We were looking at advanced non-small cell lung cancer, 495 patients in total were treated, 101 of those were treatment naïve and the remaining patients were patients who had had prior chemotherapy. In fact, the majority of the patients had had multiple prior lines of chemotherapy for their non-small cell lung cancer.

What was your reason for using pembrolizumab?

Pembrolizumab is a PD1 inhibitor and there was early data in non-selected trials, early phase I trials, that indicated that patients with non-small cell lung cancer could have a response when treated with a PD1 inhibitor.

Can you explain why the PD1 inhibitor, the checkpoint inhibitor, was a good candidate then?

The checkpoint inhibitors were being evaluated largely for the treatment of melanoma but in the early cohorts patients were evaluated who had other histologies and there were patients with non-small cell lung cancer included in that.

Of course, pembrolizumab is not the only checkpoint inhibitor available, is it? So why didn’t you go for ipilimumab for example?

That is correct, there are multiple inhibitors of PD1 or PDL1. Ipilimumab has been evaluated in non-small cell lung cancer. There has been some intriguing data looking at ipilimumab combined with chemotherapy. However, as a single agent, the ipilimumab has been associated with certainly some toxicity and there has not been a great deal of data indicating benefit for single agent ipilimumab in non-small cell lung cancer to date.

So what happened in the study?

Of the 495 patients the response rate was 19.4%. Among the patients who had a response the duration of the response exceeded a year. We also, in a rigorous way, looked at the biomarker PDL1. We underwent a training set validation set approach in which we took 182 patients and used them to try and define the level of PDL1 standing that would predict for outcome.

And the PDL1 is related to PD1, it’s the ligand to PD1, the target of your drug.

That is exactly right. So PDL1 is a ligand for PD1 and has been evaluated as a potential predictive marker.

What did you find when you looked for this marker, then?

Then after selecting expression of PDL1 in at least half of the tumour cells as our marker in the training set, we looked at an independent validation set of 313 patients, 223 of whom were previously treated and 90 of whom were treatment naïve. In that group of patients in the validation set we found the response rate for patients who had staining for PDL1 in at least half of their cells was 45.2%.

This is looking very interesting. Do you think it could point to this therapy becoming mainstream?

I think that this is certainly exciting data. This is a large group of patients; among our screened patients approximately a quarter of them did have PDL1 staining at this level and, again, amongst that group of people progression free survival and response rate looked quite good compared to what we are used to with cytotoxic chemotherapy. Perhaps most strikingly with 10.9 months of follow-up the median overall survival for patients with staining in at least half of their tumour cells for PDL1 has not been reached.

What about toxicity with pembrolizumab?

Overall the drug was quite well tolerated. The most common treatment related toxicity, as assessed by the investigators, was fatigue and that was seen in less than 20% of the patients. In terms of grade 3 or greater toxicities, those were seen in less than 10% of the patients on trial.

And in comparison with other checkpoint inhibitors and other therapies? How do these toxicities look?

I have been very cautious to compare amongst the PD1 and PDL1 inhibitors. I think it is very hard to compare amongst single agent studies without comparisons that are formal. However, when compared to ipilimumab which has shown tremendous benefit in melanoma patients, the toxicity profile compared to that, this has generally been an easier drug to tolerate compared to that.

So what is the practical bottom line message for cancer doctors everywhere?

The bottom line message is that this is a very exciting class of drugs. Nivolumab is now proved in squamous cell carcinoma of the lung; there is no drug yet approved in adenocarcinoma of the lung which is the most common histology. But the message is that these are very exciting drugs that are generally well tolerated and that with this biomarker you can also identify a group of patients who are particularly likely to respond to the drug.