Panelists: Ghassan K. Abou-Alfa, MD, Memorial Sloan Kettering Cancer Center; Richard S. Finn, MD, Geffen School of Medicine at UCLA; Masatoshi Kudo, MD, PhD, Kindai University Faculty of Medicine; Arndt Vogel, MD, Hannover Medical School
Ghassan K. Abou-Alfa, MD: With this said, however, let’s go back to our basis and start with talking about systemic therapy. Because understandably, for all of us, for the past 11 years or so, we started with 1 trial that was reported at that time: sorafenib, a multi-TKI that was actually in the early days when our group at Memorial Sloan Kettering Cancer Center led the first effort back in the beginning, like in the 2000s. It was not even understood to be antiangiogenic, but, nonetheless, we got it at that point, and we showed great results that were promising enough to go for a phase III trial that did actually randomize patients to sorafenib versus placebo in the first-line setting for HCC. And the trial was positive. It showed an improvement in survival that really was close to about 11 months in regard to the sorafenib versus the placebo. It was statistically and clinically significant and became the standard of care in no time in 2007.
For 10 years or so, we tried everything under the sun to try to see how we can make it better, either by doing other drugs or even trying to combine it with other drugs, and nothing worked. And really, we were left in a very easy field where sorafenib is the only standard of care that, to some extent, we were satisfied with. But, of course, also we’re dissatisfied in regard to adverse events or in regard to what can it lead to, because there were no responses, but, nonetheless, people did survive longer. And then, out of nothing, our colleagues in Japan come up with a drug called lenvatinib. So, let’s see. Dr. Kudo, tell us, what did you do?
Masatoshi Kudo, MD, PhD: In Japan, after the trial, lenvatinib was approved on March 23. And already for more than 1300 patients, lenvatinib was prescribed. So, that’s more patients than were prescribed regorafenib for 2 months. Currently, the first-line treatment option is lenvatinib, which seems to be replacing the sorafenib right now.
Ghassan K. Abou-Alfa, MD: Fascinating.
Masatoshi Kudo, MD, PhD: Yes, because the Japanese patients more frequently have hand-foot-skin syndrome, and there are elderly patients and their body weight is light. So, less-than–60 kg patients are more frequent. In the real-world setting, we prescribe to the patients, they do not complain, there’s almost nothing. Because we are used to the management of the hand-foot-skin reaction for sorafenib or other AEs related to a TKI, it is very easy to manage. So, it’s a perfect trial to show response. We see the patients every week for 1 month and then every 2 weeks. But 2 weeks later, the AFP value drops, of course, and the responders drop—not dramatically, but there’s a significant drop. And then we take a CT scan at 2 weeks. Only at 2 weeks, the tumor became necrotic or there’s a low-density change. We can see the response, and the patients are happy, and the physicians are also happy and continue with lenvatinib. So, it’s very different from sorafenib.
Ghassan K. Abou-Alfa, MD: No doubt that we’re all happy to see the data. But, Rich, here in the United States, and understandably probably anywhere else, people will ask us about the design of that study and the outcome of that study. So, can you take us through the statistical design as well as the outcomes?
Richard S. Finn, MD: Yes, it’s an important question, because up until this time, we had about 5 randomized studies in the frontline setting and they were all negative. This is the first phase III study to meet its primary endpoint, which was noninferiority. So, first, if we talk about lenvatinib versus sorafenib, obviously they’re both small molecules, multikinase inhibitors. Lenvatinib does have a very potent pan-VEGF receptor profile, but it also hits the fibroblast growth factor receptor family very well. FGFs 1-4 are important in proliferation, as well as maybe angiogenesis resistance. In the design of this study, they took patients like others who were Child-Pugh A, good performance status, and who had not been treated in the frontline setting.
They excluded some patients who hadn’t traditionally been excluded from other studies, such as patients who had more than 50% of their liver involved or invasion of the main portal vein branch. So, we know that liver cancer likes to grow into the vasculature, and when we do imaging and we see that, that’s called macrovascular invasion. It’s macrovascular because we can see it on a scan. And often deep in the liver, you’ll see a tumor invading a branch, but in this study, they excluded patients where that tumor is growing into the main portal vein, to outside of the liver. And again, historical studies didn’t exclude those patients, and the guidelines from Japan would indicate that these patients shouldn’t be treated with sorafenib. And that became the basis for the design.
Patients were stratified typically by region, extrahepatic spread, macrovascular invasion, performance status, and by weight. Again, unique to this study and unique to the dosing of lenvatinib is it’s dosed—greater or less than 60 kilos—12 mg versus 8 mg, which I’ll remind everybody is much less than the approved doses in thyroid cancer and kidney cancer, which are 24 mg. So, this study randomized about 950 patients in an open-label fashion—so several things to discuss about the design—to sorafenib or lenvatinib, with the primary endpoint being noninferiority. And if the noninferiority endpoint was met, then we looked at superiority and other secondary endpoints. And, as we know, the study met its endpoint. Median OS went from about 12.3 months to 13.6 months with lenvatinib. That was a hazard ratio of about 0.92, and the upper limit of the hazard ratio was 1.06. That was below the cutoff of 1.08 to decrease noninferiority.
We’ve seen, in some of the radioembolization studies, for example, that the curves overlap, but those studies were designed for superiority, so they’re negative studies. And the upper limit of the hazard ratio in some of those studies is 1.43, so very wide. This study was designed to come up with a hazard ratio that is fairly tight, in that if we are under 1.08, we can say with very high confidence that these drugs have an equivalent effect on overall survival.
Ghassan K. Abou-Alfa, MD: So, this is very important that you bring up a concept. We all hear about it now, but we do not necessarily understand fully noninferiority, which means, in other words, the aim of the study was not really to look if one drug is better than the other. Really the purpose and the objective were to look if they are equal. And, as such, you cannot answer a superiority question. But, nonetheless, as we heard from Dr. Finn, if anything, the fascinating part is that the hazard ratio cutoff on the top level of the noninferiority was actually very close to where the superiority would have started kicking in. It was 1.06?
Richard S. Finn, MD: 1.06.
Ghassan K. Abou-Alfa, MD: And if anything, the cutoff was 1.08. In other words, anything that is above the 1.08 would have been a superior study. It’s fascinating that we were really very close to really making it even superior.