George Netto, MD (Host): Welcome to a new podcast of Precision Pathology podcast series. Today, we're going to dive into a new area, which is precision screening. And for that, I chose an amazing paper, a lot to unpack in this paper that just came out a couple of months ago in Nature Cancer in the February issue. It's based on a multinational, a huge group of authors. And I'll start by mentioning first author is, Jason Vassy, and last author is Tyler Seibert. A lot of work from the group in Boston, but also Penn and Europe and Australia. And you can see the paper for the complete list. But I chose a colleague of mine and a collaborator, Dr. Kara Maxwell, Assistant Professor at University of Pennsylvania. She's a faculty member at the Philadelphia Veteran Affairs Medical Center. But also, a member of our Abramson Cancer Center, and the Department of Medicine at University of Pennsylvania.

Kara is one of our young stars who are working a lot trying to help prostate cancer patients, which is really the topic of this manuscript that we're going to use as a platform to discuss in general how we're moving away from preventive healthcare screening being one-size-fit-all. You do a mammogram, you do a PSA, to really a more refined system where it's more precise. We can use genetics, we can use other factors in deciding what's the population at risk and how we can best follow that population and prevent cancer, significant cancer in it.

And as Kara in a second is going to tell you, no disease is a better model than prostate cancer, giving all the controversy about is PSA helping us or hurting us, doing PSA test. So welcome, Kara, and thank you for accepting my invitation.

Kara Maxwell, MD, PhD: Of course. Thank you so much for having me.

Host: So, let's start with a little background. I kind of started the conversation about prevention, more precise screening.

Kara Maxwell, MD, PhD: Yeah.

Host: Tell me what led to that in more details?

Kara Maxwell, MD, PhD: I mean, you know, as you are well aware, and thank you so much again for having us on this podcast or having me on this podcast. It's really exciting to be able to talk about this work. It's really quite in the press after ASCO this year with precision screening, particularly with all the wonderful results coming out in the breast cancer world from the breast cancer risk and the WISDOM trial. So, we're, you know, really, excited to be working on this in prostate as well.

And as you've already alluded to, prostate cancer screening has been so controversial just within my career. You know, when I trained as an internal medicine resident, we did PSA on everybody. That was what we thought we had to do. And then, the pendulum just really swung, you know, as you're well aware, to really this becoming a category D recommendation from the USPSTF and, you know, thinking that we shouldn't screen anybody. And clearly, neither of those extremes were at all the right approach, right? We know that there are subsets of men that get aggressive early-onset prostate cancer, aggressive prostate cancer as they get older, and we really need to understand who those men are, who are at more risk for actually dying of prostate cancer than the general population. And that was really the impetus, you know, for this study.

Host: Thank you for that nice introduction. Let's talk a little bit about—you have a figure 1 in that manuscript, which really summarized the whole process. And what I like about it as a pathologist is to highlight how hard it is to develop a test, to validate in cohorts, and to make it a clinical test, and then now taking it into a clinical trial. So, it would be great if you guide our audience through the different steps. It sounds like it's nested in the Million Veteran Program. So if you can describe to us the different population that it was developed and tested on.

Kara Maxwell, MD, PhD: Of course. I really think that, you know, we're finally in the era of really being able to use all of the genome-wide association studies that we've all been aware of and many, many people participating over the years to find, you know, really define the genetic architecture of various diseases like prostate cancer. Prostate cancer is a very heritable disease, and we know that there's a significant number of loci throughout the genome that have associations with prostate cancer. This is all great. However, if you're going to validate a score like this, you do need to validate it in the population that you're targeting.

And so, because this was a VA study, we really were very keen to open this in our veteran population. This is a population at very high risk for prostate cancer. And so, we started with the Million Veterans Program. So, the Million Veterans Program started a number of years ago now and has really been an amazing demonstration of the willingness of people and be involved in research and also the strength of really these national collaborations.

And so, Million Veteran Program has enrolled over a million veterans now with a blood test, with survey results, and with linkage to the really extensive electronic health record in VA. So, this is the oldest electronic health record in the country. It's a beast to work with, as many of the physicians on here are well aware. However, what it means is that all that data is collected. And then, with linkages to other things such as SEER data, Medicare data, we can get good pictures of many of our veterans' trajectories.

And so, the Million Veteran Program did some genetic analyses. And I want to be very specific here about this, given the pathology audience. You guys are well aware of these different tests. So, there were two tests that were done in the genetic space on the veterans. One was a SNP genotyping array of approximately 600,000 men. And then, in addition, there's been whole genome sequencing, you know, separate from a SNP genotyping array on a subset of men. This study just used the SNP genotyping data. And this is because most of the loci we found in GWAS prior have come out of SNP genotyping arrays.

And so, the model, though, the decision was made instead of using prior published loci, that we would look to see if there were additional loci that might be important in our veteran population. That was really the sort of the impetus for the internal development of the model with VA data, and then validation to come up with pCare in general.

Host: So, this is the polygenic score when you're talking about the side, 601 Yes top polygenic score.

Kara Maxwell, MD, PhD: Yes, correct. Exactly. Identifying the loci was really the first step for developing the pCare model. And then, as we are well aware in cancer risk, there are many other factors that lead to increased cancer risk in an individual beyond tjust their genetics. And so, this study also wanted to develop other markers of things that could cause people to be at risk. For example, we know that African American individuals are at a higher risk for prostate cancer. So in this model, what we chose to use are called the genetic principal components. These are a measure of sort of the population substructure that is within, the population that you're developing your model on.

I think it's important to talk about the family history piece because this is one of the things many of these polygenic models are being developed. And unfortunately, a lot of times it's a black box of what exactly is going in. You maybe understand which loci are going into the model, but particularly the family history variable is extremely different in different studies.

So, I think it's important to point out we don't have the most ideal family history variable. We just did a binary presence or absence of a first-degree relative with prostate cancer. This is possibly going to need to be refined in the future, but I think it's important to sort of make that note that family history and genetic ancestry are incorporated into the model in addition to the polygenic genetic score.

Host: And that's what you're referring to. So, started with a polygenic score, which sounds like included 601 loci in the MVP, the Million Veteran Program. And then, you added the family, a dichotomous, as you said. And also, you added some genetic ancestry. And that's the principal component. And those three components created what is called PCARE, which Prostate Cancer Integrated Risk Evaluation. So, we'll refer to it as PCARE. And that's gives you a score now, and that went to the validation step.

Kara Maxwell, MD, PhD: Correct. Exactly

Host: So, my understanding, you validated it in four previous cohort, called practical cohort, including the Swedish, the African, the Asian.

Kara Maxwell, MD, PhD: And Protect.

Host: And Protect. Thank you. So after the validation, now you're ready to make a clinical test out of this.

Kara Maxwell, MD, PhD: Correct. Correct.

Host: So, can you expand a little bit about this?

Kara Maxwell, MD, PhD: Yes, of course

Host: This combined genetic and exomic test that you were developing?

Kara Maxwell, MD, PhD: Of course. Just to make a quick note about the validation, you know, this was a really important component of this because one of the major concerns with polygenic risk scores in general is whether or not these are going to perform well across different genetic ancestries. So, I think it's important to note that at least in prostate cancer, I think, enough of the genome-wide association studies, including MVP, have been done across ancestries, that we're pretty confident that the score is actually working across ancestries. I'll say that that is not necessarily the case for all cancers. And so, that point is very important with the validation than with the other cohorts of other mixed genetic ancestries.

The next part was then to say, "Let's bring this into a clinical test," right? And that actually ends up being the tough part. One of the main things that we had a lot of debates on is we know that there are also monogenic causes of prostate cancer as well. These probably only account for 3% to 4% of prostate cancer, but those individuals are at very high risk for a totally different genetic reason. And other studies have shown that if you have a rare monogenic disorder, your polygenic risk score may not be as high. It might actually be low. And so, we were very keen to make sure that we were going to be testing the broad array of genetic risk.

Now, it may seem weird why am I spending so much time on this. Well, this is basically population-level screening for rare monogenic disorders, which is not something we routinely do, right? We don't have any sort of guidelines to suggest that we should be on a population level testing for BRCA1 or 2 or Lynch syndrome. And that is essentially what we're doing in this study.

So, it was important to us that when the study was being built that this was being done, but that also that that would be an important component of the consent that the individuals coming into the study would be understanding that they were signing up for either their polygenic risk, but also this potential to find a rare monogenic disorder.

So, how is this going to get done, right? Because this gets into the weeds, and again, as pathologists, I know you love this. You know, SNP genotyping is done completely different than exome sequencing, right? So, how are we going to actually get a test that combines the both? And so, working with Broad Clinical Labs, this was really a great collaboration where they developed probes then across the individual loci that are associated with the polygenic score so they could be done in one NGS test versus having to do two separate tests.

The clinical report was really scrutinized, by a lot of individuals, a lot of genetic counselors, providers like myself that do see individuals for genetic risk, to try to come up with a pCARE report that would be reasonably interpretable, right? It's kind of hard when you have two different components of this test, the rare variant versus the polygenic score, sort of how we were going to come up with that. We feel that we have a nice test report, and then that has led to the ability to have the PROGRESS trial started.

Host: So, these monogenics, so in addition to the polygenic and family and ancestry like we talked about to create the pCare, now you're adding, you mentioned, the monogenic. And those are the same ones listed in the NCCN guidelines. So, the BRCA, BRCA1, BRCA2, ATM, PALB, CHEK2, and what have you, HOX, and like you mentioned, the Lynch syndrome, p53, and EpCam.

So, around 12 genes that you're adding. And then, Broad developed this test, and now that test led to a report that is every provider like you, is going to use as you're implementing, and we'll talk in a second about the clinical trial that is undergoing based on that test. So, the report usually mentioned if you have one of these monogenic, you're high risk.

If you don't have any of these 12, then that's where the pCare comes, and I split it into three groups. You're a low risk, you're a normal risk, you're a high risk.

Kara Maxwell, MD, PhD: Correct

Host: And that's how you explain to your patient, and then you—I guess, this is a randomized trial. So, it's blinded. You're going to put your patient on a different—so let's finish by talking a little bit about the PROGRESS study.

Kara Maxwell, MD, PhD: Yes. Yes. Because that decision right there was a difficult one, right? So, do you bucket? Do you bucket these scores or do you give a numerical value, right? And that's certainly something that is debated in this field, and there are obviously companies out there that are giving polygenic risk scores. And, you know, this sort of decision of how do you communicate that risk to somebody is really important. Do you give them a number? Or do you give them both a number and do you bucket them? And we made the decision to do high intermediate, and low because we were really trying to guide sort of the decision-making process, right?

If you are at high risk, there is a thing that we're going to implement, right? And then, if you are at low risk, that means you're at lower than the general population risk. That also, we are hoping, will lead to a physician instigating perhaps, you know, a different track saying, "Well, maybe we don't need to do PSA screening," right? "Maybe this is going to potentially cause harm for you." And so, we wanted to make sure that it wasn't just, you know, you're getting a test that shows high risk, but there is also that low risk category as well.

Host: ' Because ultimately, the whole study is trying to triage PSA and to answer that controversy about PSA in black men versus PSA in the rest of population. Are we discovering a lot of 3+3 = 6 that we shouldn't even be calling cancer probably down the road? So now, you get that score, you get that report, and how you're doing it. And these are, by the way 5,000 patients, right? Veterans who are going to use this system in deciding how they're screened with PSA and how they're followed up. We'll talk the MRI in a second, too, how this plays in. So, how do you practically use the data now, that report.

Kara Maxwell, MD, PhD: Correct. You know, this, Dr. Netto, is such a controversy in our field right now, right? Is that sort of who are the right providers to deal with the genetic revolution, right? And so, you know, it's not just prostate cancer, right? There's the WISDOM trial that just came out in breast cancer. EMERGE has been looking at polygenic scores for cardiovascular disease in addition to cancer. And so, this is going to hit us in the next—or it's actually hitting us now, I guess I should say, right? And so, who are the providers that are going to be able to handle this? And the fact of the matter iis that, you know, it's going to require reeducation no matter sort of who it is.

Most oncologists don't deal in cancer risk. Most primary care doctors do deal in risk, but they don't necessarily deal in genetics, right? So, you know, we made this decision that in this trial, Dr. Vassey is an internist himself. We've decided to implement this in a primary care setting. I think that, you know, in the end of the day, this is going to probably be the right setting to implement genetic scores. And it may mean that we need a whole separate category of sort of internists that are learning about genetics and doing these kinds of clinics. But this really is the type of provider that, you know, most people will be seen. It's going to be very tough to sort of, you know, get somebody into more specialized genetics clinic, which frankly most VAs don't have.

Most, you know, most practices in the country don't have, right? So, the decision here was to really go through the primary care network. This is a study where letters are being mailed out to veterans, and then they're deciding to get the test, and then they're bringing it to their doc at their, you know, VA saying, "This is what I got. What do you think that we should do?"

Part of the study is also just going to be to understand how this works in real clinical practice. How many people are going to be willing to say not do a PSA, right? We've already learned from the WISDOM trial in breast cancer that a lot of women aren't really willing to give up their mammograms, right? Even if they find out they're at low risk, right? So, are we actually even going to be able to move the needle on that side, I think is one point, not just from the patient perspective, but from the provider perspective as well.

Host: That's such a great point. There's that culture and there's human being behind and culture of education and practice. This is wonderful. It's a lot of things, we're going to learn from this trial.

Kara Maxwell, MD, PhD: Yes.

Host: What is the expectation? When do the results conclude of the progress?

Kara Maxwell, MD, PhD: You know, so we've been really recruiting without, you know, at all of the goals. And so, you know, I don't have the exact timeline in my head. This is a seven-year study altogether, probably in the next couple of years we should have some sort of interim results. As the paper showed, you know, the model works well. It really, you know, well delineates individuals at very high risk. And I just think it's important to make the point that if you have the highest level of polygenic risk, you're on par with a monogenic risk score as far as your lifetime prostate cancer risk.

The only thing that I think is important to point out is that we are talking just about risk of development. We're not talking about necessarily risk of aggressive disease or dying. And that is also a second controversy in the prostate cancer world, is are all these genetic things that we're developing shouldn't just say prostate cancer world, any cancer world. These risk scores that we're developing, are they predicting for aggressive disease or just disease or not? And that's a, you know, a fine point that I think does need to be discussed. Regardless, we do know that these scores do predict well your risk of disease. And like I mentioned, the high sort of polygenic risk score really is at the same level of risk as a rare, monogenic disorder. And we already knew that early screening worked to find clinically significant prostate cancers early in that population.

And then, we were really excited, the BARCODE study came out from Raziel's in New England Journal just recently, and that showed us that using polygenic risk, using the high polygenic risk score also to direct screening did identify clinically significant prostate cancers at a higher rate. There's a lot of work that still needs to be done to make sure that we are actually improving mortality, right? That is our bar that we always want to meet. However, I think that the studies are suggesting, you know, a move in that direction.

Host: And that's really based on and as eloquently shown in the paper that it does pan out for metastatic disease, for death of disease the way you design the test. But what you're saying right now, it's early to see if it's really going to—because you're trying to predict you're at high risk or low risk of being diagnosed with cancer. That's the point you're testing right now. But hopefully, in the same time, it will, because the ultimate goal is we want to find the aggressive cancer, the lethal phenotype.

Kara Maxwell, MD, PhD: Exactly. Exactly. Well, and I think it's always important to sort of make that point that if you have a genetic risk factor that increases your risk, by definition, you are going to have a higher chance of developing metastatic disease and dying from that disease. But it doesn't necessarily mean that the marker predicts for aggressive disease.

So, there's still a lot of work in that space to determine that. But it is, you know, it's a fine point because a lot of patients will say to me, they're like, "Well, look, I hear what you're saying. But if I'm at higher risk, then that by definition means I'm at higher risk of dying of the disease. That's still relevant, right?" So, you know, I think it really depends on— sometimes we get ourselves wrapped up in semantics and forgetting, you know, the result that's in front of us.

Host: And of course, the next step would be, so this is probably integrated also with the MRI needle-guided biopsies. And hopefully, based on that study, we have to mention AI down the road where there'll be a multidimensional, where maybe this one be one of the dimension, the polygenic score and the monogenic or the blended genomic-exomic BGE test score can be one component in addition to all the others.

Kara Maxwell, MD, PhD: Yeah, Dr. Netto, I'm really glad you mentioned that because I think that, you know, the MRI is just a fascinating piece here. So, in the rare genetic carriers, we have another study, open here at Penn that's multi-institutional called PATROL, where we're looking at earlier lower PSA values in rare genetic carriers, so rare monogenic carriers.

And the role of the MRI is also being looked at at that study, because what's been really fascinating is that you will identify clinically significant prostate cancer at lower than recommended PSA thresholds, say four for the general population, three for monogenic carriers, that have a completely, you know, obviously abnormal MRI, but then also ones where the MRI is completely normal.

And then, the flip side, we'll have, you know, an abnormal MRI with a very low PSA. We're seeing this in PATROL. I think that this is being seen across other studies as well. And so, just having the genetic risk and then directing towards PSA only is probably not enough. We probably need MRI as well. However, this is a big shift. We have never used MRI screening in prostate cancer before. We rely on our blood tests. So, who are we going to do this more expensive screening test in, I think, is important. And so, you know, the thought being, again, in these high polygenic risk carriers, perhaps the MRI is an important component.

That's not being studied formally in progress at all. That's certainly something that's going to be coming, and I think that's where AI is going to be important, right? So, why are some of those MRIs that are PI-RADS 1, but the gentleman has a, you know, clinically significant prostate cancer, what are we missing in the read? And I think, you know, there's a lot of excitement on looking at those MRIs for that particular component.

I will use the opportunity to shout out another VA study that I'm involved in that is using AI, led by Dr. Anant Madabhushi at Emory and Atlanta VA, and Matt Rettig in UCLA and the LA VA. And we at Philly are involved where we're consolidating MRIs and pathology slides from as many men with prostate cancer as we can get across the VA system and using AI to try to predict who's going to be at higher risk for recurrence. And so, combining that sort of finding with these polygenic risk scores, I mean, I think that's where we really need to go.

Host: Awesome. A lot to do. Thank you very much. I really enjoyed the conversation and we really appreciate everything you do, and I look forward to bringing you in the future to discuss the other studies that you hinted to. Thank you so much, Dr. Netto. Thank you.