PD-L2 expression and its relevance to anti-PD-1 therapy in cancer

Dr Jennifer Yearley - Harvard Medical School, Boston, USA

Jennifer, you’re here all the way from California, welcome to Europe. You’ve been looking at an extremely crucial topic because you’re looking at anti-PD-1 therapy, this cutting edge therapy, checkpoint inhibition and all that. But why have you been looking at PD-L2 expression, why is that so important?

That’s an excellent starting question because the PD-1 axis, which is the particular biology that’s being targeted with the anti-PD-1 and anti-PD-L1 agents, PD-1, the receptor on T-cells, it’s an inhibitory receptor on T-cells has two know binding partners, PD-L1 and PD-L2. Currently there has been extensive investigation of the distribution, the prevalence, the amount of PD-L1 that’s expressed in various human tumours, the types of cells it’s expressed on and there have been correlations that have been drawn with PD-L1 to clinical response. But, to date, there has been almost no evaluation of PD-L2 which as a binding partner for PD-1 is also able to suppress T-cell activity. So if PD-L2 is present in significant quantity in tumours and we are not looking for it we are not necessarily picking up patients who are going to respond to the anti-PD-1 therapies.

So doctors considering using PD-1 need to be looking for biomarkers then?

We all need to be looking for biomarkers, we want to predict who is going to be responding to these treatments. That’s a very high priority.

So what have you done in the study?

What we did, we started with putting together sets of archival tissues from seven different indications, including renal cell carcinoma, melanoma, non-small cell lung cancer, triple negative breast cancer – an array of tumours that are already being treated with anti-PD1 inhibition in the clinic to try to get a sense of what the relative prevalence of the PD-L2 marker was in those tumours and whether that might play a role in clinical response because it hasn’t been investigated before. So first we did an initial screen through sets of tumours, it was about 70 samples per indication, and then we expanded that to apply the findings that we had generated to test in a set of 144 head and neck squamous cell carcinomas from patients that had actually been treated with pembrolizumab to see whether we could actually correlate the findings with clinical outcome.

What did you find about the expression of PD-L2 and its importance?

We were actually very surprised by the amount of PD-L2 that we found expressed in tumours. Every tumour type that we looked at did show PD-L2 expression. It did vary in overall quantity by the indication but it was present in all of them and it was present in at least three different cell types: it was present in immune cell infiltrate in stromal cells, it  was present on tumour cells and in some tumours it was actually present in high quantity on endothelium, which may be especially relevant because T-cells that are trafficking out of the blood vessels into the tumour tissue are going to encounter that on their way in to the tumour.

That’s present and expressed?

Yes, yes. And there’s literature that supports the ability of endothelial PD-L2 to down-regulate CD8 T-cell activity in cytolytic behaviour. So that process of engaging PD-L2 as T-cells cross into the tumour across the endothelial barrier could, in fact, basically downregulate their activity in their ability to kill the tumour in that very process of entering it.

Now, this is a very exciting stage, a very exciting form of therapy because extended survivals have been noted in a number of cancers using checkpoint inhibition. So what do you think doctors might get out of this in terms of clinical application?

At this point this is still very much early exploratory research and we need to expand our numbers, we need to definitely be looking at larger cohorts of patients that have been treated to see if we do in fact… to see if the initial findings that show some degree of significant predictivity of PD-L2 in patients, whether that bears out when we look at larger numbers of patients. But if in fact it does then the implications very strongly are that in patients in whom PD-L2 is present an anti-PD-1 targeted agent is much more likely to be efficacious than an anti-PD-L1 targeted agent, simply because anti-PD-L1 targeted agents aren’t going to have any effect on the PD-L2 that also can shut down the T-cells.

Can you target PD-L1 and PD-L2 separately by two different agents?

That certainly could be done, nobody is doing it at present. So it would entail the development of novel therapeutics for the PD-L2 component.

So what’s the take-home message for doctors at this point?

I think that the take-home message is keep your eyes peeled and we’re going to have more data on this coming because there’s very active ongoing work to evaluate whether the initial very exciting findings we’ve come up with are going to pan out in the long-run. But right now it does look like we may have a molecule that has additional predictive benefit for clinical response in patients.

And the impact of that clinical response and in which cancers, do you think?

So this is data that we don’t have yet. It certainly is the case that there are some tumour types where we seem to see more PD-L2 expression than in others and there are some tumour types where we seem to see more expression of PD-L2 in the absence of PD-L1, that’s something we are particularly interested in following out the relevance of. But at this point it really remains to be seen. We need to see how much this is going to add to our ability to predict patient response.

But it all points towards more individualised therapy?

It certainly points to needing to have a deeper understanding of what’s going on in each individual patient’s tumour, yes.

Is it moving away from the particular tumour type and more towards the nature of the expression of different biomarkers?

I would say that that is very much the case. It’s not so much the histology of the tumour, it’s the particular expression patterns that are present in the tumour, regardless of histology or cell type of origin.

Jennifer, thank you very much.

Certainly.