A new study finds that precision medicine for oncology — genetic testing to determine the best drug treatment for each cancer patient — is not always so precise when applied to people of non-European descent.

In precision medicine for oncology, scientists identify mutations that transform healthy cells into tumor cells. In an ideal world, they would then be able to compare genetic variations from the tumor to the patient’s normal tissue. However, a normal tissue sample is often not available. So scientists use a stand-in: population databases that filter out benign genetic changes from those that may cause cancer.

Patients whose tumor-cell sequencing cannot be compared to their normal-cell sequencing run the risk of being misdiagnosed, researchers said.

“The field of precision medicine isn’t taking into account population differences. The approaches being used are imprecise when you look at very specific populations,” said John Carpten, PhD, professor and chair of translational genomics at the Keck School of Medicine of USC, director of the Institute of Translational Genomics, and one of the study’s lead authors.

The study, published Oct. 19 in the journal BMC Medical Genomics, found that precision medicine using a tumor-only approach to guide therapeutic intervention is less precise for people whose ancestors are from Latin America, Africa and Asia.

“You might be getting the wrong therapy simply because of our lack of understanding of the genetic architecture of your ancestry,” said David W. Craig, PhD, professor of translational genomics at the Keck School, co-director of the Institute of Translational Genomics and the study’s senior author. “These findings argue that we’re really not doing a very good job of doing precision medicine for many populations.”

A multi-pronged problem

Many hospitals, especially in underdeveloped nations, collect tumor tissue for research purposes without collecting normal tissue for comparison. Without normal cell samples, it is difficult to determine which mutations potentially cause cancer and which are benign variants in the human genome.

“It is very difficult to identify a somatic, or potentially cancer-causing, variant when you don’t have a germline, or normal, sample,” said Rebecca Halperin, PhD, an Arizona-based Translational Genomics Research Institute (TGen) assistant research professor and the study’s other lead author.

Another part of the problem is that most of the tens of thousands of individuals worldwide who have undergone whole-genome sequencing — the spelling out of the nearly 3 billion chemical bases in their DNA — are of European descent. This creates a bias in existing databases, which are used to exclude potentially inaccurate results called false-positive variants.

European ancestry is among the least diverse genetically. This population, particularly Scandinavians, has the fewest genetic variants, the study said.

There is a need to sample more people from more diverse parts of the world, Carpten said.

People whose ancestry can be traced to less developed parts of the world — areas that have experienced the most rapid population increases in recent history — have the most genetic variants. Bangladeshi people, for example, possess one of the world’s most diverse genomes.

The scientific community is beginning to see the shortfalls of precision medicine, said Rick Kittles, PhD, a premier scientist in population genetics and cancer.

“This study goes beyond the barriers to participation and provides insight on the lack of genetic data from diverse populations and its impact on the value and utility of precision medicine,” said Kittles, a founding director of the Division of Health Equities at City of Hope. “There is still much work to be done in order for all communities to benefit from precision medicine.”

Lighting up the problem areas

USC and TGen are trying to move precision medicine forward. To help researchers sort out potentially inaccurate results, USC and TGen researchers created a computational tool called LumosVar: Lumos means light and Var refers to genetic variance. LumosVar “lights up” the genome’s potentially cancer-causing genetic mutations.

“Simply sequencing more individuals from various populations is not enough,” Craig said. “We really need access to the cells that were passed on from previous generations. But when those aren’t available, we need better tools. LumosVar is one such tool.”

Craig and Halperin are the co-creators of LumosVar, an open-source tool to search for potentially cancer-causing mutations. LumosVar is available for download at https://github.com/tgen/LumosVar.

The study was funded by The Ben & Catherine Ivy Foundation and the Multiple Myeloma Research Foundation.