Update on Systemic Therapy for Advanced Liver Cancer - Episode 8
Transcript: Nikolaos Pyrsopoulos, MD, PhD: Let me go back to the appealing immuno-oncology agents. What’s your take on nivolumab as the second-line for therapy?
Andrew Zhu, MD, PhD: I think clearly, in addition to the tyrosine kinase inhibitors, we also have the checkpoint inhibitor option. We have both pembrolizumab as well as nivolumab being tested in the second-line setting. I think, as we all know based on the durable responses of both agents, these are actually independent of the underlying etiology. In other words, responses we definitely observe regardless of the underlying hepatitis B, hepatitis C etiology. And also in the setting of CheckMate040 for nivolumab, it’s also occurring regardless of the prior sorafenib exposure. For that reason, both drugs got the accelerated approval obviously from the FDA.
If you look at the KEYNOTE-240 study, you definitely see the overall survival benefit trend. In other words, pembrolizumab definitely improved the survival, and the P value actually reached the statistical significance. On the other hand, it’s just not meeting the prespecified statistical boundary. It is actually requiring a more stringent threshold. For that reason, it should be viewed as a statistical negative study as well.
In KEYNOTE-240, the response rate was also at 18%, also very durable. So I think this really brings the challenge, how do we position checkpoint inhibitors in the treatment options for advanced hepatocellular carcinoma [HCC]? Because this is in the same context of what we discuss earlier for CheckMate 459. In other words, you have durable response data, but in the end, I think the phase III studies did not actually achieve the statistical boundary for improved survival.
Nikolaos Pyrsopoulos, MD, PhD: That’s an excellent point. The community has some concerns right now because pembrolizumab has obtained FDA approval.
Andrew Zhu, MD, PhD: Yes, correct.
Nikolaos Pyrsopoulos, MD, PhD: What is the future for you? What do you predict?
Andrew Zhu, MD, PhD: I think that’s actually an incredibly important question. My bias is that I definitely think it would be very nice for us to have a good predictive marker so that we know who are those 15% of the patients benefiting from this class of drug? And we know that testing by PD-L1 [programmed death-ligand 1] expression is definitely not going to be the predictive biomarker that’s useful for clinical decision. Actually, both in KEYNOTE-240 as well as KEYNOTE-224, there are biomarkers that are being built into the study, and in the end I don’t think we have a definitive answer choosing any particular biomarker, particularly PD-L1. Likewise, CheckMate 459 as well as CheckMate040, they also look at PD-L1 expression as well.
I think PD-L1 perhaps will not be actually used very successfully as a predictive biomarker. Also, Nik, as we all know, mismatch repair [MMR] is definitely a very good marker selecting patients likely to respond to this class of drug. Unfortunately, in HCC, we know the frequency for the MMR-deficient phenotype is occurring very infrequently. For that reason, that’s not going to be a very practical biomarker. My sense is that we definitely have about 15% of patients clearly benefiting from PD-1 [programmed cell death protein 1], but I think with the phase III data it will be very interesting to see how the regulatory agency will take into the whole consideration of the clinical benefit but also, at least up to this point, the lack of true survival benefit shown in CheckMate 459 and KEYNOTE-240.
Transcript Edited for Clarity