NEW YORK (GenomeWeb) – Researchers at the annual meeting of the American Association for Cancer Research in Washington, DC earlier this month addressed the lack of diversity in many large genomic datasets, presenting findings which showed that this lack could be hindering efforts to lower the rates of cancer-related deaths.
In one presentation, Icahn School of Medicine researcher Eimear Kenny noted that despite studying millions of people at once in massive studies being powered by ever newer, faster, and cheaper sequencing technologies, the representation of non-European, non-East Asian people in those studies has been flat at 4 percent in the past 10 years. Individuals of European descent make up about 76 percent of current genomic databases, while East Asian genomes make up the other 20 percent.
In a plenary session on the opportunities of precision cancer medicine, Chanita Hughes-Halbert, a professor at the Medical University of South Carolina's Hollings Cancer Center, said that the most recent data on cancer death rates by race and ethnicity from the US Centers for Disease Control and Prevention shows that "significant" disparities still remain, even with all the advancement that has been made in early detection and treatment. This illustrates that much more work remains to be done to "promote cancer equity," she added.
Consequences of research disparities
Several researchers provided findings that showed the negative effects of such disparities. At a symposium on minorities in cancer research, Case Western Reserve University researcher Kishore Guda presented research on the mutational landscapes of colon cancers in African Americans, who, Guda noted, exhibit the highest colorectal cancer incidence rates and mortality rates in the US, in both men and women. African Americans are 25 percent more likely than whites to develop CRC, and 50 percent more likely to die from it.
And although one cause of disparity in health outcomes is lack to access to cancer screening, Guda cited one study that showed disparities in mortality rates were still high even for African Americans who had the same screening as whites. He also presented findings from a study of CRC in African Americans and whites with access to military system healthcare. Even in this context, African Americans had worse survival rates than whites. Further, whites within the military system had a lower rate of CRC incidence than whites in the general population, but there was no such decrease in incidence rates for African Americans.
This led Guda to attempt a comprehensive molecular analysis of CRCs in diverse ethnic groups, but he found that African American CRCs were notably underrepresented in public CRC datasets. So he and his team undertook a study of African American CRCs, performing whole-exome sequencing on 29 late-stage tumors to look for somatic mutations.
They found 2,900 protein-altering mutations in 2,100 genes, most of which were present in data from The Cancer Genome Atlas. But when they looked closer at a set of 385 genes that were mutated in at least two tumors from their discovery set, the researchers found mutations in 52 genes that had not been previously reported.
They developed targeted capture array for the 52 genes in question, resequenced these in the original discovery set, and then validated them in 74 other African American CRC tumors. From that analysis, they found that 20 of the 52 genes were significantly mutated. In order to confirm that these were not platform-based artifacts, the team sequenced 129 late-stage CRC tumors from Caucasians, and concluded that 15 of the 20 genes were predominant in the tumors from African Americans.
Further analysis showed that mutations in these 15 genes were associated with poorer overall survival in terms of either metastasis or recurrence, and were even more pronounced in tumors from Stage 3 patients. Of these 15 genes, four were selectively mutated in African American CRC tumors, and were not present in any CRC tumors from white people.
Two of the four mutations were in EPHA6 and FLCN. Mutations in these genes are considered highly deleterious. In fact, EPHA6 is part of pathway that has been identified as significantly mutated in CRC tumors in whites, and FLCN germline mutations can cause Birt-Hogg-Dubé syndrome, a rare disorder that affects the skin and lungs and increases the risk of certain types of tumors such as familial renal and thyroid carcinomas.
Much of this work was published last year in the Proceedings of the National Academy of Sciences, but Guda and his team are now planning several studies to look for these mutations in bigger data sets, and to determine if the mutations are associated with any histology or driver genes, or environmental exposures. But it is clear, he noted, that looking at predominantly Caucasian CRC tumor data such as is present in most major databases has caused a lack of understanding of the genomic drivers of CRC in African Americans, which could have contributed to the disparity in outcomes seen today.
In a separate symposium on international cancer genomics studies, University of Southern California Keck School of Medicine researcher John Carpten demonstrated that not only can diversity in genomic studies highlight genetic differences in how people of different races respond to cancer, but also that the lack of it can obscure other causes of health outcome disparities.
Carpten and his team conducted a study of multiple myeloma in 45 African Americans and 196 Caucasians. Early observations from study identified a gain of chromosome 1q in most of the patients. While this is a known marker for outcome, Carpten found that it appeared more frequently in the Caucasian tumors than in the African American tumors.
The team then enrolled 1,000 untreated, newly diagnosed patients for a new study, and split them into three different treatment groups. The researchers plan to follow the patients for three to four years to see whether there's a difference in how Caucasian and African American patients react to the various treatment regimens.
The team is also conducting mutation analyses, and although the study is still ongoing, it has identified key genetic mutations involved in the disease that seem to appear in whites more than in African Americans. For example, TP53 mutations seem to be associated with European ancestry in the study. Further, a copy number analysis of TP53 found that biallelic alterations are rare in African Americans, but common in whites.
While the data is not yet complete, it seems to indicate that certain mutations that confer a worse outcome appear more often in whites. This indicates, Carpten said, that worse outcomes for African Americans may indicate a lack of access to healthcare, rather than biological causes. In the case of multiple myeloma, it's possible that equal treatment could improve outcomes in African American patients, though the team needs to finish its study before making conclusions.
A second speaker at the same symposium, Duke-NUS Medical School's Bin Tean Teh, demonstrated the granularity of diversity that is sometimes needed to understand the cause of a particular kind of cancer.
Cholangiocarcinoma (CCA) accounts for 5 percent to 10 percent of cases of primary liver cancer in most of the world, he said. In some parts of Southeast Asia, a major risk for CCA is infestation by a common parasite called the liver fluke. The parasite is relatively rare in the West, but common in the northwestern areas of Thailand. Farmers there tend to eat uncooked fish and get infected with the fluke. Though the parasite is treatable, continuous exposure can lead to chronic inflammation, which is a risk factor for CCA.
But CCA is a heterogeneous cancer. Molecular characterization of CCA shows recurrent frequent mutations of TP53 and KRAS, as well as SMAD4, ROBO2, and various other genes. CCA cases from northwest Thailand specifically show higher rates of TP53 mutations. Focusing on these cases, Teh said, could cause researchers to miss other possible gene mutations involved in other subtypes of the disease.
In fact, integrated analyses conducted by Teh and his group have shown that there are four distinct subtypes of CCA, each with different therapeutic implications. If a researcher were to use samples from the northwest of Thailand exclusively to represent Asian CCA incidences, he or she could miss the mutations involved in the other subtypes that appear in Asia, or even in different areas of the same country.
What causes disparities?
Health disparities and a lack of diversity in research are not new problems. For years, scientists have been discussing ways to include a larger and more diverse swath of individuals in genomic sequencing and clinical studies.
In fact, the NIH has said that it wants to make sure that 75 percent of the cohort for the Precision Medicine Initiative represents groups historically underrepresented in biomedical research.
But it's not just as easy as going out into different neighborhoods and asking people of different races if they want to join in. Determinants of health disparity include biological, social, and psychological factors working independently and interactively. And although many are hoping the PMI can indeed begin to address the problem of diversity, some doubt the agency's ability to accomplish its goal in the context of the current political environment.
In fact, South Carolina's Hughes-Halbert said during her talk that some believe the program actually has the potential to exacerbate disparities — as treatments become more advanced but also more expensive, it will raise challenges for providing up-to-date care to populations that have trouble getting even basic cancer care.
In a forum on moving toward precision medicine in the context of health disparities, University of California, San Francisco researcher Laura Fejerman noted that many different factors drive health disparities besides inclusion in genomic databases, such as public policy, sociocultural and environment factors (access to screening, access to healthy food, lifestyle, pollution, stress, education, reproductive behavior), and biological factors.
For example, white women are typically diagnosed with breast cancer at Stage 1, but this is not the case for African Americans, Hispanics, and Southeast Asians. Although the screening for breast tumors is basic and works relatively well, there is an issue with access. In most cases, people don't get screened because they can't afford it, or they don't know where to get it for free. "No one is making it easy for them," Fejerman said, adding that the biggest disparity now exists between people who have insurance and those who don't. The uninsured, she noted, are disproportionately minorities.
Researchers need to start building adequate datasets, Fejerman said, to recognize who's not getting what they need and to give it to them. That includes access to care, to healthy food and exercise, and to education.
Mount Sinai School of Medicine population health scientist Carol Horowitz agreed. Though not specifically a cancer researcher, Horowitz said she faces many of the same obstacles in her research as the ones Fejerman noted.
"We say that there aren't a lot of black people in these datasets. Why? Because no one's trying to put black people in these datasets," she said.
One way to ameliorate the problem is for researchers to come out of their silos, according to Horowitz. Researchers must recognize the problems faced by their target patient populations, and must work with as many stakeholders as possible to not only recruit a diverse set of people into studies, but to also clearly communicate how these studies could possibly help the participants themselves, or help their communities.
"Participants may be skeptical. So we need to build a system where people are more involved. Ask the community what they want and use that to design research," Horowitz said. "The community may ask questions that researchers or clinicians haven't thought of because they live with certain circumstances that researchers and clinicians don't experience."