While PD-L1 has been perceived as an important biomarker in the realm of lung cancer, Keith Kerr, BSc, MB, ChB, FRCPath, FRCPE, says it still has a long way to go.
In an interview with Targeted Oncology, Kerr, of the University of Aberdeen, Aberdeen, Scotland, discusses the possible difficulties of testing for a wide variety of biomarkers in patients with lung cancer, as well as the importance and general perception of PD-L1 as a biomarker.
TARGETED ONCOLOGY:What are the biggest challenges in determining which patients receive which immune therapy?
Kerr:In terms of selecting patients for immunotherapy, there are a lot of challenges and a lot of controversy. The community is concerned that we have the correct biomarker, or if we have any biomarkers at all that works. Many oncologists would prefer not to have to select patients because they would like to give these drugs to everyone for a variety of reasons, both scientific and for medical reasons, but also for financial reasons. If you think about global healthcare, it's probably not going to be feasible or practical to give these drugs to everyone, and it's probably not safe or desirable. So we have to select the patients.
At the moment, the only biomarker that we have any validated data on is PD-L1 immunohistochemistry. PD-L1 immunohistochemistry is perceived as being not a terribly good biomarker, but I think it gets a bit of an unfair reputation because I think it's compared with things like EGFR mutations or ALK translocation, where the performance of the biomarker will always be better because of the biological nature of the system that we're talking about. PD-L1 expression is not an all-or-nothing phenomenon and it's not the only controlling factor in the immune response or response to the drug. Therefor, it will never have the predictive power of the EGFR mutation, for example. It's perceived in the wrong light. Our expectations need to be downgraded.
The other problem with PD-L1 immunohistochemistry, of course, is that each drug has been developed with its own biomarker test. To a greater or lesser extent, these tests are not the same, and therefore we cannot think generically about PD-L1 immunohistochemistry. We have to be much more specific in our thinking, but at the same time it's not practical for us to do 4 or 5 PD-L1 tests so we can decide which drug we're going to give the patient. We need to have some simplification, but what that simplification would look like right now is very hard to say. We need more data.
TARGETED ONCOLOGY:What are you thoughts on PD-L1 as a dynamic biomarker?
Kerr:Much of the evidence will suggest that it is dynamic and that it is inducible, and will be up- or down-regulated according to stimuli, including immune stimuli, in the tumor. This is probably one of the other reasons why the biomarker doesn't look at good as it otherwise might be, because the test is performed at one time point, and when the patient is actually receiving the therapy, the PD-L1 stages of the tumor may have changed. There are very little data out there and I think what data have appeared might suggest that it's a little less dynamic than we thought it might be. Certainly, expression levels will change.
TARGETED ONCOLOGY:In starting to understand all this, how can oncologists think about PD-L1?
Kerr:I think that the practice of testing will be driven by 2 things. It will be driven by a requirement mandated by whatever license the drug has in whatever health system or country the oncologist is working. It may be necessary to test to prove a certain level of expression to have a reimbursable prescription. In health systems where testing is not required, the oncologist may still wish to know what the PD-L1 expression is, because there is still no doubt in my mind that most of the data indicate that the chances of response to these therapies is determined by the level of expression.
If you are a very low responder, or a negative responder, your chances of responding to the drug, which is expensive and has side effects, are significantly less. Within that group of patients, we're now beginning to be concerned that some of those patients actually progress rapidly when they're given immunotherapy. So that's another concern about the liberal use of the drug. We haven't quite learned the best way to use it yet.
TARGETED ONCOLOGY:What other biomarkers are currently, or will become, important?
Kerr:The list gets longer by the day, but in practice for most of us, we still really only have two biomarkers that are associated with approved and reimbursed therapies. That's EGFR mutations and ALK translocation. We've been talking about these for a long time, and we've been talking about a long list of other biomarkers for a really long time, but very few of them have gotten over the line and become completely approved.
The next one around the corner, and it's already around the corner in some places like in the US, is ROS1. In other countries treatments for ROS1 are not yet approved, so testing for ROS1 is still something of an academic exercise. If you find a patient with ROS1, you either have to get compassionate use of the drug because it's not on hospital lists, or you have to get patients into trials. So it's still a challenge in that sense, but I'm sure within the next 12 months, let's hope at least, that ROS1 will become part of our routine testing portfolio.
After that, there will be other things, possibly HER2, and we have MET beginning to make a resurgence through the EXON 14 skipping mutation story, and those patients appear to be sensitive to some of the MET inhibitors. There are many other potential drugs out there with potential biomarkers, and of course that's going to throw up all sorts of challenges for us going forward.
TARGETED ONCOLOGY:What sorts of challenges do you perceive will present themselves in the realm of biomarkers?
Kerr:If there is a demand for a whole multiplicity of tests on one single sample, a pathologist like me is always moaning about having tiny amounts of tumor to work with. So we're going to have to be smarter about our testing strategy, and part of that may be the better use of next-generation sequencing, which can give us a tremendous amount of information about a large number of targets and mutations, and probably possibly gene copy number. That will help, but it won't provide the only answer.