In a three-day meeting starting Tuesday, a panel of external experts convened by the FDA will discuss whether the agency should consider clawing back accelerated approvals from three immuno-oncology agents.
All six conditional nods, shared among Merck & Co.’s Keytruda, Bristol Myers Squibb’s Opdivo and Roche’s Tecentriq, were based on early data showing they could shrink tumors or slow disease progression. But the companies have since failed to show the treatments could help patients live longer in confirmatory trials. Industry watchers have been closely following the event to gauge the FDA’s standard and level of leniency on go-aheads based on surrogate trial markers that didn’t translate into a survival benefit.
Here, we summarize the positions of the drugmakers and the FDA for the six indications to be discussed, offering a preview of what to expect from the upcoming discussions.
April 27 afternoon: Tecentriq
Indication: Tecentriq in combination with Abraxane (nab-paclitaxel) in PD-L1-positive metastatic triple-negative breast cancer (TNBC)
Background: The original IMpassion130 trial in patients who hadn’t received prior chemotherapy for metastatic disease showed the combination of Tecentriq and Bristol Myers Squibb’s Abraxane cut the risk of disease progression or death by 40% over solo Abraxane in patients whose tumor expressed PD-L1 levels of at least 1%.
In all enrolled participants, which included PD-L1-negative patients, the therapy only cut the risk of death by 13%. Because those data didn’t reach statistical significance, the analysis of survival in the PD-L1-positive subgroup couldn’t be performed. In the PD-L1-positive group, the estimated reduction of death risk reached 33%, as patients on the Tecentriq combo lived a median 25.4 months, versus 17.9 months for Abraxane. However, as the FDA staffers noted, the agency cannot determine whether this improvement of 7.5 months is due to chance alone.
Another phase 3 study dubbed IMpassion131 could confirm the original conditional nod. It tested Tecentriq with the original paclitaxel formulation. But an overall survival analysis of the PD-L1-positive patients in that trial showed that the combo did even worse than solo paclitaxel. That finding prompted an FDA alert, which warned physicians not to replace Abraxane with paclitaxel in metastatic TNBC.
Arguments: Roche argues that the survival benefits seen in the IMpassion130 trial in the PD-L1-positive population were clinically meaningful, while the FDA thinks the lack of statistical significance casts doubts on those purported benefits. Roche also thinks that the problems with IMpassion131 shouldn’t be used to minimize the Tecentriq-Abraxane regimen.
A replication of the IMpassion130 study, which was statistically powered to examine PD-L1-postive patients, would be ideal to confirm the indication. But Roche noted that Tecentriq and Abraxane are already widely adopted and included in all major oncology guidelines, and therefore giving patients in the control group chemo poses an ethical question. Plus, Roche obtained feedback from investigators, patients and advocates suggesting there would be reluctance that would make recruiting participants for such a study difficult.
Roche has three ongoing phase 3 studies of Tecentriq in TNBC. The IMpassion132 trial is testing Tecentriq with chemo for advanced or metastatic TNBC that has relapsed within 12 months after prior therapy. While noting that the study is enrolling a different patient population and has a different chemo backbone, the FDA agrees that IMpassion132 could help confirm Tecentriq’s current indication in TNBC with high PD-L1 expression.
April 28 morning: Keytruda
Indication: Monotherapy for patients with previously untreated bladder cancer who are not eligible for cisplatin-containing chemotherapy and whose tumors express PD-L1 at a combined positive score of at least 10, or patients who are not eligible for any platinum-based chemotherapy regardless of PD-L1 status
Background: The confirmatory phase 3 Keynote-361 trial tested Keytruda with or without platinum-based chemo. But an independent data monitoring committee noted during the study that patients with PD-L1-low expression had worse outcome on Keytruda versus chemotherapy. So the trial halted enrollment of those patients for Keytruda monotherapy, and the FDA slapped a restriction on the label that forbids use in PD-L1-low patients.
The final analysis didn’t show a significant benefit for the Keytruda-chemotherapy arm. It only cut the risk of disease progression or death by 22%, or the risk of death by 14%, over solo chemo. After that fail, all other analyses wouldn’t hold statistical significance and were therefore only exploratory.
Meanwhile, the FDA approved Pfizer and Merck KGaA’s Bavencio as a maintenance therapy to prevent tumor progression in patients who respond to one cycle of chemo. It was a regular approval based on data from the Javelin Bladder 100 trial showing that Bavencio slashed patients’ risk of death by 31% compared with best supportive care.
Arguments: Merck suggests the Keynote-361 trial wasn’t specifically designed to test the currently limited indication, which represented only a subgroup of patients. Merck did flag some exploratory analysis. For example, in a subpopulation of trial participants, Keytruda monotherapy was mostly comparable to carboplatin in patients with PD-L1 expression of at least 10, showing a slight death risk reduction of 18%.
Merck argues that because Keytruda as a monotherapy produced similar survival data to chemotherapy but was better tolerated is meaningful. It also pulled out some real-world data showing median survival for first-line patients with non-cisplatin-based chemotherapy was 6.9 months, which was lower than the 11 months Keytruda monotherapy showed in another real-world data analysis.
But the FDA maintains those details are insufficient to assess whether this real-world data would be appropriate to infer patient outcomes, noting there are simply too many concerns for such non-randomized trials.
The agency also suggests that an estimated 75% of cisplatin-ineligible patients can now be treated with Bavencio maintenance therapy after one round of gemcitabine and carboplatin chemotherapy. But Merck argues Bavencio only addresses a subgroup of patients and doesn’t offer a solution for patients who are ineligible for any platinum-based chemo or who want to avoid platinum-related toxicity.
Merck proposes that alternative trials could confirm the indication. One of them is LEAP-011, which pits Keytruda and Eisai-shared Lenvima against solo Keytruda in cisplatin-ineligible patients whose tumors express PD-L1 or patients who are ineligible for any platinum chemo regardless of PD-L1 status. But the FDA thinks the study design doesn’t make it possible to isolate the effect of Keytruda or to compare it with the current standard of care, which is carboplatin-based chemo.
April 28 afternoon: Tecentriq
Indication: Monotherapy for first-line bladder cancer patients who aren’t eligible for cisplatin-containing chemo with PD-L1 expression covering at least 5% of tumor-infiltrating immune cells
Background: Early review of the confirmatory IMvigor130 trial found Tecentriq monotherapy decreased survival for PD-L1-low patients versus chemo, leading to an FDA restriction in those patients. Eventually, Roche reported that the Tecentriq-chemo combo cut the risk of disease progression or death by 18% as patients on the Tecentriq regimen lived 8.2 months without disease progression versus 6.3 months for solo chemo.
At two interim analyses, overall survival didn’t achieve statistical significance, as the Tecentriq-chemo cocktail reduced the risk of death by 17% and 16%, respectively. Final analysis is expected in 2022, when the trial is expected to reach 667 patient deaths.
Meanwhile, the IMvigor211 trial in post-platinum treatment failed to confirm a benefit for Tecentriq, and Roche has recently withdrawn the drug’s second-line indication. Plus, the IMvigor010 trial testing Tecentriq versus simple observation for high-risk muscle-invasive bladder cancer after surgery also showed no benefit for the PD-L1 inhibitor in delaying disease recurrence.
Arguments: The progression-free survival benefit in IMvigor130 was statistically significant, but the FDA doesn’t consider it clinically meaningful.
Roche notes that the overall survival data for Tecentriq monotherapy in PD-L1-high patients, including those who are ineligible for cisplatin, showed benefits over chemo. In cisplatin-ineligible patients with PD-L1 expression of at least 5%, patients on Tecentriq lived a median 18.6 months versus 10 months for chemo. But those trials weren’t powered to detect statistical significance.
The FDA is now asking its external experts to evaluate whether Tecentriq’s current indication should be maintained while the company waits for the final survival results from IMvigor130.
April 29 morning: Keytruda
Indication: Monotherapy for gastric or gastroesophageal junction (GEJ) adenocarcinoma with PD-L1 at combined positive score of at least 1 after two or more prior lines of therapy
Background: The original Keynote-059 trial showed Keytruda monotherapy shrank tumors in 13.3% of stomach or GEJ adenocarcinoma patients after two previous rounds of treatment. The Merck PD-1 inhibitor also has two tissue-agnostic indications for all microsatellite instability-high (MSI-H) or tumor mutational burden-high (TMB-H) tumors regardless of their location. Keytruda’s tumor response rate dropped to 7.1% in a subsequent exploratory analysis that counted only PD-L1-positive participants that didn’t bear those two biomarkers.
The phase 3 Keynote-061 trial failed after showing Keytruda only pared down the risk of death by 18% in second-line treatment. The other confirmatory trial, Keynote-062, in newly diagnosed patients, also failed. Keytruda monotherapy showed a risk reduction of 9% over chemo. The results would have suggested Keytruda monotherapy was non-inferior to chemo, but the FDA didn’t sign off on a prespecified non-inferiority margin. The combination of Keytruda and chemotherapy wasn’t superior to solo chemo, as the death risk reduction reached only 15% for patients with a combined positive PD-L1 score of at least 1 or at least 10.
Meanwhile, Bristol Myers Squibb’s Opdivo, used in tandem with chemotherapy, just became the first immunotherapy to win an FDA nod for previously untreated gastric/GEJ cancer and esophageal adenocarcinoma. Adding Opdivo to chemotherapy cut the risk of death by 20% in patients with those diseases in the CheckMate-649 trial.
Plus, Taiho Oncology’s Lonsurf is also approved for third- or later-line gastric/GEJ adenocarcinoma after showing it reduced death risk by 31%.
Arguments: In a briefing document, Merck didn’t spend much time trying to find excuses for the two confirmatory trial flops. Instead, it focused on Keytruda’s potential in filling an unmet need. Its main argument? Keytruda monotherapy offers a chemo-free regimen for third-line stomach cancer.
The New Jersey pharma noted that current National Comprehensive Cancer Network guidelines only recommends Lonsurf for patients with low-volume gastric cancer who have minimal or no symptoms. Merck suggested chemo's use may be limited given patients who reach third-line treatment are usually very sick and have a poor prognosis. This heavily treated population may not be able to tolerate the toxicities of chemotherapy, either, Merck added.
For Keytruda’s existing nods in all MSI-H and TMB-H tumors in second-line patients, Merck noted these biomarkers only represent about 17% of gastric cancer, and many of these patients aren’t identified in clinical practice. By comparison, the combined positive score method “identifies a much larger patient population,” it said. “Removing the 3L+ gastric cancer indication will limit access for these patients who have a persistent unmet need and limited effective treatment options.”
As for Opdivo, Merck argues says it’s likely that not all first-line patients will get the regimen “due to tumor PD-L1 and/or HER2-expression levels, physician preference, or inability to tolerate chemotherapy.” Therefore, it suggests some patients may still benefit Keytruda monotherapy in later lines of treatment.
The FDA did acknowledge there’s an unmet need, but it simply isn’t sure Keytruda monotherapy is the drug to fill it. Nevertheless, the agency said it’s willing to consider other trial results evaluating Keytruda as part of combination regimens. As the landscape of gastric cancer treatment changes, “it is unclear what role if any [Keytruda] as a single agent should have in unselected patients with CPS>=1 disease,” FDA staffers wrote in the briefing document.
During a meeting in June 2019, the FDA had said that any one of the three ongoing phase 3 studies in gastric cancer—Keynote-859, Keynote-811 or Keynote-585—could serve as the new confirmatory study for the current indication, Merck said. Results from those trials are expected within the next one to three years, according to the company.
Of those, the FDA considers Keynote-859 the most relevant; it tests whether adding Keytruda to chemotherapy in first-line HER2-negative gastric/GEJ adenocarcinoma could extend patients’ lives in all-comers, in PD-L1 CPS of 1 and above, and of at least 10. Merck said it modified the sample size and statistical analysis plan of the study based on findings from the Keynote-062 trial.
April 29 afternoon: Keytruda
Indication: Liver cancer patients who’ve been previously treated with Bayer’s Nexavar
Background: The original Keynote-224 trial observed a response rate of 17% for Keytruda in post-Nexavar liver cancer patients. But the confirmatory Keynote-240 trial only showed a death risk reduction of 22%—which wasn’t statistically significant—over placebo.
Meanwhile, Roche’s Tecentriq-Avastin combo has earned a go-ahead for previously untreated liver cancer. It reduced the risk of death by 41% over standard-of-care Nexavar. And the FDA noted that there are no data supporting the use of Keytruda after progression on that Roche regimen. Bristol Myers’ dual immunotherapy Opdivo and Yervoy also carry an accelerated approval for post-Nexavar patients based on tumor response data.
FDA recently rejected Merck and Eisia’s front-line application for their Keytruda-Lenvima combo because it only had tumor response data. Now, the phase 3 LEAP-002 trial is studying that combo in the first line for survival data.
Arguments: While new agents are available, patients still need the option of a single-agent PD-1 inhibitor in second-line liver cancer, Merck said in a briefing document. Some people may not be eligible for Tecentriq in the front line and not appropriate for anti-angiogenic therapies such as Exelixis’ Cabometyx and Bayer’s Stivarga in the second line, it said. Opdivo and Yervoy also had serious side effects such as bleeding, and therefore require prior patient screening.
But the FDA’s position was clear: Although some patients aren't eligible for those drugs, Keytruda doesn’t look like the drug to fill the unmet need without showing a survival benefit.
Merck also suggested it had assumed an “unrealistically high” magnitude for an overall survival effect in Keynote-240—at 35% based on results seen in melanoma and lung cancer—which led to the disappointing readout. It also pointed to a higher use of post-progression therapy for placebo versus Keytruda, saying that could explain why the placebo patients lived a median 10.6 months, which was 2.5 months to 3 months longer than the survival records in placebo arms of similar trials run at the same time.
The FDA acknowledged that the trial might have been successful if assuming a smaller survival difference, but it rejected the idea that post-trial therapy caused the negative readout, given only a small number of placebo patients reached for a checkpoint inhibitor after disease progression.
After the Keynote-240 flop, the FDA agreed Keynote-394 and LEAP-002 could potentially serve as alternative confirmatory studies. But the agency noted that Keynote-394 is only conducted in Asia, where patients might have different underlying liver functions than the Western population.
April 29 afternoon: Opdivo
Indication: Liver cancer patients who’ve been previously treated with Bayer’s Nexavar
Background: In the CheckMate-040 trial, Opdivo triggered a response in 14.3% of patients who had previously received Nexavar. But in the randomized CheckMate-459 trial in newly diagnosed patients, the drug only pared down the risk of death by 15% versus Bayer’s standard of care med.
Meanwhile, Roche’s Tecentriq-Avastin combo has been approved for previously untreated liver cancer, with a death risk reduction of 41% over Nexavar.
Arguments: Bristol Myers Squibb argued it might haven’t followed the confirmatory CheckMate-459 trial for long enough. It noted an extended time to separation of the survival curves between the two trial arms than it had previously expected.
The company had anticipated a delay of five months after treatment to start to see an average death risk reduction of 26%, but instead only reached that difference after eight months. The New York pharma suggested the late separation was likely caused by a higher-than-expected use of subsequent therapy, including the use of immuno-oncology agents in the Nexavar arm after disease progression.
The FDA instead generally considered those post hoc analyses as hypothesis-generating only and therefore “generally are not helpful.” It also noted that only about 20% of patients in the Nexavar arm were treated with I-O after their disease progressed.
Bristol Myers also pointed to improvements in safety and quality of life for Opdivo, but the FDA didn’t see them as extra credits for the PD-1 inhibitor, given the trial’s main endpoint of survival wasn’t met.
The company also pointed to the fact that it’s the only drug among existing therapies mentioned by the FDA to have clinical data in patients with Child Pugh cirrhosis B; compared with the A type that’s commonly enrolled in clinical trials, patients who score B on the cirrhosis severity marker have bad liver function and therefore have poorer prognoses. In these sicker patients, Opdivo triggered a response of 10.2% as estimated by investigator, but the FDA noted the number was only 6.1% by blinded assessment.
Bristol Myers has proposed CheckMate-9DX to serve as the new confirmatory trial. The study is testing adjuvant Opdivo use versus placebo in people at high risk of recurrence after curative surgery. It finished enrollment in December, and the final readout is expected in the first quarter of 2023.
The FDA has its concerns. “There would be some uncertainty with respect to whether an effect in patients with low burden of disease would translate into benefit in the advanced disease setting,” the FDA staffers wrote in the briefing.
Other Opdivo liver cancer trials include CheckMate-9DW, which is evaluating the pairing with Yervoy and is meant to confirm the combo’s second-line accelerated approval.