ASCO 2015

Genomic marker predicts anti-PD-1 response in several cancers

Dr Dung Le - Johns Hopkins Kimmel Cancer Center, Baltimore, USA

You were looking at something called mismatch repair deficiency. It sounds very interesting because it was being described as a way of actually signalling which cells are cancer and then how to go and get them. What were you actually doing there?

Mismatch repair deficiency just means that these tumours are deficient in proteins that are involved in mismatch repair, so DNA repair. So because they are deficient in these proteins their tumours generate hundreds of thousands of mutations that can then be recognised by the immune system, almost like a foreign protein. So it is a red flag to the immune system, this is something different that can be targeted.

So what is it about the immune system that notices these different things? Is it because they are foreign, basically?

They look a little bit more foreign, they’re what we call wild-type which is the normal. So the T-cells are in the body that can recognise it but they go into the tumour and then they’re turned off. So giving it an immune therapy such as a PD1 blocker can take the brakes off of those T-cells and then can cause an immune reaction to the tumour.

Can you tell me what you did in your study?

What we did is we tested the drug pembrolizumab, which is an anti-PD1, and we tested it in three cohorts of patients. The first two cohorts of patients were colorectal cancer patients and in the first cohort they were colorectal cancer patients with mismatch repair deficiency and in the second cohort they were colon cancer patients with mismatch repair proficient tumours. In the third cohort really we were asking the question about mismatch repair deficiency and so it included any tumour type that had mismatch repair deficiency. So what we did is we looked at their responses and we saw zero responses in the colon cancer patients with mismatch repair proficiency but 62% response rate in the colon cancer that was mismatch repair deficient and 60% response rate in the non-colorectal patients with mismatch repair deficiency. So it really was limited to the mismatch repair deficient cohort.

You almost never hear of such a sharp difference between a tumour positive of something and a tumour negative of something. What did you make of these findings?

Because I really think we were testing that hypothesis that the high mutational load is what is predicting the response to this therapy. We know that the deficient tumours are very high in mutational burden compared to the proficient tumours.

And you used a PD1 targeting agent?

We did.

What happened then? That was it, basically, pembrolizumab?

Yes, so the patients in the deficient cohort responded, a good number of them responded and for a very long time. So many of their responses are lasting over a year.

What’s the clinical significance of that?

This is a patient population that’s had mostly two or more prior therapies. Their survival is measured in months, maybe somewhere between 4 and 6 months, and we’ve been following patients with a median of 7-8 months follow-up for the entire group. So this is really making a difference in survival of these patients because we’re able to stop what’s happening and regress it. These are all treatment refractory progressive patients, meaning their cancers were growing before coming into this study.

What could this mean in the real world? You had a very refined experiment there but could clinicians perhaps think about using these sorts of processes?

So I envision a day where they will. Of course this is a very small study so we need to study in more patients and we are expanding those cohorts in our study and Merck is also planning subsequent industry sponsored studies in these various diseases. So I think we need more data but the testing is clinically available, commercially available as part of the standard of care so it’s not something where tests necessarily need to be developed.

Is it something, then, that should be routinely carried out, do you think right now?

I think it’s a bit early. Again, it’s a very small study, I think the data is very promising and I think we just need to study more patients with the agent before doing it routinely.

So what’s the clinical bottom line message to take home?

The bottom line is if we can prove this with a larger number of patients that eventually it will be part of the armamentarium for this subset of patients across disease types based on a genetic marker as opposed to based on a histology.