Round 1: Treatment for Patient Case 1

The debate opens with a case exploring first-line treatment for BRAF wild-type metastatic melanoma, weighing the long-term benefits of ipilimumab+nivolumab against the tolerability of nivolumab+relatlimab.

To hear arguments related to treatment for patient case 2, see Round 2.

To hear arguments related to treatment for patient case 3, see Round 3.

Dr Jose Lutzky: So, this is a 74-year-old woman with stage IIC melanoma that was resected two years ago. The patient decided not to receive adjuvant therapy but presented with weakness, lack of energy that was limiting the daily activity.  

So, the ECOG performance status was 1. The physical exam was normal. She had imaging with a PET-CT scanning that showed liver, lung, and cutaneous metastases. They were biopsy proven. And an MRI of the brain revealed no evidence of metastatic disease to the brain.  

The molecular testing showed that she was a BRAF wild type, and the LDH was 350 units per liter, slightly above the reference range, with the upper limit of normal being 225. In terms of past medical history, the patient had hypertension that was managed and controlled.  

So, I want to start with Dr Tarhini, who will defend the IPI + NIVO field. 

Dr Ahmad Tarhini: Thank you, Jose. And I want a thank you to the organizers for this opportunity to be here with Dr Smithy and Dr Lutzky to have this discussion.  

So, clearly this is a patient with metastatic melanoma with elevated LDH but no brain metastases, good performance status, meaning ECOG performance status of 1. It’s a BRAF wild type, we have a lot of treatment options; If we look at the NCCN guidelines or other national guidelines, these may may include anti-PD-1 as monotherapy, anti-PD-1 plus anti-CTLA-4 (IPI + NIVO), anti-PD-1 plus LAG-3 inhibition (NIVO + RELA), among other options listed there.  

If you look at anti-PD-1 as monotherapy and data from the KEYNOTE-006, now we have 7-year data, and more recently, last year, published the 10-year data. There are durable benefits in a subset of patients. We look at, let’s say the 7-year progression-free survival data, this is about 24% of patients may have PFS at about 7 years. Those that are alive, overall survival is about 38% of patients. 

The CheckMate-067, now we have 10-year data. Obviously, this is the study that tested ipilimumab plus nivolumab, with IPI, 3 milligram per kilogram, and NIVO, 1 milligram per kilogram, or nivolumab as compared to ipilimumab as the control arm. This study was a major advance obviously in the treatment of metastatic melanoma in terms of the likelihood of response, but also the durable benefits now up to 10 years. Looking at the 10-year PFS data, 31% of patients with IPI + NIVO treated are expected to be progression-free and alive at about 10 years.  

There is an overall survival rate, the number is 43%. Now, if we look at the melanoma-specific survival, not shown here, but it's 52% of patients. Obviously, MSS is an estimated, exploratory analysis, but also gives us an idea about the likelihood of potential cure, let’s say, with immune checkpoint inhibition in patients. So it is a, let’s say a regimen that has been well proven, well demonstrated, and has long-term benefits. Now, the caveat here is, or let’s say one of the limitations, is the toxicity profiles. We look at the likelihood of grade 3-4 toxicity with IPI + NIVO, with IPI-3 NIVO-1, it's close to 60% in the latest 10-year paper. Those that have to discontinue treatment because of toxicity are about 45%, from this paper as well. So we have good efficacy, concerns about toxicity.  

Now, with regard to the flip dose—so 1 milligram per kilogram IPI, and NIVO 3 milligrams per kilogram—that was tested in the CheckMate-511 study. The CheckMate-511 study looked at IPI-3 + NIVO-1 as compared to IPI-1 + NIVO-3 with a primary endpoint of actually toxicity, of safety, so it was not powered to look at efficacy. It numerically reported the response rates, but it was primarily a toxicity study. It is less of an interest to me now when I'm concerned about toxicity because of the NIVO + RELA data. So, if I have to use IPI + NIVO, I usually use IPI-3 + NIVO-1. So I’ll leave it at that, and Dr Smithy. 

Dr James Smithy: Great. So, thanks so much for teeing this up. I think Dr Tarhini made an excellent case for the kind of rationale for combination checkpoint blockade as a benefit over PD-1 monotherapy in these patients, but the question really is, which combination is better and can we get similar efficacy with a more favorable toxicity profile.  

So, I definitely want to introduce the data from RELATIVITY-047. This is a randomized, phase 3 study of patients that were previously untreated, comparing nivolumab monotherapy with nivolumab and relatlimab, so relatlimab, novel checkpoint, which is LAG-3, and this is a fixed-dose combination. So, there are 355 patients randomized to NIVO + RELA and 359 randomized to NIVO here. This population reflects very much the patient that we're talking about in this case. There's a mix of different M stages, which have different prognostic features. But there's a significant number of patients with M1c disease in this trial. Everyone had an ECOG performance status of 0 or 1, and we do see some patients with elevated LDH numbers, about a third of patients, 36% of the NIVO + RELA group, had a greater than upper limit of normal.  

And so the primary endpoint of the study was progression-free survival by blinded independent central review. We have the main headline here, and what we can see is that there is a statistically significant PFS benefit with NIVO + RELA over NIVO monotherapy, the median PFS being 10.2 months with NIVO + RELA versus 4.6 months with nivolumab. You can see that difference in progression, with the progression curves kind of persisting out. We now have data that's 4 years updated with this trial. We don't have the 10-year data yet because it's a newer combination, but hopefully we'll see with further updates of this durability can persist over time. 

Then we have an overall survival curve as well where we see not a statistically significant trend result here, but definitely separation of the curves, which I think is promising as well. You can see the curves flattening out as we get out to 48 and 60 months, where it looks like there is a trend towards improved overall survival of NIVO + RELA compared to NIVO monotherapy. Melanoma-specific survival, as Dr Tarhini mentioned as well, more specific here for our melanoma endpoints, and this looks promising, too.  

But I think the response rate is going to be different depending on how this is adjudicated for different trials. And I just want to point out that for RELATIVITY-047 in this initial trial, this was by blinded independent central review, which was different than CheckMate-067. So there's a response rate of 43% for NIVO + RELA, is significantly better than the 33% for NIVO, and it's just hard to compare apples to oranges across the two trials, because they were adjudicated differently.   

Ultimately, not so relevant for our case here, but we're thinking about, if a patient did have brain metastases, could there be activity of NIVO + RELA in the brain? This analysis here, just looking at CNS-progression free survival is a promising, I think, endpoint as well. So you can see patients in both arms, NIVO + RELA with the light blue and then NIVO monotherapy, tracked over time as the likelihood of being free of a CNS progression event, and there is a separation here of these curves. It appears that there are fewer events for CNS progression with NIVO + RELA. I think there are ongoing, active efforts, including the BLUEBONNET study, looking at whether NIVO + RELA can be used to treat active brain mets. It's a little bit out of the scope of this debate, but I think some early signals that it could be helpful in that setting as well.  

This slide here is looking at an indirect comparison between NIVO + IPI and NIVO + RELA. What we wish we had to answer this question, like the definitive thing, would be a phase 3 trial that's randomizing people with similar characteristics to either treatment. We don't have that, so the best we can do is compare patients that were treated in CheckMate-067 and in RELATIVITY-047. 

And this is one of the, I think, key findings from this analysis. This was presented at ASCO and then published recently in JCO, where taking patients from each arm, so from the NIVO + IPI arm from CheckMate-067, and then the NIVO + RELA arm from RELATIVITY-047, and first correction that was done was making sure everything was adjudicated the same, so using a blinded, independent, central review for both sets of data was important. Then the second piece was looking at the weighting of the patient characteristics too, because whenever you make a cross-trial comparison, you're worried about whether the features of one group reflect those of the other group. So what you can do is assign a propensity score for each patient and make sure that you reflect the baseline characteristics similarly across the 2 groups. So with all that said, once accounting for all those differences in the 2 trials, if you overlay the PFS curves from the 2 arms, they look very strikingly similar. So I think this is the best data we have to say that for at least patients without active brain mets at the time of starting treatment, it doesn't seem like NIVO + IPI is associated with a significant PFS benefit one way over NIVO + RELA. If we look at OS, it's a very similar curve as well. The NIVO + RELA, actually if you look, there's a little bit of a hint of daylight where the NIVO + RELA curve looks slightly better, but I think we can both say that these don't look dramatically different one way or the other. Melanoma-specific survival, very similar result.  

Okay, so that's the efficacy data. I think it's a bit weaker than a standardized clinical trial, but I think the difference in safety is really what's going to seal this argument, and I think this slide is probably the most important of all of these in saying that even if we're saying we don't think one is markedly more effective than the other, I think we can say that NIVO + RELA has a significantly lower risk of severe IREs compared to CheckMate-067. And this would be the crux of why I would recommend this regimen over NIVO + IPI for this patient.  

As Dr Tarhini mentioned, the rates of grade 3 or 4 IREs with NIVO + IPI is approaching 60%. And then with NIVO + RELA, we see that number at about a third, in the 22% to 23%, and this is with or without weighting; the numbers look very similar.  

I think this is something not to be underplayed for patients because I think these types of severe IREs can have patients end up in the hospital and you have to discontinue therapy. And really we have emerging data now that high-dose steroids can impede the efficacy of ICI responses over time as well.  

So I think with all of this considered, if you don't have active brain mets, I think we have a pretty strong case for trying NIVO + RELA first before turning to NIVO + IPI. 

The next question is, okay, what if we don't use NIVO + RELA right now, we want to save it for later? Is that an effective strategy as well? So before it was studied as a first-line regimen, NIVO + RELA was studied in the later line, like PD-1 refractory setting. This is the RELATIVITY-020 study. There were two different populations in this trial. The upper panel is looking at patients who had 1 line of anti-PD-1 therapy, and then the other, the one below, is allowed to have more than one line. The results look very similar between the two, where the median PFS for both of these is only around 2 to 3 months. So I think these data bring into question how effective can NIVO + RELA be in the later-line setting? We do have some emerging real-world data sets, which are international and going through the review process now. Maybe it's a little bit better if you don't have such a heavily treated population as RELATIVITY-020, but I still think this gives me pause to say, "Oh, I can definitely use NIVO + RELA later on if I don't use it up front." I think that's more or less the initial comparison between these. 

I just want to point out one other data set before I wrap up. We talked a lot about the 067 and 047 data, but how does this actually look for patients treated in the real-world setting as well?  

And so there's this Flatiron database, which uses claims codes, and a real-life EHR aggregated database, and essentially, a very similar analysis showed a very similar result using these non-trial patients as well. This was presented by my colleague, Mike Posto, at ASCO earlier this year and took 408 patients with NIVO + RELA and 600 patients with NIVO + IPI, looked at OS and real-world progression-free survival on these patients. And again, the efficacy curves basically overlay one another. So, recapitulating what we saw in that matched indirect treatment comparison as well.