Linking DNA Profiles
To Diseases May Not
Lead to Prevention

Listening to scientists and government officials wax euphoric last week over the latest milestone in human genetics, a thought came to mind: twins.

The milestone was the completion of the first phase of what is called the human Haplotype Map, a $138 million, three-year undertaking by more than 200 scientists in six countries. A haplotype is a group of genetic rarities that sit on the same chromosome and tend to be inherited together, passed from parent to child as a chunk.

The so-called HapMap catalogues and locates what Francis Collins, director of the National Human Genome Research Institute, calls "differences in the human instruction book." Although any two unrelated people are 99.9% identical in their DNA, the 0.1% where they diverge explains, in part, who develops common diseases such as cancers, cardiovascular illness, diabetes and arthritis -- as well as how.

The "who" part, scientists hope, will emerge from studies that link common haplotypes to common diseases. By "testing each common genetic variant in the human population and matching it to the existence of disease," scientists might have a powerful new tool for diagnosis and even prevention, said geneticist David Altshuler of the Broad Institute of Harvard University and MIT. If having a particular haplotype means an elevated risk of, say, type-2 diabetes, someone with that haplotype might take steps, such as exercising and dieting, to keep risk from becoming reality.

The HapMap team is already well along in this goal. The Japanese aim to study 300,000 people to find haplotypes that match 47 diseases. The British are looking for clues to the genetic basis of diabetes, bipolar disorder, rheumatoid arthritis and cardiovascular disease by genotyping 2,000 people with each.

I started to think of twins here. Identical twins develop from a single fertilized egg, giving them identical DNA sequences and identical haplotypes. But twins are not identical in how they look, how they think and certainly not in what diseases they get -- not even in what diseases with deep genetic roots they get.

Identicals are only 50% "concordant" for schizophrenia; that is, in only half of twin pairs does the second sibling have the disease if the first does. They are 40% concordant for hypertension, 30% for diabetes, 25% for multiple sclerosis, 20% for breast cancer, and just under 20% for stroke and arthritis, according to a 2005 review by Irving Gottesman, University of Minnesota, and colleagues at the University of Toronto, in the journal Human Molecular Genetics.

If the exact same genetic profile in identical twins sometimes leads to disease and sometimes doesn't, the power of haplotype prediction is, shall we say, limited. Prof. Gottesman and his colleagues note that scientists tend to have blind spots. The persistent belief that DNA sequence anomalies explains which diseases people get, they say, "may be one such blind spot among geneticists."

Even if some genetic data are truly predictive, it isn't clear what good that will do anyone. "You can't unload your genotype," says David Goldstein of Duke University. "If you have a higher genetic risk for a disease, it is not clear what changes you can make in the way you live that will decrease that risk." Even seemingly obvious changes -- staying slim if you have a genetic risk of diabetes, eating heart-healthy if you have a genetic risk of heart attack -- might not actually decrease the risk.

"A typical outcome will be more like APOE4," a gene that raises the risk of Alzheimer's disease, says Prof. Goldstein: "You have a genetic risk factor and there's nothing you can do. We have no examples where identifying genetic risk factors helps with prevention, let alone improves health." (Only regular screening for colon polyps in people with a genetic risk for colon cancer comes close.)

Coincidentally, exactly a week after the HapMap announcement, hundreds of scientists have gathered at Duke University to explore glitches very different from those in the HapMap. Rather than focusing on changes in DNA -- the sequence of molecules that make up genes -- as the HapMap does, the new field of epigenetics looks at changes on DNA, such as the addition or removal of little molecular bundles that turn genes on or off. Genes can acquire these bundles by chance, or as a result of diet, environment or experiences, and they seem to trigger several forms of cancer and, probably, mental illness and other diseases.

In part, the research is driven by a recognition that "although the human-genome sequence has been complete for several years, few common diseases have been explained by common variants" in DNA sequence, geneticist Andrew Feinberg of Johns Hopkins University said in remarks prepared for the conference. That suggests reining in expectations for using the HapMap to predict who will succumb to diseases.

The HapMap scientists expressed hope that in addition to identifying who is likely to develop particular diseases, the data will reveal how disease occurs and lead to the development of drugs that block the pathway. Next week, I'll explore why this is as iffy as using genes to predict who will develop a disease.