At ODAC without sponsors, FDA highlights safety signals in PI3K inhibitor trials

Why is the April 21 ODAC different from all the other ODACs?

At all other ODACs, a commercial sponsor gets to make a case. On April 21, only FDA’s voice will be heard. 

The reason for this unprecedented change of format is not of the happy-making sort. The agency wants the world to know about potential decrements in overall survival that have been observed in six post-marketing randomized controlled trials of phosphoinositide 3-kinase (PI3K) inhibitors in indolent NHL or CLL.

The drugs in question demonstrated durable overall response rates or improvements in progression-free survival, but post-approval studies have demonstrated potential detriments in overall survival, which the agency believes are most likely due to toxicities. 

“This is a forward-thinking discussion of how these products should be developed, based on what has been observed so far, and will focus on four major issues,” Richard Pazdur, director of the FDA Oncology Center of Excellence, said to The Cancer Letter.

“First, potential detriments in overall survival observed in six randomized trials in indolent non-Hodgkin lymphoma or chronic lymphocytic leukemia,” Pazdur said. “The second issue is the toxicity noted with excessive dose interruptions, dose delays, and treatment discontinuations, and the third issue, related to toxicity, is the lack of careful dose exploration prior to conducting registration trials. 

“The last issue is the limitation of single-arm trials that led to many of the initial accelerated approvals and whether the single-arm trial design remains a viable option for registration in these diseases.”

The agency is focusing on drugs that inhibit the PI3K signaling that promotes proliferation of malignant lymphocytes. The drugs in question are:

• Idelalisib (Gilead Sciences Inc.) 
• Copanlisib (Bayer Healthcare Pharmaceuticals Inc.) 
• Duvelisib (Secura Bio Inc.) 
• Umbralisib (TG Therapeutics Inc.)

After devoting the April 21 session to  reviewing the problems seen in PI3K inhibitors, ODAC was scheduled to turn its attention to a specific application—a combination of ublituximab and Ukoniq (umbralisib) for the treatment of adult patients with chronic lymphocytic leukemia and small lymphocytic lymphoma. 

I hate to sound like a broken record, but we have discussed all of the above issues on multiple public occasions—inadequate dose exploration, challenges with single-arm trials, consideration of dosing for more chronic malignancies, patient-focused drug development.

Richard Pazdur

However, on April 15, the sponsor, TG Therapeutics Inc., announced its decision to voluntarily withdraw the application for the combination referred to as U2.

The company also withdrew Ukoniq from sale for the approved indications of treatment of adult patients with marginal zone lymphoma who have received at least one prior anti-CD20-based regimen, and for the treatment of adult patients with follicular lymphoma who have received at least three prior systemic therapies. Ukoniq was granted accelerated approval in these indications in February 2021. 

“The decision to withdraw was based on recently updated overall survival data from the UNITY-CLL phase III trial that showed an increasing imbalance in OS,” the company said. The decision to withdraw Ukoniq from sale was primarily based on the withdrawal of the BLA and sNDA for U2 in CLL.

In a statement, Michael S. Weiss, chairman and CEO of TG Therapeutics  said:

In an interview, FDA officials said the development plans for PI3K drugs were fundamentally flawed, some in ways that also affect other cancer drugs.

One such problem is reliance on the outdated concept of maximum tolerated dose. 

“For the initial approvals, there was limited dose exploration before embarking on trials for approval. In many instances, doses were determined using a maximum tolerated dose approach,” Nicole Gormley, director of the Division of Hematologic Malignancies 2, Office of Oncologic Diseases, said to The Cancer Letter.

“All of these products have an exposure-response relationship for safety, with minimal exposure-response for efficacy. If lower doses had been studied more, they may have identified doses that have less toxicity but similar efficacy.

“Currently, with OCE’s Project Optimus, we advise companies to rigorously characterize dose response and safety evaluations, including potentially randomized dose trials. While there is concern that robust dose exploration may delay drug development, in the long run investing in these efforts early can result in more successful drug development programs and safer, more tolerable drugs for patients.”

Agency officials also published a paper on the subject in the April 14 issue of The Lancet Oncology. 

Pazdur and Gormley spoke with Paul Goldberg, editor and publisher of The Cancer Letter.

Paul Goldberg: What’s so special about the ODAC on April 21?

Richard Pazdur: We usually have meetings with a commercial sponsor presenting an application and have a separate FDA presentation about the application. For the meeting on April 21, only the FDA will be presenting, and there will not be any sponsor presentations. 

We want to discuss safety problems encountered with the PI3K inhibitors in hematological malignancies. 

This is a forward-thinking discussion of how these products should be developed, based on what has been observed so far, and will focus on four major issues. First, potential detriments in overall survival observed in six randomized trials in indolent non-Hodgkin lymphoma or chronic lymphocytic leukemia. 

The second issue is the toxicity noted with excessive dose interruptions, dose delays, and treatment discontinuations, and the third issue, related to toxicity, is the lack of careful dose exploration prior to conducting registration trials. The last issue is the limitation of single-arm trials that led to many of the initial accelerated approvals and whether the single-arm trial design remains a viable option for registration in these diseases. 

On the following day—April 22—there will be a more familiar ODAC meeting with sponsor and FDA presentations. The April 22 meeting will focus on umbralisib plus ublituximab in CLL and the existing umbralisib indications under accelerated approval for iNHL. 

[On April 15, the agent’s sponsor, TG Therapeutics Inc., withdrew the application, citing overall survival data that showed “an increasing survival imbalance in favor of the control arm.”] 

What’s unique about the PI3K inhibitors and how many has FDA approved?

Nicole Gormley: Four PI3K inhibitors have been approved for relapsed or refractory indolent non-Hodgkin lymphoma or chronic lymphocytic leukemia: idelalisib (Gilead Sciences Inc.), copanlisib (Bayer Healthcare Pharmaceuticals Inc.), duvelisib (Secura Bio Inc.), and umbralisib (TG Therapeutics Inc.). While these all inhibit the PI3K delta form, some of these products also inhibit other isoforms.

What these drugs do is inhibit the PI3K signaling that promotes proliferation of malignant lymphocytes, which is the rationale for using these in hematologic malignancies.

Are there different types of PI3K inhibitors? One was approved the first week of April. How is that different from the others?

RP: Yes. Alpelisib is another PI3K inhibitor. It is an alpha-specific inhibitor, and is not part of this discussion. Alpelisib was recently approved for PIK3CA-related overgrowth spectrum, and previously was approved in breast cancer.

Years ago, when Richard Schilsky was chairman of ODAC, he came up with the term “toxic placebo.” Are the PI3K inhibitors turning out to be toxic placebos?

RP: No, it’s far more complicated. These drugs are not “placebos”—they have activity in these diseases. It’s about how these drugs were developed. 

In general, these drugs demonstrated durable overall response rates or improvements in progression-free survival, in their trials submitted for initial approval, but post-marketing trials demonstrated potential detriments in overall survival, most likely due to serious toxicities. 

The dosing of many of these drugs were not adequately evaluated prior to leaping into single-arm trials or randomized trials with PFS as primary endpoints. Because of the relatively long natural histories of these diseases and subsequent therapies, overall survival may be challenging to evaluate. 

Many confirmatory trials did not have a formal statistical plan to evaluate OS. Nevertheless, for drugs approved on a PFS endpoint, FDA has always asked sponsors to provide us overall survival data in the original submission and more mature collections of OS post-approval. 

In these situations, we evaluate OS descriptively, looking to ensure there is not a potential for harm based on the results. 

Although we frequently focus on OS as an efficacy endpoint, we should not forget that it is also a safety endpoint. OS is both a safety and efficacy endpoint making it a unique endpoint.

What were the major problems with these drugs?  

NG: There have been six randomized controlled trials in indolent NHL or CLL that have raised concerns for potential decrements in overall survival for the PI3K inhibitor arm due to increased toxicity (see Table below). 

Two drugs of this class, idelalisib and duvelisib, were granted regular approval in CLL based on improvements in PFS in RCTs. The approvals for indolent NHLs were accelerated approvals based on durable overall response rate in single-arm trials, which required post marketing trials demonstrating improvements in PFS. 

While sponsors don’t need to demonstrate an improvement in overall survival for continued marketing, survival data, as previously mentioned, was required to be submitted for drugs approved using PFS endpoints. 

While there is concern that robust dose exploration may delay drug development, in the long run investing in these efforts early can result in more successful drug development programs and safer, more tolerable drugs for patients.

Nicole Gormley

Eventually, three RCTS involving idelalisib in CLL or indolent NHL were terminated early due to potential detriments in OS in the idelalisib arms. The main cause was infections, including sepsis, pneumonia, and opportunistic infections. Safety alerts were issued, and warnings and limitations were added to the drug label. 

Gilead withdrew the indolent NHL indications for idelalisib earlier this year because of poor accrual to another confirmatory trial. Similarly, Secura Bio withdrew the FL indication for duvelisib in December 2021 because a confirmatory trial was never initiated.

Also, the recent final OS analysis of the duvelisib RCT that supported the initial approval in CLL demonstrated a potential decrement in OS favoring the ofatumumab arm. 

The additional RCTs of copanlisib and umbralisib in relapsed or refractory indolent NHL and CLL, respectively, demonstrated potential detriments in OS that were associated with more severe toxicities in the PI3K arm compared to the control arms. 

Throughout the history of these drugs, multiple risk mitigation strategies were used. These included boxed warnings (idelalisib, duvelisib), communications REMS (idelalisib, duvelisib), FDA safety alerts (idelalisib, umbralisib), limitations of use in product labeling (idelalisib) and restricted indications (duvelisib). 

Randomized Trials with Concerning Overall Survival

Abbreviations: 
RCT: Randomized controlled trials; OS: Overall Survival; CLL: Chronic Lymphocytic Leukemia; iNHL: Indolent Non-Hodgkin Lymphoma

Is this unprecedented in oncology in your experience?

RP: Yes. I have not seen this before. 

When one observes six randomized trials raising concerns for potential decrements in OS—albeit not formally statistically tested—one cannot assume that these finding are simply due to chance. 

While many of the OS results represent early findings and definitive conclusions would be challenging to interpret if observed from a single trial, the pattern here has been observed in multiple trials. We are looking at the entire picture—these potential decrements in OS—coupled with the toxicity profiles. Up to 65% of patients in some trials had serious adverse events and up to 85% of patients in some trials had grade 3 or higher adverse reactions.

Dose interruptions were encountered in up to 64% in some trials; as high as 29% of patients and 35% of patients had dose reductions or permanent discontinuations in selected trials. 

The survival findings were frequently associated with sepsis, pneumonia, and opportunistic infections, diarrhea, colitis, hepatotoxicity, as well as dermatological adverse events. Immune toxicities are partly due to decreased regulatory T-cell activity associated with this drug class.

You mentioned the issue may be dosing. Why do you think that?

NG: For the initial approvals, there was limited dose exploration before embarking on trials for approval. In many instances, doses were determined using a maximum tolerated dose approach. 

All of these products have an exposure-response relationship for safety, with minimal exposure-response for efficacy. If lower doses had been studied more, they may have identified doses that have less toxicity but similar efficacy.

Currently, with OCE’s Project Optimus, we advise companies to rigorously characterize dose response and safety evaluations, including potentially randomized dose trials. While there is concern that robust dose exploration may delay drug development, in the long run investing in these efforts early can result in more successful drug development programs and safer, more tolerable drugs for patients. 

Sounds like a lot of problems in drug development have come together with these drugs?

RP: Yes, this is a “perfect storm” involving many issues in oncology drug development that we have repeatedly cautioned sponsors about. 

The diseases tend to have long natural histories and patients can remain on these drugs long-term. Patients may even cycle back to drugs that they previously used. Having high rates of dose interruptions, dose reductions, and treatment discontinuations secondary to adverse events is self-defeating and sponsors should reconsider the dosing strategies.

First, as already mentioned, there was very limited dose exploration performed without careful consideration of tolerability. We have been actively encouraging sponsors to focus on randomized early dose exploration taking into account exposure-toxicity and exposure-response relationships rather than just jumping to the MTD. As Dr. Gormley mentioned, this is the major objective of Project Optimus.

Secondly, when sponsors select single-arm trials as a registration strategy, many focus primarily on the dose that yields the highest response rate without adequate consideration of the dose’s adverse event profile. 

Sponsors realize that efficacy is a high hurdle in drug regulation, but this must be tempered by safety findings. This singular focus aimed at maximizing the response rate using the MTD may not be the best plan, especially when drugs may be used for prolonged time periods. 

The words “tolerable” and “manageable toxicities” are frequently used by sponsors even in situations where there are unacceptably high safety issues. I have previously publicly addressed the need to provide the patient’s perspective on what is “tolerable”—not what the drug company considers “tolerable.” 

The OCE has spent considerable time and effort in addressing patient-focused drug development programs and these need to be more fully embraced by drug developers. 

Thirdly, we cannot adequately characterize the drug’s toxicity in single-arm trials since a control arm is not present. Hence, it is not until later when we perform randomized trials that we fully comprehend the entire risk-benefit picture.

This current experience must lead us to question the continuation of single-arm trials in these diseases. Single-arm studies can provide data on response rate and response duration; however, other important efficacy endpoints, such as PFS or OS, cannot be evaluated. The use of single-arm trials as registration strategies may result in delays in fully understanding the risk-benefit of the drug.

Lastly, when randomized trials were conducted for many PI3K inhibitors in hematological diseases, the comparator arm—rituximab, ofatumumab, or bendamustine—lacked the class-specific toxicity profile of the PI3K inhibitors and generally were better tolerated.

These comparators with their more favorable safety profiles provided an optimal comparative background to assess these toxicities and their potential adverse effect on overall survival. This set up a perfect storm for demonstration of a survival disadvantage.

So, what has FDA learned, what can the oncology drug development industry learn from this?

NG: It’s time to step back and re-examine drug development for hematologic malignancies. We would be reluctant to proceed with single-arm studies for initial approval in indolent lymphomas and similar diseases—because you won’t have comparative safety information until relatively late with accelerated approval based on a single-arm trial. 

RP: We realize the sponsors are under pressure to develop drugs as quickly as possible. We understand that patients want innovative drugs as soon as possible.

However, I hate to sound like a broken record, but we have discussed all of the above issues on multiple public occasions—inadequate dose exploration, challenges with single-arm trials, consideration of dosing for more chronic malignancies, patient-focused drug development. 

Unfortunately, what we saw here with the saga of the PI3K inhibitors was the confluence of the failure to address many of these issues.

If sponsors do not carefully examine dose and schedule prior to embarking on registration trials, the development plan will potentially suffer the consequences of the proverbial “building a house on quicksand.”