Prevalence of Germline Variants in Prostate Cancer and Implications for Current Genetic Testing Guidelines - Oliver Sartor

May 21, 2019

Oliver Sartor, MD discusses his contributions to the recent publication in JAMA Oncology on the prevalence of germline variants in prostate cancer and the implications for genetic testing guidelines.  3600 patients had various forms of germiline genetic testing performed and were available for this analysis.  In the dataset, 17% of patients had germline variants and 30% of those were BRCA1/BRCA2.  In addition, 4.5% of these patients had HOXB13 which is an important factor for prostate cancer risk and 9.6% had ATM variants.  Regional differences are also observed in the prevalence of these mutations. Dr. Sartor stresses the importance of germline testing, despite whether or not there is a positive family history. There are implications for individual patients as well as family members who may be at risk for developing certain cancer types due to these germline mutations. 


A. Oliver Sartor, MD, Professor of Medicine and Medical Director, Tulane Cancer Center; C. E. and Bernadine Laborde Professor of Cancer Research, New Orleans, Louisiana

Alicia Morgans, MD, MPH, Associate Professor of Medicine in the Division of Hematology/Oncology at the Northwestern University Feinberg School of Medicine in Chicago, Illinois.

Read the Full Video Transcript
Alicia Morgans: Hi, my name is Alicia Morgans and I am an Associate Professor of Medicine at Northwestern University where I'm a GU medical oncologist. I am delighted to have here with me today, Dr. Oliver Sartor, who is a Professor of Medicine and a Medical Director of the Tulane Cancer Center as well as being a GU Medical Oncologist. Thanks so much for joining me today.

Oliver Sartor: Glad to be here, Alicia.

Alicia Morgans: Great. Oliver, you've had many, many fabulous contributions to the prostate cancer literature, but I wanted to speak with you about one of your more recent contributions, really exciting paper published in JAMA Oncology, looking at germline variants in prostate cancer. You have a large dataset, and you've come to some really important, I think, clinical and guideline-type implications with the data that you found. Can you give us an overview of your study and just tell us a little bit about your work.

Oliver Sartor: Absolutely. I'd first like to give a little bit of credit to Colin Pritchard, who published an amazingly influential New England Journal paper a couple of years ago. When he did that, he and the many colleagues who contributed to that study, it really had implications for me and my patients. I started to look into the germline testing of our advanced prostate cancer patients. Of course, immediately, there were troubles and trying to get all the genetic counselors and everything, but we'd want to go into that. But, I ended up working fairly extensively and beginning to see the real implications of the germline changes that could occur in these patients.

I started to work with a company called Invitae, in particular. The genesis for this paper was really that I asked the company for access to their entire dataset on prostate cancer patients. They had about 3600 patients who have had various forms of germline genetic testing. They actually agreed that I would have access, be able to guide the analysis of these patients. We get a really nice data set, I mean up to 3600 patients for BRCA testing. Then, less tests for some of the other testing. But anyway, we looked at these men and characterize them to the best of our abilities.

We found that about 17% had germline variants, and it turns out that about only 30% of those variants were BRCA1/BRCA2, so that's kind of implication number one. Everyone thinks of BRCA1/BRCA2, but there's more to the story here. Previous large testing sets on the germline DNA repair, it's kind of neglected the HOXB13, which is unequivocally important for prostate cancer risk. We were able to identify 4.5% of our patients have that. Then, we had mismatch repair, which has therapeutic implications for ... and even the FDA says that pembrolizumab is indicated for those with mismatch repair, that'd be somatic, actually, are MSI high.

We found other things, like the family history, which everybody sort of intuitively thinks is really important, turned out not to be so important. Anyway, it was kind of interesting for us to carry out the study and the limitations, which I can't go into of course. But, the bottom line is we found a lot of pathogenic variants in the germline of these patients, and there were a lot of implications for those findings.

Alicia Morgans: Absolutely. I think just to take a step back from those really important and pivotal findings and kind of to go back to what you said initially, which was that you referenced the Pritchard paper. Since that paper came out, there have been multiple guidelines written about how we as prostate cancer physicians, whether we're urologists treating localized disease or medical oncologists or urologists treating more advanced disease, how we should think about doing germline testing in those patients. What I found, really one of the things I found very fascinating about your paper is that a number of the patients who were identified as having these potentially actionable germline mutations would not have actually been picked up with the current guidelines for germline testing as they are written. I think your paper said 37% of these patients would actually not have been really eligible for testing. Can you comment on that and also comment on the various recommendations that are currently out there because honestly, it's a very confusing field to sort through for all of us I think?

Oliver Sartor: Boy, it sure is. When you sort of make guidelines, you almost have to ask like what guideline because NCCN actually had three different guidelines that would potentially apply. They have some prostate guidelines, then they have the hereditary breast/ovarian cancer guidelines and then they have the Lynch syndrome guidelines. Each of those were actually inconsistent with one another when it came to prostate cancer. Then, of course, you move outside of the NCCN and it gets even crazier. 

Now, we wrote this paper in 2018, and it was kind of interesting that the 2018 guidelines, what, of course, we're referring to because that's when we submitted the paper. It turns out the 2019 guidelines incorporated many of the mistakes that we had actually identified in this manuscript. For instance, if you go back to the 2018 guidelines, there was a limitation on things that should be tested for, there was a limitation on this stage of the patients, there was a limitation with regards to DNA repair genes only, leaving off important things, again, like HOXB13.

When you begin to look at this, what we found is that if you've got advanced cancer, it really doesn't matter a lot about your family history. You really just need to get tested because you don't have other markers for doing so. The Gleason score, people said, "Oh, well, nothing happens with a Gleason 6." Well, turns out that even in Gleason 6 cancers you can run up a sort of surprising number of positives. I'm not talking about the localized ones. One of the deficiencies, and if you don't mind, Alicia, let me point out one of the deficiencies in this paper because it's very well. That is we did not have clinical annotation. This is coming out of a commercial database and we did not have individual assessments for all the staging that would have been important and would have been optimal. That was a shortcoming of the paper. But, the bottom line is if you've got an advanced cancer, we actually believe that panel testing looking for relatively broad number of genes is the best way to go.

Alicia Morgans: I would have to agree with you, and I appreciate that limitation. It was also a group of patients who had been referred to get germline mutation testing, so it wasn't just everyone in the population had their testing. These were patients who were included in the database because some clinician sent it off, sent off their tissue to be tested or sent to be tested. That is something, but generally, the levels that you found, and the lack of association with things like family history, least of the information that you had, I think was really, really important.

The other thing I thought was important and was mentioned in the paper, and as you can see in publications that are already out there, is that geography, and the way that we cluster as populations also can really alter the risk of individuals having these mutations. It was interesting in part of the paper you mentioned, that if you look back at the Pricthard paper, if your risk of having one of these germline mutations was 8.8%, if you were treated at the University of Washington; you could actually see a stark contrast to the risk of having one of these mutations if you were treated at Memorial Sloan Kettering, I think you said 18.5%. It's really interesting to see too how we geographically separate ourselves. I mean it makes you think back to old lessons in genetics with the founders effect and how different populations can kind of grow up with these certain mutations. I wonder if you have any comments on that, too.

Oliver Sartor: I do. I concur with the point made, that there's going to be geography, and there's actually some preliminary data out of Utah that makes Utah look a little bit different than many other areas of the country. I think there's ethnicity differences that we really need to explore. We have a fair number of African Americans whom we're actually able to report the African Americans as a distinct subset. You may not have noticed that part of the article because it's in the appendix, but it turns out that we really don't know a lot about African Americans and how the germline alterations might be present in that population. The Jewish does seem to be a little bit higher. But, as we think about geography, we also need to think about ethnicity. We're just scratching the surface, Alicia. There is so much more work to do here.

Alicia Morgans: Well, I'm glad that you've started us on the way. I just wonder if you can comment on a couple specific alterations that have been maybe under-discussed, and one in particular that's also maybe been over-discussed, so just two alterations. I'd love to hear more about the HOXB13 because I think that, as you said, has been really, you identified it. Not hugely prevalent, but it was definitely present. Then, also ATM because there's been some preliminary data that suggests that some of these PARP inhibitors, for example, may not have the same effect with that mutation. Love to hear your thoughts on those two alterations.

Oliver Sartor: Yeah. HOXB13 is pretty interesting. First of all, it's not DNA repair, it's a HOX gene involved in developmental processes, but the HOXB13 is uniquely expressed in the prostate. I think there's still a little bit of debate exactly over how it contributes to cancer, but there's absolutely no doubt. That was present in 4.5% of alterations. That’s not a huge number, but it has a pretty strong predilection. Interestingly, however, it doesn't necessarily predispose to bad prostate cancer as, say, BRCA2 does. BRCA2 is associated very clearly with more aggressive disease. HOXB13 is not, so it doesn't have a lot of treatment implications right now and doesn't have a lot of adverse prognostic event implications. But, if you've got a HOXB13 in the family, you're going to be quite likely to develop prostate cancer, so that's number one.

The other is the ATM. ATM, was of the positive variant, was 9.6%, so relatively prevalent. It was number three. We had number one in our group was BRCA2, then CHEK2, then ATM, and kind of order of number of variants we detected that were pathogenic. ATM is a very, very, very interesting gene, but the PARP inhibitors, which I think if you go back an original paper in The New England Journal with olaparib out of the Royal Marsden group, sort of implied that ATM might have predictive consequences for PARP inhibition. I'm not convinced that that's the truth today. There was a presentation at ESMO that seemed to indicate sort of to the contrary, that the ATMs were not responding to PARP inhibition.

Now, we need to have more data, but what we know from the PARP perspective is that all these are not created equal. BRCA1/BRCA2, quite PARP sensitive. I might mention another one, CDK12, which, by the way, is not a germline variant. When you get CDK12, it's always somatic, but CDK12 does not respond to PARP inhibition either.

Alicia Morgans: So much more to learn and I really appreciate you kind of walking us through this. If you had one message to take from this, though, and really thinking about how do we incorporate these pretty striking findings in our day to day clinical practice, what would that message be?

Oliver Sartor: I think it's really for the clinicians who are not testing to think about the implications of not testing. When you find a variation of importance, you're not just identifying it for the patient, you're identifying it for their family, the patients’ sisters, the patients’ daughters, and having the cascade testing that follows is incredibly meaningful and impactful for many, many, many people. If you're not testing, you actually are not taking, in my mind, full responsibility for the patient and all the positive actions that can come from a good physician and their approaches. I think germline testing is incredibly important. We're incorporating it into our clinic. The positive implications for daughters that we've been able to identify with BRCA lesions that they didn't really know were there is pretty incredible. Think about not just the patient, but for the family as well.

Alicia Morgans: Great. Well, thank you. Thank you for your message and thank you for walking us through this work that you've done. I really look forward to having future conversations where we can answer some of those unanswered questions that we kind of brought up to today. Thank you so much, Oliver.

Oliver Sartor: Thank you, Alicia. Glad to be with you.