A new study shows how hard it is to predict the health risks of healthy people from their DNA. (NHGRI)
In a new study, scientists have shown just how hard it will be to one day use your DNA to predict your risk for illnesses.
As expected, it wasn’t the reading of the DNA that was the tricky part -- that gets cheaper and easier all the time.
Instead, the difficulty was figuring out which DNA differences mattered and which didn’t. Four groups of scientists couldn’t agree on which differences predicted an increased risk for an irregular heartbeat and which didn’t. And even the differences they compared didn’t turn out to predict much of anything.
Until scientists work this out, the dream of truly personalized medicine will remain a distant one.
Harder Than We Thought
Listening to the news, you can be forgiven for thinking that sooner rather than later doctors will be able to predict most anything about your health from a look at your DNA.
In fact, right now, they can’t tell you a whole lot.
But the day when you can go to your doctor and she can correctly tell you only from your DNA which ailment you are at a higher risk for is a long way off. This becomes clear in a new study in the Journal of the American Medical Association (JAMA).
In this study researchers couldn’t use DNA to reliably predict who, from a group of 2,202 patients, was at a higher risk for arrhythmia, an irregular heartbeat. This is surprising because the two genes they looked at are supposedly two out there that are better understood.
If we can’t make good predictions on the few genes we understand really well, then odds are we will do even worse with the ones that are still more of a mystery. And these are the genes people will be most interested in, involved in heart disease, type 2 diabetes and so on.
This was a small study and it may be that with a larger group of patients, scientists will be able to make better predictions. But regardless it does give us an inkling of what we are up against in the upcoming genomics revolution.
It is going to be a hard slog. But in the end, the hope is that it will be worth it because it will make us all healthier.
The Pitfalls of Personalized Medicine
The key to predicting people’s health risks based on their DNA is knowing what to look for. This is even harder to do than you might think because everyone’s genetic make up is so unique.
The first step in finding disease-causing DNA differences is to gather up two groups of people and compare their DNA. One group will have the disease and the second, the control group, will not.
Any DNA differences more common in the disease group could be implicated in that disease. Lots of follow-up work is done until scientists are left with what they think are the really important variants.
Surprisingly, it often isn’t the variants themselves that are important for personalized medicine. Instead, it is the genes they affect that are the real prizes.
This is because many of the variants they find in the original study will be rare—not many people will have them. So just looking for them would not be that useful.
What they will look for are DNA differences in these genes that no one has seen before. Turns out that there are a lot of these.
The final step is to predict which of these other variants will be important for the disease. (These are called “variants of uncertain significance” or VUS.) This is what this study could not do very well.
Genes and an Irregular Heartbeat
A number of previous studies have shown that certain DNA differences in two genes, SCN5A and KCNH2, increase a person’s risk for having an irregular heartbeat. These are the two genes the researchers focused on.
Not surprisingly, the researchers didn’t find these DNA variants in any of the 2,202 patients they looked at. Remember these kinds of variants are usually pretty rare.
But they did find lots of other differences. They were prepared for this and had lined up four different groups to help them predict which variants might increase a patient’s risk for arrhythmia.
They came up with 42 DNA variants in 63 patients that looked like they fit the bill. Unfortunately, none of these DNA differences predicted anything about a patient’s risk for arrhythmia. There was no significant difference in cases of arrhythmia between the group that had these variants and the group that didn’t.
In some ways this wasn’t surprising. After all, 63 people is a pretty small group.
What is more troubling for personalized medicine is that the four groups making the predictions agreed on very few of the 42 variants. Making accurate predictions is an incredibly important step, and it failed in this study.
To get to a future where genetics can help the maximum number of people, we are going to need to be able to use a patient’s DNA to figure out his or her health risks. We’ll get there, but it may take longer than we thought.
Get the best of KQED's science coverage in your inbox weekly.