How Can Treatment Be Individualized By Using Genomic Classifiers? APCCC 2022 Presentation - Felix Feng
August 11, 2022
Felix Y. Feng, MD, Vice Chair of Faculty Development, Director of Translational Research, Department of Radiation Oncology, Director of the Benioff Institute for Prostate Cancer Research at the University of California of San Francisco
APCCC 2022: How Can Treatment Be Individualized By Using Genomic Classifiers?
How Genomic Testing Impacts NCCN Risk Classification of Black Men with Low and Intermediate Risk Prostate Cancer - Andrew Winer
Treatment of Unfavorable Intermediate Risk Prostate Cancer - Ashley Ross, Jason Hafron, and Katie Murray
Almudena Zapatero: Okay. Our next speaker is professor Felix Feng, who's joining us remotely. And Felix has been attending for the last... Since we started. So it's about the nice hour of... I think it's 3:00 in the morning for you, Felix. I think you're recorded, so we'll present it. We're falling a little bit behind. So Dr. Zapatero and I are failing in our moderator cowbell utilization. So we're going to speed up. We're going to put you on like 76 speed, Felix, in your recording. Just kidding.
Felix Feng: Hello. My name is Felix Feng, and I'm a professor of radiation oncology, urology, and medicine at the University of California at San Francisco. And today I'll be talking about how can treatment for biochemical recurrence be individualized using genomic biomarkers. And I want to begin by thanking [inaudible] for inviting me to present. And at the same time, I want to apologize for not being able to be there in person. Unfortunately, six hours before my flight was scheduled to depart to Zurich, my father fell gravely ill, and he is now unfortunately in the intensive care unit at our local hospital here. So I wish I could be there with you in person, but this recording will have to do for now. So here are my disclosures. I'm on these scientific advisory board for several biomarker companies.
And I want to begin by commenting that genomic biomarkers are now included in certain treatment guidelines for the management of PSA recurrences. And this is the 2022 NCCN guidelines. And you can see that for patients with PSA persistence or recurrence, risk stratification with Decipher is included as a consideration, a category 2B consideration. And one of the reasons why Decipher made it into these guidelines is that Decipher has been extensively validated as a prognostic biomarker in the post-prostatectomy setting. And this was a study that Dan Spratt and I did a number of years ago, where we showed that Decipher was prognostic of metastasis across a wide range of subsets of patients who had underwent prostatectomy for prostate cancer.
Now, it's one thing to have a prognostic biomarker, which is a biomarker of aggressive disease. It's another question to know what to do with that biomarker. And so one question we can ask ourselves is, what treatment decisions can be guided using genomic classifiers? And today I'm going to talk about three specific clinical scenarios. The first is adding androgen deprivation therapy to salvage radiation. The second is deciding between adjuvant versus salvage radiation. And the third is perhaps modifying salvage radiation dose based on genomic biomarkers.
So starting with the first question, and in that context, I want to highlight a recent study where our team at NRG Oncology used Decipher to identify which patients should receive anti-androgen therapy with salvage radiation. And to do this, we took samples from the RTOG 9601 study, which was a phase III study that took patients with PSA recurrences after prostatectomy and these clinical features that you see here and randomized them to salvage radiation with two years of placebo versus salvage radiation with two years of bicalutamide. Overall, we were able to obtain 352 samples from the 760 patients on the study. And the median followup was 13 years.
And the first major finding was that the Decipher classifier risk stratifies all outcomes. And so here you can see cumulative incidence plots for distant metastases, prostate cancer specific mortality, and overall survival Kaplan-Meier curves. And you can see that patients with high Decipher score in red did worse than patients with intermediate and low Decipher scores shown in tan and blue. And what was striking was that Decipher remained a significant predictor of outcome in a multi-variable model accounting for standard clinical pathologic features, as you see here.
And the most interesting finding was the finding that the absolute benefit from hormone therapy is smaller in patients with lower Decipher scores. And so here you can see the entire patient population from RTOG 9601 that was profiled. And what these bar plots represent is the absolute difference in the predicted rate of a particular endpoint, let's say 12-year distant metastases on the left, in patients who did not get the anti-androgen therapy versus patients who did get the anti-androgen therapy is a function of Decipher score. So what you can see here is that patients with low Decipher scores are shown in blue. And the fact that these bars are much smaller than the patients with intermediate and high Decipher score shows that the patients with intermediate to high Decipher score derived a much larger benefit from the addition of anti-androgen therapy to radiation compared to the patients with a low Decipher score.
Now, if we focus on patients who received early salvage radiation, which we define as a PSA less than 0.7, and we used this cutoff because this was the median PSA of the patients who were enrolled on this trial, what we can see now is that patients with a low Decipher score, again shown in blue, derived almost no benefit at all in terms of 12-year distant metastases and 12-year prostate cancer specific mortality from the addition of anti-androgen therapy to the radiation. And in fact, numerically they had a worse overall survival when receiving anti-androgen therapy compared to no anti-androgen therapy with the salvage radiation. And so while these results are quite striking, I want to point out that we had too few samples to demonstrate a statistically significant interaction between Decipher status and the treatment, which in this case was bicalutamide versus no bicalutamide. But the fact that there was really almost no benefit in the vast majority of these endpoints for the addition of anti-androgen therapy suggests that patients with a low Decipher score probably shouldn't be receiving anti-androgen therapy in this setting.
Now, another point I want to bring up is that genomic classifiers might actually predict nodal disease as detected by PSMA PET imaging. And I bring this up because I'm sure that Dr. James just gave a wonderful talk on using PSMA PET imaging to personalize therapy in the context of PSA recurrence. And so at UCSF, we did a study on 91 patients who both had Decipher testing and PSMA PET at the time of recurrence. And we found that patients with a higher Decipher score were associated statistically with having PSMA-avid nodal disease. And so here you can see percentages on the left, with patients with high Decipher score shown in purple, and you can see the percent of positive nodes at each of these nodal stations compared to patients with low Decipher scores shown in the orange. And so perhaps this explains why we can potentially see a benefit from adding anti-androgen therapy in patients with high Decipher scores compared to those with low Decipher scores.
So I think that there is actually reasonable evidence for adding ADT to salvage radiation based on a genomic classifier like Decipher. Let's talk about other potential clinical scenarios, specifically adjuvant versus salvage radiation. So as you know, the radical study and the [inaudible] study demonstrated that there was no benefit to adjuvant versus salvage radiation in the context of PSA, in the context of post-prostatectomy patients. But a number of years ago, Bobby Den and I and Adam Dicker did a study where we demonstrated that Decipher scores may identify patients who actually benefit from adjuvant versus salvage radiation. And on the left here, you can see the incidence of metastasis in patients with low Decipher score versus those with high Decipher score on the right. And I think the take-home is that patients who received adjuvant radiation had a lower incidence of developing metastatic disease compared to patients who got salvage radiation. These patients were matched by clinical pathologic features.
But this is a retrospective study, small numbers of patients. And so while provocative, I'm not sure we can say it's definitive. So I think that it's still questionable, whether we can use a Decipher score to select patients for earlier receipt of adjuvant radiation compared to salvage radiation. I will say that for younger patients with aggressive features who present in my clinic, sometimes I'll use a Decipher score to lean a little bit towards adjuvant radiation. But I would say that you could argue that both ways.
And so the last point I want to bring up is whether we can modulate salvage radiation dose based on genomic classifiers. And for this, I want to highlight the SAKK 09/10 study. And I bring this study up because this conference is in Switzerland. And so this study randomized patients to receiving either 64 gray or 70 gray of radiation for biochemical recurrence after prostatectomy, and they all received radiation alone. And Alan Del Pra and others led a study basically looking at the impact of Decipher score on outcomes in these patients, and what he and his team found was that Decipher is associated with freedom from biochemical progression in this study. And here you can see Kaplan-Meier curves as a function of Decipher score. And this held up on multi-variable analysis. But I think the key finding is that similar estimates for Decipher risk groups were observed in the 64 versus 70 gray, meaning that I don't think we can use Decipher to select who should get the higher radiation dose versus not. And so I think in the context of trying to decide on salvage radiation dose, we currently can't use genomic biomarkers to do that.
And so this leads me to my conclusions. Number one, that genomic classifiers can be used to help individualize treatment decisions in the post-prostatectomy PSA recurrent setting. The strongest evidence for a genomic classifier is the use of Decipher to identify which patients should not receive androgen deprivation of therapy with salvage radiation for a PSA recurrence, but additional evidence needs to be generated to support other potential treatment decisions in this setting. Again, sorry that I can't be there in person, and I hope you enjoy this wonderful conference. Thank you.