does race exist

78 S C I E N T I F I C A M E R I C A N D E C E M B E R 2 0 0 3

IF RACES ARE DEFINED AS GENETICALLY DISCRETE GROUPS, NO.

BUT RESEARCHERS CAN USE SOME GENETIC INFORMATION TO

GROUP INDIVIDUALS INTO CLUSTERS WITH MEDICAL RELEVANCE

By Michael J. Bamshad and Steve E. Olson

DOES EXISTRACE

COPYRIGHT 2003 SCIENTIFIC AMERICAN, INC.

from a biological standpoint? Do physi-cal features reliably say anything infor-mative about a person’s genetic makeupbeyond indicating that the individual hasgenes for blue eyes or curly hair?

The problem is hard in part becausethe implicit definition of what makes aperson a member of a particular race dif-fers from region to region across theglobe. Someone classified as “black” inthe U.S., for instance, might be considered“white” in Brazil and “colored” (a cate-gory distinguished from both “black”and “white”) in South Africa.

Yet common definitions of race dosometimes work well to divide groups ac-cording to genetically determined pro-pensities for certain diseases. Sickle celldisease is usually found among people oflargely African or Mediterranean de-scent, for instance, whereas cystic fibro-sis is far more common among those ofEuropean ancestry. In addition, althoughthe results have been controversial, ahandful of studies have suggested thatAfrican-Americans are more likely to re-

spond poorly to some drugs for cardiacdisease than are members of other groups.

Over the past few years, scientistshave collected data about the geneticconstitution of populations around theworld in an effort to probe the link be-tween ancestry and patterns of disease.These data are now providing answers toseveral highly emotional and contentiousquestions: Can genetic information beused to distinguish human groups havinga common heritage and to assign indi-viduals to particular ones? Do suchgroups correspond well to predefined de-scriptions now widely used to specifyrace? And, more practically, does divid-ing people by familiar racial definitionsor by genetic similarities say anythinguseful about how members of thosegroups experience disease or respond todrug treatment?

In general, we would answer the firstquestion yes, the second no, and offer aqualified yes to the third. Our answersrest on several generalizations about raceand genetics. Some groups do differ ge-

netically from others, but how groups aredivided depends on which genes are ex-amined; simplistically put, you might fitinto one group based on your skin-colorgenes but another based on a differentcharacteristic. Many studies have dem-onstrated that roughly 90 percent of hu-man genetic variation occurs within apopulation living on a given continent,whereas about 10 percent of the variationdistinguishes continental populations. Inother words, individuals from differentpopulations are, on average, just slightlymore different from one another than areindividuals from the same population.Human populations are very similar, butthey often can be distinguished.

Classifying HumansAS A FIRST STEP to identifying linksbetween social definitions of race and ge-netic heritage, scientists need a way to di-vide groups reliably according to theirancestry. Over the past 100,000 years orso, anatomically modern humans havemigrated from Africa to other parts ofthe world, and members of our specieshave increased dramatically in number.This spread has left a distinct signature inour DNA.

To determine the degree of related-ness among groups, geneticists rely ontiny variations, or polymorphisms, in theDNA—specifically in the sequence ofbase pairs, the building blocks of DNA.Most of these polymorphisms do not oc-cur within genes, the stretches of DNAthat encode the information for makingproteins (the molecules that constitute


■ The outward signs on which most definitions of race are based—such as skincolor and hair texture—are dictated by a handful of genes. But the other genes of two people of the same “race” can be very different. Conversely, two people of different “races” can share more genetic similarity than twoindividuals of the same race.

■ Nevertheless, scientists can use genetics to sort most large populationsaccording to their ancestral geographic origin. This approach does not work aswell for populations resulting from recent mixing with other groups, however.

■ The medical implications of racial genetic differences are still under debate.

Overview/Genetics of Race

Look around on the streets of any major city, and you will see a

sampling of the outward variety of humanity: skin tones ranging from

milk-white to dark brown; hair textures running the gamut from fine

and stick-straight to thick and wiry. People often use physical characteristics

such as these—along with area of geographic origin and shared culture—to

group themselves and others into “races.” But how valid is the concept of race

PH

OTO

ILLU

STR

ATIO

NS

BY

NAN

CY

BU

RSO

N (

pre

ced

ing

pa

ge)


much of our bodies and carry out thechemical reactions of life). Accordingly,these common variations are neutral, inthat they do not directly affect a particu-lar trait. Some polymorphisms do occurin genes, however; these can contributeto individual variation in traits and to ge-netic diseases.

As scientists have sequenced the hu-man genome (the full set of nuclearDNA), they have also identified millionsof polymorphisms. The distribution ofthese polymorphisms across populationsreflects the history of those populationsand the effects of natural selection. Todistinguish among groups, the ideal ge-netic polymorphism would be one that ispresent in all the members of one groupand absent in the members of all othergroups. But the major human groupshave separated from one another too re-cently and have mixed too much for suchdifferences to exist.

Polymorphisms that occur at differ-ent frequencies around the world can,however, be used to sort people roughlyinto groups. One useful class of poly-morphisms consists of the Alus, shortpieces of DNA that are similar in se-quence to one another. Alus replicate oc-casionally, and the resulting copy splicesitself at random into a new position onthe original chromosome or on anotherchromosome, usually in a location that

has no effect on the functioning of near-by genes. Each insertion is a uniqueevent. Once an Alu sequence inserts it-self, it can remain in place for eons, get-ting passed from one person to his or herdescendants. Therefore, if two peoplehave the same Alu sequence at the samespot in their genome, they must be de-scended from a common ancestor whogave them that specific segment of DNA.

One of us (Bamshad), working withUniversity of Utah scientists Lynn B.Jorde, Stephen Wooding and W. ScottWatkins and with Mark A. Batzer of Lou-isiana State University, examined 100 dif-ferent Alu polymorphisms in 565 peopleborn in sub-Saharan Africa, Asia and Eu-rope. First we determined the presence orabsence of the 100 Alus in each of the 565people. Next we removed all the identify-ing labels (such as place of origin and eth-nic group) from the data and sorted thepeople into groups using only their ge-netic information.

Our analysis yielded four different

groups. When we added the labels backto see whether each individual’s groupassignment correlated to common, pre-defined labels for race or ethnicity, wesaw that two of the groups consisted onlyof individuals from sub-Saharan Africa,with one of those two made up almostentirely of Mbuti Pygmies. The other twogroups consisted only of individuals fromEurope and East Asia, respectively. Wefound that we needed 60 Alu polymor-phisms to assign individuals to their con-tinent of origin with 90 percent accura-cy. To achieve nearly 100 percent accu-racy, however, we needed to use about100 Alus.

Other studies have produced compa-rable results. Noah A. Rosenberg andJonathan K. Pritchard, geneticists for-merly in the laboratory of Marcus W.Feldman of Stanford University, assayedapproximately 375 polymorphisms calledshort tandem repeats in more than 1,000people from 52 ethnic groups in Africa,Asia, Europe and the Americas. By look-

w w w . s c i a m . c o m S C I E N T I F I C A M E R I C A N 81

Individuals from

different populations

are, on average, just

slightly more

different from one

another than are

individuals from the

same population.

MICHAEL J. BAMSHAD and STEVE E. OLSON have come to the subject of human genetic vari-ation from different directions. Bamshad, a geneticist at the University of Utah School ofMedicine, studies the genetics of populations to better understand human history and theorigins of disease. Olson, a science writer who lives outside Washington, D.C., is the authorof Mapping Human History: Genes, Race, and Our Common Origins, which was nominatedfor the 2002 National Book Award in nonfiction. They met during the writing of MappingHuman History and have been working together ever since to explore and explain patternsof human genetic variation worldwide.

THE

AU

THO

RS


ing at the varying frequencies of thesepolymorphisms, they were able to distin-guish five different groups of peoplewhose ancestors were typically isolatedby oceans, deserts or mountains: sub-Sa-haran Africans; Europeans and Asianswest of the Himalayas; East Asians; in-habitants of New Guinea and Melanesia;and Native Americans. They were alsoable to identify subgroups within each re-gion that usually corresponded with eachmember’s self-reported ethnicity.

The results of these studies indicatethat genetic analyses can distinguishgroups of people according to their geo-graphic origin. But caution is warranted.The groups easiest to resolve were thosethat were widely separated from one an-other geographically. Such samples max-imize the genetic variation among groups.When Bamshad and his co-workers usedtheir 100 Alu polymorphisms to try toclassify a sample of individuals fromsouthern India into a separate group, the

Indians instead had more in commonwith either Europeans or Asians. In other words, because India has been sub-ject to many genetic influences from Eu-rope and Asia, people on the subconti-nent did not group into a unique cluster.We concluded that many hundreds—orperhaps thousands—of polymorphismsmight have to be examined to distin-guish between groups whose ancestorshave historically interbred with multiplepopulations.

HUMAN GENETIC DIVERSITY


NAD

IA S

TRAS

SER

RESEARCHERS OFTEN USE SHORT PIECES of DNA called Alupolymorphisms to determine whether various populations arerelated to one another. Alus have no known function, yet theycopy and insert themselves at random throughout a person’sgenome. Because previously inserted Alus do not excisethemselves, Alu patterns can be used as yardsticks to estimatehow close two people—and, on average, two populations—aregenetically. For example, an Alu polymorphism on chromosome1 occurs in roughly 95 percent of sub-Saharan Africans whohave been sampled, 75 percent of Europeans and northern

Africans, and 60 percent of Asians, whereas a different Alupolymorphism on chromosome 7 is carried by approximately 5percent of sub-Saharan Africans, 50 percent of Europeans andnorthern Africans, and 50 percent of Asians. Some individualscarry both polymorphisms. No single polymorphism can, byitself, distinguish all the members of one major human groupfrom all the members of another group, but by analyzinghundreds of these polymorphisms, scientists can groupindividuals sampled from different locations on the basis of their genetic profiles. —M.J.B. and S.E.O.

Chromosome 1 Alu

Chromosome 7 Alu

Both Alus

EUROPEANS ANDNORTHERN AFRICANS 75% Chromosome 1 Alu 50% Chromosome 7 Alu (25% Both)

SUB-SAHARAN AFRICANS95% Chromosome 1 Alu5% Chromosome 7 Alu

ASIANS60% Chromosome 1 Alu50% Chromosome 7 Alu(10% Both)


The Human RaceGIVEN THAT PEOPLE can be sortedbroadly into groups using genetic data, docommon notions of race correspond tounderlying genetic differences amongpopulations? In some cases they do, butoften they do not. For instance, skin col-or or facial features—traits influenced bynatural selection—are routinely used todivide people into races. But groups withsimilar physical characteristics as a resultof selection can be quite different geneti-cally. Individuals from sub-Saharan Afri-ca and Australian Aborigines might havesimilar skin pigmentation (because ofadapting to strong sun), but geneticallythey are quite dissimilar.

In contrast, two groups that are genet-ically similar to each other might be ex-posed to different selective forces. In thiscase, natural selection can exaggeratesome of the differences between groups,making them appear more dissimilar onthe surface than they are underneath. Be-cause traits such as skin color have beenstrongly affected by natural selection, theydo not necessarily reflect the populationprocesses that have shaped the distribu-tion of neutral polymorphisms such asAlus or short tandem repeats. Therefore,traits or polymorphisms affected by nat-ural selection may be poor predictors ofgroup membership and may imply genet-ic relatedness where, in fact, little exists.

Another example of how difficult it isto categorize people involves populationsin the U.S. Most people who describethemselves as African-American have rel-atively recent ancestors from West Africa,and West Africans generally have poly-morphism frequencies that can be distin-guished from those of Europeans, Asiansand Native Americans. The fraction ofgene variations that African-Americansshare with West Africans, however, is farfrom uniform, because over the centuriesAfrican-Americans have mixed exten-sively with groups originating from else-where in Africa and beyond.

Over the past several years, Mark D.Shriver of Pennsylvania State Universityand Rick A. Kittles of Howard Universi-ty have defined a set of polymorphismsthat they have used to estimate the frac-tion of a person’s genes originating fromeach continental region. They found thatthe West African contribution to thegenes of individual African-Americansaverages about 80 percent, although itranges from 20 to 100 percent. Mixing ofgroups is also apparent in many individ-uals who believe they have only Euro-pean ancestors. According to Shriver’sanalyses, approximately 30 percent ofAmericans who consider themselves“white” have less than 90 percent Euro-pean ancestry. Thus, self-reported ances-try is not necessarily a good predictor of

the genetic composition of a large num-ber of Americans. Accordingly, commonnotions of race do not always reflect aperson’s genetic background.

Membership Has Its PrivilegesUNDERSTANDING the relation betweenrace and genetic variation has importantpractical implications. Several of the poly-morphisms that differ in frequency fromgroup to group have specific effects onhealth. The mutations responsible forsickle cell disease and some cases of cys-tic fibrosis, for instance, result from ge-netic changes that appear to have risen infrequency because they were protectiveagainst diseases prevalent in Africa andEurope, respectively. People who inheritone copy of the sickle cell polymorphismshow some resistance to malaria; thosewith one copy of the cystic fibrosis traitmay be less prone to the dehydration re-sulting from cholera. The symptoms ofthese diseases arise only in the unfortu-nate individuals who inherit two copiesof the mutations.

Genetic variation also plays a role inindividual susceptibility to one of theworst scourges of our age: AIDS. Somepeople have a small deletion in both theircopies of a gene that encodes a particularcell-surface receptor called chemokine re-ceptor 5 (CCR5). As a result, these indi-


Genetic analyses

can distinguish

groups of people

according to their

geographic origin.

But caution

is warranted.


viduals fail to produce CCR5 receptorson the surface of their cells. Most strainsof HIV-1, the virus that causes AIDS,bind to the CCR5 receptor to gain entryto cells, so people who lack CCR5 re-ceptors are resistant to HIV-1 infection.This polymorphism in the CCR5 recep-tor gene is found almost exclusively ingroups from northeastern Europe.

Several polymorphisms in CCR5 donot prevent infection but instead influ-ence the rate at which HIV-1 infectionleads to AIDS and death. Some of thesepolymorphisms have similar effects indifferent populations; others only alterthe speed of disease progression in se-lected groups. One polymorphism, forexample, is associated with delayed dis-ease progression in European-Americansbut accelerated disease in African-Amer-icans. Researchers can only study suchpopulation-specific effects—and use that

knowledge to direct therapy—if they cansort people into groups.

In these examples—and others likethem—a polymorphism has a relativelylarge effect in a given disease. If geneticscreening were inexpensive and efficient,all individuals could be screened for allsuch disease-related gene variants. Butgenetic testing remains costly. Perhapsmore significantly, genetic screening rais-es concerns about privacy and consent:some people might not want to knowabout genetic factors that could increasetheir risk of developing a particular dis-ease. Until these issues are resolved fur-ther, self-reported ancestry will continueto be a potentially useful diagnostic toolfor physicians.

Ancestry may also be relevant forsome diseases that are widespread in par-ticular populations. Most common dis-eases, such as hypertension and diabetes,

are the cumulative results of polymor-phisms in several genes, each of whichhas a small influence on its own. Recentresearch suggests that polymorphismsthat have a particular effect in one groupmay have a different effect in anothergroup. This kind of complexity wouldmake it much more difficult to use de-tected polymorphisms as a guide to ther-apy. Until further studies are done on thegenetic and environmental contributionsto complex diseases, physicians may haveto rely on information about an individ-ual’s ancestry to know how best to treatsome diseases.

Race and MedicineBUT THE IMPORTANCE of group mem-bership as it relates to health care has beenespecially controversial in recent years.Last January the U.S. Food and DrugAdministration issued guidelines advo-cating the collection of race and ethnicitydata in all clinical trials. Some investiga-tors contend that the differences betweengroups are so small and the historicalabuses associated with categorizing peo-ple by race so extreme that group mem-bership should play little if any role in ge-netic and medical studies. They assert thatthe FDA should abandon its recommen-dation and instead ask researchers con-ducting clinical trials to collect genomicdata on each individual. Others suggestthat only by using group membership, in-cluding common definitions of race basedon skin color, can we understand how ge-netic and environmental differencesamong groups contribute to disease. Thisdebate will be settled only by further re-search on the validity of race as a scien-tific variable.

A set of articles in the March 20 issueof the New England Journal of Medicinedebated both sides of the medical impli-cations of race. The authors of one arti-cle—Richard S. Cooper of the LoyolaStritch School of Medicine, Jay S. Kauf-man of the University of North Carolinaat Chapel Hill and Ryk Ward of the Uni-versity of Oxford—argued that race isnot an adequate criterion for physiciansto use in choosing a particular drug fora given patient. They pointed out twofindings of racial differences that are both


CO

UR

TESY

OF

NAN

CY

BU

RSO

N

NEW YORK ARTIST Nancy Burson used her invention, called the Human Race Machine,to generate the series of composite photographs that appears on page 79 (two alsoappear elsewhere in this article). The machine takes a photograph of an individual—in this case, a white woman—and adds and subtracts various outward features ofracial identity to show what a person might look like if he or she were a member ofanother race. Burson says she seeks to use her work to underscore the commonalityof humanity. And indeed, outward features often say little about the bulk of aperson’s genetic makeup. The Human Race Machine was one of the most popularattractions at the Millennium Dome in London, where it drew hundreds of thousandsof people. More about the artist and her work can be found on her Web site,www.nancyburson.com

About the Photoillustrations


now considered questionable: that acombination of certain blood vessel–di-lating drugs was more effective in treat-ing heart failure in people of African an-cestry and that specific enzyme inhibitors(angiotensin converting enzyme, or ACE,inhibitors) have little efficacy in such in-dividuals. In the second article, a groupled by Neil Risch of Stanford Universitycountered that racial or ethnic groupscan differ from one another geneticallyand that the differences can have medicalimportance. They cited a study showingthat the rate of complications from type2 diabetes varies according to race, evenafter adjusting for such factors as dispar-ities in education and income.

The intensity of these arguments re-flects both scientific and social factors.Many biomedical studies have not rig-orously defined group membership, re-lying instead on inferred relationshipsbased on racial categories. The disputeover the importance of group member-

ship also illustrates how strongly the per-ception of race is shaped by different so-cial and political perspectives.

In cases where membership in a geo-graphically or culturally defined grouphas been correlated with health-relatedgenetic traits, knowing something aboutan individual’s group membership couldbe important for a physician. And to theextent that human groups live in differ-ent environments or have different expe-riences that affect health, group mem-bership could also reflect nongenetic fac-tors that are medically relevant.

Regardless of the medical implica-

tions of the genetics of race, the researchfindings are inherently exciting. For hun-dreds of years, people have wonderedwhere various human groups came fromand how those groups are related to oneanother. They have speculated about whyhuman populations have different physi-cal appearances and about whether thebiological differences between groups aremore than skin deep. New genetic dataand new methods of analysis are finallyallowing us to approach these questions.The result will be a much deeper under-standing of both our biological natureand our human interconnectedness.


NAD

IA S

TRAS

SER

Mapping Human History: Genes, Race, and Our Common Origins. Steve Olson. Mariner Books, 2003.Human Population Genetic Structure and Inference of Group Membership. Michael J. Bamshad et al. in American Journal of Human Genetics, Vol. 72, No. 3, pages 578–589; March 2003.The Importance of Race and Ethnic Background in Biomedical Research and Clinical Practice.Esteban González Burchard et al. in New England Journal of Medicine, Vol. 348, No. 12, pages 1170–1175; March 20, 2003.Race and Genomics. Richard S. Cooper, Jay S. Kaufman and Ryk Ward in New England Journal of Medicine, Vol. 348, No. 12, pages 1166–1170; March 20, 2003.

M O R E T O E X P L O R E

People from Europe/Middle East

People from Africa

AAAGAAAGAAAGAAAGAAAGAAAGAAAGAAAG

AAAGAAAGAAAG AAAGAAAGAAAGAAAGAAAGAAAGAAAGAAAG

Carries 2 and 3 repeats: Likely to have

African heritage

Carries 6 and 8 repeats: Likely to have ancestors from Europe or the Middle East

Number of Repeats on a Chromosome

Rela

tive

Num

ber o

f Car

riers

Chromosome inherited from mother

Chromosome inherited from father

2 3 4 5 6 7 8

Unit of repeat

DNA

Tracing Human Origins

COUNTING THE NUMBER of DNA units called short tandemrepeats on chromosomes can allow scientists to groupindividuals according to probable ancestry. One such repeat,AAAG, occurs between two and seven times in people withAfrican heritage but between five and eight times in thosewhose ancestors came from Europe or the Middle East. (Everyperson inherits one set of repeats from their mother and one

from their father.) Accordingly, someone who carries two andthree repeats is likely to have African heritage, whereassomeone with six and eight repeats probably has ancestorsfrom Europe or the Middle East. People with between five andseven repeats occur in both populations, however, makingthese individuals more difficult to classify using only thisparticular repeat. —M.J.B. and S.E.O.


does race exist

Documents