KEYNOTE-427 was a study from several years ago looking at pembrolizumab in patients with metastatic kidney cancer, both clear cell and non-clear cell kidney cancer, which showed that single-agent pembro was actually rather active in both of those scenarios. At the time we stored both blood and tumour for subsequent analysis to try to understand who was responding and why. We’ve reported some of the sequencing data and the immunohistochemistry data in the past, which looked at different signatures which might enrich for patients more likely to respond.
This analysis we’re reporting at this meeting looks at the blood that was sampled from patients. We didn’t get matched tumour and blood on all patients, and we were not able to do this analysis of ctDNA on all patients. So for example, one of the weaknesses is of the 110 clear cell patients where we’re looking at ctDNA, we’re actually only able to get results that we can interpret on about 58 of those patients.
How was the ctDNA collected and analysed within the KEYNOTE-427 trial framework?
We took blood at several time points, both at baseline before treatment, and at several other time points. We’re reporting here baseline and what happens at cycle two day one. We saw some interesting results at baseline. For example, of the almost 60 patients who had ctDNA testing done, almost a third of them were negative by the Natera 16-bit test, which is admittedly an older Natera test. That suggests that up to a third of patients who had metastatic disease tested negative which is obviously not ideal. Now you might ask, is that a problem with the test or is that a problem with the biology of kidney cancer? In this case it’s almost certainly partly related to the biology of kidney cancer because, unlike other tumours like bladder cancer, kidney cancer doesn't shed a lot of DNA in the circulation, making ctDNA analysis difficult. This was one of the first studies to explore whether that analysis was useful in the context of immune therapy for kidney cancer.
What were the results?
So the results come in in two buckets. Essentially, what did the baseline ctDNA help us with? And what did the change in ctDNA help us with? When you look at baseline ctDNA values, for those patients who are positive, they had a worse overall survival. Not surprising, probably because they had more tumour bulk. When you look at the predicted value of ctDNA baseline, it wasn’t super predictive of either overall response or PFS. Those numbers were similar, those values were similar, whether they were positive or negative, so at least in this study, it wasn’t a great predictive pre-treatment biomarker.
More interestingly though is the changes in ctDNA over time, so if you were negative at baseline and stayed negative, or if you were positive and became negative over time, you did better based on just response, PFS, and overall survival. Now the numbers are small, but for example as far as response rate goes, if you look at the patients who started positive and became negative, almost two thirds of those patients had responses which is an amazing number, except when you learn that the total number of those patients in this study was only six. That sort of speaks to one of the things we need to do next, which is we need to take this preliminary study and expand it to much larger groups of patients.
How might ctDNA guide treatment decision or early identification of responders and non-responders?
It’s an exciting time for kidney cancer biomarkers, particularly blood-based biomarkers, both DNA, in this case ctDNA, but also protein-based biomarkers like KIM-1. One of the things you’re starting to see even at this meeting is combining those two technology platforms into one assay to try to improve sensitivity. To try to address questions like, in the metastatic setting, is a patient responding getting a better sense of their response before you see changes in the scans, for example. But even probably more important is in the adjuvant setting after they have a surgery, what are their protein levels? What are their ctDNA levels? Should they get adjuvant treatment? Because that’s a very important question for our patients, where we’re probably over-treating patients with these exciting new therapies because many of them have already been cured by surgery. Could we reduce that exposure by looking at blood-based biomarkers?
One of the exciting things about the ctDNA world is it’s improving, the technology has improved a lot since this study. So I mentioned that this was a Natera platform. Natera, for example, has improved their platform several times since this study was performed. There are other companies in this space, other investigators in this space that are coming up with new technologies and new ways of looking at what the tumour is shedding in hopes of improving our ability to predict outcomes for kidney cancer.
