Thank you, Suzanne. On behalf of the study investigators I’m very pleased to present the first disclosure of overall survival results from the phase III CHECKMATE 067 trial which explores the efficacy of nivolumab, ipilimumab or the combination of both in patients with previously untreated metastatic melanoma. These are my disclosures.
Manipulation of immune checkpoints such as CTLA4 or PD-1 with targeted antibodies has recently emerged as an effective treatment strategy for multiple cancer types. In the phase II randomised CHECKMATE 069 study the combination of nivolumab, an anti-PD-1, and ipilimumab, an anti-CTLA4, resulted in high rates of response, a median duration of response of at least two years and a favourable two year survival rate of 64%. Back in 2015 we presented the initial results of the present phase III study which showed significant improvements in the co-primary endpoint of progression free survival as well as significant improvements in response for both nivolumab containing arms versus the ipilimumab control group. CHECKMATE 067 is the first phase III study to report survival outcomes with the combination of anti-PD-1 and anti-CTLA4 therapies.
In CHECKMATE 067 945 patients with unresectable stage 3 or 4 melanoma were randomised 1:1:1 to either a combination arm, nivolumab or the control group, ipi. Randomisation was stratified by BRAF mutational status, M stage and PD-L1 expression. Patients were treated until progression or unacceptable toxicity. It is important to note that this study was not designed to compare between both nivolumab containing arms but descriptive analyses will be presented.
The co-primary endpoints were progression free survival and overall survival in the intent to treat population. Secondary and exploratory endpoints included response rate by RECIST version 1.1, correlation of PD-L1 with survival endpoints and safety. As specified in the protocol, the OS analysis was to occur after all patients had 28 months of follow-up at which time a total of 644 deaths were projected to be observed. However, the actual number of deaths across the study at the time of the September 2016 database lock was 467 which was 28% lower than projected. As per the study design the study was not powered for a comparison between nivo and the nivo/ipi combination but descriptive analyses again will be presented.
Overall the updated progression free and overall survival results with 28 months minimum follow-up are consistent with the primary nine month analysis, as reported in 2015. It should be noted that the complete response rates have increased in all three treatment arms with longer follow-up, specifically the complete response rate increased from 12% to 17% in the combination arm and from 10% to 15% in the nivo alone arm. This slide also shows impressive median durations of response with 28 months minimum follow-up has still not been reached for the combination.
The formal analysis of the co-primary endpoint of overall survival demonstrates a significantly lower risk of death in the nivolumab monotherapy group by 37% when compared to ipi, as well as with the combination of nivo plus ipi versus ipi. The medians have still not been reached for either nivo containing arms whereas the median was 20 months in the ipi comparator. In descriptive analyses the reduction in risk of death with nivo plus ipi relative to nivo alone was 12%. It should be noted that the OS analysis, as specified in the protocol, was to occur after all patients had 28 months of follow up, at which time a total of 464 deaths were projected to be observed.
Two year survival rates were 64% for the combination and 59% for nivo alone which is similar to recent updates for the phase II CHECKMATE 069 trial and the phase III CHECKMATE 066 nivolumab monotherapy trial. Finally, in the middle of the slide you can see that subsequent systemic therapies were received by 62%, 44% and 32% of ipi, nivo and combination treated patients respectively. The median time to any subsequent systemic therapy has still not been reached in the combination group and was 27 months in the nivo alone arm. Remarkably 66% of combination treated patients remain free of treatment at two years.
As stated in the study design, patients were stratified by several factors, including BRAF status and PD-L1 expression. Notably for both nivo containing arms in the BRAF mutated subgroup median overall survival has still not been reached with a descriptive hazard ratio between nivo and nivo plus ipi of 0.71. Two year OS rates were 71% for the combination, 62% for nivo and 51% for ipi.
In the subgroup of patients whose tumours had PD-L1 expression less than 5% the median overall survival for the two nivo containing arms has still not been reached with two year OS rates of 63% and 55% respectively. The descriptive hazard ratio of 0.84 is consistent with the main ITT findings. Similar to what we’ve reported previously with PFS, overall survival appeared comparable in the nivo and the nivo plus ipi patients with tumour PD-L1 expression of at least 5% with a hazard ratio of 1.05. In contrast to the survival results it’s clear that treatment with the combination resulted in a numerically higher response rate versus nivo alone, regardless of PD-L1 expression. For example, in the subgroup of patients with tumours expressing PD-L1 of at least 5% response rates were 74% with the combination in comparison with 59% for nivo.
The safety profile for the three groups was consistent with earlier experience. The incidence of high grade adverse events was highest in the combination group and lowest in the nivo alone group. Despite this, most select immune mediated adverse events were manageable and the vast majority resolved in 3-4 weeks. Close to 40% of patients treated with nivo plus ipi discontinued treatment, primarily within the combination phase of treatment. Discontinuation of treatment did not appear to preclude benefit with 70% of patients developing a response and, impressively, their median OS has still not yet been reached.
Finally, the last row and the footnote provide details of the four deaths were attributed by investigators as being treatment related. The one death in the nivo and the one death in the ipi arm were previously reported in our initial NAJM publication. For nivo plus ipi two new late treatment related deaths were reported greater than 100 days over the last dose of study treatment. The first death occurred in a patient with an extensive history of cardiac disease who died due to cardiomyopathy following a re-challenge with an anti-PD-1 agent outside this trial. The second death occurred in a patient with persistently elevated liver enzymes in which the patient was not adherent to the recommended safety algorithms.
In summary, based on both the current OS data and the previously reported PFS results, CHECKMATE 067 was a positive study that met both of its co-primary endpoints. Descriptively the combination showed better overall OS, PFS and more durable response outcomes than nivo alone with consistent results seen across clinically relevant subgroups such as those with low PD-L1 expression, patients with elevated LDH and those with tumours that were BRAF mutated. In the roughly one-quarter of patients with tumours having PD-L1 expression greater than or equal to 5% both nivo and nivo plus ipi resulted in a similar prolongation of survival although the response rate remained numerically higher with the combination. The safety profile of the combination is consistent with earlier experience with the majority of immune mediated adverse events resolving within 3-4 weeks. Even in patients who discontinued the combination due to toxicity an impressive survival benefit and responses over 70% were observed.
In conclusion, considering all of the study findings, first line nivo plus ipi may represent a means to improve outcomes versus nivo. Although the data by tumour PD-L1 expression are intriguing, the role of PD-1 as a predictive biomarker is not yet fully understood and the data is still immature. Therefore, decisions about the optimal first line treatment choice should be made on an individual patient basis with all factors taken into consideration. Thank you.
Thank you James. So, we’ve just heard about the results of combining drugs that block two distinct checkpoints, CTLA4 and PD-1. When used in combination these drugs have a powerful anti-tumour effect in melanoma; this also carries with it an increased risk of toxicity, as you’ve just heard. This combination is the first immunotherapy combination to be approved by the FDA. It reflects a very active area of research now in immuno-oncology with several hundred different clinical trials of various combinations ongoing. In a situation where there is benefit but there is also some risk of the therapy we like to try to find biomarkers that would help us select the patients who are most likely to respond and, on the other hand, least likely to incur serious side effects. In your presentation, James, you showed some intriguing data on patients who had the common melanoma BRAF mutation where it appeared as if the combination therapy did a lot better than nivolumab monotherapy and also in the PD-L1 negative melanomas it seemed as if there might be an advantage. What work needs to be done now to further explore this? That’s my first question, the second question is if you have a newly diagnosed stage 4 melanoma patient in your clinic today what does that conversation look like?
So the first part of that question, what work needs to be done, if we pick both of those examples, PD-L1 expression and tumour BRAF status, these are very intriguing analyses that are presented here but understanding the biology fundamentally is what we really need to do to really see, for example, if what’s been shown with BRAF mutational status is a real effect or not from a biological perspective.
PD-L1 expression is obviously not an ideal marker, we can see that. Work has been done to try and look at different thresholds of PD-L1 expression, Georgina Long presented that at ESMO last year, and clearly it’s not binary in the sense that something like BRAF mutation status is. I’m not sure there is necessarily a clinically useful cut-off for use in the clinic, which brings me to the second part of the question – what do you do? If you’re sitting down with a patient in clinic, they’ve got advanced melanoma, you’re discussing the treatment options, there’s a lot of different factors that you’re going to think about. You’re going to think about the patient’s general level of fitness, the tumour may have a BRAF mutation so you might be thinking about BRAF targeted therapy, you might be thinking about side effects. I’d say there’s a whole constellation of factors certainly that I’d discuss with my patients in clinic when we’re formulating a treatment plan. So at the moment I don’t think there really is any one thing that says we’re going to go down this route or that route; I think we have to put everything together.
Open it to the floor for questions.
[Audience member] Nick Mulcahy from MedScape. Dr Larkin, I’m surprised you’re so subdued about the BRAF as a biomarker and an indicator because that kind of jumps out, there’s a big difference there between the survival benefit compared to the single agent nivo. Otherwise I can’t see why you would want to do the combination first with the expectation of more toxicity and cost. Do you want to comment on that?
Yes, sure. At the moment these are observations from a phase III study and it’s part of what I just said to Suzanne. So it’s interesting but I think still speculative at this stage and it’s not ready, for me at least, to use for my clinical practice based on those kinds of analyses. So I would really want to understand better what that meant and certainly I’d discuss these kinds of things with my patient but it’s not strong enough for me to say we should do this or that currently based on what we’ve seen.
Fair enough, thanks.
[Audience member] Peggy Eastman with Oncology Times. Immunotherapy drugs are extremely expensive so my question is if you had a biomarker that proved true that was validated do you think payers would be more inclined to want to pay for this combination of therapy rather than just one alone?
I think that’s a great question. Working in the UK and in Europe we have real reimbursement fragmentation and payers have different rules from regulatory bodies sometimes. But ultimately we all want to have better biomarkers so that we can rationally select patients for treatment in terms of efficacy, in terms of side effects and also in terms of financial consequences. So we all want that but the biomarkers need to be robust if they’re being selected by anybody – clinicians, regulators and payers. Just on cost I’d just make one point – in the UK NICE, the National Institute of Health and Clinical Excellence, a body which basically fundamentally works out whether the efficacy of a new treatment represents value for our state-funded health system, all of the recently approved melanoma treatments, pretty much without exception immunotherapy and targeted therapy, have been looked at by NICE and have been felt to represent sufficiently good value for money to be approved.
[Audience member] Charles Bankhead, MedPage Today. Is there a standard or recommended test, PD-L1 test, for nivo as there is with pembro?
At the current time the answer to that is no. In the labelled for nivo and nivo/ipi combination for melanoma in the US and Europe there is no stipulation in the same way that I understand for non-small cell lung cancer in pembro but there’s a need to measure PD-L1 expression in order to identify patients’ first line lung cancer. I don’t treat lung cancer but I think that’s the situation. So the short answer is no.
[Suzanne Topalian] May I just add to that? It’s not a required test so it’s not a companion diagnostic but the FDA calls it a complementary diagnostic. So there is a test, the 28-8 test it’s called, which is a PD-L1 immunohistochemistry test that can be used based on the early results from this trial to make decisions between the combination therapy or nivo monotherapy. Where that will stand now with these new overall survival results we’ll have to see.
[Audience member] What seems to be the emerging practice for testing with avelumab and atezolizumab and the others that are coming along? Is there a practice that is emerging at all for how to test for this? Also I notice that in one of the studies you were looking at a cut-off of greater than 50%, in this one greater than 5%. It seems like it’s going to be difficult to sort this out until there is some sort of standardisation somewhere along the way.
I think that’s going to be necessary, some sort of standardisation but I don’t think it’s going to be straightforward. You’re going to get variation between diseases; you’re going to get different thresholds; you’ve got different antibodies, different assays and so on and so forth. So, yes, we’re going to need some standardisation but it’s going to take a bit of time to do that. The other problem is with PD-L1 expression it can vary over space and time; it can be inducible. It really is different from BRAF mutational status, for example, which is relatively straightforward to measure and relatively fixed over space and time. So there are some technical challenges there as well.
