Last month, researchers at King’s College London and the University of North Carolina at Chapel Hill published a paper demonstrating that anorexia is likely not only a psychiatric condition. The results are groundbreaking, and, as the researchers noted, they go some way to explain why treatment outcomes for anorexia are so poor: anorexia has one of the highest mortality rates of any mental illness.
The researchers analyzed the DNA of nearly seventeen thousand people with anorexia as well as more than fifty-five thousand control subjects and discovered eight “risk loci” that were more common among anorexia sufferers than the control population. A locus is a position on a chromosome which correlates to one or several genes, and hence to a genetic attribute. In other words, the researchers effectively found a set of genetic attributes that anorexia sufferers were more likely to have. Depression, anxiety, and obsessive compulsive disorder were among them, which was expected. But so was the ability to burn fat more quickly, the natural propensity to do more physical activity, and increased resistance to type 2 diabetes. These metabolic traits are all healthy attributes on their own, but combined with genes that are linked to other mental illnesses, they increase the risk of anorexia, suggesting the disease has a metabolic component too.
This study is just one of many recent developments in a very promising method of biomedical research known as the genome-wide association study (GWAS), in which scientists take a large sample of genetic data provided by healthy volunteers and people with a particular disease and analyze it to find genetic variations that are common in those with the disease. The aim is to develop a profile of the genetic traits that make an individual more likely to develop certain illnesses, which is called a polygenic risk score (PRS). In the future, many researchers hope it will be possible to screen people for hereditary illnesses early on and treat them preemptively using PRS.
In addition to anorexia, similar progress has been made for conditions like schizophrenia, heart disease, type 2 diabetes, skin cancer, breast cancer, prostate cancer, depression, and obesity, among others. PRS research has led to the prescription of statins, or cholesterol-lowering drugs, for patients at greater risk of heart disease before they begin to show symptoms. And recently, genome-sequencing was used to screen 350 babies in the intensive care unit of a Cambridge Hospital for genetic defects: a quarter of them were found to have underlying conditions which could then be treated, and overall, the care plans were adjusted for 75 percent of the babies screened. The eventual routine implementation of PRS research into clinical practice will likely improve diagnosis times and change treatment plans, as these examples demonstrate.
But there is a catch. This groundbreaking research is relevant mostly to people of European descent. In short: white people. While every GWAS uses a vast amount of genetic data, it is often of European origin only. And while it is well known that findings from the genetic data of one ethnic group cannot be easily generalized to another population, the practice persists. This is acknowledged as a flaw by researchers in almost every study I consulted, and yet none had taken steps to mitigate it. Additionally, the disparity has hardly been reported on.
In the second paragraph of the July anorexia study, the researchers state plainly that only genetic data of European origin was used—but not one of the mainstream news outlets celebrating this groundbreaking scientific advance included any mention of this. It wasn’t in the Guardian, which ran a selection of letters and news pieces on the study, or in the Wall Street Journal. Neither CNN nor even New Scientist mentioned it. Both the BBC and Sky News cited that the study had used data from participants across seventeen countries but failed to mention that all participants were of European descent, even though these two pieces of information are found in the same sentence of the paper. Only Scientific American specified that “all of [the subjects] had European ancestry.”
A few outlets reporting on the study nonetheless found space to run hand-wringing comment pieces discussing the nuances of what can be considered, philosophically, to constitute a mental illness. But surely the fact that this research, which often receives public (read: tax payer) funding, is of limited utility to people who aren’t white is more pressing.
These omissions are sadly understandable. It is, unfortunately, normal to report on a scientific paper without having read it in its entirety. It is normal for journalists to be asked to find a topline in a complex piece of information that is not in their field. And in a fundamentally racist society, it is normal not to question the use of Europeans as a baseline.
One paper on polygenic risk scores, published in March by researchers at Massachusetts General Hospital’s Analytic and Translational Genetics Unit, traces the origins of the racial bias inherent to this field of research so far. It concludes, in plain terms, that “clinical uses of PRS today would systematically afford greater improvement [in healthcare] for European-descent populations.” The researchers describe this issue as the “major ethical and scientific challenge surrounding implementation of PRS,” although it is worth noting that this has not prevented them from running their own studies.
The prediction accuracy of PRS is far lower, the paper explains, for non-Europeans: 1.6 fold lower for those of Hispanic/Latino and South Asian descent, 2.0 fold lower for those of East Asian descent, and worst for those of African descent, at 4.5 fold lower. The root of the problem is a historic tendency to over-recruit white participants for genetic research studies: nearly 80 percent of participants in GWAS groups are of European descent, even though they make up only about 16 percent of the global population. Interestingly, the researchers note, even when preexisting datasets include a significant proportion of ethnic minorities (the UK Biobank is one example), researchers still tend to use exclusively European data because studying an ethnically homogenous set simplifies statistical analysis.
It is somehow very hard to imagine a group of elite scientists making the decision to exclude all genetic data of European origin from a large medical research project.
This is not a new problem. The Massachusetts paper details how “all studies to date” have observed that PRS are much more accurate for Europeans than non-Europeans, an “entirely predictable” consequence of using majority European data. The researchers call for the prioritization of greater diversity in PRS research, but the disparity has been understood for at least a decade, and no meaningful change has yet been made. Indeed, the proportion of non-European individuals used in these studies has stagnated or declined since late 2014, suggesting, as the researchers dryly put it, “the absence of a trajectory to correct this imbalance.”
I came across the research highlighting the racial inequalities in this bold new area of biomedical research while reading up on PRS, and I can’t understand why I had to do so much digging around to find it. This paper was published in Nature Genetics, a well-known and well-respected journal, and it is both comprehensive and easy to read; there are helpful diagrams throughout, and all technical points are clearly explained, none of which is a given in even high-quality research papers. It is hard to argue this study is not groundbreaking, since it makes some fairly devastating claims about health inequalities. It is, in other words, great fodder for reporters, yet I did not find it reported anywhere other than in specialist publications.
One interesting point about all of this is that the disparity in PRS research could have easily gone the other way. Genetic data of African origin is much more diverse than European data—it is, as the Massachusetts researchers put it, “disproportionately informative for genetic analysis.” Researchers opting to use only one ethnic group for the sake of statistical simplicity might have done so with African data instead, and yet it is somehow very hard to imagine a group of elite scientists making the decision to exclude all genetic data of European origin from a large medical research project.
There are some tangible plans to try to address this fundamental inequality in genetic research. The UK’s Department of Health & Social Care recently announced that it will recruit five million healthy participants for a new genetic research undertaking called the Accelerating Detection of Disease (ADD) challenge. After scrapping initial plans to include only those who could pay for the privilege, it has committed to prioritize enrolling under-represented groups. However, there has been no explanation of how this recruitment will work, and the timeline for the project’s completion is five years from now—a significant time in research terms, when many GWAS are already up and running. The same press release claims that children and adults who are likely to have a rare genetic disorder, or some forms of cancer, will receive genome sequencing as part of their care from this year onwards. It did not specify whether the efficacy of this care would be dependent on their ethnicity.
A few months ago, a tweet by presidential hopeful Andrew Yang went viral: “My father has a PhD in Physics. I believe in science.” This is clearly an annoying and embarrassing thing to say, especially from the mouth of a techno-deterministic faux-populist like Yang, and he was roundly mocked. But the sentiment was also liked by around eleven thousand people. It perfectly encapsulates the way a certain type of person thinks about scientific research: as factual, objective, “correct.” And, despite its reliance on funding from the government as well as large corporations and philanthropies, this research is often portrayed as apolitical.
Why do we do experiments in a lab? I still remember this question, posed during a particularly awkward lecture in a final-year philosophy class I took while studying for my degree in physics. The format of the class was flexible and open, exactly the sort of environment that overachieving science types tend to find alarming. We all sat in a circle and were told to speak whenever we wanted, which in this case meant we were silent. The professor kept trying, though, and eventually the question, “Why does nobody do an experiment standing in the middle of the road with traffic coming?” elicited the suggestion, “A car would clearly just drive through it.” Finally, this marked the start of a stunted hourlong discussion about the nature of experiments and scientific research.
We do experiments in a lab because creating a sterile environment is the easiest way to investigate the behavior of one component in a system, to test how altering a single variable affects all of the others. An experiment conducted in lab conditions can be repeated by another group of people, who will hopefully find similar results. And it is more cost effective to create an artificial environment over which you have complete control than to try and account for other issues on the fly (see: conducting your experiment far away from rogue cars, rather than trying to prevent them from driving on a particular stretch of road).
The downside to lab science is that when you start making decisions about the conditions in which you should study something, you have already put your own imprint on the results—something that is widely understood in social sciences like anthropology. The relationship between results from research conducted in a lab and the real world is always heavily caveated. In that philosophy of science seminar room, it was clear that not one of us had given much thought to the decisions that had gone into constructing lab environments, despite spending most of our time in them. Manchester, where I studied, is one of the more ethnically diverse cities in the United Kingdom, but when I think of that uncomfortable lecture, the people sitting in that circle—every face was white.