Imagine taking a genetic test that could tell you your personal risk of developing complications and dying from a particular disease, such as cancer, heart attack or even COVID. A version of such a test exists—albeit an imperfect one.

Genome-wide association studies (GWAS) are becoming an increasingly common avenue to assess COVID risk. The approach holds potential for fighting the disease by identifying the locations, or loci, on the human genome that put an individual at higher or lower risk for severe disease. Scientists hope it may eventually unlock the door for new forms of treatment.

“Whole-genome sequencing allows you to check every single base pair in the genome,” says Athanasios Kousathanas, principal genomics data scientist at the London-based company Genomics England. “And this allows you to find the particular genes that might be involved with higher precision.”

Some experts warn, however, that GWAS alone are insufficient for accurately assessing COVID risk. They say that genomic analysis may be difficult to disentangle from social risk factors and could leave health systems open to discrimination.

Manuel Ferreira, a researcher at genetics company Regeneron, is part of a team using GWAS to hunt loci related to COVID risk by sifting through thousands of genomes from four aggregated databases. In their most recent study, published in March in Nature Genetics, Ferreira and his co-authors crunched the numbers and found that individuals with a rare variant of the ACE2 gene seemed to be at nearly 40 percent lower risk than the general population of developing severe COVID. It is “what we call a ‘strong effect,’” Ferreira says.

The ACE2 gene encodes a specialized ACE2 protein located on a cell’s surface. Normally the protein helps to regulate things such as blood pressure and inflammation by allowing specific protein fragments into or out of the cell. But it also gives SARS-CoV-2, the virus that causes COVID, a cellular entry point for infection. When the virus comes into contact with the ACE2 protein, it latches on with its exterior spike protein like a burr snagged on a sock. From there, the virus enters its target cell.

But Ferreira found that people who carry a specific variant of the ACE2 gene have about 39 percent fewer receptors for the protein studding their cellular surfaces. The researchers hypothesize that, as a result, fewer SARS-CoV-2 viruses are able to gain entry into these individuals’ bodies, significantly reducing their risk of severe COVID. “In a sense, it’s not totally surprising, since we know that the virus requires [these] receptors to get into the cell,” Ferreira says.

Kenneth Baillie, a research clinician at the University of Edinburgh, recently collaborated with Genomics England’s Kousathanas on a study that identified 16 new loci linked to severe COVID risk. Some, Baillie believes, are potential targets for new drug therapies. “I’m sure there are more that are targets for therapy that we haven’t understood the biology [of] well enough yet,” he says.

But other researchers caution that when it comes to predicting severe COVID, it is nearly impossible to disentangle genetic risks from social risk factors such as access to health care and working conditions, even using genome-wide analysis.

Elsie Taveras is a pediatrician at Massachusetts General Hospital. But when the pandemic struck, she—like many others in her field—was pulled onto the intensive care unit floor to help treat the influx of patients. Right away, she noticed a pattern among those with severe COVID: Most were people of color from low-income communities. Many did not speak English.

“I never would have thought that the most important thing I can bring to a care team wasn’t so much my medical expertise,” Taveras says. “It was being able to be there because I could help that team with my Spanish language.”

Between navigating language barriers and limited financial resources, many of Taveras’s patients avoided seeking treatment until their illness had worsened. Others lived in multigenerational homes or worked frontline jobs in which isolation was all but impossible. These social pressures put them at higher risk of severe COVID—not because of genetics but simply because of circumstance.

Geneticists do their best to account for such disparities in their analyses. “Epidemiologically, the way you can better understand the extent to which genetics [versus social risk factors] is driving the severity” of disease, Taveras says, is to “adjust for some of those variables.” By comparing individuals of similar ancestry, socioeconomic status, gender or medical history, scientists can establish a baseline for a patient’s odds of developing severe COVID. But even with these controls in place, “it’s imperfect,” Taveras says.

An earlier genetic analysis, for example, linked high COVID risk to having type A blood and low risk with blood type O. But subsequent research found the association between type O and COVID risk to be negligible, while the connection to type A blood was nonexistent.

Ferreira’s research drew from a database containing hundreds of thousands of genomes. These data gave the researchers a clear picture of the subjects’ ancestry and medical records but next to no context for their income level, housing situation or primary language.

Ferreira and his colleagues found that individuals with European ancestry had about one-in-200 odds of carrying the COVID-risk-reducing ACE2 variant. In people with African ancestry, the odds were around one in 100, whereas people of South Asian descent had about a one-in-25 chance (although this last sample was very small, and the result was not statistically significant). But even these estimates can be fraught.

“We have this long and complicated history about biological race as a contested category,” says Azita Chellappoo, a philosopher of medicine at the Open University, based in England. “It’s kind of not surprising that that’s been something which geneticists have taken up in the context of COVID-19,” she says, even though ancestral categories often paint incomplete pictures of the diversity within a population. For example, Ferreira’s study looked at the genomes of nearly 45,000 people with European ancestry but only about 2,500 people of African descent and 760 of South Asian descent.

Furthermore, Chellappoo argues, focusing on individual loci misses the way that genes interact with their environment and one another in context. “My genes don’t do anything by themselves,” she says.

But other researchers still see enormous value in searching for specific COVID-related loci. “We sort of kick the tires on the analysis,” says Edinburgh’s Baillie, “and we just keep getting the same results. So we’re very confident that these [effects] are real.”

GWAS have also been used to pinpoint loci associated with loss of taste and smell in COVID patients, as well as markers associated with developing pneumonia after COVID infection. Future GWAS investigations may shed light on the mysteries of the lingering symptoms collectively known as long COVID.

Ultimately Chellappoo, Baillie and others agree that genomic analysis holds potential for developing the next generation of COVID treatments. Ferreira’s research into the ACE2 protein, for instance, could yield a new avenue for preventing SARS-CoV-2 infection: blocking the receptors rather than attacking the virus itself. Current ACE2-blocking drugs, which are commonly prescribed for blood pressure control, have so far been ineffective against COVID. But Ferreira believes that a blocker specifically developed with COVID in mind could be more viable. “Our genetics suggest that blocking [ACE2] would be useful,” Ferreira says. And with vaccines, antiviral drugs and monoclonal antibodies still in short supply globally, new therapies are desperately needed.

When it comes to assessing severe COVID risk, the key is balancing internal and external factors. “For sure there’s value in understanding the genetic contribution,” Taveras says, so long as we keep in mind “that there’s also a relative contribution to the severity of illness from these social risk factors that we can’t measure as precisely as a genetic mutation.”