Decipher Scores and PSMA PET Imaging: A Correlative Study in High-Risk Prostate Cancer - Amar Kishan

October 13, 2023

Daniel Spratt engages with Amar Kishan about the role of transcriptomic profiling and PSMA PET imaging in prostate cancer. Dr. Kishan discusses a study involving 18 trials and 12,500 patients, which suggests that extra-prostatic disease is often the driver of distant metastasis in high-risk patients. The study also explores the correlation between Decipher scores and the risk of upstaging on PSMA PETs. Dr. Kishan emphasizes that while Decipher is easily available, PSMA PET is not universally so, advocating for its use in high-risk patients with high Decipher scores. Both experts highlight the importance of ongoing prospective trials and the potential for these tools to refine treatment pathways.


Amar Kishan, MD, Professor, Vice-Chair of Clinical and Translational Research, Chief of the Genitourinary Oncology Service, Department of Radiation Oncology, UCLA, Los Angeles, CA

Daniel Spratt, MD, Chair and Professor of Radiation Oncology, UH Cleveland Medical Center, Seidman Cancer Center, Case Western Reserve University, Cleveland, OH

Read the Full Video Transcript

Daniel Spratt: Hi, my name is Dr. Dan Spratt. I'm the Chair and Professor of Radiation Oncology at UH Seidman and Case Western Reserve. I'm here with UroToday, and very lucky to have Dr. Amar Kishan, who is a professor and Vice Chair of Clinical Research at UCLA. Thank you so much for being here.

Amar Kishan: Thanks very much for the kind invitation.

Daniel Spratt: So, you've done just incredible work in prostate cancer, in transcriptomic profiling, and in PSMA PET imaging. And recently you did a study that's really trying to assess how do transcriptomic profiles and PET imaging maybe correlate or be additive. Can you just tell me more about this study?

Amar Kishan: Sure, absolutely. And thanks for your kind words. So, this goes back to the fundamental question, what is the driver of failure after radiation? And we've looked at it, meaning, really us, the two of us, and others in a consortium group we built. When someone has a recurrence after radiation that's a metastatic recurrence, is it really that the primary treatment to the prostate failed and then the cancer spreads, or that these are disease sites that occurred and existed but were occult before the treatment was done? And PSMA PET would give us an idea potentially to identify some occult disease that's extra-prostatic in nature.

So, we looked at 18 trials, 12,500 patients, and found that most patients with high-risk disease that develop a distant metastasis, or are developing a metastasis before a clinical relapse in the prostate is identified, which suggests that extra-prostatic disease is the driver of distant metastasis. If that were true, tools like the Decipher test, which are highly prognostic, may be telling us that the disease has already spread. That's one possible interpretation. As opposed to just being a very aggressive primary tumor, perhaps a patient with a high Decipher is at high risk of occult cancer that has spread.

As you very well know, the original discovery and validation series for Decipher obviously were performed before PSMA PET was widely available and therefore it is truly possible that high Decipher is indeed predicting for this, or prognostic of this. In fact, there are several nationwide trials that are investigating this question in NRG-GU009 and high risk disease, and O10 and intermediate risk disease. Does having a high Decipher serve as a predictive biomarker for intensification of systemic therapy?

So, it was with that backdrop that we looked at this question, patients at risk of upstaging on PSMA PETs, do they also have a high Decipher score? Could there be a relation between these two things? Now of course, PSMA PET is so new it is not widely adopted. And so we don't have thousand patient data sets where patients have had a PSMA PET, what we do have is a nomogram that we developed at UCLA based off of 213 patients with high risk disease that had a PSMA PET where we estimated the risk of up staging the risk of a patient coming in with conventional imaging confined to the prostate. You get a PSMA PET and it is indeed node positive or metastatic. And we took the commercially available Decipher tests, they have a database called the grid database, and we obtained the Decipher scores from those patients as well as their risk of upstaging using our custom nomogram. This is 4,625 patients with high risk disease.

And the bottom line result is that we found that Decipher score is correlated with the risk of up staging, with a correlation coefficient of around 0.42. So, not a perfect linear correlation, but a significant association. And moreover, on multi-variable analysis, when we control for Gleason score, PSA, stage, age, and Decipher, there's still a significant association between Decipher and a risk of upstaging on PSMA PET. And the strength of that association, if we try to quantify it, is actually very similar to PSA levels and similar to earlier Gleason grade group, and outperforms T stages in terms of the power of that. Now a related question is in very high risk prostate cancer, or let's say the STAMPEDE criteria for high risk prostate cancer, and this is something I know that you have talked about and presented on and lectured on.

In that specific group of patients where it has been shown that there is a benefit to systemic therapy intensification, is that really just coming from this effect? Is it because these patients that get a benefit from adding abiraterone to ADT that's driven by the fact that they have extra prostatic disease? In that cohort, the median Decipher score was 0.9, this is about 426 patients or so. And we still saw that correlation, which is pretty strong. So, our overall conclusion and our hypothesis here is that the Decipher score, it may be predicting for aggressive primary tumor biology, but is also predicting for a ability to spread. And many patients with a high Decipher may have extra-prostatic disease at the time of their presentation.

Daniel Spratt: It's really amazing because as you know, Decipher was trained to predict metastatic disease, obviously with conventional imaging as you said, and the fact that now years, many years later with much more sensitive and specific imaging with PSMA PET that your nomogram predicted probability of extra prostatic disease that this is predicting, and there's a strong correlation, or a correlation between them. In your practice, or is there something clinically today, if you were to tell the audience out there, how do you think people could use this information in their practice?

Amar Kishan: Yeah, great question. Right now we know that Decipher is pretty easily available, if a patient has tissue that's available for testing, that can be obtained. PSMA PET though it is approved, FDA approved, it's not universally available, and different centers might not be as comfortable obtaining and interpreting a PSMA PET. So, my takeaway would be, someone with high risk disease and a high Decipher, we really ought to do a PSMA PET to identify if there already is extra-prostatic disease, and that can have implications for their treatment pathway.

Daniel Spratt: Great. And then in terms of next steps, because obviously it's been shown by your group and others, PSMA PET is prognostic as well. And Decipher obviously has been shown repeatedly to be prognostic, and as a lot of people, sometimes, not always, but both together may even improve prognostic performance more. So, curious in terms of what next steps do you think now that both sets of data from these modalities, or these biomarkers, is becoming more readily available, what do you think you can do next to make this really help patients more?

Amar Kishan: Yeah, great question. So, first of all, the prospective trials are underway, which is great. So, the two that I mentioned, and others that are being launched as well. Our next step is to actually see if we can validate this in patients who have had a PSMA PET, because of course the limitation of what we did was we applied this nomogram, no nomogram is perfect. So, what we're working on right now, what we're presenting some preliminary data at this meeting, and we're hoping to have more robust findings early next year, is looking at a multi institutional data set of patients who actually had a PSMA PET.

So, they truly know we can detect that they have node positive disease, M1A disease, M1B, M1C disease, in the definitive or post-op setting. Maybe we can link that with Decipher data and possibly look at what pathways might be activated or more prominent in a patient that has metastatic disease at presentation versus at recurrence. Or is there a difference between propensity for nodal recurrence versus bony recurrence? And that would be the next step. Beyond that, if that type of effort is positive, then we might be able to design further trials that are a little bit more refined to certain patient populations.

Daniel Spratt: That's really amazing. And as we conclude, just thinking back 10 years ago, not that long ago, a conversation like this, we would dream of having this level of precision where we're talking about genomic classifiers, molecular imaging, using them together, the ongoing NRG trials. It's just amazing for patients. And thank you so much for all the work you're doing. Thank you personally. It's always great getting to work with you. And thanks so much for being here today.

Amar Kishan: The pleasure is all mine. And thank you very much for the invitation.