THOSE OF US who have felt a racist鈥檚 stare know that it scalds our skin and triggers emotions ranging from anger to shame. Last year, that anger boiled over in race riots that rocked the US and Britain. Of course, those riots sprang from a complex web of causes, but it鈥檚 also true that opponents on each side had one simple thing in common: the colour of their skins.
It might seem a long way from the streets of Cincinnati or Bradford to the hospitals and research institutes where scientists seek to explain the diversity of human biology. But well-meaning doctors say they, too, need to pay attention to race. They argue that it鈥檚 a useful indicator of people鈥檚 predisposition to disease or response to drugs. Meanwhile, other scientists think that this is nonsense. Geneticists know that the racial differences most of us perceive are little more than skin-deep. They say that medical research and treatment based on visible racial characteristics is a waste of time at best and downright dangerous at worst.
Even so, the latest genetic data has added a further twist to the argument. It turns out that you can group people into genetically similar clusters after all鈥攊t鈥檚 just that they don鈥檛 correspond to conventional racial groups. But they may provide doctors with better categories for research and treatment until the time when diagnosis and therapy can be tailored to an individual patient鈥檚 genes.
Advertisement
For most people, geneticists鈥� talk about the non-existence of races runs up against the evidence of our own eyes. The people you meet on the street are easy to classify as being of African, South Asian, European, or some other origin. How do we explain these apparently clear differences between people of different 鈥� well, races? Don鈥檛 external features such as skin colour, hair type, eye shape and body stature separate us?
They do, but only on the surface. Many of these dramatic differences are adaptations to different climates and don鈥檛 imply any deeper genetic differences, says Luigi Luca Cavalli-Sforza, a population geneticist at Stanford University in California. Black people are black because of melanin, the pigment that protects their skin from the damage caused by strong sunlight. White people are white because they lack melanin, since the pigment would prevent their skin from making enough vitamin D under the weak Sun of high latitudes. The tall, thin stature that helps the Masai in eastern Africa stay cool would be a disadvantage in the Arctic where the Inuit evolved with stout torsos and short limbs.
But while the Masai and the Inuit represent two extremes of external appearance, in fact there are no abrupt boundaries between human populations. Features that appear unambiguous when you see an individual on the street look a lot less clear-cut when you view a wider spread of humanity. 鈥淚f you walk from Senegal to Japan and you are asked to mark a line where the African characteristics stop and the Oriental characteristics begin, that鈥檚 a very hard task and one that genetics cannot help solve,鈥� says Guido Barbujani, a geneticist at the University of Ferrara in Italy.
The notion of genetically meaningful races began to crumble in 1972, when Richard Lewontin, a geneticist at Harvard University, analysed variations in blood proteins taken from populations around the world. His conclusions came as a shock: humans from different 鈥渞aces鈥� are not as genetically different as their appearance would suggest. He found that nearly 85 per cent of humanity鈥檚 genetic diversity occurs among individuals within a single population, such as the Swedes. Another 8 per cent occurs between populations of the same race鈥擲wedes and Spaniards, for example. Only 7 per cent was accounted for by consistent differences between races. In other words, two individuals are different because they are individuals, not because they belong to different races.
At the time, some researchers doubted Lewontin鈥檚 findings, pointing out that variation in proteins did not accurately reflect variation in DNA. But nearly 25 years later, Barbujani and his colleagues surveyed DNA sequence diversity directly, and their results鈥攑ublished in 1997鈥攚ere nearly identical to Lewontin鈥檚. Other labs have also replicated these results.
Lewontin and Barbujani鈥檚 research provides a way of measuring our diversity that does not depend on our external appearance. And it shows that as a species we are unusually homogenous. In biology, a 鈥渞ace鈥濃�攐r subspecies鈥攊s defined as a population that is geographically isolated and genetically distinct from others of the same species. To determine if a species has races, geneticists first quantify the genetic diversity of the species on a scale of 0 to 1. A 0 means that individuals from different populations are no more different than individuals from the same population, while a 1 means that each population is made up of genetically identical individuals, and all the genetic differences exist between populations. For races to exist, a species should have a genetic diversity number of at least 0.25 to 0.3. Mammals such as coyotes, elephants, gazelles and grey wolves have genetic diversity numbers that range from 0.3 to 0.8, and geneticists recognise subspecies within them. But for humans, the number is 0.15. Compared with other animals, we don鈥檛 meet the threshold.
Moreover, the genetic make-up of human 鈥渞aces鈥� overlaps so broadly that you can鈥檛 accurately predict someone鈥檚 race by their genes. Barbujani and his colleagues analysed a set of 21 DNA sequences from 1330 individuals from 32 populations worldwide. Using the most sophisticated statistical software available, they asked a computer to assign each individual to his or her continent of origin. The results, to be published in the journal Genome Research, show that the computer couldn鈥檛 do it. With a single DNA sequence, the computer got it wrong 80 per cent of the time. As more sequences were added, the computer got better, but even at its best the computer still failed 30 per cent of the time.
So the races we think we see have little relevance to biology. But is there a better way to get at humanity鈥檚 underlying genetic variations? David Goldstein of University College London thinks so. 鈥淭here is quite a simple alternative to racial labels in representing the genetic structure,鈥� he says.
Goldstein and his colleagues analysed DNA samples from people in eight populations from Asia, Africa and Europe, and used statistics to sort the individuals into genetically similar groups. They found that the people divided into four clusters, broadly corresponding to four geographical areas: Western Eurasia, sub-Saharan Africa, China and New Guinea. But the clusters did not follow established racial lines. For instance, 62 per cent of Ethiopians were assigned to the cluster containing most Norwegians, Ashkenazi Jews and Armenians, and 21 per cent of Afro-Caribbean individuals were grouped alongside West Eurasians.
It鈥檚 tempting to regard these clusters as the 鈥渞eal鈥� races of humanity, but things may not be so simple. Barbujani and his colleagues performed a similar analysis using two entirely different sets of genetic markers and samples from over 30 populations鈥攁 much larger group than Goldstein鈥檚. To their surprise, they found that the two sets of markers yielded two different clustering patterns, both different from Goldstein鈥檚. One set of markers broke people down into one largely Eurasian group, plus two other groups in which individuals came from all over the world. The other set led to four groups: one made up of African and Oceanic people, one containing Asians and native Americans, and two other groups that were mainly Eurasian. The differences in the two groupings are so large that Barbujani concludes there is no obvious way of classifying humans into a few, well-defined groups.
Goldstein, however, still believes there is an underlying set of clusters that represent the inherent genetic structure of humanity, and that finding it is mainly a matter of studying the right genetic markers. 鈥淲e are all using too small a number of markers,鈥� he says. 鈥淢y guess is that when we use a large enough set of markers and an exhaustive enough set of individuals, the results will stabilise. In fact, I鈥檓 quite sure that they will.鈥�
Whatever the outcome of this dispute, one thing is clear: conventional notions of race are鈥攐r should be鈥攄ead. Even where doctors find race useful in predicting the risk of certain diseases, they would do better to abandon it in favour of the more general notion of ancestry. Ashkenazi Jews, for instance, are prone to a rare genetic mutation that causes breast cancer, a mutation not shared by other white people. A focus on race can also blind us to real groupings that cut across conventional racial lines. Sickle-cell anaemia, for example鈥攐ften regarded as an African disease鈥攐ccurs in people from all over the world, including those from India and the Mediterranean, because a single copy of the sickle-cell gene protects individuals against malaria, which is common in all those regions.
And when it comes to the more complex diseases such as diabetes, hypertension and coronary heart disease, the situation gets even murkier. Take hypertension, for instance. African-Americans suffer higher rates of high blood pressure than white Americans. But are the underlying causes genetic, environmental, socio-economic or a complex mixture of all three?
No one knows, says Michael Stein of the Vanderbilt University School of Medicine in Nashville, Tennessee. Race, when used as a proxy for our genetic make-up, is a crude classification that glosses over the real causes of complex diseases such as hypertension, he says. For instance, stress鈥攎ental, physical, or financial鈥攃an cause hypertension, and black people in the US are more likely to suffer socio-economic stress than white people. 鈥淏y focusing on race we鈥檝e really been very simplistic in our approach to understanding hypertension,鈥� Stein says. 鈥淲e鈥檝e done ourselves a disservice.鈥�
Yet medical studies continue to regard race as an important predictor of disease risk or response to drugs. Last year, for example, Derek Exner of the University of Calgary in Canada and his colleagues reported that a drug called enalapril lowered blood pressure more effectively in white patients than in black patients. The researchers suggest that the findings may have a genetic basis, because other studies have shown that black and white people differ in their ability to metabolise drugs such as enalapril. In the light of what geneticists are saying about race, many have criticised such studies. 鈥淎ll race-based biological research should be phased out, because racial biology is an oxymoron,鈥� says Robert Schwartz, a former chief of the department of haematology and oncology at the New England Medical Center in Boston. Nevertheless, some physicians are already claiming that enalapril should not be prescribed to African-Americans.
Also, race-based research can reinforce stereotypes and encourage the blaming of patients for their illnesses. For instance, in Britain, South Asians are seen as a 鈥減roblem鈥� population because of their higher risk of diabetes and coronary heart disease. Medical literature often refers to the 鈥淪outh Asian health problem,鈥� says Nish Chaturvedi, an epidemiologist at the Imperial College School of Medicine in London. 鈥淚f you read that too often, it does sound like South Asians themselves are the problem, rather than some of the diseases that occur,鈥� she says.
Still, some researchers maintain that races are useful categories. 鈥淐omparing incidence rates between races may provide some insights as to the cause of disease. When you are thinking about allocating health resources and developing prevention programmes, then looking at disease by race is very important,鈥� says cancer researcher Ray Merrill of Brigham Young University in Provo, Utah.
There is a way to appease both camps, says Goldstein. Genetic clusters of the sort that he and Barbujani have identified may give doctors the hints they need, while avoiding racial baggage. Take drug response, for instance. Researchers normally study variations in genes that code for drug metabolising enzymes (DMEs). Mutations in the DMEs can affect how we process drugs, turning a dose that is ineffectively low for one person into a fatal overdose for another. Traditionally, researchers have tried to correlate these variations to race鈥攁nd the traditional races do in fact differ in the DME variations they display. But Goldstein thinks we can do better. He argues that DME variations don鈥檛 correspond very closely to racial categories. 鈥淲e鈥檒l see that maybe a drug works well in a higher fraction of individuals from one part of the world than another part, but it鈥檚 not going to be the case that a drug works terrifically in China and terribly in the UK,鈥� he says. 鈥淭here just aren鈥檛 those kinds of sharp differences.鈥�
In contrast, variation in six genes that code for DMEs was strongly correlated with the four genetic clusters he found in his latest study. 鈥淭here are some cases where the racial classification fails spectacularly,鈥� says Goldstein. For instance, Chinese and Papua New Guineans, who would both fall into the 鈥淎sian鈥� racial group, ended up in separate clusters鈥攁nd their DME variations were significantly different.
As useful as all this may be, it still falls short of the ideal. Knowing someone鈥檚 genetic cluster still gives you only a rough guess as to a given drug鈥檚 potency鈥攎ore accurate than if you were using racial labels, but still not a firm prediction. 鈥淭he long-term goal must be to find the individual mutations that matter, and genotype people,鈥� says Goldstein. 鈥淔orget about what race they are, or even what genetic cluster they are from, just genotype them directly and find the drug that works best on that genotype.鈥�
But that鈥檚 still some years off. In the meantime, some sort of ancestry-based clustering may provide the tool medicine needs to get past the concept of race once and for all.