Study: Knowing Genetic Makeup May Not Help Predict Disease Risk

The great promise of the Human Genome Project is that if we can crack the genetic code in each of our cells, we may be able to predict what diseases we might get — and prevent them. But more than a decade into this project, no medical miracles have been produced. Now, a new study by the Harvard School of Public Health has more disappointing news. WBUR's All Things Considered host Sacha Pfeiffer spoke with the study's senior author, Peter Kraft, an associate professor of epidemiology at Harvard.

Sacha Pfeiffer: Your study looked at one of the possible key reasons for why simply mapping the human genome — as huge a scientific accomplishment as that is — might not alone be enough to start curing or preventing diseases. What else have researchers thought might be necessary to do that?

Peter Kraft: We've actually been fabulously successful, in the last five years especially, in finding genetic variants that are associated with disease risk. But when people looked and asked the question, 'Do these actually help us predict who's going to be at high risk?' the answer was mostly no. And one of reasons that might have been is that people looked at these variants in isolation, one at a time. But if you considered how they work together, and how they work together with the environment, to influence cancer risk or disease risk generally, people thought that might help boost the predictive ability.

So in terms of how genes work with other genes, or how genes react if you smoke, or if you're overweight, or if you've taken hormones — that kind of thing?

Right, exactly. So the models up 'till now have assumed that a gene is a gene and its effect is the same whether you smoke or not. But, of course, that may not be the case and in fact probably isn't the case.

And so in your study you took those factors into account — environmental and lifestyle factors. What did you find?

We sort of played a thought experiment and said, 'What if we knew how actually these things worked together? And given that information we tried to predict who was at high risk and who was at low risk. And we found that even knowing that information, which we're a long way from knowing and understanding — but even if we knew it, the change in the risk estimates would not be all that great. It's giving us a 1 to 3 percent increase of our ability to detect people who are at high risk.

Is that not a very useful increase?

Well, it depends on the context, but not necessarily. It seems to be in the range where your decision as a patient and your clinicians' recommendations wouldn't really change that much. So given what they knew before they drew your blood and looked at your genetics, their recommendation would probably be the same.

So your study tells us that if we get our genes mapped, we might learn a little bit more if we're at risk of a disease, but not very much to help our doctors. So where does that leave us in terms of our hopes for the Humane Genome Project and this idea that we could create "personalized medicine" customized for every individual?

What we found is that genetics will not give us a crystal ball that we can look into and say, 'Aha, you are going to be diagnosed with heart disease at age 55 or younger, and therefore you should do this or that.' That's not how it's going to play out.

If we wait patiently and let the researchers keep researching, could we get to that point?

Our studies and others suggest maybe not. There will be some diseases where that will work out. But for many — in fact, many of the ones where we really would like to be able to do that — unfortunately it won't.

There is some positive news in your study in terms of what you describe as clues we could learn from this. Tell us what we have to be hopeful about.

It's not only can we predict who's at risk, but by understanding the genes that are involved in disease, that actually tells us something about the causes of disease as well. A good example is prostate cancer, where people have looked and looked for environmental factors that might influence prostate cancer risk — diet, tobacco smoke, things like that. And generally we haven't found anything. What gives men prostate cancer, and some men don't get prostate cancer, and why some men get an aggressive form of cancer that will kill them and some men don't, we really don't know. So the genetics is actually giving us a clue as to what's driving that process.

So the idea is that even if we don't learn about disease risk, we might find some clues about disease causes, and that's a good thing for scientific research?

Exactly.