Masatoshi Kudo, MD, PhD: In general, AFP [alpha-fetoprotein] is elevated in advanced HCC [hepatocellular carcinoma]. Maybe 80% of advanced HCC is above normal level. However, the AFP greater than 400 nanogram, that may be around 50%. AFP is a diagnostic biomarker and also prognostic marker. The high AFP value is a very poor prognostic marker, so we can monitor the treatment response as well.

R. Kate Kelley, MD: I guess the next question really brings us to how to use AFP as a serum biomarker for HCC at this point, talking about ramucirumab but also in the context of prognosis.

Josep Llovet, MD, PhD: Yes, AFP has been in the field for 40 years, has been used in defining at-risk populations of developing HCC, has been used in the setting of surveillance, and certainly is a prognostic factor. For instance, a patient with high AFP of both 200 and 400 is a well-known predictor of dropout in the waiting list of transplant, for instance, or recurrence after transplant. It’s also a prognostic factor in patients undergoing radiofrequency or chemoembolization. And certainly now we know that in advanced HCC, both the 200 and 400 threshold define patients with aggressive tumor. But since 400 is the one that is the most prominent now in trial design, I think that the AFP should be incorporated in the trials as a prognostic factor and certainly should be a stratification factor in the trials, both at intermediate and advanced stage of the disease, I would say.

R. Kate Kelley, MD: There’s one thing that I’m curious to get your take on. We see that in the overall REACH population, ramucirumab didn’t achieve an overall survival benefit, but in the high AFP, it did. It does appear to be a predictive biomarker, perhaps signaling some underlying biologic pathway or expression signature related to angiogenesis. But when we look at the subgroup analyses of high AFP for TKIs [tyrosine kinase inhibitors], it’s also antiangiogenic in large part. We see from the REFLECT trial, the high-AFP and low-AFP cohorts both benefit from sorafenib and lenvatinib in that trial. And, similarly, we looked at cabozantinib, and baseline AFP is not a predictor of response, meaning that low-AFP patients benefit as well as high-AFP patients, even though, again, they’re largely antiangiogenic drugs. Do you have any clue why that might be?

Josep Llovet, MD, PhD: I think that all the other drugs—sorafenib, regorafenib, cabozantinib, lenvatinib—are blocking several kinases, so they are not just drugs blocking VEGF receptor 2. But for instance, lenvatinib is blocking FGF receptor 1, 2, 3, 4, and regorafenib is also blocking Tie2. So there are other targets out there. And, needless to say, cabozantinib is blocking MET. So the beauty of ramucirumab is that it’s only blocking 1 driver, and as a result of that, provides survival differences. And this is novel and quite unique in HCC by blocking just 1 driver. So, of course, you can ask what is the relation between high AFP and effective drugs blocking the VEGF receptor 2, and there are not strong data about that. But certainly, there has been an association that those patients with high AFP levels may have more activation of the VEGF receptor 2 pathway, and this may be the association that explains that.

Arndt Vogel, MD, PhD: I think it’s really important to have positive biomarkers, prognostic biomarkers, and predictive biomarkers. First of all, prognostic biomarkers are important to understand the natural history of the disease, and I think it was really important to understand the poor survival of patients with a high AFP level, for example. These are important certification factors for future clinical trials, which do not only include, of course, AFP but also vascular infiltration and extrahepatic metastases.

And then when we have a couple of drugs available in first- and second-line settings. We somehow need to decide which drug should we use. And it would be nice if you have biomarkers that clearly would indicate which drug you should use in first-line, second-line, or third-line treatment or how we should sequence these treatments depending on biochemical, clinical biomarker.

Having more options is good because we can really sequence treatment. We can really decide, and we can switch treatment if the treatment is not tolerated. But on the other hand, of course, we would also like to have the best sequence for our patient and therefore biomarkers actually needed in this field.

Transcript Edited for Clarity