Alicia Morgans, MD, MPH, and Susan Halabi, PhD, on Prostate Cancer: New Findings on Classifying Patients Into Risk Groups

Alicia: Susan, thank you so much for joining me today. You gave an incredible discussion of the work that you've done within the Alliance study looking at metastatic castration resistant prostate cancer, trying to understand using ctDNA, how we can prognosticate which patients may do better or worse. I wonder if you could share a little bit about the parent study that led to this investigation. Susan: Thank you, Alicia, first for having me here. It's always a pleasure to touch base with you and expand on some of my research. Specifically, what I presented yesterday was a correlative science study that was part of the Alliance phase three trial, AO3-12-01 that randomized first line MCRPC patients to Enzaludomide versus Enzaludomide plus abiraterone and prednisone. So what we did, we collected plasma specimens from patients who were enrolled on the trial and for those who had at least two milliliters of plasma were eligible for the, uh, ctDNA sequencing. Of those, 776 patients were sequenced. And, what's interesting is that, uh, the number of, uh, patients on the trial was pretty large. So that really is the largest study to date that I'm aware that had sequenced so many patients, 776, as I said. Alicia: It's incredible. And I think one of the things that you've contributed to our field has been these prognostic algorithms, sort of thinking about clinical predictors of which patients are going to have more aggressive disease, and we sometimes use them clinically, and we use them frequently in clinical trial design. This is an MCRPC population. So, how is this different, and how did you actually add to those clinical parameters? Susan: Yeah, that's an excellent question. You know, uh, the first time I did the model was in 2003, and because the patients are getting better in MCRPC, it's incredible to know the stage migration. So when I first did the first model in 2003, we had the same variables, PSA, alkaline phosphatase, uh, we had site of metastases. But what we found out that with state migration, patients are living longer, so we really need to update those prognostic models. But, so far, none of the prognostic models that have been validated included any ctDNA, uh, genetic variant. So we had this great opportunity, uh, where Dr. Scott Tam ran the ctDNA, which we call it, uh, ctDNA AR CT detect. And that was based on a customized, really, assay , uh, that he extracted cell-free DNA from one to three milliliters of plasma. And then, they run, uh, the, uh, preparation for, you know, the library, and all the preparation was done at the University of Minnesota Genomics Center. So then, once we have done that, what we did is, we included that data with the clinical data, and we did modeling. And we had a hypothesis, because our hypothesis was, while we know genetic variants are important based on what's been published there, we know on the other hand that all the clinical variables are established prognostic factors of overall survival. So, by including some of these AR and non-AR ctDNA alteration, and if we include ctDNA, uh, what we call N-Euclid diffraction, definitely our, uh, prognosis is going to be much better. And indeed, that's what we did. We, uh, looked at only the established clinical variables, and when you look at that, the time-dependent area under the receiver operating characteristic curve was 0.72. But then, when we added 16 other genetic factors, that increased to 0.77, and that was statistically significant. Alicia: That's fantastic. And, you know, obviously this is a lab-based assay. It's not a CLIA approved assay that we could use in clinic, but, you know, I anticipate at some point we might be able to use some of these ctDNA findings, these genes, in our clinical prognostication. Do you see it going there quickly, or do you see this more as something that might be useful as we continue to design clinical trials that are trying to identify these different subgroups of patients with different risks? Susan: Yeah, that, that's really a very good question, Alicia. What I think is going to happen is, before we can implement it in clinical practice, I think the first step, it has to be validated. So we did not have an external data set. We are in the process of finding external data sets to validate, uh, this classifier, if I may call it this way. And, uh, definitely there is a huge need for, uh, selecting patient based on their, uh, risk factors. Uh, and, of course, a patient having a genetic variant may be predisposed at a higher risk of death than one who doesn't. So, uh, like, just to be specific, when we look at the top genetic variants that contributed to the model, they were AR enhancer gain, M gain, and then we had RSPO-II. I think we are the first to show that AR enhancer gain is like the most influential predictor of overall survival. Uh, but again, we cannot yet implement it in the clinic until it's been validated. And once it's val- it's been validated, I think people will be open to using it, because, when we created those prognostic groups, you could see a very clear separation in terms of overall survival. Like, I mean, to be specific, patients who were in the poorest group had the median survival around 17 months, versus patients in the low risk group, their median hasn't been read. So clearly there is a need because, as you know, the patients in MCRPC, they're very heterogeneous. Alicia: They are. Well, I so appreciate you talking with me about this, and I look forward to seeing the way that this, this clinical classifier mixed with a ctDNA classifier helps us to make better clinical trials and ultimately, someday, will help us in clinic. Thank you so much. Susan: Thank you for having me.