AACR 2014

Benefits found for lung cancer patients with high levels of PD-L1 with MK-3475 treatment

Dr Leena Gandhi - Dana-Farber Cancer Institute, Harvard Medical School, Boston, USA

This is, again, a study of MK-3475 which you already heard about, the description of, from Dr Daud. This is looking at cohorts of non-small cell lung cancer patients treated on the same phase I trial. This study was supported by Merck, otherwise I don’t have any disclosures.

So I will skip the description of the compound itself since you’ve already heard about that, and talk a little bit about the rationale for this particular subset of patients that were evaluated in looking at relationship between outcome and biomarker expression. So this study, KEYNOTE-001, is an ongoing phase I study and it includes multiple different cohorts of patients with non-small cell lung carcinoma. My colleague, Dr Garon, previously presented at the World Lung Cancer Conference on one of these cohorts which were previously treated patients who had mandated tumour biopsies and demonstrated a best overall response rate of 21% by RECIST, 24% by IRRC, which was investigator assessed, RECIST was centrally assessed. What was notable in those patients was that higher levels of PDL1 expression did appear to correlate with an increased response rate to the MK-3475. In addition, there were very durable responses noted with a median PFS of responders not reached at 62 weeks, which is something really not heard of in a population of previously treated non-small cell lung cancer patients. So these findings really formed the basis for looking more closely at that biomarker expression of PDL1 and how it really influences outcome and whether patients should be selected on the basis of that.

So for this particular study a training set of 146 non-small cell lung cancer patients was chosen from multiple different cohorts of patients that were enrolled to this study. That means they had different treatment histories, they were treated at different dose levels and different schedules on the current study but they all had a mandated tumour biopsy and were evaluated for expression of PDL1. An optimal cut point was explored to determine relationship with outcome which will be further validated in subsequent studies.

This just shows you the general design which you heard a little bit about, similar to what Dr Daud presented, but a mandatory biopsy was required within 60 days of enrolment onto the study. Patients were treated again at different doses and at different dosing schedules but all were treated consistently until evidence of progression. Assessments were made regardless of dosing schedule every nine weeks. The primary endpoint here was RECIST 1.1 evaluation of response rate, however, investigators were allowed to make clinical decisions based on IRRC. This, again, was a fresh biopsy which was formalin fixed and paraffin embedded for the purposes of evaluating PDL1 expression.

There were two different assays that were used in this non-small cell lung cancer study. One was a laboratory developed assay which was used to select patients for eligibility on certain cohorts and this looked at both intensity of staining as well as stromal staining and tumour cell staining of PDL1. The clinical trial assay, which was ultimately used to determine a cut point and relationship to outcome, really ultimately used percentage of tumour cells with membranous staining as the scoring modality. But here again you can see that there is variability in terms of overall staining as well as intensity of staining across different tumours.

This slide shows you the overall response rate and you can see that there are different total ends for RECIST 1.1 versus IRRC because of the central review that was required for RECIST 1.1. But the overall response rate for both groups was 19% but you can see that there’s a clear differentiation when you stratify patients based on strong positivity, defined as over 50% of tumour cells expressing PDL1, versus either negative, complete negative, or weakly positive which included patients that had up to 49% staining. What isn’t shown on this slide but will be presented this afternoon is actually separating out that completely negative population from the weakly positive population, in essence they had very similar response rates and very similar outcomes overall. So what you really see is that a 50% cut-off really separates out those who are much more likely to respond from patients who have more weak positive or negative staining.

When a 50% cut-off is used, that translates into a prevalence of about 25% who are PDL1 positive and that prevalence was based not on just these patients here but on all the patients who were screened for the study so that’s approximately 300 patients.

This slide really just shows you that there is a clear separation in terms of outcome. Progression free survival was markedly improved for those with strong PDL1 staining of over 50% with a hazard ratio of 0.53 and a p-value of 0.004. Although the median there isn’t that much separation, what you really have to see is that the curves greatly separate out over time because of the durability of responses that are seen in responders. These numbers are based on still very small numbers of patients with long term follow up so we expect that these medians and the final PFS differences may change over time with additional patients.

In terms of overall survival, as you saw with the melanoma group, there is not a statistically significant difference at this point but there certainly is a trend towards improved overall survival with a hazard ratio of 0.65. Again, this median and the basis for the separation is a very, very early look at this point, it’s based on very few numbers of patients who have had long-term follow-up so I expect that that may change over time as well.

So, in summary, I have told you that there is a clear separation of outcome both based on response rate and progression free survival with a trend towards improved overall survival for those who have high levels of PDL1 expression of over 50%. These data will be further validated in ongoing studies which include both additional patients that are going to be enrolled to this current phase I study, about approximately 300 patients, but also an ongoing randomised study of 2mg versus 10mg/kg versus docetaxel in previously treated non-small cell lung cancer patients. That study is enrolling patients already based on PDL1 expression using the 1% cut-off as an enrolment eligibility criteria. However, patients will be stratified based on the higher cut point so that we can get additional data as to the utility of really separating out patients on high versus low expression.

In addition, MK-3475 is certainly being evaluated in multiple different contexts as a single agent but also in combination with other types of chemotherapy. So this may give us some additional data in different settings of different expression levels how the drug can be most effectively utilised.

I’d just like to thank my co-investigators as well as all the patients and families that participated in this trial and the fact that they underwent these mandatory biopsies which for a lung cancer patient population is really a very difficult thing to ask of patients. Thank you.