Read the Full Video Transcript

Andrea Miyahira: Hi. I'm Andrea Miyahira at the Prostate Cancer Foundation. With me is Dr. Tim Rebbeck of Dana-Farber Cancer Institute. Dr. Rebbeck will discuss his group's most recent paper, "Heterogeneous genetic architectures of prostate cancer susceptibility in sub-Saharan Africa." This was just published in Nature Genetics. Dr. Rebbeck, thanks so much for joining.

Timothy Rebbeck: Thanks. Great to be here. Thank you for having me here today. I'd like to discuss the work that we've recently published from the Men of African Descent and Carcinoma of the Prostate, or MADCaP, Network. This study involves about 7,000 individuals from across Africa, cases and controls, cases with prostate cancer and controls. You can see on this slide the number of centers in Africa that were involved in this research. And we've been funded by a number of organizations, including the NIH and the AACR.

This study involved genomic characterization of the prostate cancer cases and controls in order to understand a little bit more about the heterogeneity that we observe in prostate cancer, and particularly to address the disparity that we have observed in prostate cancer risk and outcomes around the world.

As you all know, men of African descent have the highest rates of prostate cancer. And as far as we know, the risk of developing prostate cancer is strongly genetically driven. There are relatively few epidemiological risk factors that play an important role aside from race, age, and family history. So we think that genetics is incredibly important.

And so we've begun to characterize this complex and heterogeneous genetic architecture across Africa. And in this slide, I show just a summary of, on the top, the genetic ancestry estimates among West, East, and Southern African men in our study. And as you can see, there are very clear distinctions in genetic ancestry and genetic architecture across these different populations, although they are not completely unique. There is some variation that is common across all of these areas.

In this study, we also then identified a number of different loci associated with prostate cancer. And this shows you, across the genetic variants that we reported, the genetic variance proportion, that is, the proportion of variation that was attributable to each of these particular loci.

And as you can see, there were a few that popped out, including loci at 6q, 8q, and 11q, which were previously reported as important prostate cancer susceptibility regions. But as you can see, there's a huge amount of variation in the proportion of the variation in prostate cancer risk according to each of these areas.

And particularly, 8q24 is an important area and seems to have a much bigger role in prostate cancer in West and East Africa and substantially less in South Africa. So we're beginning to see some heterogeneity, even within the African continent, around what kind of genetic contributions there are to prostate cancer risk.

We also observed some interesting phenomena, including the observation that effect sizes on risk vary more than allele frequencies. So on the top set of panels, you can see the comparison between West-East, West-South, and East-South, the different comparisons across regions. And you can see that the frequency of alleles that we observed are very similar across all of these regions. There's a lot of correlation.

In contrast, the effect sizes, the odds ratio effects contributed by these loci vary quite a bit more. And they're not the same in each part of Africa. So we're beginning to get a sense of why we see differences in risk or why the genetic contribution is different. And this seems to be in part due to the fact that the effect sizes in different parts of Africa vary.

We also evaluated a little bit about the nature—the reasons why we see these variations across Africa, and we found three interesting observations. One is that many of the alleles that confer risk are private alleles, meaning they're only found in Africa. So these are alleles that would not have been identified in studies of European or Asian descent populations. They're only seen in Africa.

And so this helps us understand why it's important to study very diverse populations, because these private alleles contribute something to our knowledge about prostate cancer risk, but we only see them if we're studying populations within Africa. And that's something you can see on the left-hand side, the effect size allele frequency estimate in Africa on the x-axis and the allele frequency estimate in Europe on the y-axis. And you can see, down on the bottom left-hand corner, there are a number of alleles that are essentially missing in Europe that are very important based on our association studies in Africa.

In the middle panel, we also identified a number of younger SNPs, that is, SNPs that arose after the out-of-Africa migration and/or are unique to Africa. So if you look on this middle panel, you can see that the number of generations, younger alleles on the left to older alleles on the right, the gray dots represent alleles that are rare in Europe, and the blue dots represent alleles that are relatively common in Europe. And these represent the alleles that we found in our association studies in Africa.

So you can see the newer alleles, younger alleles, are not seen in Europe, whereas the older alleles that probably arose a long time ago in Africa, but then were transmitted through the out-of-Africa migration to Europe and other places, those are very old alleles. So we're beginning to get a pattern of where these alleles arose and at what time.

And finally, on the right-hand side, we have evidence for neutral evolution. And this is maybe completely understandable, that the data that we have on the right-hand panel suggest that the reason that we have differences in frequency in Europe or in Africa, or particularly in Africa, based on our data, is that there is no natural selection occurring, that the reasons for variation in the allele frequencies are due to genetic drift or mutation, which is what you kind of expect, because we wouldn't think that prostate cancer would have a huge impact on reproduction given how late it tends to be diagnosed.

But this confirms that the frequencies that we're seeing really are a function of neutral evolution. And maybe that's a little bit different in Europe, because you can see the alleles in the European side are a little bit skewed to the right, suggesting that there might be some additional evidence of natural selection in Europe. And we've previously published evidence for a number of alleles that seem to be under natural selection in some way in Europe.

So in summary, Africa-specific variants are associated with prostate cancer, we knew that before, but this is the largest study that really confirms that and that confirms that there are alleles in Africa that are not found anywhere else that confer prostate cancer risk. We've identified clinically relevant haplotypes—that is, chromosomes that are African that have very large effects, odds ratios greater than 2—which might be clinically relevant, and that the genetic architectures vary substantially across populations both within Africa and between Africa and other parts of the world.

So we think these data, in addition to informing prostate cancer risk, really provide a motivation for the need for research in diverse human populations to provide novel insights into cancer etiology, risk, and prevention. Thank you.

Andrea Miyahira: So thank you so much, Dr. Rebbeck, for sharing this with us. So did you evaluate the genes near African-specific prostate cancer risk loci, such as the long non-coding RNAs near the 8q24.21 locus, to determine if there are any variants with functional alterations that may explain the association with prostate cancer risk?

Timothy Rebbeck: So we did some deep characterization of 8q24 in general. We did an analysis of the linkage disequilibrium patterns and haplotypes that are specifically associated with risk. And we identified African-specific lead SNPs and lead haplotypes that are unique to Africa.

So it's very clear that we have unique genetic architecture of risk in Africa compared to other places. And some of these loci include regions that have been previously identified as being important for the functional regulation of prostate cancer risk, maybe [INAUDIBLE], upstream of [INAUDIBLE], and things like that. But we, in this analysis, didn't specifically evaluate functional associations.

Now, we've just developed a study where we are looking at expression patterns in our prostate cancer cases. We have gene expression data. We have the germline data that I just presented. And we're going to be doing TWAS, RWAS, CWAS—some studies that will help us understand the reason for variation in gene expression. And so that's the next step.

Andrea Miyahira: OK. Thank you. And considering Africa is highly diverse with many populations, do you see any variance in prostate cancer risk in different populations?

Timothy Rebbeck: Yeah. First of all, we identified the lead SNPs in most of our associations were unique to Africa, so again, things that would not have been identified had you only studied European or Asian populations. These are African-specific. And the effect sizes are different across Africa. So we did identify things that were unique to East, South, and West Africa, varying across the different regions, and things that are specifically unique to African populations.

Andrea Miyahira: OK. Thank you. How far do you think we are from identifying all of the prostate cancer risk genes in Africa and other populations?

Timothy Rebbeck: Yeah, I think we have found a lot of the genetic variation that is detectable by things like case-control studies. So as you know, the recent work of Chris Haiman's group, that we were a part of, has identified 451 prostate cancer susceptibility genes. And these are starting to be quite small in effect so that there are probably more loci out there that have very small effects. We would have seen them if they have large effects at this point.

We have pretty good polygenic risk scores based on the identified loci. Based on what we found right now, in our African set, any novel loci that would be identified in Africa are going to have very small magnitudes of effects.

And again, that might be biologically interesting. But for our clinical prediction, risk prediction, risk stratification, we believe that the effects of any genes that have yet to be identified, any loci, are likely to be quite small. And I think that's what we've seen from the very large studies in the multi-ethnic populations that have been done to date.

Andrea Miyahira: OK. Thank you. And speaking of the polygenic risk score, how do you think these scores should be developed and applied globally, considering many populations remain understudied and admixing increases with each generation?

Timothy Rebbeck: Yeah, we have computed polygenic risk scores. And we're actually going into more detail in the polygenic risk score calculations and assessments in Africa. And one of the reasons that's of interest to us is because many of these cases are not screened.

They're definitely associated with more aggressive disease. They're not associated with PSA and screening. So there's some unique information that I think we can still learn about how PRSs will impact risk prediction and identification of aggressive disease. But as you all know, the PRSs so far have not been able to be translated into clinical practice very readily. And I think we still have a little ways to go before we know whether these PRSs will have clinical utility and how.

Clearly, there are people who can be identified as having extremely high polygenic risk based on their PRS, but it's a rare group of individuals, and we don't really know if that level of risk can be intervened on in the same way that we might with, say, a BRCA2 mutation. So I think there's quite a bit of work left to be done in figuring out the translation of this information into either risk assessment and management and monitoring, as well as clinical practice of other kinds.

I think the one thing we have learned from our study and from many others is that the polygenic risk scores don't translate well from population to population. So if a PRS has been generated in Europe, it's not going to predict very well the polygenic risk in Africa, et cetera. So I think it is critical, if we're going to be able to use polygenic risk scores, that they be developed and applied and implemented in diverse populations or in the population relevant to what that clinical translation might be.

Andrea Miyahira: OK. Thank you. And what are the next steps for these studies?

Timothy Rebbeck: So we are now in the process of creating a complete list of genetic variation genomics in a larger sample set of prostate cancer cases. So we're in the process of developing—funded by NIH—germline genetic variation through whole-exome or whole-genome sequencing.

The same in our tumor sample sets. So we'll have a tumor mutational assessment. We're doing expression analyses in our tumors. And we're going to—we have a long-range clinical follow-up among all of our cases. So we have very extensive risk factor pathology, medical records information, treatment information, genomics that have multiple different levels, and outcomes, mortality, and other outcomes.

So we feel like we're going to be able to characterize—like has been done in TCGA or other larger studies in the United States—we're going to be able to characterize the genomics and their impact on both risk and biology of prostate cancer, but also on clinical outcomes.

Andrea Miyahira: OK. Well, thanks so much for sharing this with us today.

Timothy Rebbeck: Thank you. It was great to be here.