UC Berkeley Students Denied Genetic Test Results
futurepundit.com
futurepundit.com
First, there is the issue of our limited genetic knowledge. Genetic tests are currently based on a sample of 1 out of every ~3000 nucleotides in your genome. Therefore, although they do cover the vast majority of common variants, they cannot tell the full story. You may have a detected SNP that predisposes you to some risk of disease, but at the same time you may have a rare variant - not picked up by the tests - that is virtually guaranteed to protect you from the same disease. This is not trivial, as rare variants are common (everyone has lots of rare variants; however, few people share the same rare variants).
Furthermore, most SNPs that are associated with disease have not been demonstrated to be causal for disease. In fact, I'm aware of only a small number of SNPs that are not in exons (coding regions of genes) that are nevertheless known to be causal: one is rs12740374, which causes a decrease in LDL production via a novel pathway.
I suppose my point is that there is a large ethical debate that hinges on our scientific knowledge of the limitations of genetic testing. This author seems not to know about this, or not to care, and instead has chosen to write a polemic. I find this unfortunate.
An extremely one-sided discussion. Genetic testing is already very useful, and it's only going to get more useful.
First, there is the issue of our limited genetic knowledge.
Our knowledge is incomplete, but it's far from useless. Genetic testing lets us know what foods or drugs are particularly harmful or beneficial. It lets people know of future disease risks. Knowing about these risks helps people change their lifestyles to reduce risk, or at least plan for the future. While today's test results aren't 100% accurate, they give the best probability estimate based on currently available science. To say that no information is better than this is absurd.
People don't want to restrict other diagnostic tests, even though they can be similarly inaccurate. It's simple technophobia that causes newer tests to be regarded with suspicion.
Also, there's the issue of rights. It's my DNA. I have trillions of copies of it in my body. I am literally made of this stuff. If human beings have any right over their own bodies then they should be able to read their DNA.
Getting GWAS data from 23andMe, on the other hand, is generally not so much useful as it is entertaining.
The best probability estimate for most adult diseases still comes from history and the physical exam, not from genetic tests.
> Also, there's the issue of rights. It's my DNA. I have trillions of copies of it in my body. I am literally made of this stuff. If human beings have any right over their own bodies then they should be able to read their DNA.
I am most sympathetic to that argument.
I mostly agree, although 23andMe did tell me something very useful. I have the two G alleles for rs4680. (http://www.snpedia.com/index.php/Rs4680) One research paper said I am likely to respond well to Modafinil. I now find Modafinil extremely useful. While 23andMe is mostly for the coolness factor right now, it's only going to get better. For a small fraction of people, data from services like 23andMe is undeniably useful, since it will tell them things we do know with high probability (drug interaction/metabolism, early-onset Alzheimer's from ApoE4, etc).
The best probability estimate for most adult diseases still comes from history and the physical exam, not from genetic tests.
I meant that genetic testing along with a physical exam is more useful than just a physical exam.
We are getting closer to clinical utility of these tests. I'm no Luddite. But I think that these tests need to be thought about like every other medical test. (See http://www.nejm.org/doi/full/10.1056/NEJMoa0706728 for why I say that we're not quite there yet, but getting closer for some phenotypes.)
It's one thing to say that you have a right to your raw genetic data; like I said, I'm sympathetic to it from a rights-based standpoint, though not a clinical one for reasons that bhickey has brought up. It's another thing to say (not implying that you are saying this) that companies should be allowed to make clinical claims about you based this data. That interpretive step merits thoughtful consideration, and probably regulation.
The false positive rate for some Down syndrome tests runs at around 5%. True negative is around 80% (Quad test). The rate of Down syndrome pregnancies amongst 40 year olds is approximately 1%. If the quad test says that a fetus has Down syndrome, the probability that it actually does is only 14%. The story is even worse amongst younger mothers (0.1% disease incidence) -- 1.6% true positive rate. Given that the pregnancy termination given a Down syndrome diagnosis is 93% we should care about these numbers.
I can go on. Breast cancer testing in women who aren't in a risk population is similarly dangerous. Biopsies do carry small risks. Most men with prostate cancer will probably die of something other than prostate cancer. If we treat it, it often leads to urinary incontinence and erectile dysfunction.
Bad tests are frankly dangerous and should be subject to regulation.
A 14-16x greater risk of Down syndrome seems to me like a good reason to terminate a pregnancy. I agree that more accurate tests would be better.
Breast cancer testing in women who aren't in a risk population is similarly dangerous. Biopsies do carry small risks.
I do not know enough about this topic to form a useful opinion. That sounds reasonable though.
Most men with prostate cancer will probably die of something other than prostate cancer.
Isn't this a problem with overzealous treatment, not inaccurate testing?
Bad tests are frankly dangerous and should be subject to regulation.
I'm pretty hands-off about genetic testing, since there's not much of a direct risk involved in spitting in a tube. Most of the risk is from decisions based on currently-incomplete scientific knowledge. As long as people are properly educated about how to judge risks and uncertainty, I think letting them see their own genes is fine.
Even then, why get the test at all if it won't change management? In other words, if you're not going to do anything differently based on a positive or negative test, the test is simply a waste of money, and potentially harmful.
Well, quite a few genetic tests do change management (Warfarin sensitivity, antidepressant response, various differences in activity of various enzymes.), but I even want to know about test results that don't. I want to know if I have ApoE4 or some other gene that highly correlates with currently-incurable diseases. If I have a better idea of my life expectancy and my future quality of life, I can change my behavior today. I would live my life quite differently if I thought I my brain would rot in 20 years.
Our knowledge can also be harmful. I don't know that much about these tests but I would make the point that unreliable tests in general are not something to screw around with.
If a cancer test produces enough false positives, it can do much more harm than good. It's easy to take a position for your self that "I can make a rational decision no matter how much noise I'm getting on a sketchy subject". But if you start giving out sketchy data and interpretations to a large number of people, the statistic say you screw people's live up.
... Uh, what they are withhold isn't "your DNA" but their very crude analysis of it. Thus I don't see rights coming into this... but I'm sure others will.
Just some background in brief: Single nucleotide polymorphisms (SNP in the singular) are single base changes in the genome. Given the size of the genome (approximately 3 billion bases), the mutational rate, and the time since all modern human populations diverged from a single ancestral group (sorry, no number off the top of my head) we should expect that each position has mutated no more than once. So, for the most part SNPs come in two flavors, a minor allele and a major allele. If the SNP is associated with disease, one allele is called the risk allele, while the other is the protective. Down to business.
For a SNP and a disease, we can compute the odds ratio or relative risk (there is a difference) for that SNP. Suppose you have the risk allele, the relative risk is 2, and the frequency of disease is 1% -- congratulations, you have a nominal 2% chance of developing the disease. Alright, onto the problems: For most complex diseases, this doesn't help us in the slightest and is non-informative. There are a lot of markers, these diseases are rare, genetic risk might well be subsumed by environmental effects. (Good rule of thumb: Don't smoke cigarettes, don't be obese. Eat your vegetables. I'm a scientist, not a doctor.)
If you test someone, you'll invariably get a pile of risk alleles for some disease, just because there are a lot of diseases and a lot of SNPs. This will invariably cause distress in at least some people. The harm done to the subjects would clearly exceeds the benefits to them. Given the study design there's no reasonable way to obtain informed consent from the subjects if they're to be informed of the results. Without informed consent, you can't gather the data in the first place.