Weber Discusses Sequential Versus Concurrent Immunotherapy in Melanoma

Jeffery S. Weber, MD, PhD

Nivolumab (Opdivo) and ipilimumab (Yervoy) have demonstrated considerable success in the field of metastatic melanoma as both single agents and in combination. However, questions remain regarding sequencing the agents and the high toxicities that often occur when the 2 immunotherapies are used together.

To explore these challenges, OncLive spoke with Jeffery S. Weber, MD, PhD, deputy director of the Laura and Isaac Perlmutter Cancer Center, co-director of the Melanoma Program, and head of Experimental Therapeutics at NYU Langone Medical Center.

OncLive: Could you describe the goals and design of the CheckMate-064 trial?

Weber discusses findings from the phase II CheckMate-064 trial, in which patients were randomized to receive either nivolumab followed by ipilimumab followed by nivolumab maintenance therapy, or ipilimumab followed by nivolumab and maintenance therapy with nivolumab. He also places the results from this sequential study in the context of findings from the phase II CheckMate-069 trial, which examined the combination of nivolumab and ipilimumab in patients with advanced melanoma.Weber: CheckMate-064 was a trial that was written by F. Stephen Hodi, MD, of Dana-Farber Cancer Institute, and myself, which became a Bristol-Myers Squibb—sponsored trial, that ultimately evolved to a randomized phase II trial. This was a trial in which 140 patients were randomly allocated to receive sequential immunotherapy. This means that patients were randomized 1:1 to receive 12 weeks of induction nivolumab in what we called cohort A, followed by a forced switch to 12 weeks of ipilimumab and then maintenance nivolumab every other week at 3 mg/kg until progression, toxicity, or refusal.

In cohort B, the opposite sequence was given: 12 weeks of ipilimumab at the standard dose at a schedule of 3 mg/kg every 3 weeks for 12 weeks and then a forced switch to 12 weeks of nivolumab and then maintenance nivolumab until progression, toxicity, or refusal. Cohorts A and B varied only in the reversal of the sequence, the total treatment would have been the same in either cohort; the timing was just different.

The primary endpoint was grade 3, 4, and 5 adverse events (AEs) by the end of the second induction cycle at week 25. Secondary endpoints included response rate, PFS, and a variety of biomarker studies. It was also a very important biomarker study. Patients on this trial predominantly have leukapheresis at time 0, after the first cycle at week 13, and after the second induction cycle at week 25. We had biopsies before and after in a fair number of patients.

The exploratory endpoint also included survival comparisons between the arms. The demographics showed that while there was a fairly good balance between the 2 arms, there was actually an imbalance in the PS1 patients who favored cohort A. There was also a slight imbalance in patients who had brain metastases that went the other way. There was also what appeared to be a significant imbalance in the PD-L1 population. There were more PD-L1—positive patients by the 5% in cohort A than cohort B. Therefore, there were 2 prognostic factors that appeared to favor arm A—but one that appeared to favor arm B—which was fewer patients with brain metastases.

What were the findings regarding toxicity?

The eligibility was very straightforward. You could have failed a systemic therapy, but you could not have had a previous PD-1/PD-L1 antibody or a CTLA-4 antibody. Most of the patients were treatment-naïve, but a good proportion of them had 1 prior systemic regimen. If you had brain metastases, you had to have them treated and be stable for at least 4 weeks. It was a metastatic cohort that one would see in any registration study, with the exception of that slight imbalance between the arms.In the first induction cycle, there were no surprises as they were only receiving 1 drug. There was a 7% grade 3/4 AE rate in the nivolumab arm and about a 20% grade 3/4 AE rate in cohort B that received ipilimumab.

The surprise was that, by week 25, we actually saw more toxicity in cohort A at about 50% versus about 42% in cohort B. That is getting up to the rates that we are seeing with concurrent therapy. Please be reminded that the reason we did the trial was to try and achieve equal efficacy of concurrent therapy by sequencing the drugs to lower the toxicity.

What were the efficacy findings?

If you look overall—meaning beyond week 25 to include maintenance therapy—about half of patients received maintenance treatment in either arm. There was well over a 50% rate of grade 3/4 adverse events—no deaths—but a 50% rate in arm A and a slightly lower rate in arm B. This suggests that, in terms of toxicity, we did not do better than concurrent therapy.The best overall response rates were surprisingly different. Arm A, at the end of the day, had a best overall response rate of about 54%. It was only about 30% in arm B. That is a big difference, as if somehow getting the induction of ipilimumab with the forced switch to nivolumab compromised patients’ ability to respond to the subsequent nivolumab.

We were very surprised. This translated into a very impressive difference in survival. If you look at the survival curves at 1 year, with about 18 months of total follow-up, there was a very big difference between cohorts A and cohort B. Cohort A had a survival somewhat similar to the concurrent arm of the CheckMate-069 trial, whereas cohort B was probably not even as good as the crossover arm that received ipilimumab on the CheckMate-069 trial, when many of them crossed over to nivolumab.

What should be taken away from these findings?

The hazard ratio for the difference between arm A and arm B, in terms of survival, was about .5. That is a very significant hazard ratio. If you look at the curves, they split apart very early—literally before the first evaluation at week 12, and they stay apart and appear to almost plateau. For arm A, the survival was around 60%, which is not that much different than the 69% seen in the concurrent CheckMate-069 trial that led to the approval of the concurrent regimen.Our conclusions were that there was not a very big difference in toxicity; both arms were very similar to concurrent therapy. The hypothesis that we had when we set out was not satisfied. We could not reduce toxicity by giving sequential versus concurrent ipilimumab and nivolumab.

The other major conclusion was that one sequence was clearly superior in terms of response and survival. It is much better to get nivolumab followed by ipilimumab versus the reverse. The numbers were not that big, and there were some imbalances between the 2 arms, but those tended to cancel each other out. My impression was that I would absolutely discourage patients from receiving frontline ipilimumab and then, at some point, going to nivolumab.

It simply reinforces the opinion of many investigators in the field that you should go with nivolumab first and then, maybe, if you need to switch to ipilimumab, there is no advantage of doing that over concurrent therapy—in terms of toxicity.

Weber JS, Gibney G, Sullivan RJ, et al. Sequential administration of nivolumab and ipilimumab with a planned switch in patients with advanced melanoma (CheckMate 064): an open-label, randomised, phase 2 trial. Lancet Oncol. 2016;17(7):943-955.