How Merck hopes to win in I/O combos: Q&A with Roy Baynes
Earlier this month, immuno-oncology's dizzyingly rapid expansion was on display at the annual meeting of the American Society for Clinical Oncology in Chicago.
With approval of the fifth PD-1 axis inhibitor in May, what was once a showdown between Merck & Co. and Bristol-Myers Squibb has become an increasingly crowded field. ASCO highlighted both the maturing clinical profiles of Keytruda (pembrolizumab) and Opdivo (nivolumab), as well as early data on some of the most promising combinations pairing checkpoint inhibitors with novel agents.
The industry's headlong plunge into combination approaches reflects a central puzzle to immunotherapy: some patients experience remarkable responses but many still don't see a benefit. Adding in new agents, it is hoped, can broaden efficacy to a wider pool of patients without compromising on safety.
To that end, nearly 800 clinical trials are currently ongoing involving combinations of PD-1/L1 inhibitors, more than triple the numbers seen in 2015, according to data compiled by EP Vantage. Some of those attempts won't stick. Indeed, Roche recently handed back the rights to NewLink Genetic's IDO inhibitor after poor data.
Each player involved hopes to unlock the next advance, but it's still not clear what that might be. Roy Baynes, head of global clinical development at Merck Research Laboratories, sat down with BioPharma Dive at ASCO to explain the pharma's approach and how it hopes to stay ahead of the competition.
Editor's Note: The following interview has been condensed and lightly edited for clarity.
On Merck’s approach to combinations
Roy Baynes: We don't know what the best combination is going to be. What we do know is that monotherapy is a very effective treatment for a large number of patients — not everybody.
We are trying very hard to try to identify the patients for whom more is needed. That involves obviously biomarker and patient selection type of efforts. When it gets down to the combinations, we have no idea what the best combination is going to be. It is early days in this. Anyone who tells you they have the ideal combination is probably misleading you.
"Anyone who tells you they have the ideal combination is probably misleading you"
SVP, global clinical development, Merck Research Laboratories
As we think about combinations, we think about them in really three key mechanistic buckets.
Firstly, can we increase the amount of antigen being presented? Antigen is the protein that activates that T-cell receptor and activates the killer T-cells.
One of the mechanisms that is being envisaged is if you kill a lot of cancer cells and don't suppress the immune system at the same time, it is possible — just in that act of killing — you will have macrophages digest that material and present the peptides which are derivative from that as antigens. This is a process that we call antigenic cell death. One of the ideas there is that chemotherapy, radiation therapy, targeted therapy could all do that.
The second mechanistic bucket is: Can we activate the passage of T-cells into the cancer? Can we get that T-cell population to be more active in the cancer? This is where we look at manipulating a lot of the stimulatory and checkpoint inhibitors to try to do that.
Between those two is this action of priming the T-cell. Can we actually do something to actually rev the T-cell up? That, again, involves targeting surface molecules such as other checkpoint inhibitors and using cytokines and also exploring potentially co-stimulatory molecules.
On combining Keytruda with IDO1
Baynes: IDO1 it turns out is an inhibitory molecule. What we are trying to do here is actually inhibit the effects of IDO1. Epacadostat is obviously the furthest along. We partnered with Incyte on that. We did an initial Phase 2 study looking at a number of different tumor types and you have seen quite a lot of those data at this meeting
There we have got nice evidence of what looks reasonably good efficacy with head and neck cancer, bladder cancer, kidney cancer, lung cancer, melanoma. We have taken all of these now forward into Phase 3 trials.
The whole goal [in adding new agents to Keyturda monotherapy] is to identify combinations which increase efficacy but without increasing toxicity. One of the commonly described combos is the IO/IO combo of CTLA4 plus a PD-1 antibody. There is quite some evidence that in fact this may improve efficacy at least in the short term.
The long term data are somewhat equivocal at this time. These are all cross-treatment comparisons but it is really not clear that that combo is adding much efficacy but it certainly seems to be adding toxicity. So the goal here is to actually get combos which don't increase toxicity but do enhance efficacy.
Some have zeroed in on the IDO pathways as a solution to that problem. Is Merck seeing it the same way?
Baynes: Yes, that is why we took it forward. We saw evidence that there was increased efficacy. Remember, these are all response rate data. So that is short term data. Obviously, the big question is does that translate into a long-term benefit. But we did not see an increase in toxicity. The toxicity profile looks very much the same as pembrolizumab alone.
Do you think that the primary activity you'd expect to see with adding IDO1 would be in cancers that Keytruda is already most active in?
Baynes: Firstly, the monotherapy data are still emerging. We are quite early in this course. We have shown across 30 major cancer types that we have efficacy from monotherapy in 25 of those. That is pretty amazing and makes it a broad spectrum anti-cancer drug.
We have approval for a small subset of those and there is a lot of work ongoing around monotherapy. Monotherapy has been explored, at least initially, in the most advanced cases who have failed all treatments. Now we are moving earlier in treatment, so we are going to second-line, first-line, adjuvant settings as monotherapy.
"The monotherapy story is by no means over"
SVP, global clinical development, Merck Research Laboratories
The monotherapy story is by no means over and we do think that for a very large number of patients monotherapy is actually what you need.
On biomarkers and identifying the best responders
Baynes: We are doing a lot of biomarker work to try to identify those patients most likely to respond and those who should probably be explored with other therapies. Our first foray into this was with PD-L1 staining. PD-L1 is the ligand and just on first principle seems logical that if is what you are trying to inhibit, go for the tumors that have a lot of PD-L1. That has panned out reasonably well.
The next signature that we are looking at is the gene expression profiling where we are looking at essentially the degree of inflammation within the tumor. This is looking at a discrete set of genes that seem to be a very nice inflammatory signature. That also seems to be quite predictive.
The other nice thing about it is it allows us to look at those whose tumors have inflammation but don't actually mount a nice immune response - trying to identify a specific genetic expression which may actually be mediating that resistance.
The third big group we are looking at is the mutational burden, that is to say the number of mutations in the cancer. I think there is good evidence to suggest that mutational burden is important.
Keytruda recently won approval for site-agnostic tumors with a certain biomarker. How large of a step forward is this for precision medicine? In the nearer term, do you think this type of approval may become more common? (Read about the approval here.)
Baynes: Precision medicine within site-specific tumors has been around for a while. But in fact this notion of mechanism-based approvals which transcend the actual histology really is a watershed moment. For many years, there has been a belief that ultimately we will get to the notion of can we approach disease based upon understanding of the mechanism rather than what organ it arises in. This is really the first coherent proof that this is, in fact, achievable.
We certainly have a lot of other ideas about other approaches that could utilize this type of idea where we actually identify a mechanism and then try to explore a tissue-agnostic, mechanism-based indication. Obviously, we are not disclosing all of those now.
For identifying these biomarkers to predict response across sites, has the ability to identify and validate that signature kept up with advances in immuno-oncology drugs?
Baynes: How much of what has happened in the field has been empiric and how much has of it actually has been driven by informed biology? The answer is mixed. I think it is the key question.
In some circumstances, the clinical experience has outstripped the understanding mechanistically. That is not unusual but, in some circumstances, the mechanistic understanding has really advanced the field. MSI-high is a great example of that.
I do think the precision medicine is important. It is not going to work every time but it definitely has to be pursued because it actually allows for smarter development, a clearer value story and, frankly, it is good for patients. It identifies patients that are most likely to benefit.
Are these trials more difficult to run, to select and accrue patients into these trials?
Baynes: When you go to a cancer center, you go either to the lung cancer clinic or the colorectal cancer clinic or the genitourinary clinic. You don't go to the MSI-high clinic. The big question is who is the investigator that is going to champion this?
It does create some challenges but I think there is enough molecular understanding at the oncology practice level that people recognize this is really a big deal. Interestingly, these trials have enrolled quite well.
It is does produce logistic challenges because medicine is not organized by biology. It is organized by taxonomy, basically.
Follow Ned Pagliarulo on Twitter