Raoul S. Concepcion, MD, FACS: Mike, at Johns Hopkins, you’re at one of the premiere institutions that has looked at this—androgen receptor variance, CTCs [circulating tumor cells], those types of things. Can you give a little commentary about where we’re moving in terms of trying to figure out this type of testing?
Michael A. Carducci, MD, FACP: I think, as Neal said, we have this paradigm that we treat things in sequence. We decide something is no longer benefitting the patient, then we move. And yet, when you look at where you squeeze in sipuleucel-T, you know to put it in early. But then, if you go to your AR- targeted agents, abiraterone or enzalutamide, you have no data to pick one or the other. Some data, however, is going to come out about potentially combining them.
There’s a negative study, from a British Columbia group, that said if you layer abiraterone and enzalutamide together after progressing, it only adds toxicity—no benefits. So, this whole idea of layering still has refinements. But, I think it is an important finding. What we’ve done, at Johns Hopkins and other groups, is try to figure out who’s going to respond to the AR targeted drugs. A study in 2009 showed, in one of the early AR variants, that splice variants basically keep the AR on.
Other groups have shown a similar variance, but the main one coming out is AR-V7 [androgen receptor variant-7]. Early data suggested that if you had one drug or the other, and you became AR-V7—positive, it comes in 3 flavors: you don’t have any circulating tumor cells; you have circulating tumor cells but they don’t make AR-V7, those are negative; or you have circulating tumor cells that are AR-V7–producing, they’re positive. That’s the worse prognosis group—they do poorly. And yet, we also showed that you are less likely to benefit from the prior hormonal therapy or AR-targeted drugs in general.
When you take that out to the real world, in practice, we are finding that folks that are AR-V7—positive can respond. It’s low—15%, 20% maximum. Does that change your therapy? In our practice, we check. If someone’s progressing rapidly, and I want to use a drug that’s going to give them benefit, it tells me that I might want to switch to something like chemotherapy or radium-223. Whereas, if they’re negative, I might try the other AR targeted agents from a side effect portfolio. We’re still trying to figure out that aspect. Assays across the country vary. Again, some centers can’t find it, but others can. There still has to be a lot of work in terms of it just becoming a standardized test.
Raoul S. Concepcion, MD, FACS: Evan, the traditional splice variant is a CTC. We’re using CTCs. Again, I think Mike brings up a great point about reproducibility. What’s been your experience on the West Coast?
Evan Y. Yu, MD: We’re using both the EPIC and the QIAGEN assay ones—protein-based ones, PCR-based—similar to what you do at Johns Hopkins. We haven’t published all our data, yet. We haven’t found it to be as commonly available. That being said, there were just recent data presented at the 2017 ASCO Annual Meeting for a trial where patients were being selected for an AR-splice variant and then were going on to galeterone. They found that only 8% of patients benefitted. Now, again, these patients hadn’t been treated with prior abiraterone or enzalutamide, so the rates were on the lower side to being able to find it.
I think another question is, is it really a driver? The biology is certainly interesting and nice. It can heterodimerize and homodimerize with other splice variants, or with a full-length AR, and be constitutively active. But, at the present time, it seems like the full-length AR is still much more present than AR-V7. And my understanding is that if you have AR-V7, you usually have full length AR and, likely, AR amplification. Could AR amplification really be the driver and be, potentially, the epiphenomenon? Again, I think the story is real. I think the biology is real. I just don’t know how common it really is, and I don’t know whether or not it’s really the driver. Some could argue that, maybe, it doesn’t matter if it’s driver. If it truly pans out as a predictive biomarker, that’s fine. But the reason it matters is just that there are companies designing drugs that are specifically targeting splice variance. And so, will those pan out?
Michael A. Carducci, MD, FACP: When you talk about prediction, it can predict 2 things: whether or not you’re likely to respond to another AR targeted agent—that continues to be evaluated—and, does it predict for a much shorter prognosis? When you’re looking at trials, how do you handle them? Or, if they’re going to progress too fast, can immunotherapy studies keep up with it? What is it really telling you about that patient population other than that you need to be on alert because these are folks who are going to do poorly?
Evan Y. Yu, MD: I don’t disagree with you, but it’s just nomenclature. One’s prediction, the other one’s prognosis—nomenclature.
Transcript Edited for Clarity