Precision medicine research must lose its Eurocentricity

Precision medicine research relies on genetic testing at scale, but how do we reach more diverse patient populations and collect nationally representative samples? Patrick Short, CEO and co-founder of Sano Genetics, and Natasha Ratcliffe, director of community engagement and partnerships at Couch Health, explain how Eurocentricity is holding back genetics, and what can be done.

Humans across all nationalities, ethnicities and ancestries may share 99.9 percent of the same DNA, but it’s that 0.1 percent that holds the key to many secrets – such as why some people are more prone to certain diseases than others, or why some respond better or worse to particular treatments.

Yet it’s long been a major problem when it comes to genetic studies that the majority of data sets come from people of European descent, who make up a mere 16 percent of the global population. In 2009, a study found this was as many as 96 percent of participants. A decade later, this figure was 78 percent, which – while a step in the right direction – remains woefully inadequate. And in 2015, the U.S. Food and Drug Administration (FDA) found that fewer than 20 percent of clinical trials reported treatment benefits or side effects from Black patients.

This has real-life implications in the here and now. Polygenic risk scores (PRS), which give an estimate of an individual’s liability to common and complex traits or disease like coronary artery disease, Alzheimer’s disease, or breast cancer, are “several times more accurate in individuals of European ancestry” than in others, researchers have found.

In other words, health advice based on an individual’s PRS has the potential to be wildly incorrect, or even downright dangerous, for certain groups of people. The “Eurocentricity” of genetics studies, unless addressed, will mean that new advances in personalized medicine will disproportionately benefit people of European ancestry and exacerbate existing health disparities.

For rare disease studies too, diagnosing patients relies heavily on open databases like ClinVar, but the biased data held in these means lower rates of diagnosis for populations that are less well represented.

Why is genetics research so Eurocentric?

According to one analysis of genome-wide association studies, almost three quarters (72 percent) of participants came from three countries alone: the U.S., the U.K. and Iceland. This is partly historical, as the first major genome projects were conducted in these countries. But, even as genomics has grown, disproportionate funding has continued to flow to the U.S. and Europe  (though the African Society of Genomics has been crucial for driving forward increased genetics and genomics research in African countries).

Whilst populations in the U.S. and Europe are certainly not representative of the world, diversity does exist amongst country populations, which provides an opportunity to collect data from people from non-European ancestry. 

Some large-scale genetics studies like All of Us have made efforts to be reflective of society, but most countries have tended to neglect their often diverse populations. Other studies like East London Genes and Health have focused on underrepresented and underserved communities, and made significant contributions to genetics and community-based research practices. But, unfortunately, the exceptions continue to prove the rule, with most large-scale genomics studies not putting enough emphasis on representativeness and inclusive design.

Researchers have been aware of this bias for years. So why, despite incremental improvements, does the problem of Eurocentricity in clinical trials persist?

A key reason is that, despite recognition that genomics data sets should be more representative, how we’ve historically recruited for studies simply hasn’t changed very much. Most research studies are still largely dependent on healthcare professionals who, aside from being time-poor and overburdened, can hesitate to put patients forward because they feel ill-equipped to help them understand genetic test results and their implications. 

This approach also takes for granted a level of trust between a patient and their doctor which doesn’t always exist, particularly for patients from underrepresented communities who are more likely to experience negative healthcare experiences due to systemic biases and lack of cultural competency.

For example, research in 2020 found that 64 percent of Black Britons thought the U.K.’s National Health Service (NHS) did less to protect their health than that of their white counterparts. Meanwhile, in a 2021 Deloitte focus group made up of U.S. citizens who identified as Black, Hispanic, Asian or Native American, 55 percent said a negative experience had made them lose trust in a healthcare provider. 

The answer is not to bypass the healthcare system; healthcare providers play a critical role in research. Instead, researchers should embrace additional strategies that are more community- or patient organization-led. Co-developing the research and outreach approach with patients and community members leads to much better outcomes and avoids costly mistakes down the line. 

Furthermore, taking part in research needs to be easier. Embracing digital tools and non-invasive testing will make it easier for people to take part. Of course, not all participants use the internet, so an omni-channel approach that combines healthcare sites, digital, at-home and pop-up community sites will be needed here as well.

Similarly, we must also remember language is a barrier. Making sure participation is possible not just in English but other languages as well, removes another hurdle to informed consent and taking part.

Companies, funders and regulators need to reflect this priority

Perhaps most importantly of all, funders in the private and public sector need to prioritize solving this problem. Our experience is that most are reluctant to fund deep community engagement, as they don’t always see it as critical to science and R&D. But, because most studies are conducted in European ancestry populations, and trust with healthcare providers is higher, recruiting people in this demographic tends to be easier and cost less, while recruiting a more representative population costs more due to the necessary investment of time that is needed to build trust with communities who face disproportionate barriers to engaging in research. This leads to a negative feedback loop that perpetuates the problem and makes the science less impactful.

Funders should make explicit commitments to fund recruitment and community engagement costs, translation and decentralizing participation to include not just healthcare sites, but online, at-home and pop-up community clinics.

Regulators also have an important role to play. For novel diagnostics and precision medicines, we predict that regulators will start to take a firmer stance requiring representative populations in trials and supporting data. In April 2022, the U.S. FDA released guidance for industry on improving diversity in trials, and this is likely just the beginning. Dana Goldman, Edith Perez and Carlos Del Rio make a suggestion in a 2022 STAT News article to take lessons from the Orphan Drug Act of 1983 which dramatically increased incentives for developing drugs to treat rare diseases, and apply a similar incentive structure to clinical trials that are representative.

But as an industry, we can’t wait. Companies like BenevolentAI, Genomics England and VariantBio are leading new initiatives to generate more diverse data sets, and large pharma companies like AstraZeneca are taking representativeness seriously. Funders and regulators will help, but change needs to be accelerated by companies large and small taking the initiative to set clear objectives to make studies representative, and give their teams the resources and tools to make it happen.

Imbalances in research will always exist to some extent, but by improving our research practices and setting clear priorities, we can take huge strides forward.