The Uromigos Episode 112: Genomic Classifier of High-Risk African American Men With Prostate Cancer Identifying a High-Risk Population

Dr. Kosj Yamoah of Moffitt Cancer Center describes the work he presented at ASCO 2021.

Episode Transcript

Brian:
I’m here with Tom, and we have a special guest, Kosj Yamoah from Moffitt Cancer Center. He’s going to fill us in on this really interesting abstract around decipher score, and some racial disparities in use. And I think it’s a topic we haven’t really covered on the Amigos podcasts. So, Casa, if you could just… Kosj, excuse me, if you could just introduce yourself, tell us where you’re from, what you do, and then maybe just a little bit of background of how this study came about.

Kosj Yamoah:
Thank you very much for having me on your podcast. So, my name is Kosj Yamoah. I’m the chief of genital urinary radiation oncology at Moffitt Cancer Center. I also direct the radiation oncology cancer health disparities research, and I’m fortunate to chair the national veterans, the first prostate cancer health disparities group. So, with my focus on prostate cancer health disparities, this work is really an important testament to a very collaborative work that really speaks to the heart of some of the issues we face in prostate cancer, health disparities.

And so, before we get into this, I’ll call it a groundbreaking work, I just want to first talk about what we know about prostate cancer in men of African origin. Number one, the incidence is higher, meaning that the ability to get prostate cancer is high in among men of African origin, it didn’t matter what you’re in the United States or on the African continent or the islands, the incidence is higher.

Also, a case at a younger age, at a much younger age, and that their mortality could be worse. However, from developing the disease to mortality a lot happens, and this is where prostate cancer disparities get very complicated because the intermediary steps are complex of various things, including access and socioeconomics and structural racism, and also some biological components that we as scientists wanted to explore. So, that’s how this study really started, that was the backdrop with which this study was designed.

Brian:
And so because you told us about this decipher score. I think most people are familiar with genomic classifiers, but maybe just 30 to 60 seconds on what it is, how it’s used in practice. I think that’ll be a nice backdrop for your data.

Kosj Yamoah:
Right. So, the decipher score is really a genomic classifier, and its origins was really a way for us to understand when we look at localized prostate cancer, how following treatment or in terms of the tumor genomics, how the risk of metastatic disease to develop within five years, right.

That’s really how their tool was validated. So following treatment, you can actually modify that, or you can actually inform you on how you can actually use that information clinically, right. So, it’s a clinical estimate based on your tissue, either following surgery or now we have the biopsy component as well to tell you how aggressive your tumor is and the chance of it having metastasized in five years. So that’s the decipher testing. Now …

Tom:
What are you actually measuring?

Kosj Yamoah:
So, it’s transcriptomic data from the RNA transcripts that are generated from the tumor tissue itself. So, it’s not a genetic test. It’s a genomic test looking at tumor tissue RNA transcriptome, it’s a hood the entire transcriptome there, including the coding and non-coding regions.

Tom:
In terms of the gene expression, are there specific gene signatures which you’re interested in or in prostate cancer, is that right?

Kosj Yamoah:
So, this is a developed up to a 22 gene panel to give you that specific decipher score. However, as part of the great protocol, you can actually request the entire 46,000 transcripts that are out of that AFFYmetrix platform. But the genomic classifier itself is based on 22.

Brian:
22 gene score like another tumors. And so, this is a tumor tissue, and this is used currently after definitive local therapy, is that where you’re coming from?

Kosj Yamoah:
Actually, that’s also used in biopsy tissues as well. So, the application has now been expanded to include biopsy before any definitive therapies administer. So, you can actually get it on a biopsy samples. And that’s how we’re able to apply to this clinical trial we’ll be talking about in a little bit.

Brian:
Okay. And so, it’s basically an adjunct to clinical characteristics that PSA and Gleason score, et cetera?

Kosj Yamoah:
Yes. However independent characteristic of the genomic lab that I think makes it stand out is that it does not depend on clinical characteristics. And what I mean by that is, it gives you an independent data point, you don’t need to know the patient’s PSA or Gleason score or whatever. Once you have that team of tissue, it can independently tell you the aggressive features that you can now use to add them to clinical characteristics in terms of managing your patients.

Brian:
So, tell us about the data that you just presented at ASCO.

Kosj Yamoah:
Right. So basically, this is the first prospective validation study of the genomic classifier to ever be conducted in African American man. So, because most of the time, what we’ve had to do was to go retrospectively, so go back and take two more tissues and look for how they compare. And that is always fraught with uncertainty as to whether you’re comparing apples and oranges. And it doesn’t even you to give you a true idea of what is going on.

So, we wanted to validate the genomic classifier. Like any other genomic work precision medicine, these tools that are here to stay have not been developed nor validated in different ethnic groups. And for us because prostate cancer has a higher incidence and mortality, we thought that that’s a group of patients that deserve to have these precision modalities or genomics applied to their tumor tissues to give us a better understanding in a prospective fashion.

So, we use the very unique enrollment strategy that I think is really important here, that we actually enrolled African American men first and then match them consisting of clinically matched non-African American counterpart so that we are matching based on the Gleason scores, the number of course positive, the PSE range and all the things that we look for to estimate disease burden.

And then all of them did get decipher testing upfront. And then it was dealer’s choice in terms of treatment options, whether they choose surgery or radiation was a physician patient decision-making. And then we followed them for two years. That’s the design of the study. And I think that was a very unique design here, which has never been done before.

Tom:
And how many patients do you roll?

Kosj Yamoah:
We wanted to have available at least 100 in each group. And so, we grew, our target was 240 patients and to make sure that we have 100 patients available for each group to have a statistical analysis.

Brian:
So, when I have a couple of questions, but maybe just finish up sort of the main findings of the study.

Kosj Yamoah:
So, the first things we realized, as expected that these were, and to clarify, these are localized disease, low risk, majority, intermediate risk favorable, unfavorable intermediate risk. And for the low-risk category, we excluded patients who are going on active surveillance. So, these patients have to have had higher volume disease and should not be an active servant so that they are committed to definitive therapy with either radio therapy or radical prostatectomy.

The sights and [Wally 00:07:49] were included with the obviously Moffitt Cancer Center, as well as their VA centers. And what we found was that with the match cohort, clinical matching was incredibly good. Obviously, that’s that a study was designed. There was no significant difference in any characteristics except age.

So, age had about a six-year disparity in terms of the age gap. And that’s something that we felt like was important because they are getting the disease that a lot younger. And perhaps this is something that would impact guidelines in terms of when we started screening men of African origin.

And aside from that, when we looked even in there, we recognize that in the so-called low risk favorable intermediate risk patients, for which sometimes even active surveillance is recommended, highly recommended that we had a 49% chance of having a high genomic risk as in the African American cohort versus 10% in the non-African American cohort.

Now, this is to say that we’re not suggesting that these patients should not be, not all [inaudible 00:09:01], should not be basically put on anti-surveillance or put on less aggressive therapy. What this is saying is that we need to add these genomics to clinical characteristics because our current clinical staging criteria may be under staging a subset of men of African origin.

It’s not all of them because when we look at the data, those were low genomic classifies where the proportions were the same. It was the high genomic fraction that we saw that it was an enrichment of high genomic features by decipher in a subset of African American men. And I think those subsets and those that need to be managed a little differently.

Tom:
And you don’t know who those subsets are when you start out?

Kosj Yamoah:
You don’t know that because the clinical agent based on NCCN doesn’t give you that information.

Tom:
But that subset also there is a smaller group, there is a subset that also applies with other ethnicities at a lower rate.

Kosj Yamoah:
Exactly. So, we have 10% of that in the low and favorable necessarily risk group versus 49%. So yes, and that’s the beauty of these types of studies where you have a well-matched group that you recognize that this is enrichment, right? So, do we found the same substance in non [inaudible 00:10:13] enriched [inaudible 00:10:15]? Yes.

Brian:
So almost half the men who would have been characterized as lowish risk based on NCCN and could have even been surveilled actually had high-risk genomic features, is that right?

Kosj Yamoah:
Right. So, that’s a combined. So, I’ll say probably more like 25% in each group. Yes, it’s combined. Yes, 25%.

Tom:
What happened to the patients? Did they get the same sort of treatment with the treatment matched or was there an imbalance in there?

Kosj Yamoah:
This is fascinating. So, this was not a treatment intervention study. So, 2.0 might be what we start to do to intervene in treatment wise. However, there was mentioned data from clinical trials that African American men or men of African origin might benefit from some form of treatments that were a bit more intense, basically a combination of radiation and ADT versus radiation alone.

And so perhaps whether the use of ADT, which is the homeowner treatments could be what is really accounting for some of that aggressive biology and to be actually improving that response rates in the African American patients that is seen clinically. We yet to investigate that, but that’s the next thing we’re going to be doing to see whether this data translates to adequate treatment response in those cohorts that we see clinically.

Brian:
So, it stands to reason that the genomic classifiers can sort of unearth more men with higher risks that could impact treatment although he didn’t study that here. But what, I don’t know if you know, what percent of men of African origin actually get a decipher score. And I assume it’s a fairly low percentage, but what are the strategies to increase that? Because it seems like an important component of at least for the stratification.

Kosj Yamoah:
And, you bring up a great point. This is why this study is some ways being looked at very, very seriously because of the fact that we’ve had a dearth of information when it comes to men of African origin in genomic tests and across the board, whether it’s next gen sequencing or it’s [inaudible 00:12:11] or any other.

And this is probably the one of the rigorous evaluations in a prospective manner. And I don’t want to emphasize the prospective nature of this. And that is really looking at a patient African American patients get in genomic testing upfront. So, they’re…. Yes?

Brian:
Do you know currently what percent get it?

Kosj Yamoah:
From the initial analysis we’ve done, it’s probably in the 4% range. PCG also describes 4%. And so that’s been mostly what’s happening. Now the strategies improve that…

Tom:
Let’s make the assumption that no one in Europe’s getting this test, just make the assumption. Should this change your approach to the low-risk African American population? How does this …

Kosj Yamoah:
No, not really, not necessarily because when we look at low risk patients and again, remember that these low risks were low-risk, high-volume. So, let’s say we have low risk patients [inaudible 00:13:07] civilians for the low-risk patient regardless of ethnicity.

But what I would do differently is that I will make sure that I’m adhering to the guidelines, meaning that if they need to have PSA’s every six months on an MRI, which I would always recommend, especially for African American men, based on what we are learning, and biopsies as recommended.

Those are the very important protocols that will safeguard against the missing that 25% fraction of them that might have aggressive disease. So, I would just say have a knee-jerk reaction to just change treatment just because of the cutoff patient.

Tom:
So, this is a difficult question, but in my interpretation of that means that actually even in this high-risk group, you’re not actually doing anything differently.

Kosj Yamoah:
Well, we’ve not done the analysis yet. So, what we need to do next is to look among African American patients or patients that did fail localized treatment very early, how many of them have this high genomic signature? Was it consistent across both ethnic groups? That’s the question that we can ask them.

Brian:
[crosstalk 00:14:13] asked another way, are there data in a broad population, not this population about how decipher score impacts management?

Kosj Yamoah:
Yes. So, there is this tons of data. In fact, that’s what caused the decipher to really get them to clinical approvals, that tons of data that we, and actually there a bunch of clinical trials going on right now using decipher now to impact management.

So that is proven, but again, overwhelming majority 90% plus was not in African American men. So, this is saying that we need desperately to include disciple testing in men of African origin as much as we do for all other risk groups.

Brian:
So how do you achieve that? What are the barriers? Is it just access or there are other barriers that might be prevented in this population, prevent testing?

Kosj Yamoah:
I think it’s also both ways. So, it’s access and as well as education. So, historically, African American men or men of African origin have been very hesitant to go on checkout trials because of historic maps, right?

So, I’m doing both, right. So, in Moffitt, in Tampa, and in other areas, I advocate, I go into the community, I speak to them about importance of owning the process, getting involved in their healthcare really, and being advocates for themselves and recognizing that things have changed.

So, that’s one angle. The other angle is expanded access to these genomic testing and making sure that reimbursements are adequate. And I have to say decipher has to be very good at that in working with patients whose insurance may not even want to pay for it and making that happen.

So, I think that the companies must also with the patients to have adjusted rates that could allow them to pay out of pocket if the insurance flooded because of setting insurance premiums. Together, that’s how we can change that.

Tom:
What are the big flaws in the work you did, what are the shortcomings? And how can you correct them?

Kosj Yamoah:
We try to use real world application of this. So, we really didn’t, in some ways, evaluate whether a physician is using a decipher test to manage treatment or not. So that could be a potential for, for instance, I don’t know if my colleague who got a decipher test for his favorable intermediate risk patients decided to add short-term ADT just because he saw that the patient had a high decipher test or not, I don’t know, right.

So, because we wanted it to be independent of the test and allow the physician, a patient decision process to proceed, we are not quite sure how that decision of the provider based on the decipher [inaudible 00:16:44] impacted treatment choices and hence overall outcomes at the end of the study.

So that is something that will be real life data points that we will collect. But that could be one of the drawbacks here in terms of being able to say, well, this physician decided to add ADC or not, or the [inaudible 00:17:05] was done this way. And then immediately thereafter adjuvant radiation was recommended because of the disadvantages that came back, we cannot know that information.

Brian:
What are your next steps? Is it a separate prospective study? Is it longer-term, follow-up with the treatment data you mentioned, where do you go from here?

Kosj Yamoah:
So, first of all, a couple of things we can learn from this is that their current guidelines, NCCN guidelines, do not adequately risk stratify patients. And there has been a study out here in JCU that is proposing we use in what we call the clinical genomics restratification that puts in both the genomics and the clinical features into a different restratification.

I will be a strong proponent of this. And I’ll say that should be something that should be the next thing that clinicians should start to look at, because in the era of physician medicine, there should be no way that a prostate patient, we should only be relying on crude estimates of DRES, biopsies, and PSA alone.

So that’s one thing that we can start to really make change here based on a prospective data that we might be missing. And these subsets and that our current NCCN restratification may not be adequately doing the job.

So, I think that’s an easy takeaway point here that we can fix. The second thing is, can we use this information to manage patients? And that’s a separate question that we’ll need a clinical study in the short term to answer that. And we’re already working on that.

Brian:
So, Tom, I have one more question. Do you want to do a final one before we wrap up?

Tom:
No, I think this is really, it’s slightly different from what we normally do, these cohort type trials. I think it’s really interesting, it’s obviously extremely important. And this sort of testing in early prostate cancer is something that we’re not doing enough and maybe some of the other cancers that I’m involved with. Why don’t you put your last question together?

Brian:
So, I sort of just to go backwards a little bit, I mean, you have this big dataset in men of African descendants, the decipher score, and it’s a 22 gene classifier. Is there reason to think that maybe those 22 genes aren’t as applicable in this cohort of men?

Meaning is it another set of 22 were part of these and part of other ones, is there a reason to develop a separate test in this cohort? Or is it, I mean, that may not be practical, but is the biology different enough that that might be a worthwhile endeavor?

Kosj Yamoah:
Right. So, we’re doing that as well, like I mentioned earlier that we have access to the 46,000 transcripts. So, we’re going to be doing that. We already have a publication that came out last year in clinical cancer research that suggests that African American biology might actually have increased immune infiltrates, more of an immunosuppressive microenvironment.

So, we are trying to understand that as well, but ultimately, I think it goes down to your point is that we need to study all populations. And with that would identify enrichment of biology in subsets. And with that, we can actually personalize the treatments. And so, I’m a big advocate of that.

Tom:
This has been terrific. Listen, congratulations. It’s our first year Amigos ASCO podcast, 21. Next year, we’re going to see each other face to face and brighter, higher up a ballroom in a hotel, and you’d be very invited. I hope your work continues to go from strength to strength. It’s certainly an important topic. Thanks so much for joining us today.

Brian:
Thanks, Kosj K.

Kosj Yamoah:
Thank you. Yeah, thank you.