Ethnicity Matters in Research: A Case Study in Prostate Cancer

By Angela Liu and Shirley Wu

Nearly one in six men in the United States develops prostate cancer. This sobering statistic has spurred tremendous advances in research over the years, such that prostate cancer is now one of the most treatable and survivable cancers. But there is still much we don’t know about prostate cancer and the factors leading to its development. These factors include non-genetic factors such as diet and lifestyle, but also include genetic factors — variants of DNA that differ between individuals and are associated with susceptibility to the disease.

With new ways of probing the human genome, our knowledge of the genetic factors contributing to risk for prostate cancer is growing. Researchers have identified dozens of genetic associations over the last five years related to prostate cancer. But is the current pace of research enough?

We would say no. Consider this: rates of prostate cancer in African American men are 1.5 to 3 times higher than for men with non-African ancestry, and African Americans have a two-fold higher mortality rate from the disease compared to European Americans. Non-genetic factors surely play a role in these differences, and chances are that genetic factors contribute, too. But nearly all of the genetic research in prostate cancer has been in populations of European descent. Our knowledge of prostate cancer genetics in African ancestry individuals specifically — a group especially affected by the disease — is progressing much too slowly.

The few studies in African Americans have mostly been unable to confirm genetic associations identified in European populations, suggesting that different genetic factors influence risk in these two groups. A large study published recently in PLoS Genetics showed systematically that genetic risk variants identified in Europeans are not representative for African Americans. Looking at more than 10,000 African American men, these researchers found that only half of the European risk variants were associated with prostate cancer in their study group. Even when the same region of the genome was associated with prostate cancer in both populations, the exact location or variant often differed.

As the PLoS study illustrates, findings from one population can’t simply be applied to other populations, emphasizing the need for more research in different populations. A recent study published in Nature Genetics, identifying a new risk variant associated with prostate cancer in African Americans, is an important step in the right direction, but even more studies are needed. The especially high rate of prostate cancer in African American men illustrates how crucial it is to deepen our understanding of prostate cancer specifically in African Americans. And changing research generally to focus on more diverse groups has the potential to improve how we treat and prevent disease in all populations.

At 23andMe, we recognize that some of the biggest impacts in healthcare will come from changing how research is done. We recognize areas where traditional research models are failing and are committed to engaging patients, scientists, and the every day consumer to make a difference. 23andMe’s innovative research platform has already transformed research in Parkinson’s disease, and it can do the same to accelerate research in diverse populations. Keep on the lookout for upcoming projects from 23andMe that focus research where it is needed most!






  • Sarah

    Are you differentiating between “race” and “ethnicity” here? I get the two confused, especially when talking about biology. Sociologists and anthropologists say the biology of race is a “myth,” and that race is a social and historical construct. Do these researchers in this study agree? How can you more clearly present these findings without reifying the layperson notion of race as a biological reality? I see biomedical and genetics research as potentially problematic when it comes to this issue. The scientific community doesn’t seem to speak with a unified voice about these issues, and so it’s nearly impossible for the community to hear a consistent, clear message.
    Thanks for considering ~

  • Joanna Mountain (23andMe)

    @Sarah:
    You raise some very important questions. You may have noticed that this blog post doesn’t include the term race. This absence reflects the lack of agreement about the meaning of the term, and the changing use of the term over time. The term “ethnicity” is generally understood to be applicable to a socially-defined group, and is used in the title. I do not presume to speak for the authors of the original studies, but here are some of my thoughts regarding your questions.

    For any number of reasons, diseases occur at different frequencies in different groups. If that weren’t the case, there would be far less need to label groups within the medical context. One can consider these disease frequency differences to indicate a “biological difference” between the groups, since diseases are biological. However, such differences can exist even when there are no genetic differences between the groups. For instance, Chinese American women and Chinese women living in China experience different rates of breast cancer, probably because of differences in diet and lifestyle. So biological differences don’t necessarily imply genetic differences. That said, because different human populations were separated, somewhat, for thousands of years, there are different frequencies of some genetic variants in different populations. For instance, the variant associated with sickle cell disease is more common in populations with roots in western Africa than elsewhere, and the main variant associated with cystic fibrosis is more common in populations with roots in Europe. To the extent that ethnic groups correspond with geographic origin (social identity often correlates with geography), ethnicity can be correlated with genetics. Hence the continued use of group labels within the medical genetics literature.

    Someday the need for the group labels within the biomedical context may disappear. Once we understand the genetic and non-genetic factors associated with disease more fully, we can move beyond the group labels. Today, however, our understanding is far from complete, and so the labels remain useful in this context.

  • Sarah

    Thanks Joanna – that was very helpful and clearly presented! It’s such a nuanced topic. In the research context, I’m used to seeing “ethnicity” checkboxes as “non-Hispanic” and “Hispanic.” Then “race” is “white,” “black,” etc. Is this because the US census hasn’t gotten up to speed with the genetic perspective on geographic origin and ethnicity? I work in genetics research, and we often have studies where participants have dealt with similar checkboxes. We usually don’t use the self-reported “race” and “ethnicity” in the actual genetic analysis. Rather, we estimate their “genetic ancestry” by comparing their genome to so-called defined groups. Now we have three nebulous divisions: self-reported race, self-reported ethnicity, and (I don’t know what to call it) – genetic ancestry? It seems rather convoluted to me.

    I think it’s critical that the public not interpret these types of studies in a way that entrenches racism. You make such an excellent point about the differences between diseases between “genetic” vs “biological.” Even “genetic” components of diseases (cancer, for example) aren’t always inherited. I just think the media needs to be more careful when writing about these types of findings. Most people aren’t going to go read the Nature article, and it’s so easy for someone skimming the Sunday morning paper to go “disease -> races -> genetic differences -> inherent inferiority -> dismissal of socio-economic causes for health disparity.” (Or something along those lines). I appreciate your blog’s attention to detail, but just wanted to bring this up as a white (black?) elephant in the room.

  • http://www.monasheikh.com Mona Sheikh

    I’m excited about the Roots initiative, and this article further highlights its necessity. Are there any other papers written on disease genotypes that differ between European and non-European populations?

    • http://23andme.com Shwu

      Hi Mona,

      There are at least a few other diseases where the genetic risk factors appear to be slightly – or sometimes very – different between ethnicities. Coronary heart disease is one (see here for a recent study) and a few new studies just came out on asthma (see two published this week in Nature Genetics). In other cases, there are specific variants that have a relatively strong effect on risk in specific populations where that variant is more common, e.g. kidney disease.

      For the most part, we just don’t know enough about disease genetics in non-European populations for most diseases to really say. Hopefully the Roots initiative and similar projects will help change that!

  • ResourceDragon

    In a sense, I see this article as a promo for 23andme’s services. This is because while some people belong to just the one ethnic group, most people are a mixture. The problem then is: which genes have you picked up from which sources and how do they interact? (Then epigenetics and environment get involved, too.) That can only be answered with a personal gene map!

Return to top