Kevin Shee: It's an honor to be here. Thank you for the invite. It's always nice to be in the city for GU ASCO.
Zachary Klaassen: And your career's on a meteoric rise. I know you're going to stay here for fellowship, but you've done some great work and we're going to talk about some of that today.
Kevin Shee: Awesome.
Zachary Klaassen: So, maybe just line up for our patients, why active surveillance in general is so important, and historically how Decipher genomic classifiers helped us risk-stratify these patients.
Kevin Shee: Perfect. So, the goal of active surveillance is really to avoid overtreatment in patients who don't need it. And I think with PSA testing, many patients were being diagnosed with prostate cancer and immediately we're getting jumps to surgery or radiation. And the goal of it is to use the armamentarium of tools that we have and we know that work well, to just closely monitor the prostate cancer over time. And for low-risk prostate cancer, it is endorsed by guidelines as the primary option, management option for low-risk prostate cancer. And it is an option for favorable-risk intermediate prostate cancers as well. So, UCSF has been part of this journey since the beginning.
Zachary Klaassen: Long time, yeah.
Kevin Shee: Since the beginning. And I think using PSA testing, MRI, MRI-targeted biopsy, and now with the role of genomics, I think we're going to continue to do better and better in keeping the patients away from overtreatment.
Zachary Klaassen: Yeah, absolutely. I think I've used Decipher in my clinic for exactly that reason, but we're going to talk a little bit more about Decipher GRID. Just lay out the background from the study, maybe highlight some key points on Decipher GRID.
Kevin Shee: Yeah, sure. So, for those of you who aren't familiar with Decipher GRID, every Decipher test that is sent from a clinician actually comes with all of this Decipher GRID data in the background. And what it is, is it's the gene-level whole-transcriptome data and is available for each sample that is sent into Decipher. And so this goes beyond the individual score from zero to one that you normally get from Decipher. It goes beyond the reports that Decipher also gets. And what is also included in the GRID is validated prognostic signatures that have been published, or some of them are unpublished, as well as other types of signatures that may be relevant for different aspects of cancer biology. So in particular, what we looked at from our study was the cancer hallmarks, the hallmarks of cancer, which were famously published by Weinberg. And there's a signature score for each of those, and we were wondering, are any of these potentially relevant for active surveillance? As you know, the Decipher test was born out to look at post, like surgery patients and to predict metastasis, and it may not be optimized for active surveillance patients. And I think that's the goal of our study, we think there's a lot of data in the GRID that can be utilized to potentially come up with something better.
Zachary Klaassen: Yeah, it's a great background. And I think you've got some great results to share. Just go through some of the high-level points from your study.
Kevin Shee: Yeah, perfect. And actually, I'll backtrack a tiny bit.
Zachary Klaassen: Sure.
Kevin Shee: And a couple months ago, we actually published in European Urology Oncology, our sort of experience with Decipher in an active surveillance population. So using our institutional cohort, we had 500 patients that had Decipher testing in addition to multiple biopsies on active surveillance. And interestingly, we did find that high risk based on the manufacturer cutoff being 0.6 was significantly associated with risk of progression on active surveillance. And this is for major upgrade events being transitioned or upgrade to greater than or equal to grade group three prostate cancer. And in addition, it was also significantly associated with development of what we consider unfavorable histology patterns. So, expansile or large cribriform and intraductal carcinoma, we found that the current Decipher score is actually fairly good at predicting which patients are going to develop that in our surveillance cohort.
So, with that sort of background in mind, what we did is we did a deep dive into the GRID. We didn't actually, for this particular project, look at the individual gene-level data, but we looked at validated prognostic signatures. So, there's 18 signatures available in the Decipher GRID that we were able to use in our cohort. And we used a Cox proportional logistic regression model. We basically used a stepwise approach, and so some of the signatures came in and out once they were added to the model. And of note, it did include the original Decipher score, so the zero to one initial Decipher score. But we found that there were additional signatures in particular for progression for upgrade and major upgrade, like Long and Yu, which are two of the individual signatures in the Decipher GRID, were significantly associated with risk of those upgrade outcomes. And that's after controlling for CAPRA, which is a well-developed risk calculator score from UCSF.
Zachary Klaassen: Yeah, no, it's a great background and a good summary of your data. What's interesting to me is the fact that you had CAPRA in there, which as you mentioned, is very well validated and a great tool. But even adjusting for CAPRA, some of these hallmarks still had more prognostic significance. So, what's the impact of that? How should we be approaching these patients for active surveillance going forward?
Kevin Shee: Yeah. I'll preface my answer now by just saying that this is preliminary data.
Zachary Klaassen: Sure.
Kevin Shee: We're just scratching the surface of what's available in the Decipher GRID. That being said, what it means is that there are biological underpinnings of how the cancer behaves that is not captured by our traditional CAPRA variables. So age, grade, PSA, the things that we've been looking at for many, many years. And so, there really is a role for trying to figure out which of these biological underpinnings are more relevant for surveillance patients, and just seeing how we can better optimize and maybe create a future test that works best for surveillance patients.
Zachary Klaassen: That's awesome. Congratulations on the great work. Anything we haven't hit on, any take-home or wrap-up points for our listeners?
Kevin Shee: Yeah. I think the one sort of caveat that I want to stress, Decipher, it's not for every patient. I think the patient's coming in with low-risk prostate cancer by all of our known variables, I don't think is someone that necessarily needs to get a Decipher test. I think it may give us some additional information if a provider may be on the fence, say there's a borderline case. I know high volume 33 is something people used to worry about. Our institution, we still put these patients on surveillance, but I think it's something to keep in mind, because I think I don't want audiences to see our paper that was just published and say that, "I want a Decipher test no matter what." I think it's really relevant for the patients that could go either way, or just may inform whether the patient needs an earlier or later biopsy. I think really finding that balance and where Decipher really fits in, I think is really important. And I think there's so much more to do with this data, and I'm excited to work on it for the next couple years.
Zachary Klaassen: Awesome. Great job, Kevin. Thanks for coming on UroToday.
Kevin Shee: Yeah, thanks for having me.