NEW YORK – During the European Society for Medical Oncology's Virtual Congress on Monday, a pair of Phase III studies demonstrated that first-line treatment with immune checkpoint inhibitors combined with chemotherapy could significantly prolong survival for patients with advanced gastric and esophageal cancers.
In both studies, the overall population of patients receiving the combination of a checkpoint inhibitor and chemotherapy benefitted, but the subpopulations with higher degrees of PD-L1 expression in their tumors lived longer. The detail left some experts at the meeting dissatisfied with the design of these studies, pointing out that although the investigators were claiming benefit in all comers, patients with higher levels of PD-L1 expression may have been carrying the survival advantage seen in the trials. The percentage of PD-L1-high patients enrolled in the trials, they further noted, may not reflect that of patients with gastroesophageal cancers in the real world.
The presented studies included the randomized Phase III CheckMate-649 trial, which evaluated nivolumab (Bristol Myers Squibb's Opdivo) plus chemotherapy versus chemotherapy alone as first-line treatment for patients with advanced, HER2-negative gastric cancer, gastroesophageal junction cancer, or esophageal adenocarcinoma; and the Phase III KEYNOTE 590 trial, which evaluated pembrolizumab (Merck's Keytruda) plus chemotherapy versus chemotherapy alone as first-line treatment for patients with locally advanced, unresectable or metastatic esophageal adenocarcinoma, esophageal squamous cell carcinoma, or Siewert type 1 esophagogastric junction adenocarcinoma.
"It's a good day for patients with gastric cancer," said Elizabeth Smyth of Cambridge University Hospitals in a discussion following the presentations, both of which showed that the addition of immune checkpoint inhibition to chemotherapy prolonged overall survival versus chemotherapy alone. The results of the trials were historic in that the cancers evaluated — gastric, gastroesophageal junction, and esophageal cancers — have, for the most part, been treated with marginally effective chemotherapy strategies in the first-line setting since the 1990s, and the five-year survival rates for advanced stages of esophageal cancer and gastric cancer hover around five percent.
In the CheckMate-649 trial, which randomized 1,581 patients, those who received nivolumab plus chemotherapy experienced a median overall survival of 13.8 months versus 11.6 months among patients who received chemotherapy alone. Patients whose tumors expressed a PD-L1 combined positive score (CPS) of at least 5 lived an average of 14.4 months on the checkpoint inhibitor plus chemo versus 11.1 months for those on chemo alone.
In the KEYNOTE-590 trial, which randomized 749 patients, those who received pembrolizumab plus chemotherapy lived for an average of 12.4 months versus 9.8 months for patients who received chemotherapy alone. Broken down by histology and PD-L1 expression, patients with esophageal squamous cell carcinoma and a PD-L1 expression CPS of at least 10 derived the greatest degree of overall survival benefit, living an average of 13.9 months on pembrolizumab plus chemo versus 8.8 months with chemo alone.
Jaffer Ajani of the MD Anderson Cancer Center said in an interview that while he agreed that the study results were no doubt practice changing, he felt uncertain about the subgroup analyses according to PD-L1 expression, and he noted that neither study identified responders and non-responders with the specificity that he would have liked to see.
Survival according to PD-L1 expression
During their presentations, both Markus Moehler of the University Medical Center of Johannes Gutenberg University Mainz in Germany, who led CheckMate-649, and Ken Kato of the National Cancer Center Hospital in Japan, who led KEYNOTE-590, said that the combination of an immune checkpoint inhibitor plus chemotherapy should be the new standard of care in the first-line setting for these patients regardless of PD-L1 expression.
In the CheckMate-649 trial, the dual primary endpoints were overall survival and progression-free survival specifically in patients whose tumors had a PD-L1 CPS of 5 or higher. As secondary endpoints, researchers also considered both the overall survival and progression-free survival benefits of the nivolumab-chemotherapy regimen in the overall population. The investigators did see a clinically meaningful benefit in both the higher PD-L1 expression group and the overall group, but they did not specifically evaluate the population of patients whose tumors had a PD-L1 CPS of between 1 and 5.
Because of this blind spot in CheckMate-649, Smyth questioned whether oncologists can truly trust that all of their gastric and esophageal cancer patients can benefit from the nivolumab-chemotherapy. In her review of the data, she pointed out that 60 percent of patients in the all-comer group had a PD-L1 CPS score of 5 or greater, and as such, the overall survival benefit seen in the study could be driven mostly by the patients with higher PD-L1 expression.
The all-comer group may be "enriched with highly immunogenic tumors to a degree which would not be seen in an unbiased, non-trial population," Smyth said, explaining further that, while the exact percentage of patients in the real world whose tumors express a PD-L1 CPS of greater than 5 is not known, it is unlikely to exceed 40 percent.
As a result, the study population, weighted toward those with higher PD-L1 expression, likely doesn’t resemble real-world gastric and esophageal cancer patients, and as a consequence, she said that the benefit seen with nivolumab in the study may very well be overestimated. "We may not see these overall survival results in [patients with a] CPS greater than 1, or all patient populations, outside of this trial," she noted.
In the absence of an analysis detailing responses specifically in the lower-PD-L1 expression group, it is not possible to say for certain whether this is the case. As such, Smyth said "we may need further analysis to generalize the results to less immunogenic tumors."
Andres Cervantes of the University of Valencia in Spain raised a similar concern with regard to the KEYNOTE-590 trial. While the results of the trial showed an overall survival benefit in the all-comer group with pembrolizumab plus chemo, the patients with tumors of squamous histology and a PD-L1 CPS of at least 10 derived the greatest benefit. Although the KEYNOTE trial enrolled patients with both squamous cell carcinoma and adenocarcinoma histological tumor subtypes, only 27 percent had adenocarcinomas.
As such, Cervantes similarly recognized that there were gaps in the design of this study, and that those with squamous cell carcinoma — particularly those with a PD-L1 CPS of at least 10 — were carrying the weight of the observed overall survival benefit in the all-comer population.
Both pembrolizumab and nivolumab, it's worth noting, have already been approved by multiple regulatory agencies as a second-line treatment for esophageal cancer patients specifically with the squamous cell carcinoma histology. In the US, pembrolizumab is approved in this setting specifically for patients with a PD-L1 CPS of at least 10. Accordingly, Ajani said that the KEYNOTE-590 results "were not a surprise" for this subset in particular.
"I don't have any doubt about the squamous [population] in KEYNOTE-590," he said, guessing that regulatory agencies would likely be willing to expand the indication to the first-line setting in this subgroup. But the benefit of checkpoint inhibition and chemo for adenocarcinoma patients — particularly those with lower PD-L1 expression — is less certain.
"What we are not given here is the granularity," Ajani said. "If you exclude the high [PD-L1] CPS population, the benefit might disappear."
In other words, in both the CheckMate and the KEYNOTE studies, questions remain as to whether a subset of patients with low PD-L1 expression will truly derive a survival benefit from the addition of immune checkpoint inhibitors to chemo. "When [Merck and BMS] go to the FDA [seeking approval], the FDA is going to ask all of these questions and tease it all out," Ajani predicted, "and they might limit approval to higher [PD-L1] CPS groups."
Merck, BMS vying for a win
How regulators decide to parse this data for these two competing immunotherapies will decide which drug firm, Merck or BMS, captures the biggest slice of the gastroesophageal cancer market. The two trials do not lend themselves to perfect, direct comparison due to differences in the types of patients enrolled. The CheckMate study enrolled patients with gastric, gastroesophageal junction, and esophageal cancers, and limited the histology to adenocarcinoma, while the KEYNOTE trial enrolled patients with both squamous cell and adenocarcinoma subtypes and did not include patients with gastric cancers.
Nonetheless, in assessing the two trials side-by-side, Ajani said that he views the CheckMate-649 data as slightly "more solid" than the KEYNOTE-590 because the former enrolled more patients. If, based on the CheckMate-649 results, BMS scores a regulatory approval for front-line nivolumab plus chemo for patients with adenocarcinoma of the stomach and gastroesophageal junction, Ajani mused, "I don't know what will happen to pembrolizumab … We probably wouldn't use it anymore if we had nivolumab and chemo upfront." (Merck's pembrolizumab is currently approved in this type of cancer, but in the later-line setting.)
However this plays out from a regulatory standpoint, Ajani said, "There's no question this is going to make the landscape more complicated for drug development." Fortunately for patients, a competitive landscape usually means more treatment options.
"This is a big landmark for patients," Ajani acknowledged. "This is the first time in a global study [of advanced gastroesophageal cancers] where the experimental arm has a median survival of higher than 12 months."