NEW YORK – Several months after heeding an independent data monitoring committee's recommendation and calling it quits on the Phase III Keynote-598 study, Merck has gone public with the data backing the decision.
According to the interim results of the terminated study, advanced, previously untreated, PD-L1-high non-small cell lung cancer patients did not benefit from the addition of the CTLA4 inhibitor ipilimumab (Bristol Myers Squibb's Yervoy) to the PD-1 inhibitor pembrolizumab (Merck's Keytruda).
The results of the Keynote-598 trial were presented Friday during the International Association for the Study of Lung Cancer (IASLC)'s World Conference on Lung Cancer (WCLC), and simultaneously published in the Journal of Clinical Oncology.
The Phase III trial pitted the pembrolizumab-ipilimumab combination against pembrolizumab monotherapy in the first-line treatment setting for NSCLC patients whose tumors express PD-L1 in at least 50 percent of their tumor cells and who did not harbor EGFR or ALK alterations. Pembrolizumab monotherapy is already approved in the US for NSCLC patients with PD-L1 expression in at least one percent of tumor cells, and as such, patients enrolled in the Keynote-598 trial were already eligible for pembrolizumab monotherapy. But according to Merck's Senior VP and Head of Global Clinical Development Roy Baynes, until this head-to-head comparison, the jury was still out as to whether there were added benefits to combining pembrolizumab with a CTLA4 inhibitor.
"There have been a number of trials of CTLA4 [inhibitors] added to a PD-1 [inhibitor]," Baynes said in an interview. "But the majority of them have not directly asked the question of whether the CTLA4 [inhibitor] adds efficacy to the PD-1 [inhibitor]. Most of the trials have been indirect comparisons where you have combination [immunotherapy] versus standard-of-care, and monotherapy versus standard-of-care … but no direct comparison of combination versus monotherapy."
Merck's researchers wanted to parse what, if anything, CTLA4 inhibition was contributing in these combination immunotherapy studies, in which PD-L1-positive NSCLC patients had these high response rates. The only way to do this was a head-to-head, randomized trial like Keynote-598.
In total, there were 568 patients enrolled in Keynote-598, 284 receiving pembrolizumab and ipilimumab, and 284 on pembrolizumab and a placebo. After a median follow up of 20.6 months, the overall survival and progression-free survival comparisons between the randomized arms painted a clear picture: the ipilimumab addition did not drive up efficacy. Median overall survival was 21.4 months versus 21.9 months, and median progression-free survival was 8.2 and 8.4 months in the combination and monotherapy arms, respectively. Overall response rates were identical in the two treatment arms — 45.4 percent — and the duration of response was 16.1 months versus 17.3 months with the combo and monotherapy, respectively.
Given these results and the fact that 76.2 percent of patients receiving combination immunotherapy experienced adverse events versus 68.3 percent of patients receiving monotherapy, the independent data monitoring committee suggested not only that Merck terminate the trial, but that the patients in the combination arm stop receiving their ipilimumab infusions.
"The question was directly asked whether the addition of a CTLA4 [inhibitor] would improve the outcome, and the answer was, 'No'," Baynes said.
Implications for treatment, competition
While Baynes did not discuss Merck's competitors in the NSCLC immunotherapy space by name, the elephant in the room when considering the Keynote-598 results is the fact that BMS' own PD-1 inhibitor, nivolumab (Opdivo), is approved in combination with ipilimumab in frontline NSCLC. However, unlike in Keynote-598, the study that led to the nivolumab-ipilimumab approval in NSCLC, CheckMate-227, didn't compare the combination against nivolumab monotherapy but against chemotherapy. This raises questions as to whether ipilimumab could synergistically improve upon the efficacy of one company's PD-1 inhibitor but not another's.
If adding ipilimumab to pembrolizumab did little to improve patients' outcomes over single-agent pembrolizumab in Keynote-598, some oncologists are undoubtedly wondering whether they'd be meaningfully benefitting their lung cancer patients by giving them ipilimumab-nivolumab or just increasing the risk of adverse events?
The answer has competitive implications for the two companies, since Merck and BMS have been vying for market share with their checkpoint inhibitors for years, with lung cancer being a particularly profitable space. Though neither company has released full-year 2020 earnings, in the third quarter, Keytruda brought in $3.7 billion for Merck, and is projected in 2020 to well exceed the $11 billion dollars it netted in 2019. BMS' ipilimumab sales were up 26 percent to $446 million in Q3, but nivolumab revenues were down 2 percent year over year at $1.8 billion. At the JP Morgan Healthcare Conference last month, BMS CEO Giovanni Caforio said that nivolumab would return to growth and that sales in first-line NSCLC, where the drug is paired with ipilimumab, is "going very well."
In BMS' view, the Keynote-598 data and the data backing its own ipilimumab-nivolumab combo cannot be compared, not only because the drugs are different, but because of key variations in study design and patient populations. Keynote-598 restricted patient enrollment to those with very high PD-L1 expression, whereas BMS' immunotherapy combination is approved for those with PD-L1 expression above one percent tumor proportion score.
"Keynote-598 is one trial, which only tells us about Keytruda plus Yervoy — a combination that is not approved — in patients with metastatic NSCLC whose tumors express PD-L1 greater than or equal to 50 percent," Sabine Maier, BMS' VP and head of oncology clinical development, said in an email. "The results have no bearing on the combination of Opdivo plus Yervoy, which has clearly shown long-term survival benefits in Phase III trials across lung cancer, melanoma, renal cell carcinoma and mesothelioma." In a three-year follow-up of CheckMate-227, the overall survival rate among patients receiving BMS' nivolumab-ipilimumab combination was 33 percent compared to 22 percent in the chemotherapy comparator arm. The progression-free survival rates were 18 percent versus 4 percent, respectively.
"The most important benefit observed with Opdivo plus Yervoy is increased durability, including more durable overall survival," Maier wrote. "For the trials where we have seen overall survival benefit with Opdivo plus Yervoy, we now have between two and five years of data clearly demonstrating this long-term benefit."
BMS has paired nivolumab with ipilimumab in first-line NSCLC treatment and in myriad other cancer types, and time and again highlighted what it believes to be a unique synergy between the two agents. "Opdivo plus Yervoy targets two different immune checkpoints to help destroy tumor cells," Maier said. "Yervoy helps activate and proliferate T cells, while Opdivo helps existing T cells recognize the tumor. Some of the T cells stimulated in the presence of Yervoy can become memory T cells, which may allow for a long-term immune response."
BMS has garnered FDA approvals for nivolumab plus ipilimumab for metastatic NSCLC patients with PD-L1 expression levels of at least one percent TPS and together with chemotherapy for metastatic NSCLC patients regardless of PD-L1 expression. However, single-agent nivolumab is approved for later-line NSCLC treatment but does not have the first-line monotherapy indication that Merck's pembrolizumab does. The company has marketed the combination approach undoubtedly hoping for a competitive edge over Merck's pembrolizumab in first-line NSCLC.
Maier specifically referenced the results of the CheckMate-9LA and CheckMate-227 trials, which both showed that the combined immunotherapies were superior to chemotherapy alone. "We are confident in the trial designs of CheckMate-227 and CheckMate-9LA and the relevance of the results to the first-line NSCLC population," Maier said. "These trials have supported regulatory approvals for Opdivo plus Yervoy with or without chemotherapy in NSCLC in nearly 50 countries around the world."
BMS views the Checkmate-227 data to be demonstrative of the nivolumab-ipilimumab combo's benefit over chemotherapy, and while the company notes there is a suggestion in exploratory analysis in the same study that the combo is better than single-agent nivolumab, BMS maintains the study wasn't designed for a definitive conclusion on this front. According to Maier, in the exploratory analysis, among patients enrolled in CheckMate-227 with PD-L1 expression in over 50 percent of their tumor cells, the three-year overall survival in the combination nivolumab-ipilimumab arm was 43.1 percent, and the three-year overall survival in the nivolumab monotherapy arm was 35.5 percent. Patients in the chemotherapy comparator arm experienced a three-year overall survival rate of 25.6 percent.
The CheckMate-227 trial led to FDA approval last May of the nivolumab-ipilimumab combo in first-line, PD-L1-positive NSCLC. However, at the time of that approval, some oncologists were quick to point out that the comparator arm in CheckMate-227 should have been a PD-1/PD-L1 inhibitor monotherapy, or even one of these agents with chemotherapy instead of chemotherapy alone.
This criticism is rooted in the fact that by the time BMS's immunotherapy combo was approved in frontline NSCLC, the standard of care was no longer chemotherapy alone. Pembrolizumab and atezolizumab (Genentech's Tecentriq) were both already available as options for these patients. Pembrolizumab plus chemo, at that point, was the go-to, standard-of-care regimen for frontline PD-L1-positive NSCLC treatment. This was, and still is, particularly true of patients with higher levels of PD-L1 expression.
To that end, Sandip Patel, an oncologist with the University of California, San Diego, who was not involved in Keynote-598 or either of the first-line lung cancer CheckMate studies, said he doesn't feel that the latest data from Keynote-598 is practice-changing so much as practice reinforcing. "I imagine few people were routinely using ipilimumab with anti-PD-1 [therapy] for the PD-L1-[high expressing] population," he wrote in an email. He added that patients whose tumors express PD-L1 in more than 50 percent of cells "are best served" by either anti-PD-1 monotherapy or chemo with anti-PD-1 therapy and that this has been the preferred strategy for himself and many other oncologists.
Merck's Baynes likewise didn't feel that the findings would rock the boat for patients or their oncologists. He suggested that most patients with high PD-L1-expressing tumors weren't receiving a PD-1 inhibitor combined with a CTLA4 inhibitor anyway, and those already receiving single-agent pembrolizumab will likely continue to do so in the wake of Keynote-598.
Still, from a financial standpoint, it is important for patients to know if they stand to benefit comparably from two expensive agents as opposed to one. As of May 2020, the list price for an infusion of 3 mg/kg of nivolumab plus 1 mg/kg of ipilimumab was $21,583, and the list price for pembrolizumab was $9,724.08 per 200 mg dose.
Meanwhile, UCSD's Patel believes there is still a place for nivolumab plus ipilimumab. He considers a CTLA4 inhibitor, a PD-1 inhibitor, and chemo for patients whose tumors have PD-L1 expression below one percent, squamous histology, and central nervous system metastasis.
What about low PD-L1 expression?
The question as to whether the combination approach has a greater benefit than the monotherapy in patients with PD-L1 expression levels below 50 percent remains unknown. Neither one of BMS's CheckMate trials was designed to compare the combination against the monotherapy, and according to Maier, the company doesn't feel that a trial like this is necessary.
The CheckMate-227 trial showed that patients whose tumor PD-L1 expression levels are below one percent benefitted from the immunotherapy combination as well, but because the original design of the trial was not set up to examine the regimen in these patients, the FDA approval ultimately stipulated that patients must express PD-L1 in at least one percent of tumor cells.
"Interestingly, CheckMate-227 had its most intriguing efficacy in PD-L1 less than one percent," said Patel. "[Keynote-598] shows that anti-PD-1 or chemo plus immunotherapy are the best options in PD-L1 over 50 percent population."
When asked why Keynote-598 was designed with a PD-L1 expression cutoff of over 50 percent when the FDA-approved pembrolizumab monotherapy indication includes patients with PD-L1 expression in more than one percent of tumor cells, Baynes explained that this was based on the best available data at the time. When Keynote-598 was designed, single-agent pembrolizumab had just demonstrated a frontline treatment benefit versus chemotherapy in the Keynote-024 trial, which also had the higher, 50 percent or more PD-L1 expression cutoff.
The authors of the Journal of Clinical Oncology paper on Keynote-598 view this as a limitation of the study. "Keynote-598 was designed in 2017, when the only randomized, Phase III data available for pembrolizumab as first-line therapy for metastatic NSCLC were from the Keynote-024 study that demonstrated superiority of pembrolizumab monotherapy versus chemotherapy for metastatic NSCLC with PD-L1 TPS greater than or equal to 50 percent," they wrote. "Thus, KEYNOTE-598 did not assess the role of combined PD-1 and CTLA-4 inhibition in patients with low or negative PD-L1 expression."
The authors went on to write that Keynote-042 would later show that pembrolizumab monotherapy was efficacious in a broader group of patients with PD-L1 expression of at least one percent, but that the benefit in the 1 percent to 49 percent group was more modest than the PD-L1 high group. Interestingly, this same observation was made in BMS' CheckMate-227 trial; the benefit in the broader group of patients with PD-L1 expression of one percent or higher was driven by the higher response rates seen in the over 50 percent PD-L1-expressing group.
Since Keynote-598 only compared the combination to monotherapy in the high PD-L1 expression group, some oncologists remain unsure as to whether the single-agent approach would reign superior in patients with lower PD-L1 expression levels.
Amit Kulkarni, a thoracic oncologist with the University of Minnesota Masonic Cancer Center, who was not involved in Keynote-598, suspects that some patients who express PD-L1 in less than 50 percent of their tumor cells, or even less than one percent, may fare better on pembrolizumab-ipilimumab than with just pembrolizumab. "We know from [the earlier Keynote-042 trial] that the efficacy of single-agent pembrolizumab decreases with decreasing PD-L1 expression," he reminded.
Kulkarni said that the question as to whether in the lower PD-L1 expressing groups the anti-PD-L1/anti-CTLA4 combination approach is preferable over pembrolizumab plus chemo remains "one of the biggest questions in the field." A head-to-head comparison of nivolumab, ipilimumab, and chemotherapy versus pembrolizumab and chemotherapy in the PD-L1-low expression group would help answer this question, but Kulkarni does not have confidence that such a trial will happen. Instead, he is hoping that more precise biomarker-based approaches will help identify best responders to combination immunotherapy.
Patel echoed this hope for better predictive biomarkers for NSCLC in the immunotherapy realm. "We don’t understand the best biomarker strategy to help guide us toward a particular immunotherapy combination yet, unfortunately," he said. "Our results reflect clinical data due to lack of precision immunotherapy biomarkers to optimize patient selection of treatment."