the time, but it is not a realistic goal, and such a goal may actually run counter to the purpose of the accelerated approval program as this would require that we adopt too conservative an approach and not “on-ramp” a drug until we had much more data.
What do those 12% of withdrawn indications have in common in terms of study design? Were they lacking in demonstrating causal inference?
LFA: I don’t think there is a discernible pattern, and if there were one, I am not sure we would attribute the withdrawal to the design of the trial. The post-mortem for a trial intended to verify the clinical benefit of a drug that is granted accelerated approval is no different than what it would be for other trials. Trials fail for many reasons, but the design of the trial I think would not be at top of the list of reasons.
What are other common pitfalls or shortcomings that you encounter in sponsor applications in the accelerated approvals pathway?
LFA: I think the biggest pitfalls are the ones I mentioned earlier and that we described in our NEJM paper: the over-reliance on single-arm trials, the focus on refractory settings, and not initiating the trials intended to verify clinical benefit early enough to shorten the time of uncertainty regarding the benefits of the drug.
We ultimately want to provide early access to drugs but in so doing, there is inherent risk. Granting accelerated approval is a step along the continuum that ultimately gets us to a traditional approval. Therefore, we want to minimize the risk to patients and the risks are those associated with the safety of the drug and those associated with potentially having an ineffective drug on the market.
As we noted in our commentary published in The New England Journal of Medicine last September, a comprehensive strategy is necessary to ensure that at each step of the process, we can reduce the risk to patients.
In your related NEJM commentary, you explain that the difference in median time to off-ramp action was longer if the confirmatory trial was initiated after approval, and that “this difference was most striking among withdrawn indications, with a median time to withdrawal of 3.8 years if the confirmatory trial was ongoing at the time of AA, as compared with 7.3 years if such a trial had not been initiated. “Delayed withdrawals in this latter scenario represent the greatest risk to patients.” Can you speak more about this?
LFA: Yes, certainly. When there is a delay in initiating the confirmatory trial, we essentially extend the time period for when we have a drug on the market, but we have residual uncertainties regarding its clinical benefit—that is, its effects on survival. The sooner we initiate the trial, the quicker we can have assurance of the drug’s benefit or the quicker we can withdraw marketing authorization if needed.
We have to remember that the patients who receive the drugs are the ones assuming the risk. And what we are emphasizing is that we need to shorten that window of uncertainty. We owe that to our patients. The new authority granted by FDORA will now necessitate agreement with FDA on the timelines for initiating these trials.
Does FDA have a preference for a “one-trial” design or the two-trial approach? What are the advantages of having one protocol for both accelerated and regular approval?
LFA: We do not have a preference, and whether a sponsor elects to do one trial or two separate trials is really at their discretion.
It also really depends on the clinical context. In the guidance we recognize the efficiency in having one trial to support the accelerated approval with follow up to verify clinical benefit, but it may not be possible in every scenario to do this, particularly if there aren’t great intermediate clinical endpoints to rely on or the effect of the drug on that intermediate clinical endpoint is small.
With the one-trial model, we do want sponsors to ensure they do not compromise the analysis of the endpoint of clinical benefit as that is ultimately what we want to see.
Did we miss anything?
LFA: No. Thank you for the opportunity to discuss the draft guidance.
