ODAC gives thumbs-up to GSK plan to build on MSK rectal cancer study of dostarlimab that produced 100% response

Can a single drug replace a long-established curative, albeit brutal, regimen of chemotherapy, radiation, and surgery?

This unprecedented question was posed to the FDA Oncologic Drugs Advisory Committee on Feb. 9.

After more than six hours of absorbing and debating paradigm-shifting data, the committee voted 8:5 in favor of the design for two proposed single-arm trials enrolling a total of 130 patients to characterize the benefits and risks of Jemperli (dostarlimab-gxly) in the curative-intent setting of mismatch repair-deficient or MSI-H locally advanced rectal cancer. 

Yes, you read this correctly: 

There is no approval application on the table. The agency sought the committee’s guidance on a clinical trials strategy for a specific application.

Under normal circumstances, FDA staff dispenses advice on clinical trials strategies, and ODAC gets to make recommendations on data from completed trials.

On occasion—very rarely—FDA asks the committee to weigh in on issues that involve classes of drugs, and sometimes it uses the committee to air concerns about toxicities or clinical trials methodologies seen across applications. 

However, the circumstances at the core of this ODAC were anything but ordinary. Revolutionary is a better word. 

At the ODAC meeting, GlaxoSmithKline presented a plan to build on stunningly positive findings on Jemperli, a PD-1-blocking IgG4 humanized monoclonal antibody, for the treatment of dMMR/MSI-H LARC. 

In initial studies, at Memorial Sloan Kettering Cancer Center, Jemperli resulted in 100% clinical complete response rate in the 12 patients completing the planned treatment of nine cycles over six-months in a prospective phase II study. 

These results were presented at the 2022 annual meeting of the American Society of Clinical Oncology and published in The New England Journal of Medicine (The Cancer Letter, June 10, 2022).

Now, the number of responses has risen to 18 out of 18. 

Several colorectal cancer experts on the committee say that it would be unrealistic to expect to accrue patients to randomized trials in this setting.

“Some patients who received these drugs for treatment of refractory advanced disease, remarkably, remain disease free years later,” gastrointestinal oncologist Richard M. Goldberg wrote in a guest editorial in The Cancer Letter. “To borrow a phrase that I heard long ago from my fellow medical residents, ‘They beat the reaper.’ 

“Oncologists celebrate when they beat the reaper.”

Goldberg is not involved with the study and is not a member of ODAC. His editorial appears here. 

The principles established at this ODAC could be applied to other drugs that produce dramatic results in early-stage clinical studies, observers said.

The world has changed

This paradigm-shifting application required FDA to seek ODAC advice, as a means of letting it be known that the world has changed. Just as importantly, the meeting allowed the agency to present its concerns—statistical and clinical—to establish the record. 

The MSK findings required FDA to consider moving an investigational drug therapy up to an indication where curative outcomes are often achieved by standard of care—albeit with brutal side effects from chemotherapy, surgery, and radiation. 

Usually, drugs are first approved in the advanced disease setting and moved toward the frontline. To justify an accelerated approval of an investigational drug in a setting where the standard of care produces cures, the trial would have to demonstrate dramatic improvement over the standard therapy, said Richard Pazdur, director of the FDA Oncology Center of Excellence. 

Accelerated approval is appropriate because the drug is being tested in a serious, life-threatening disease, Pazdur said. 

“The uncertainty is far more acceptable when you are dealing with patients in a single-arm trial who have no other therapies available to them,” Pazdur said. “That’s a common scenario for which we use accelerated approval—the metastatic disease setting. Usually, it’s patients that have gone through the available therapies. 

“Again, this is the whole reason why we’re bringing this to the committee, is, ‘What is this risk that is tolerable here, from a regulatory standpoint, and also from a practice standpoint, because you are dealing with a curative therapy?

“So, there should be greater scrutiny here—and that’s why we’re bringing this proposal to this committee,” Pazdur said. 

FDA typically requests randomized control trials that compare an investigational therapy to standard of care, or that evaluate the investigational agent as an add-on to standard of care with approval based on established endpoints of clinical benefit, such as survival. 

Single-arm trials are discouraged, because the results are uninterpretable in the absence of a comparative group, Lola Fashoyin-Aje, deputy director in the Office of Oncologic Diseases, said at the ODAC meeting.

However, this has not been universally the case. 

“A noteworthy exception to these general principles is the use of durable complete response rate as an endpoint in single-arm trials investigating therapies for patients with BCG unresponsive, high-risk, non-muscle invasive bladder cancer with carcinoma in situ,” Fashoyin-Aje said. “In this clinical scenario, cystectomy provides a curative option, but it is associated with significant morbidity and a 90-day mortality rate that may be as high as 10 to 15% in older patients.”

This exception was considered by FDA due the lack of suitable therapy to serve as comparator in randomized clinical trials. In that case, the agency signed off on endpoint trial designs, treatment assessment, and follow-up that would be adequate for trials designed to support regulatory action, Fashoyin-Aje said.

GSK’s development plan 

The indication GSK is seeking for Jemperli is small. 

Somewhere between 2,000 and 4,000 patients in the U.S. have mismatch repair-deficient or MSI-H locally advanced rectal cancer. 

The company’s plan is to broaden the indication to mismatch repair-deficient or MSI-H locally advanced colon cancer in confirmatory trials should an accelerated approval be obtained. 

As the studies proceed—which would be a reasonable outcome of the ODAC recommendation—the committee could be expected to see the data that would be generated for Jemperli’s accelerated approval and, ultimately, full approval.

GSK proposes to conduct two single-arm trials: a continued follow-up of the original MSK Study 19288, along with Study 219369, a global, multicenter, single arm, open-label, non-randomized study that will enroll approximately 100 patients with pathologically confirmed, previously untreated LARC that is dMMR/MSI-H as assessed by local testing. 

Again, this is the whole reason why we’re bringing this to the committee, is, ‘What is this risk that is tolerable here, from a regulatory standpoint, and also from a practice standpoint, because you are dealing with a curative therapy.

Richard Pazdur

The primary endpoint in Study 219369 is clinical complete response at 12 months. Two key secondary endpoints are clinical complete response at 36 months (cCR36), and event-free survival at 3 years by investigator assessment, defined as remaining alive and free of the following: disease progression precluding surgery, local recurrence, and distant recurrence.

Following completion of treatment with dostarlimab, patients will undergo assessments with endoscopy, rectal MRI, and CT to determine clinical response. Subsequent treatment decisions will be made based upon the investigator’s assessment of clinical response, which include digital rectal examination.

GSK also proposed a phase III, randomized, confirmatory trial evaluating the use of Jemperli in stage 2-3 perioperative colon cancer, arguing that locally advanced dMMR/MSI-H colon cancer is the closest setting in which Jemperli’s benefit for dMMR/MSI-H locally advanced rectal cancer can be confirmed in a controlled study. This phase III protocol would include surgery. 

Pazdur questioned the feasibility of the company’s plan to conduct a randomized trial that would support a regular approval.

For starters, the drug has been on the market under an accelerated approval in another indication—in recurrent or advanced dMMR/MSI-H endometrial cancer. That indication is limited to patients who have progressed on, or following a prior platinum-containing regimen, who are not candidates for curative surgery or radiation.

The accelerated approval in endometrial cancer was granted in 2021. On Feb. 9, in an action that may or may not be coincidental, the agency converted that accelerated approval into a regular approval. 

The availability of information and the drug prompted Pazdur to use the “e” word—”equipoise”—in relation to the company’s plans for converting a potential accelerated approval in rectal cancer into a regular approval in rectal and colon cancer.

“I don’t want to agree, necessarily, to a trial that can’t be done. Do you actually think that over time, if this drug is approved in rectal cancer, that there will be equipoise—that people will say, ‘Okay, I’ll go on to do surgery.’ 

“Or will they just say, ‘Okay, I got a CR, I just want to take a wait and see approach to this?’ Pazdur said at the ODAC meeting. “I really think people will try to avoid any type of invasive surgery; obviously a hemicolectomy people would want to avoid.”

The ODAC votes

ODAC voted after an exhaustive discussion of concerns brought forth by the agency. These included:

• The adequacy of data derived from a small cohort of patients from two single-arm trials (with one from a single center providing a significant number of patients) to support approval.
• The proposed and unprecedented use of cCR12 as an endpoint for accelerated approval.
• The data package proposed to verify clinical benefit, including the role of a proposed study in patients with colon cancer.
• The potential impact of the variability in care, expertise, etc., across multi-disciplinary study staff and across study sites on study conduct and ultimately on outcomes.

The voting question read: 

Will the data from the proposed single arm trials enrolling a total of 130 patients be sufficient to characterize the benefits and risks of dostarlimab in the curative intent setting for patients with dMMR/MSI-H LARC?

The committee voted 8:5, focusing on the design of the trial intended to support accelerated approval in the future.

After the vote, the committee members described the rationale for casting their votes:

I voted yes. Obviously, in this setting, as has been discussed, I don’t think a randomized study is feasible, given the presentation of the existing data and patients’ overall goals and expectations. 

I will clarify, I do have some concerns about the use of complete clinical response at 12 months as the definitive endpoint, mainly because I don’t think that there’s a clear correlation—although there’s a suggestion—that there’s a correlation between complete clinical response and disease-free survival and distant metastasis rate. 

That’s my only concern in regards to the complete clinical response rate as a definitive endpoint. 

I do think the endpoint of event-free survival at three years, which is a secondary endpoint of the study, will be critically important, just to show that correlation. But overall, I believe that the study, as designed, will provide the data needed for accelerated approval.”

I voted yes. If the frequency of the disease is as small as the discussions today suggested, it’s probably difficult, if not impossible, as Dr. Lieu said, to accrue a randomized standard control arm, even a small one, as one of the public commenters suggested. 

Second, given the initial very positive data, I do think it’s difficult to get patients to enroll in a control, given the irreversible toxicities and lifestyle and extreme quality of life sacrifices. 

And finally, accelerated approval is for patients, and this feels like a condition and a potentially enormous step up in care for patients, that’s worth going with the current proposed trial design, not waiting three to five years.

I voted no, based on the fact I wish there were some of the modifications that were discussed during our extensive discussion here. In the VA, we do have ability to sequence many patients, but there is a substantial discordance. 

So, for the biomarker testing, the three year event-free survival is a more meaningful mark for these patients. And if we just rely on the 12 cCR, then we can’t go back on accelerated approval. 

That might be an issue, and that often is, with a lot of the accelerated approvals for the drugs. Then the subsequent follow up then kind of goes by the wayside. So, I did have some concerns about that. 

And possibly using this marker in other sites, although I understand it’s a different entity, what would that mean for the future and the landscape of oncology? I think this is a great opportunity, I would’ve just liked to have seen a little more regulation in terms of the design. 

But otherwise, I hope this drug is promising, and I just wish they had a higher threshold and more of the intention to treat kind of discussion that was had earlier.”

I voted yes. I think Chris Lieu very eloquently made all the comments that I would make. I do think that monitoring three year, event-free survival will be very critical. But given the compelling nature of the current data, I think this warrants moving forward. 

I voted no. I think extrapolation from the chemoradiation data to single agent immunotherapy is a little too early, and I think the cCR endpoint is also inadequate. Those are the main reasons for voting no. 

I voted no because I took the question literally and grammatically, which clearly stated whether or not this data was sufficient. And I don’t believe this data is sufficient. 

I do believe this agent has great safety, has efficacy, has a pretty impressive clinical complete response with the current data. And I do believe it has a huge opportunity for us to delay—for some, maybe never having to have—the morbidity of a surgical chemoradiation or surgical approach. 

However, I do not believe the data that we have and the data that has been proposed by the applicant is sufficient to characterize the benefits and risk in the curative intent setting for this patient population.”

I voted yes. Will the data be sufficient? Yes, I think it will. But will the analysis be sufficient? That part, I’m not so sure. For all the reasons Dr. Katsoulakis stated, there’s going to be variability in the biomarker, and we’re going to be defining the enrolled population down to the eligible population, and I think potentially misinterpreting the data that we get. 

We’re not going to be liberal in the radiographic definitions of what is persistent disease. We’re going to be very strict about that. And because of that, also, we make this study less valid to the external world if it’s interpreted that way. So, I think the design and the data we get is gonna be sufficient, but I do think we need to be very careful in the analysis that’s finally done. 

I voted yes. While the proposed studies certainly won’t answer all of our questions about the optimal use of immunotherapy in MSI high locally advanced rectal cancer, I think they’ll provide additional data to determine whether the initial pilot results are generalizable, given the multi-institutional, multinational nature of the proposed studies, the longer follow up, and increased sample size. So on that basis, I voted yes. 

I voted no. Despite the extraordinary promise of the agent and concerns about the feasibility of doing a randomized trial, I voted no because of the difficulty in interpreting results of non-comparative trials and the uncertainty around the long-term applicability of the endpoint. 

I voted no, despite the incredible response rate data and the very compelling patient testimonials. I felt that the data were not sufficient. 

Again, to sum up what Dr. Garcia said about the word sufficient, given the nature of the clinical complete response 12 endpoint, especially in this curative setting. 

I voted yes, for many of the reasons already stated. I believe in the single-arm design, and I’m supportive of cCR as an acceptable primary endpoint. 

However, I do think that we need to expand on some of the secondary endpoints, as has been previously discussed, event-free survival is already added, but adding organ preservation rate, considering adding central confirmation of dMMR/MSI-high as part of eligibility criteria, and quality of life. 

I really appreciate the robust conversation from colleagues today, but really this balancing act of identifying effective agents in minimizing morbidity and providing access. And I think that this trial will really be the first attempt to do that in this disease. So I voted yes. 

I voted yes. I think that, for me, in the context of improvement of quality of life, that the data and the design is sufficient at this stage to warrant essentially moving on. 

Again, enough to collect that data at this stage and move on. I think that it’s sufficient. Again, viewing that largely in the context of the promise of improving the quality of life aspects of treatment of the disease. 

I voted yes, but I think that this planned study in rectal cancer with dostarlimab is suboptimal. That said, I don’t think a randomized study will be feasible. 

As the discussion highlighted today, PD-L1 inhibitors will likely be used off label, and if off-label use becomes the standard practice, then there will be no way to capture data prospectively. Therefore, that makes this proposed trial important as perhaps the only means to obtain that prospective data. 

I’m not 100% confident in the one year clinical CR endpoint. That makes transparency and adequate follow-up for durability of response really incumbent upon the sponsor to share with the FDA and the public as it becomes available. 

Those endpoints must also remain at a high bar for cure rate, and not just survival, in a population that likely would be cured with standard of care. So, this trial is a potential platform to get the data, but the questions must be asked appropriately and rigorously evaluated.