Study sounds alarm on absolute representation of minorities in genome databases

("diversity of london" (CC BY 2.0) by jonrawlinson)

The racial distribution of research subjects is a long-standing, well-discussed problem, notably in the clinical trial sector. But the rise of genome sequencing has opened a new frontier for the issue that could drive divergence of the cancer care received by different races.

This worry is underpinned by emerging understanding of how the molecular biology of cancer varies from race to race, and ethnicity to ethnicity. Faced with such heterogeneity, it is important for researchers and physicians to have enough data on each race and ethnicity to identify moderately common genomic alterations. A study published in JAMA Oncology suggests The Cancer Genome Atlas (TCGA) falls short in this regard.

The JAMA study looked at approximately half of the 11,000 samples in TCGA. In this cohort, 77% of the samples came from white subjects. A further 12% of the samples were from African-Americans, while Asian and Hispanic participants each accounted for 3% of the dataset. While this isn’t a perfect reflection of the U.S. racial and ethnic composition--Hispanics are particularly underrepresented--it is a reasonable approximation and certainly closer to reality than many clinical trial populations.

Yet, while that shows TCGA has done a decent job of enrolling racial minorities, the JAMA authors report it still results in a database that contains too little information on African-Americans, Asians and Hispanics to be useful.

“Even when studies have a reasonable relative representation of racial and ethnic minorities, the overall absolute number of minorities examined may not be enough to detect small differences in the cancer’s genome,” Dr. Daniel Spratt, an assistant professor at the University of Michigan Medical School and co-author of the study, said in a statement.

Spratt and his collaborators quantified this by assessing whether it is possible to detect 10% and 5% mutational frequencies in the data specific to each race and ethnicity.

“All tumor types from white patients contained enough samples to detect a 10% mutational frequency,” the authors wrote. “This is in contrast to all other racial ethnicities, for which group-specific mutations with 10% frequency would be detectable only for black patients with breast cancer.”

- read the release
- and the abstract