FDA’s oncologic advisors support multi-arm trials for multi-phase regimens

Quick background: treatment approaches in oncology

• Cancer treatment is complex and individualized, with variations in course dependent on the cancer type, stage, biological characteristics (e.g., genetics) and more. The primary or main treatment for some cancers can be surgical resection.
• Combination therapies administered before or after this primary treatment play an important role. These can include chemotherapy, radiation therapy, immunotherapy, and hormone therapy, among others. Neoadjuvant therapies refer to initial treatment administered prior to the primary treatment to increase the chances of its success. For example, these could aim to shrink the tumor size. Once the patient has received the primary treatment (often surgery), adjuvant therapies aim to lower the chance of cancer returning. Perioperative approaches involve therapies given both before and after surgery.

On July 25, 2024, FDA’s Oncologic Drugs Advisory Committee (ODAC) convened to consider a perioperative treatment for non-small cell lung cancer (NSCLC)

• The advisors met to “discuss the supplemental application to add the regimen of IMFINZI in combination with chemotherapy as neoadjuvant treatment, followed by IMFINZI as monotherapy after surgery, for the treatment of adult patients with resectable non-small cell lung cancer (NSCLC).” The meeting also had a second, broader aim: to judge whether future clinical trials should be designed to evaluate the effect of perioperative therapy in both the neoadjuvant and adjuvant phases in patients with resectable NSCLC.
• AstraZeneca’s IMFINZI (durvalumab) is an immune checkpoint inhibitor (ICI) that targets Programmed death-ligand 1 (PD-L1). Programmed Death (PD) protein 1 and PD-L1 are proteins found on T cells that help the body check its immune response. When these two proteins are bound, it signals the immune system to stop attacking the healthy cells. According to the American Cancer Society, “monoclonal antibodies that target either PD-1 or PD-L1 can block this binding and boost the immune response against cancer cells.”
• The proposed indication for resectable NSCLC represents an area of unmet need. NSCLC remains a leading cause of cancer-related mortality in the U.S., with an estimated 5-year survival rate of just 26.4%.
• Some additional regulatory history: Durvalumab was first approved in May 2017 via the accelerated approval pathway to treat “patients with locally advanced or metastatic urothelial carcinoma who have disease progression during or following platinum-containing chemotherapy or have disease progression within 12 months of neoadjuvant or adjuvant treatment with platinum-containing chemotherapy.” This original indication was removed following failure to meet its primary endpoint in its confirmatory trial, part of a cluster of withdrawals of drugs that target PD-1/PD-L1. However, durvalumab received approval for additional indications in the intervening years and is currently labeled for five different indications, including three lung cancer indications. These include for certain patients with (1) late-stage, unresectable NSCLC, (2) metastatic NSCLC with certain genetic features, and (3) extensive-stage small cell lung cancer (ES-SCLC). The product’s non-lung cancer indications are for certain types of biliary tract and liver cancer. [ See AgencyIQ analysis for additional regulatory history.].

The resectable NSCLC development program featured positive results when comparing two treatment arms: the perioperative regimen and placebo

• The Phase 3, randomized, double-blind, placebo-controlled, multi-center AEGEAN study supports the supplemental application for the resectable NSCLC indication. The researchers randomly assigned 802 previously untreated patients “to receive platinum-based chemotherapy plus durvalumab or placebo administered intravenously every 3 weeks for 4 cycles before surgery, followed by adjuvant durvalumab or placebo intravenously every 4 weeks for 12 cycles,” according to results published in The New England Journal of Medicine in October 2023. The chemotherapy options included both cisplatin- and carboplatin-based treatments.
• The AEGEAN study had positive results for its primary endpoints: event-free survival (EFS; evaluated in a blinded fashion by independent central review) and pathological complete response (pCR; evaluated centrally), defined as absence of any viable tumor cells. The EFS duration was significantly longer in the durvalumab group than the placebo, with a stratified hazard ratio for disease progression, recurrence, or death of 0.68 [95% CI 0.53 – 0.88, P=0.004].
• Overall survival (OS), a key secondary endpoint, was assessed following AstraZeneca’s provision of updated safety and OS data during application review. With the increase in maturity, the data showed no detriment to overall survival. AstraZeneca described these results as “a trend toward improved OS favoring the treatment arm.”
• In the publication, the researchers expanded on their hypothesis behind the treatment regimen. “Perioperative regimens that combine the benefits of neoadjuvant and adjuvant immunotherapy could further improve long-term outcomes…by priming antitumor immunity while the primary tumor and lymph nodes are present and eradicating residual micrometastases both before and after surgery,” the introduction states.
• During the ODAC meeting, AstraZeneca’s Global Clinical Strategy Head for Lung Cancer LEORA HORN claimed that “In 2018 when AEGEAN was designed, the optimal trial to determine contribution of phase was not well characterized and has since become a focus with readouts of studies in early-stage disease.”

At the meeting, FDA reviewers described the two-arm study design as a “major review issue”

• FDA Clinical Reviewer BERNARDO GOULART reviewed the efficacy results, saying “AEGEAN is therefore a positive trial and met the two of the dual primary endpoints. This is not the point of today’s discussion.”
• For trials containing multi-phase regimens, a two-arm trial precludes the assessment of the individual contribution of each phase and increases the risk of overtreatment. In other words, the results do not make clear if patients need both phases of therapy. This “intensification of treatment” can expose patients to unnecessary therapy, which increases both the treatment burden and risk of toxicities. According to FDA, a meta-analysis of 28 randomized trials in solid tumors showed a greater incidence of severe toxicities associated with adjuvant ICIs.
• In two formal meetings with AstraZeneca, FDA raised concerns about the two-arm trial design. First, during the End of Phase 2 meeting in November 2018, FDA stated that “the proposed study design will not isolate the effect of neoadjuvant therapy.” Instead, the agency recommended the sponsor consider an adaptive or factorial study design.
• The sponsor opted to proceed with the two-arm study design. At the May 2023 Pre-BLA meeting, the agency again pointed out that the “design does not isolate the effect of neoadjuvant durvalumab with chemotherapy from the effect of adjuvant durvalumab monotherapy” and requested that the sponsor provide a method to assess the individual phases. “With the lack of within-trial comparisons of treatment phases, FDA has had no choice but to resort to cross-trial comparisons to refer the contribution of the new adjuvant and adjunct phases to the effects of the perioperative regimen, despite substantive limitations” of this methodological approach, said Goulart.
• Importantly, FDA has previously granted approval to entire perioperative regimens based two-arm designs like the AEGEAN trial. Reviewers pointed to two examples, one in breast cancer and one in the same disease setting of resectable NSCLC. The latter approval, granted in October 2023 to Merck, was based on the KEYNOTE-671 trial, which had positive results for both EFS and OS. However, FDA cautioned against making cross-trial comparisons to compare efficacy.
• FDA pointed out that “emerging data suggests further consideration is warranted.” Specifically, FDA pointed to a June 25, 2024, press release from AstraZeneca itself that provided updates on a phase 3 trial of durvalumab for early-stage NSCLC in the adjuvant phase alone. The announcement suggested lack of benefit in this setting, as the trial did not reach significance for its primary endpoint of disease-free survival (DFS) versus placebo.
• While AstraZeneca proposed to study the regimen’s components through cooperative group trials in the post-market setting, FDA expressed doubts on the rigor and feasibility. The two proposed trials, PROSPECT-LUNG and CLEAR-INSIGHT, are “designed to address specific questions related to contribution of phase but not the full issue.” Further, the timeline for completion of these trials makes it possible that the fast-moving field of oncology may have a new standard of care before they are done.

The future is multi-arm: FDA articulated its vision for perioperative trials

• According to the agency, trials should isolate and formally assess the contributions of treatment phases to efficacy results. While four-arm factorial designs are ideal, three-arm designs can also appropriately isolate contributions. According to the FDA’s presentation, all patients in either design would receive standard-of-care therapy throughout. A four-arm design would randomize one cohort of patients to receive the investigational drug both before and after surgery; one group to receive the drug before surgery, but not after; one group to receive it after surgery, but not before; and one group to receive standard-of-care only. The three-arm design would drop the adjuvant-only arm.
• FDA also stated its openness to alternative design such as adaptive designs and rerandomization post-surgery. In these designs, however, a modest observed efficacy difference between arms means that perioperative regimens are unlikely to be approved.
• The agency acknowledged that these designs do necessitate large sample sizes. Depending on anticipated effect sizes, FDA explained that three-arm trials can require 650-1700 patients, while four-arm trials can require 960-2400 patients. Nonetheless, these sample sizes are similar to that of both the trial under consideration (802 patients) as well as other completed trials.

ODAC members sought clarification on why AstraZeneca proceeded with the two-arm trial and discussed long-term implications of the lack of clarity

• Off the bat, ODAC Chair DANIEL SPRATT from Case Western Reserve asked why the firm did not comply with FDA’s 2018 request to design a study that could better understand the contribution of phase. He put forth some potential reasons, asking, “Is this because of cost? Is this time? Is this to maximize drug exposure? Is this lack of agreement with the FDA that this matters, if the trial ultimately is positive? Is this you don’t feel it’s your responsibility, or is it because other trials didn’t do this, so you don’t think you need to, or some other reason?”
• AstraZeneca’s Horn explained that the 2018 decision was made “following the science of what we knew about immunotherapy at the time,” referencing some early results that suggested perioperative regimens with similar immunotherapies were beneficial. Trial size and duration concerns also factored into the decision. A biostatistician from the firm explained that a four-arm trial would have required over 2,500 patients and would not have been powered to read out until late 2027.
• FDA’s ERIN LARKINS, Acting Director of the Division of Oncology 2 (DO2), pushed back on this reasoning, commenting that while the four-arm trial design would have been ideal to isolate phase contributions, the agency supported other appropriate designs with lower feasible sample sizes.
• University of Louisville thoracic surgeon VICTOR VAN BERKEL questioned how design concerns factored into the October 2023 approval of a perioperative NSCLC treatment based on the two-arm KEYNOTE-671 study. Larkins noted that KEYNOTE-671 demonstrated an overall survival benefit. While this does not mitigate the inability to parse the contribution of individual phases, she stated that her group believes it is fair “for that to be taken into the risk benefit consideration, potentially for an overall risk benefit of a regimen.” The data regarding phase contributions that emerged from trials that have read out following this approval is “the even bigger issue,” she explained.
• Later, OCE Director RICK PAZDUR spoke to this conundrum, saying “I also want to emphasize this issue about the prior approval and just be blunt — we are not a victim of our past action, so to speak. And I want to make that quite clear to everybody, that was then, and this is now. And we have new information that has come out, and we have to evaluate the situation at the current time.”
• The advisors asked for more perspectives regarding the risks of toxicity associated with potential overtreatment. AstraZeneca’s Horn critiqued the meta-analysis presented by FDA that suggested increased risk of toxicity. “It’s with studies that started a long time ago,” she said, “And with immunotherapy, when those drugs were first coming out, we didn’t understand how to manage them as well as we do today, 15 years later, with those drugs in the clinic.”
• PAMELA KUNZ, a gastrointestinal oncologist at Yale, raised the issue of “financial toxicity,” saying that “certainly a longer period of treatment for patients poses risk of lost time at work and additional issues.” Goulart set a boundary against this line of inquiry, saying that “The FDA typically does not comment on financial aspects of cancer care.”
• Similarly, as ODAC members sought to understand how a perioperative regimen fits into “standard of care,” for NSCLC, FDA steered clear of taking a stance. Goulart explained that FDA does not dictate medical practice and considers standard of care to be the regimens approved based on safety and effectiveness. Pazdur later doubled down, saying “The issue here is—what the FDA does is approve a marketing application. End of discussion.”

The discussion and the vote

Discussion Question: In light of the uncertainty around the need for both phases of treatment, discuss whether an additional trial should be conducted to clarify the contribution of treatment phase for the durvalumab perioperative regimen prior to approval.

• CHRISTOPHER LIEU, of the University of Colorado Cancer Center, led off the discussion by focusing on the positive results of the AEGEAN trial. While FDA does raise a valid point about the importance of discerning the contribution of each treatment phase to the outcome, he said, such a study could potentially take as long as six years. He compared durvalumab to Merck’s pembrolizumab (brand name: Keytruda), which the FDA approved in October 2023, also for neoadjuvant and adjuvant treatment. “And to go to the extreme, if we apply that standard to pembrolizumab, then you’re denying patients access to medications that we know work in terms of disease-free survival and maybe overall survival for six years, just to kind of figure out which phase is working,” he said. “And I think that that’s a very, very critical question, but I also don’t want to prevent access to at least what we think is the winning strategy.” Lieu pointed out that AstraZeneca is working with cooperative groups to tease out the contribution of phase in ongoing trials. Meanwhile, “I’d like to have this [durvalumab] as an option available to patients and their providers,” he said.
• ODAC Chair Spratt again raised the concept of “financial toxicity” in response. AstraZeneca, he noted, estimated that a three-arm trial would extend the study by about two years. He said durvalumab generated $4.3 billion for AstraZeneca in 2023, adding that with “about two days of revenue, you could generate probably enough to get one more arm of trial for two years of data and patients.” He added, “The financial toxicity part of this is substantial. It’s hundreds of thousands of dollars.” Because FDA had advised AstraZeneca early in the process about its desire for a trial that would isolate the effect of the treatment phases, and the applicant did not follow this advice, Spratt asked, does this change Lieu’s response? Lieu acknowledged that Spratt made a “great point” but that the therapy should not be delayed by this question because the AEGEAN trial met its primary endpoint.
• Van Berkel agreed with Lieu, saying, “I worry that trying to get an answer to that [contribution of phase] question is going to prevent people from getting care that they would benefit from.” Another committee member, RAVI MADAN, of the National Cancer Institute, emphasized this point, essentially saying that, concerning durvalumab, “whatever happened, happened, and now we’re here with data that’s pretty good for patients, and you can’t unring that bell.”
• ASHLEY ROSKO, of the Ohio State University Comprehensive Cancer Center, added that in the adjuvant phase, which she also referred to as the “maintenance phase,” clinicians and patients can decide to stop therapy if they experience toxicities. “I also know that that maintenance phase also becomes slippery in terms of the duration of therapy that a patient would benefit from, and so just lends to making sure that this doesn’t come into some type of perpetual type of adjuvant therapy,” she said.
• In response to a question, FDA’s Larkins noted that there is a separation between the durvalumab data and clinical trial design going forward. Concerning durvalumab, Larkins said, when FDA advised AstraZeneca about the AEGEAN trial design, “we didn’t have a strong scientific safety argument to say we’re going to put your study on hold if you don’t do a three-arm study design.” Now, however, “We do have enough data generated to say that this is really not the best approach to continue taking,” she said. Larkins added that the FDA is asking ODAC whether, going forward, “Do you think we should have more teeth to say we’re potentially going to put a study on hold because you can’t meet the stated objectives if you do a two-arm study design, because your stated objective should really be to prove that both parts of the regimen are having an effect?”
• At a high level, the committee was generally supportive of the strength of the AEGEAN trial’s positive results. In summarizing the discussion, Spratt said in AEGEAN, perioperative durvalumab met its primary endpoints, demonstrating significantly higher rates of pCR and EFS as compared with placebo. While durvalumab did not achieve the prespecified p-value for DFS, “it’s pretty dang close,” he said, adding that OS “is clearly not worse, and we’ll see over time” what happens. The point of the meeting, he said, is, “we can’t clarify the contribution of each phase.” He said the trial clearly showed that neoadjuvant durvalumab had an effect, but the adjuvant effect is less clear. He noted that patients in the open public hearing brought up “very tolerable” and “annoying,” but “not devastating,” side effects. Committee members, Spratt added, agree “that this is an important regimen that should be available to patients” and that the length of the durvalumab therapy can be optimized going forward. Some committee members, he said, expressed “disappointment” that the duration of the regimen “wasn’t addressed initially upon discussion with the FDA as a recommendation, but also that maybe that discussion is not always as crystal clear as ‘Do this, or else.’”

Voting Question: Should FDA require that new trial design proposals for perioperative regimens for resectable NSCLC include adequate within trial assessment of contribution of treatment phase?

• Result: 11 yes, 0 no, 0 abstain
• Before the vote was taken, Rosko sought clarification that the question pertained only to future trials in this space, not the current indication for durvalumab. Larkins confirmed that the question concerned “sponsors coming to us with new trial designs.”
• This was satisfying to many committee members, like consumer representative DAVID MITCHELL, who said that that, going forward, “FDA should be requiring that we have study designs that will answer the question” of contribution of phase.
• While the question was specific to this indication due to the remit of the advisory committee, Kunz remarked that she hopes “that the FDA considers this conversation in other solid tumors.” Similarly, Rosko noted there can be baseline differences between patients who are enrolled in neoadjuvant studies and those enrolled in adjuvant studies. It is key, she said, that trials “better characterize the type of patients” who will actually receive the therapy for the proposed indication.
• Lieu tried to present the concept as a spectrum. One end would be “an easy, nontoxic drug that we just give a little bit on both sides,” for which perioperative therapy “doesn’t really require additional data.” But the new, potent therapies already in the pipeline represent “the end of the spectrum that I think we’re all worried about.” If the contribution-of-phase question is not answered in the preapproval clinical trials, he said, “then we’re left just creating a ton of toxicity with potentially very effective therapeutics but are incredibly difficult to give – and sometimes to tolerate.”
• Lieu pointed to the cost (both in time and resources) of the additional studies needed, saying, “We’re going to cost a lot of millions of dollars by making this decision and potentially delaying drug development.” A solution, Lieu said, might be to have better surrogate endpoints, such as circulating tumor DNA (ctDNA) or pCR, but that these can also be difficult. “But it is incumbent upon this group and industry to work together to find surrogate endpoints because otherwise we’re going to start delaying drug approvals by five, six years,” he said. Van Berkel added that while stricter trial requirements will increase the research and development cost, the trials’ increased rigor “may actually behoove the companies that are doing it because they may find applications that they were not expecting that will be used by more patients in the long run.”
• Another option was supported by patient representative JIM PANTELAS, who pointed out AstraZeneca’s approach with smaller trials to try to specify contribution of each phase. He argued that “less can be more, that maybe we’re trying to accomplish too much out of one trial, and maybe what we’re looking at here is three that could run consecutively” with lower enrollment requirements for each trial.

Analysis

• For now, the discussion indicates that additional trials will likely not be required for durvalumab, potentially paving the way for an approval of the supplemental application. This will provide some relief for Merck’s similarly-studied, already-approved KEYTRUDA, and other perioperative regimens in the pipeline. That said, the committee’s vote provides strong backing for FDA to enforce a requirement for multi-arm trials for oncology products seeking perioperative indications moving forward, which could occur through clinical holds.
• During the meeting, Pazdur brought up the regimen and indication as a candidate for pragmatic trials, which have been a major talking point of the OCE leader. He said this is “an ideal situation for a large, pragmatic, simple trial, with survival as an endpoint” given the well-known safety profile prior to the adjuvant setting and elaborated that the trials would be large and “probably overpowered” to clear out noise. Pazdur pointed to the Office’s Project 5 in 5, which aims to provide clinically relevant questions for 5 pragmatic trials to the oncology community for discussion.
• An important shadow cast over the meeting was the discussions of cost—both to the sponsor and patients who may ultimately be prescribed (and need to pay for) a regimen whose longer duration doesn’t result in better efficacy. FDA forcefully declined to provide input on these occasions, which aligns with the regulatory oversight mission of the agency. As Pazdur told AgencyIQ’s ALEC GAFFNEY in an interview earlier this year, “Companies always come to us and say, ‘Oh, we don’t have enough funding to do this study.’ That argument falls on deaf ears. Don’t even bother wasting your breath discussing that with us, because we do not get involved with the finances of your particular company. Again, we have to have a level playing field whether we’re dealing with a large pharmaceutical company or a tiny biotech company. If you don’t have the funding for it, to put it in the vernacular: Maybe you shouldn’t be in the business, okay?”
• On the patient side, as committee members pointed out, longer treatment duration may also mean more toxicity and disability, so clinicians, patients, and families would benefit from more certainty of the benefit-risk balance for protracted therapy. In fact, a recent retrospective database analysis of about 1,000 NSCLC patients showed similar PFS for 9 versus 12 months of adjuvant durvalumab treatment, with only 31% of patients completing the intended course of durvalumab and over 20% discontinuing treatment because of toxicity or adverse events.
• These are sobering figures, falling in line with data about toxicity from cancer therapies in the modern era presented at the annual dose optimization meetings co-hosted by the FDA and the American Association for Cancer Research (AACR). As presenters have often noted, “dose optimization” for immunotherapies and targeted small molecule therapies needs to account for both dose and duration of treatment.
• Although the question is beyond the direct purview of the FDA, ODAC members nonetheless observed that for prolonged perioperative durvalumab treatment, someone is going to have to pay for those extra months of intravenous injections. Given the FDA’s strong posture that questions of cost are beyond its purview, payers may have more complicated work in store to recognize (and make policies regarding) the value of products with these important gaps in understanding.

To contact the authors of this item, please email Amanda Conti ( aconti@agencyiq.com) and Jason Wermers ( jwermers@agencyiq.com).
To contact the editor of this item, please email Kari Oakes ( koakes@agencyiq.com).

Key Documents and Dates

• Meeting of the Oncologic Drugs Advisory Committee (all meeting materials and recording linked here)
• July 25, 2024
• FDA Docket No.: FDA-2024-N-2559