Nightline gets caught in genetic testing trap

ABCNews Nightline last night became the latest news organization to get tripped up by the genetic testing industry. It's an irresistable story: Should I or shouldn't I find out what my genes say about my genetic destiny?

Alas, in 2008, it's the wrong story to be telling about genes. The premise of the question is false. With very rare exceptions, genes simply don't work that way. Genes are not, as we were all taught, self-contained packets of information dictating a specific blueprint design for each individual. Rather, they interact with each of our unique environments from the moment of conception to the moment we die. That's why genetic tests won't be able to tell us if are going to get a disease; all they can tell us (again, with some rare exceptions) is whether or not, statistically, we are slightly more or slightly less likely to get a disease than the average person.

So last night Martin Bashir found out that he has an increased risk of Type II Diabetes -- 19% rather than the average of 11.2%. He also has an increased chance of becoming obese.

But when you think about it, that percentage doesn't really apply to him as an individual -- it's only a population average. It's not any different from measuring the height of 100,000 people and then telling each one of them whether they are more or less likely *than average* to become a pro basketball player. There are so many other influential factors at play in any individual life, the "disposition" means almost nothing. It's useful to statisticians, useless to individuals.

I don't want this to be one of those bash-the-press blogs, because absolutely any good journalist would be expected to miss this story. That's because most of the scientists themselves have done such a poor job of correcting the extraordinary misperceptions about genetics. We're fighting an entire century of misleading metaphors about genes.

And I should point out that, toward the end of the Nightline piece, they did get the real point. After leading us in one direction for quite a long time, they closed with a quote that challenged the premise of the piece itself.

Hank Greely, a Stanford law professor, said it beautifully: "We both fear and love the idea that somehow our genomes are us, that everything about us is defined by this long sequence of as CS GS and TS. You know, it's just not true. For most of us, the events that have happened during our lifetimes  our parents' efforts for us, the friends' we've made, the jobs we've had, the good and bad luck we've had  that's more important in who we are than what genes we're born with. And that's really important to remember."

"Genes are, except for a very few unfortunate people, not destiny, they're influences, they're not destiny," he said, "and if we think too heavily that they are destiny, we are going to mess ourselves up."

Comments

[LemmusLemmus]

"That's why genetic tests won't be able to tell us if are going to get a disease; all they can tell us (again, with some rare exceptions) is whether or not, statistically, we are slightly more or slightly less likely to get a disease than the average person."

That would seem to be a rather interesting piece of information, especially if you can change your behaviour to decrease the risk of getting a certain disease. For example, if you find out that you have an elevated risk of getting lung cancer, you have a stronger incentive to quit smoking.

[Tim Dellinger]

I don't see what's wrong with the story.

If you take 1000 people with Martin Bashir's genetics and tell them that approximately 190 will get type ii diabetes, you'd be correct. It's a genetic marker. It's something observable that correlates with an outcome.

Perhaps you're concerned that the viewing public will make the "correlation implies causation" error, but the story was diligent and had someone at the end explain that there's no causation here.

Likewise, I don't understand your example. If you told Steve Kerr that, by being a mere 6 feet one inch tall, he's below average height and is thus less likely to succeed in the NBA, then he would have to think to himself: I need to do something to beat that statistic. Steve Kerr did just that, becoming a very accurate "outside shooter", i.e. he focused his efforts on being effective at a skill where height is less of a statistical marker for effectiveness. Similarly, Martin Bashir should focus on diet and exercise, where his genetic marker probably has very little statistical correlation with effectiveness in preventing diabetes.

Tim

[David Shenk]

Regarding the first comment above: do we really need a *stronger* incentive to quit smoking? It's true that a doctor could scare the crap out of a person by saying, "YOU are more likely than average to get lung cancer if you keep smoking." But that same doctor already had plenty of evidence to scare the crap out of ANY human being who is smoking. It's incredibly dangerous -- to everyone.

Further, what about the significant % of people who will take the test and find that they are LESS likely than average to get lung cancer if they smoke? They're still likely to suffer badly from smoking, mind you -- but less likely than some others.

My point is that every piece of data tells us something, and every piece of data can be isolated to tell a story. Our job is to decide which data tells the most important story. And it seems obvious to me that the most important data about health is that we ALL benefit tremendously from eating well, staying fit, exercising, reducing stress, not smoking etc. No one is immune to the dangers of those things. So if a new piece of data comes along and says, statistically, I'm 5% more susceptible to lung cancer than someone else, sure, that's added incentive, but it's got to be put in the larger perspective that I was already 100% guaranteed to suffer from smoking if I smoke.

[David Shenk]

Tim,

Of course I'm concerned about the assumption that correlation-implies-causation. Why? First, because almost every single human being alive today thinks that genes cause things directly. That's actually not true, but we're living under more than a century of dogma that says it is. For 95% of that ABC story, there is no contradiction of that idea, and there is a lot of rhetoric that reinforces it. Sticking a caveat at the end like that might make the reporter feel better, but the importance of that caveat is all but lost on almost everyone watching.

My basketball point is that no one, including Steve Kerr, needs to see a population study to quantify his or her chances of becoming something extraordinary. Kerr already knows that the odds are enormously against him, first, because only a handful of people make it into the NBA; and second, because he's shorter than most. Does he benefit by knowing those odds? Does the actual number tell him anything new? It really doesn't -- and it actually doesn't even speak to his specific chances either. A population average doesn't know Steve Kerr's particulars. Only he and his friends and family and coaches do.

There is information that will help him, as you suggest. But it doesn't come from a population study. It comes from a good coach, or a dedicated parent, or from his own observations.

Genes are important, but they aren't destiny. They don't limit us in the way that IQ mongers have implied for a century, and they don't doom us or spare us in the way that the new gene-testers are insinuating.

The information that we all need to know about how to live healthy lives, and how to achieve extraordinary success, is already available to us -- and it doesn't come from gene studies.

[Tim Dellinger]

Hi there.

Thanks for your response! I'll continue the discussion a bit because it's interesting, and hopefully we can illuminate a few points.

You're right that most people aren't especially skilled in the interpretation or use statistics, and that the difference between a "genetic marker" and a "genetic cause" is probably too subtle for the average television viewer.

That said, the studies are a valid use of time and energy that allow scientists can discern the influences of various things on the occurrence of disease. Likewise, success in any endeavor, including basketball.

Your mention of coaches/parents is interesting... I read it to say that coaches/parents have done their own informal, unquantified (sometimes intuitive) analysis of a variety of observables with respect to players. I see no difference between this and studies that use numbers. Perhaps the difference is that the coaches take into account a variety of variables (some of which are difficult to quantify), but they have nonetheless correlated those observables with performance. Although they have no hard data on height (or any other variable, like, say, determination), they all know the correlations, and they act exactly as they would act if they did have hard data. So in my mind, there's no difference between having hard data and having no hard data to support what one can merely observe without having to crunch numbers.

This now elucidates the value of the diabetes study: if a given person sees a variety of people behaving in certain ways (e.g. diet and exercise), and sees a correlation (and thus an implied risk) with respect to diabetes, then that person makes decisions based on the level of risk that they are comfortable with. But the data says otherwise: it says that there exists a certain group of people with a different destiny: they will (as is their destiny as a group, i.e. saying nothing about any individual) get more diabetes than the population at large. So the risk decision is a different one than what their informal, unquantified, sometimes intuitive (a la the basketball example) assessment would suggest. Thus the value of the genetics. Genetics *are*, in fact, destiny for the group as a whole. 19% *will* get diabetes.

Tim

[Luis Cortés Briñol]

Hello, greetings to all. Is the first time I appear. I'm Luis, a gifted boy from Spain. It was looking for pages about gifted and talented people when I found this site, that
contains articles very interesting. I am a member of several Spanish
associations of gifted people and collaborate writing about topics like intelligence, gifted minds, genius, etc. I will spend more times this web.
Thanks.

[Mitch]

David,

I agree with your thesis, which seems to be that nature and nurture interact, over the course of our lives, to produce a phenotype. But I think you're overstating your case in two ways:

First, I would love to see some data, but in my experience it's not true that everyone is a genetic determinist. I think most people understand that genes aren't destiny; the bell curve and Lawrence Summers' comments were so controversial precisely because many people recognize the importance of the environment. I'm surprised at your vehemence--do you have some personal history with this issue?

Second, I disagree that genetic information is "useless" to individuals. Why does your doctor always ask you about your family history of disease? Genetic testing is just a more precise version of that. People don't need to be told about how their height affects their chances of getting into the NBA because that's already completely obvious. But lots of things are less obvious than height, and if genetic testing can shed some light on those I'd call that useful information. Many people overstate the importance of genetic information, but countering that misconception doesn't mean that you have to say that it's useless.

Also, the genetic correlations that we have now are certainly incomplete. It may be that having a particular variant of a gene A makes you 5% more likely to have a certain condition, but it may also be the case that having a particular variant of gene B in combination with your variant of A makes you 20% more likely to have that condition. In other words, there are undoubtedly lots of undiscovered interactions between individual genetic markers, so we can expect the usefulness of genetic testing to increase.

There are even some kinds of genetic variation (e.g., copy number variation) that have significant effects but are not yet part of standard tests (as far as I know).

[Bill]

There are also significant ethical issues in genetic testing when there is no known treatment for a disease, such as ADRD (Alzheimer's Disease and Related Dementias)

Testing for those does little more than cause anxiety in the patient.

And possibly invites discrimination against them.

E.g., my mother died at an early age from a relatively rare, non-Alzheimer's dementia (frontal-temporal lobe dementia)

Will the insurance industry hold that against me when I want to buy term life, or major medical?