AACR 2016

Comparison of three different PD-L1 diagnostic tests shows a high degree of concordance

Dr Marianne Ratcliffe - Diagnostic Expert, AstraZeneca, UK

We’ve done a study to assess the agreement between the three commercially available PD-L1 diagnostic tests, namely the 22C3 Dako assay, as used by pembrolizumab, the 28-8 assay as used by nivolumab to support their clinical trials and the SP263 assay from Ventana that AstraZeneca uses in its PD-L1 clinical programme. So the reason we wanted to do this, if I just take you back a step, we’re in quite an unusual situation in the PD-L1 space that I don’t think we’ve really had before either within pharma or the clinical community. We’ve got a number of different PD-L1 PD-1 directed therapies coming to market or in the market in a very short space of time and each has their own PD-L1 diagnostic test. So that raises the question of are they measuring the same thing, how can you compare across different studies, what’s the right way to select a patient for treatment with PD-L1 monotherapy or, as we look to the near future, immuno-oncology combinations, which is an area of strong focus for AstraZeneca.

What did you find?

Each antibody is different and so it raises the question of are they measuring the same thing. What we’ve seen is really good agreement, in fact a greater than 90% agreement across multiple different cut points between the assays when they’re compared with each other. One of the key things that we’re finding from this is that PD-L1 expression, probably a simple definition of positive and negative, a binary definition like that, is not really going to provide enough detail for the complex situation within, within PD-L1 and immuno-oncology in general. Actually it’s highly likely that multiple cut points will become clinically relevant and inform treatment decisions in the future.

How will this affect clinical practice?

This is the start, really, of a debate that needs to be had with the pathology and clinical community as to what further evidence is needed before we can get to a point where we can say the assays could be used interchangeably, how should the algorithms be applied across different assays. So I don’t think we’re there yet but that’s what we would like to work towards, a situation where perhaps an assay can be used across multiple algorithms to inform treatment decisions for multiple different drugs because that will mean that patients will have a wider access to PD-L1 testing than if you were limited to just one assay.

Is there anything else you would like to add?

We know that PD-L1 is informative as a biomarker for monotherapy and we expect it to be increasingly important as we, in the near future when we’re looking at immuno-oncology combinations and that’s a strong area of focus for AstraZeneca, obviously, and what the aim is that we can bring a treatment that will benefit those patients that don’t really benefit from monotherapy.