Poring over the raw genetic data, Mark Daly noticed a startling pattern. An expert in statistical genetics and a fellow at MIT’s Whitehead Institute for Biomedical Research, Daly was scouring a region of human chromosome 5, a place that colleagues strongly suspected contained a gene that puts people at risk for a devastating digestive condition called Crohn’s disease.
The sequence spelled out in the DNA letters A, T, G, and C was almost identical in all the samples Daly examined-each from a different person. As Daly expected, sprinkled every thousand letters or so were spots where a single letter tended to vary from one person to another. Then came the surprise. Many of these single-letter variations seemed to occur together, as if they were tightly linked across long stretches of the DNA. In other words, whenever Daly looked at an individual copy of one of the sections of DNA and found an A at one of these positions, he would find a G at the next one, about a thousand letters away, a C in a third position still further down the line, and so on. After roughly tens of thousands of letters, another pattern began; the long stretches of linked variants, it seemed, divided the chromosome into neatly defined blocks. What’s more, for any given stretch of the chromosome, there were only four or five versions of these blocks that kept showing up in the different individuals Daly studied. Daly realized he was staring at evidence of an underlying structure to the human genome. He was also looking at the beginnings of biology’s next big project-and its next big controversy.
At about the same time, in the fall of 2001, several other genetic researchers reported similar findings. Much of the human genome, it soon appeared, consists of what researchers began to refer to as haplotype blocks. And as Daly had seen on chromosome 5, the blocks tend to come in a limited number of common varieties, which suggests that the genetic variants that put people at risk for common diseases might also be widely shared. Overall, the findings suggested a far simpler structure for the human genome than had previously been supposed. “It is a fundamental change in how we view genetic variations,” says Daly. “And for once, the genetics are very favorable toward performing disease studies.”
Indeed, the finding has immense implications for understanding and treating diseases such as diabetes, schizophrenia, and hypertension. Though people share roughly 99.9 percent of their genes, it is precisely that other one-tenth of a percent that plays a role in determining why one person gets schizophrenia or diabetes while another doesn’t, why one person responds well to a drug while another can’t tolerate it. If, in fact, the variable DNA letters occur in a limited number of easy-to-identify, blocklike patterns, it would give geneticists a practical way to quickly and cheaply search for the complex genetic variations related to common diseases and different drug responses. Instead of identifying all 10 million of a person’s specific single-letter variants-a time-consuming and prohibitively expensive task-researchers could simply pinpoint a telltale letter for each block and then know the other variants around it.
But first they would need a map, one that identifies the boundaries of blocks and the different versions of each block found in populations around the world. Last October, a year after Daly’s discovery, the world’s top genetic researchers-including scientists from the Whitehead, the National Institutes of Health, Johns Hopkins University in Baltimore, the University of Tokyo, the Beijing Genomics Institute, and Cambridge, England’s Wellcome Trust Sanger Institute-formed a $100 million, three-year plan to chart just such a map. It’s called the International HapMap Project, and beginning with several hundred blood samples collected from Nigeria, Japan, China, and the United States, it will use highly automated genomics tools to parse out the common haplotype patterns among a number of the world’s population groups (see “Shining Light Variations” infographic).
“This is really a natural outcome of having the sequence of the human genome,” says Aravinda Chakravarti, director of the Institute of Genetic Medicine at Johns Hopkins and a leading participant in the international consortium. “Now we want to know what part of the genome varies. Knowing the variations that enhance or retard specific diseases will be a tremendous value” for medicine, he says. “Having a catalogue of the variations will be very helpful. And the more global the catalogue is, the more helpful it will be.”
The hope is that as disease researchers and epidemiologists compare the genetics of patients with ailments such as asthma or schizophrenia to those of healthy people, the map will guide them to the differences in genes or combinations of genes that put a person at risk. Not only can such information be critical in forewarning at-risk individuals, it can also provide invaluable clues to drug developers searching for the biological mechanisms that cause the diseases.
The most immediate impact of the HapMap, though, is likely to be the prediction of how a patient will respond to a drug (see “Startups Seek Genomics’ Killer App” ). Adverse drug reactions cause more than 100,000 deaths each year in the United States alone. And, says David Goldstein, a geneticist at the University College of London, identifying the genetic factors underlying different responses to drugs could lead to quick and easy tests to screen patients. “There is absolutely no doubt that the haplotype map will help,” he says. “Even if that’s all the HapMap does, it will be a critical contribution to medicine.”
The HapMap provokes such excitement in the medical community largely because the hunt for disease-causing genes has hit a stone wall. Despite successes in finding the genetic culprits for some rare and deadly disorders, such as cystic fibrosis and Huntington’s disease, which are caused by lone genes, researchers have had a difficult time finding the genes behind common diseases, like schizophrenia, diabetes, hypertension, and alcoholism. Geneticists suspect that some combination of dozens or even hundreds of genes contributes to each of these disorders. Tracking down a single rogue gene is already like hunting for a needle in a haystack. But understanding how patterns of variations among individuals and populations correlate with common diseases is “fantastically more complex,” says College of London’s Goldstein.
The discovery of the haplotype blocks gives medical researchers a useful way to navigate this mind-boggling complexity. The evidence, says Daly, suggests that, in fact, most of the human genome consists of these blocks, which vary from 10,000 to around 50,000 letters in length. It is a structure that neatly organizes the three billion letters in the genome and one that “doesn’t necessarily have to be the way it is,” says Daly.
Those pushing the HapMap believe the orderly, blocklike structure of the genome is more a reflection of history than of any biological function. They suspect that most variations in single DNA letters date back many tens of thousands of years and have been inherited intact generation after generation, along with neighboring stretches of DNA. This explains why only a few common versions of each block are likely to be found, since humans share a limited set of ancestors. It also suggests that comparing the patterns of genetic variation found in different parts of the world can provide a remarkable history of human migration over those tens of thousand of years.
Not all geneticists, however, buy it. Some argue that much of the science behind the haplotype project remains speculative. Numerous questions still swirl around the blocklike structure, maintains Kenneth Kidd, a geneticist at Yale University School of Medicine in New Haven, CT. Doubts remain about how to define the boundaries and, even, how widespread the blocks are throughout the genome, says Kidd. What’s more, he contends, the HapMap’s premise that there are consistent patterns of genetic variation around the world is likely wrong and that there will be “tremendous” differences. “There are likely to be few universal blocks,” he says. The haplotype map, he adds, “is being touted as great for all populations, but I don’t think it will be.”
There are also doubts about whether the HapMap is even looking for the right genetic culprits. William Thilly, an MIT geneticist, says that for numerous conditions known to be caused by mutations in a single gene, there are dozens to hundreds of different mutations in that gene that have been found to cause the same disease. Thilly argues that genetic risks for common diseases are caused by a “spectrum” of relatively rare mutations scattered over unknown genes throughout the genome. He points out that many common diseases afflict diverse populations that display markedly different haplotypes. In other words, the HapMap’s effort to detail common patterns in genetic differences and link those differences to diseases is largely a wild-goose chase.
Following the Data
In his small office in a corner of a busy research lab at Boston’s Massachusetts General Hospital, David Altshuler, a physician and expert on diabetes, is full of restless energy. Six floors below, gridlock has brought the traffic coming in and out of the sprawling hospital to a maddening halt. But Altshuler, who is also the director of medical and population genetics at the Whitehead Institute and a prime mover behind the HapMap project, can barely sit still. The critics of the HapMap in the genetics community clearly have him peeved.
“They’re nihilists. All they say is, Don’t do it.’ I don’t believe it’s a panacea, but it’s a useful tool,” says Altshuler. He points to the failure of many critics to propose a feasible alternative as particularly frustrating. “Ultimately, all of genetics boils down to measuring the genetic variation in some population of people and comparing it to their characteristics and looking for correlations. That’s all genetics ever is.” And, adds Altshuler, the HapMap “is simply a tool to study genetic variation at unprecedented levels of accuracy and detail.”
Altshuler freely acknowledges that many scientific questions remain about how genes vary among individuals and populations and even about how effective looking at patterns of common genetic variations will be in tracking down risk factors for diseases. But, he adds, the HapMap offers a direct route to testing ideas about the genetics of common diseases. “For that reason alone,” he says, “it is an important investment.”
One issue to be resolved is how extensively human populations share specific versions of haplotype blocks. Geneticists do know that some differences between populations are a consequence of their migrational history. They expect, for example, that the length of haplotype blocks in populations from Africa will be shorter than those in populations of European or Asian origin. That’s because humans originated in Africa and migrated throughout the rest of the world, starting around 50,000 years ago. Thus, the genetic history of populations in Africa is older and, since their genes have had a far longer time to vary, the linked blocks have had more of a chance to break up into smaller segments. It is also likely that any migration out of Africa did not include representatives from all groups, so geneticists expect to find more diversity in Africa.
Altshuler and his colleagues found strong evidence that this is precisely the case in a preliminary study they did of the haplotype patterns of nearly 300 people, including African Americans, people from Nigeria, and volunteers with European, Japanese, and Chinese ancestry. In a paper published last summer in the journal Science, the researchers described finding most of the common haplotype varieties in all the populations, though samples from Africa showed the greatest diversity and also tended to have shorter haplotype segments. The paper’s conclusion: while there are some differences, the boundaries of the haplotype blocks and the common versions are largely shared across populations.
A Starting Point
These days, Charles Rotimi is frequently en route from his office in Washington, DC, to Nigeria to carry out delicate negotiations with community leaders and residents that will permit the HapMap project to begin gathering blood samples. The Yoruba people of western Nigeria are one of Africa’s largest and oldest ethnic groups, and a perfect starting point for the HapMap project.
Rotimi, a genetic epidemiologist at Howard University’s National Human Genome Center, is hoping the HapMap can provide genetic details that will greatly facilitate his research on how people with shared ancestry vary in their reactions to drugs and susceptibility to common diseases. Specifically, Rotimi is interested in pinning down why populations of the African diaspora in various parts of the world suffer from dramatically different rates of diseases like hypertension, diabetes, and obesity.
For example, in results gleaned from conventional epidemiological studies during the last few years, Rotimi has found that about 7 percent of blacks living in rural Africa and 14 percent of those living in urban Africa suffer from hypertension, while 34 percent of African Americans have the condition. “We see drastically different rates of disease in those that share common ancestry,” says Rotimi. “We’re seeing very clearly that the current environment is the most important factor.” But he believes the HapMap could shed new light on this result. “We’ve made assumptions about the underlying genetics” of the different populations, says Rotimi. It might turn out, he says, that the HapMap reveals previously unrecognized “subtle differences in genetic patterns” that could help him better interpret the disease findings. For example, he says, if the patterns of the haplotype blocks in the populations are sufficiently different, it could be a key clue to understanding genetic factors underlying disease risks.
It is just these types of studies that point to the ethical complexities raised by the HapMap and other new genomic methods. On one hand, looking for genetic variations among racial groups runs the danger of reinforcing old stereotypes. Yet genetic differences and similarities in populations with shared ancestry are frequently observed and can provide a powerful tool for understanding diseases; they may even help researchers pinpoint nongenetic factors, like diet and the environment, that influence who contracts a disease. At least according to some, the use of broad racial categories in genetic studies may actually help turn up social, environmental, and cultural reasons for health disparities among different groups.
Last summer, Neil Risch, a leading population geneticist at Stanford University, gained national attention by publishing a paper in an online journal called Genome Biology calling for the use of five racial categories in genetic studies. The paper attacked a growing consensus among researchers that racial classifications are neither genetically valid nor useful. Risch’s contrarian conclusion: differences in drug responses and disease risks need to be separately examined in each of the five racial groups. Otherwise, he warned, genetic research will tend to ignore issues peculiar to minority groups.
No sooner had Risch’s paper begun stirring up the race debate in the genetics community than a group of researchers headed by Marcus Feldman, a prominent population biologist at Stanford, published an article in Science that reported detailed data on gene samples from individuals from 52 populations. The research group sorted the samples using both an advanced genomic approach and self-reported ancestry. They found that the genetic samples fell generally into five geographic categories: Europe, Africa, East Asia, Oceania, and the Americas. They also found that how people categorized themselves-whether they called themselves black or white or Asian-correlated closely with the genetic categories.
Yale’s Kidd, one of the coauthors of the Science paper, notes that following its publication, some observers argued that the findings demonstrated the existence of races as biological entities, while others maintained that the data proved the opposite. “My opinion is closer to the latter,” says Kidd. The results, he explains, show that it is possible to detect very small genetic differences between different populations if you look closely enough. “There is a bit of history that is recoverable,” says Kidd. “But that doesn’t support the idea of race. It does support that when you look around there is some geographical structure that is present in the genome, though it’s extremely small.”
While the interpretation of the results might be in doubt, the paper, and today’s advancing genetic tools, clearly mark the reentry of mainstream genetics into the debate over race and how to best categorize populations.
Does Race Matter?
The U.S. Food and Drug Administration is now deciding whether to approve a controversial heart disease treatment called BiDil that is specifically meant for African Americans. (The drugmaker, NitroMed of Bedford, MA, claims that blacks are twice as likely as whites to suffer heart attacks.) This new “ethnic drug” is far from an anomaly. Earlier this year, the FDA proposed guidelines prescribing that all drugs in development be evaluated for varying effects on different racial groups.
As genomic tools improve, and there is an increasing emphasis on pharmacogenetics (the use of information about genetic variations to predict a drug’s safety or effectiveness), the debate over race and genetics will be most vigorously played out in the medical arena. Race, of course, already plays a huge role in how doctors diagnose and treat patients. Physicians are well acquainted with the idea that Caucasians with northern-European ancestry have higher rates of cystic fibrosis than Asians and blacks, while African Americans suffer from higher rates of hypertension and diabetes.
“Race is used all the time. It’s part of a doctor’s calculations,” says Mildred Cho, codirector of Stanford’s Center for Biomedical Ethics. But the downside to using race as a way to view genetic differences, she says, is that it tends to oversimplify a person’s complex genetic makeup. “It may seem like a good shortcut, but it can be misleading. It’s a shortcut to nowhere.” Most differences will be relative, she says. Imagine, for example, that researchers find that 60 percent of Asians fail to metabolize an enzyme, while 40 percent of Caucasians fail to do so. In terms of treating a particular patient, she points out, the results “are clinically not very helpful.” Similarly, she argues, new drugs like BiDil are “jumping the gun” by targeting specific races “without the necessary understanding of underlying biological causes” of disease differences.
The hope is that the HapMap and other new, advanced genomic methods will help clarify complex genetic differences and, eventually, give physicians the tools to profile the genetics of each person and use that information to guide treatment decisions. If you want to know how to medically treat a person, you need information about him or her, says London’s Goldstein. “Only in the ignorance of that,” he says, “do you think about the population and flop on a racial label and say, That’s good enough.’”
It is into these turbulent waters that the HapMap is diving. And while on one level it is only a tool to help determine specific DNA variants, the project will almost inevitably play a critical role in future debates over race, medicine, and genetics. Whether it plays a productive role-helping to destroy stereotypical concepts of race-or whether it is manipulated by those wishing to gain genetic credence for racist agendas, is still anyone’s guess.
What does seem certain is that the HapMap will produce surprises and scientific insights into human variation that both scientists and the public will struggle to understand. And like any cartographers exploring unknown geography, HapMap researchers will surely happen upon some tricky terrain. The discovery several years ago, for example, that a series of mutations in the cancer genes BRAC1 and BRAC2 were particularly common in Ashkenazi Jews raised widespread fears about how these findings could be used to stigmatize Jewish people. Imagine the potential for social harm if the HapMap produced genetic data that eventually revealed that a specific population has a propensity, say, for alcoholism or schizophrenia.
“I’m not a naysayer” to the HapMap project, says NYU’s Duster. “But I feel it is fraught with all kinds of dangers.” Those involved, he says, need to be particularly sensitive to how the genetic variations are explained to the public. “There will be differences” between populations, he says. “The wrong way to proceed is to report the differences as more profound than they are and with consequences for anything other than the particular disease.”
Participants in the project say that they are aware of these dangers, but that the potential benefits justify the risks. Altshuler points to the years he has spent treating diabetes patients and facing the frustration of not being able to offer a solution. “The reason that I do this research is that the most striking thing about medicine is how little we know, how little we have to offer patients for common diseases.”
If the HapMap fulfills its potential to help medical researchers and physicians better navigate the treatment of common and devastating diseases, like diabetes, schizophrenia, and hypertension, it will have been dangerous ground well worth exploring.
The Tool Makers INSTITUTION
TECHNOLOGY STRATEGY Illumina
(San Diego, CA) Fiber-optics-based tools to scan for genetic variations Sell instruments to large drug companies and genomics labs Genaissance
Pharmaceuticals (New Haven, CT) Mapping haplotypes and genes associated with certain diseases Market database of haplotypes and genes to large pharmaceutical makers Perlegen Sciences
(Mountain View, CA) Microarrays to scan for genetic variations Collaborate with drug companies to examine variations associated with diseases and drug responses Sequenom
(San Diego, CA) Mass spectroscopy to rapidly scan for genetic variations Sell instruments to large genomics labs; identify disease genes ParAllele Bioscience
(South San Francisco,
CA) A “single-tube” assay to scan for genetic variations Develop technologies for genomic researchers to use to scan genomes for variations