So the Genomic Dedifferentiation Signature is a ten-gene signature we developed as part of a more comprehensive microenvironmental analysis of sarcomatoid renal cell carcinomas. Sarcomatoid RCCs are an intriguing clinical paradox wherein they’re hyper-aggressive – typically patients survive for less than a year – but also preferentially responsive to immunotherapies, in particular immune checkpoint inhibitors like nivolumab/ipilimumab.  

So our goal with this was to develop a signature where we could identify patients that had a similar clinical phenotype to those patients with sarcomatoid features in their tumours, irrespective of whether or not there were sarcomatoid features present, and hopefully make it a more generalisable tool for patients with kidney cancer.  

How did this signature help differentiate outcomes between doublet and triplet therapy? 

COSMIC-313 was a phase III study that looked at nivolumab/ipilimumab plus placebo, versus a triplet therapy with nivolumab/ipilimumab with a VEGF receptor tyrosine kinase and MET inhibitor called cabozantinib. 

What we looked at was the role of the GDS in predicting outcomes to either the triplet therapy – nivo/ipi and cabozantinib – or nivolumab/ipilimumab alone. Some of our preliminary analyses of the GDS in previous reports showed that patients who had higher expression of the GDS responded better to immunotherapy-based regimens. Patients who were lower expressors of this GDS had better responses to a VEGF receptor inhibitor called sunitinib. In this randomised study where everyone got nivolumab/ipilimumab and it was really plus or minus cabozantinib, we actually ended up seeing that patients who had low expression of the signature – what we classified as GDS-low, pretty intuitively – responded better to the triplet therapy. There’s also a trend although not significant that patients who were GDS high-expressing did better with just the doublet therapy. 

What do these findings suggest about personalising treatment intensity based on tumour biology? 

One thing that the kidney cancer field lacks, particularly in the metastatic setting right now, are biomarkers to guide treatment decisions. We think that the GDS might exist as a practical approach to solving that issue. With the data we have in hand from COSMIC-313 and some of these other studies – IMmotion151 and JAVELIN-101 – which we’ve previously analysed, we have this de novo hypothesis that patients who have low expression of the GDS might benefit more from immunotherapy plus a VEGF receptor inhibitor whereas patients who are GDS high probably derive equal if not better benefit from just immunotherapy itself.  

Could this biomarker help identify patients who may safely avoid more aggressive therapy? 

I think in some ways yes, that answer is probably a little bit more nuanced. The pragmatic fact is that COSMIC-313, although there was a positive PFS result, isn’t going to lead to an approval of cabozantinib/nivolumab/ipilimumab, at least in the United States by the FDA. So clinicians are really tasked with deciding between two generalised options for metastatic renal cell carcinoma, it’s dual immunotherapy with nivolumab/ipilimumab, or immunotherapy plus a VEGFR-TKI, so either way just two agents.  

We do believe that there’s a potential that the GDS will help stratify between which treatment options are better for which patients as we alluded to before, but to say that they’re going to be able to de-escalate therapy is probably something beyond the scope of what the GDS can assess.  

What are the next steps for validating and integrating this genomic signature into clinical practice? 

I think with any of these translational and correlative studies the next step is always going to be a prospective validation. There’s a really intriguing trial going on right now from the European groups that’s being led by Laurence Albiges called the CARE1 paradigm where patients are randomised to either nivolumab/ipilimumab or an investigator’s choice of an anti-PD-1/L1 immunotherapy plus a VEGFR-TKI. Patients really are being stratified on two main factors: one is the IMDC risk status, and the second is their PD-L1 positivity in their tumours. We think that’s also a really good opportunity to investigate the role of the GDS in truly predicting outcomes to these two different dual therapy approaches. We’re really eager to await those analyses in the near future and then see how we can prospectively validate further. 

.