Hepatocellular Carcinoma: Practical Implications of Emerging Data - Episode 16
Richard S. Finn, MD: The teaser here is the responses, when they occur, can be very dramatic, and they can happen very quickly. I mean, you don’t even need imaging per se sometimes for patients who have symptoms—not necessarily decompensated, but patients who might be having pain, and they start feeling better. Increasingly, that’s more numbers of patients. It’s not the single-digit sorafenib responses or the high response rate with lenvatinib that did not have a long duration of response. It would be great if there was a biomarker or something that could help us. Certainly, we’ve seen AFP fall fairly quickly—if patients make AFP, but that’s only two-thirds of patients. Katie, you’ve done some work on AFP in the context of clinical trials, and there was some interesting biomarker work at ASCO. Could you fill us in on that?
R. Kate Kelley, MD: AFP is a really interesting biomarker in that it can be a prognostic biomarker. That’s how you know it best, that across stages of HCC [hepatocellular carcinoma], it portends a worse prognosis: patients going to surgery, getting a TACE [transarterial chemoembolization], getting systemic therapy. Having a high AFP at the outset of their therapy is prognostic of a worse outcome. It is also now a predictive biomarker, in that it can portend which patients will respond to and benefit from ramucirumab. That’s the first predictive biomarker in HCC and a landmark in that context. We think that’s because the tumors that have high AFP expression tend to correlate with angiogenesis and VEGF pathway signaling, thus their responsiveness to VEGF R2 monoclonal antibody, or ramucirumab in this case.
The third kind of biomarker that I’m particularly interested in as a tool is the role of AFP as a biomarker of response. This acknowledges the limitations of standard imaging and how difficult in some patients it can be to determine if the tumor is growing or shrinking, or where is that tumor, particularly for these infiltrative or diffuse phenotypes. We’ve seen that for several drugs, including ramucirumab, as well as now cabozantinib, both antiangiogenic agents, the changes or the kinetics of AFP after start of treatment are strongly associated with outcomes, including PFS and overall survival, as well as with radiographic response. Those are rare with those classes of drugs. For advanced patients, about 75%, 70% will have an AFP greater than 29 g/mL at baseline. For those patients, we really should try to investigate AFP as another response marker to make these early interpretations. We need to look at this by class of drugs, because the response patterns and kinetics may be different for I/O therapies than for antiangiogenics. But it is a cheap, easy, noninvasive way to augment our response assessment.
Richard S. Finn, MD: I try to focus patients away from it. Sometimes they get very nervous because, to your point about the kinetics, it might not fall from week to week. I think all of us feel that it going down is better than it going up. To your point, probably more research is needed to say that this is a criterion to change treatment. I don’t know of any other cancer type where that’s come to maturity with using tumor markers to change treatment decisions, maybe…
R. Kate Kelley, MD: [Interposing] Prostate.
Richard S. Finn, MD: PSA, I was going to say. PSA is probably the closest.
R. Kate Kelley, MD: Testicular.
Richard S. Finn, MD: Certainly. Probably there, it’s AFP as well. But in the liver cancer space, a little more work is needed. All these phase 3 studies are collecting AFP data and will give us a rich area to look at.
One area that being looked at very aggressively with various platforms is circulating tumor DNA [ctDNA], not only to get genomic DNA to identify alterations to pursue but also to look at changes in ctDNA over time as a marker of prognosis and predictive value for benefit. There was an interesting study at ASCO looking at some of the data from the single-agent atezolizumab-bevacizumab earlier studies that showed that using this Natera platform—which actually makes a personalized ctDNA probe based on a patient’s genetics and their specific tumor, and following that over time that in patients who had a very early response in their ctDNA—they seemed to be ones who had a very long-term survival. It’ll be very interesting to see how that technology matures over time. Katie, I’m pretty sure you had done some data with circulating tumor cells as well, right? Probably the precursor to circulating tumor DNA. Is that moving forward, or is it being surpassed by these new technologies?
R. Kate Kelley, MD: There are a lot of different platforms that have shown that circulating tumor cells in the peripheral blood are associated with poor prognosis, macro-vessel invasion, more advanced disease. Both in early stage patients after resection—for example, the existence thereof correlates with recurrence, and likewise also for prognosis in advanced. These are really limited so far, to my knowledge, to retrospective data sets and have not been studied prospectively. Though again, it’s an intriguing prognostic tool with a lot of studies suggesting a role.
Richard S. Finn, MD: Yeah, it’s another area in the liver cancer space that will hopefully supplement our clinical decision-making in the future.
Transcript Edited for Clarity