Thursday, January 20, 2011

Eugenics and other genetic risks, continued

Again, we return to a topic we said months ago we'd follow up on, but never did.  Holly touched on it here recently in her post about genetics and intelligence -- and, in fact, here's a quick followup on that subject; in particular, scroll down a bit in that link and read the sidebar on what happens to the brains of London cabbies -- and we wrote about it yesterday.  But, as the issues won't go away, and yet another paper is out this week in PNAS by the same characters using the same set of data, we think a topic we began to write about months ago is worth another look.  Especially given our focus this week on randomness and probability, and once again on genetic determinism and eugenics. 

Back in October, we, and many others, blogged about the paper by Fowler et al. that reported that genes largely determine political ideology.  The data set Fowler et al. mined for that paper, have mined before, and mine again for their PNAS paper this week, is the 2,600 subjects included in the National Longitudinal Study of Adolescent Health, a study of 20,000 adolescents in grades 7-12 in 1994, and followed up in 2001-2002 when students were 17-24 years old, and a subset more recently followed up with biomarkers, including DNA.  The study was designed to look at the effects of social context on health behaviors like seat belt use, drug use, sexual activity, nutrition and so on.  

The small subset who donated DNA were genotyped for 7 candidate polymorphisms
The initially targeted candidates are the dopamine transporter (DAT1), the dopamine D4 receptor (DRD4), the serotonin transporter (5HTT), monoamine oxidase A (MAOA), monoamine oxidase B (MAOB), the dopamine D2 receptor (DRD2), and the dopamine D5 receptor (DRD5).
Why these particular genes were   chosen is not something we could easily find but presumably it has to do with previous reports of their association with particular behaviors.

A cursory literature review shows that using these genetic data, researchers have, to date, found 'the warrior gene', that is, an association with DAT1 or DRD2 and gang behavior and serious delinquency, a link with low grades and DAT1, a dopaminergic gene, and that "students with a single, DRD4 variant had significantly lower grades in English and math, but only marginally lower grades in history and science", the genetic basis of victimization, the genetic basis of our place in a social network, that is, popularity, an association with genes and smoking behavior, as well as the links with political behavior and ideology reported by Professor Fowler et al.  And, the new study by Fowler et al. reports that people who share the same DRD2 allele associated with alcoholism tend to be friends.  

Ok, let's just take the DRD4 gene.  This gene that has been associated with numerous behaviors previously, including attention deficit disorder (the association is with a 48 basepair VNTR, or variable number of tandem repeats, in exon 3), novelty-seeking (that exon 3 VNTR again), externalizing behavior in toddlers (exon 3 VNTR), temperament in children with the exon 3 repeat allele if they are given poor quality parenting (or 'differential susceptibility to child rearing'), aggression at age 4 (long repeat), attachment disorganization in infants (48 bp repeat), to cite just a few of the studies reporting an association of behavior with this long repeat in this dopamine receptor gene.  And there are plenty more.

The idea, as proposed by Hamer et al. in 1996 in the precedent-setting Novelty Seeking study, was that the number of repeats in the gene affects the structure of the receptor, thus how dopamine binds to it, and so the efficiency of neurotransmission.

But now there's this, a 2010 genomewide association study by Verweij et al.:
Variation in personality traits is 30–60% attributed to genetic influences. Attempts to unravel these genetic influences at the molecular level have, so far, been inconclusive. We performed the first genome-wide association study of Cloninger's temperament scales in a sample of 5117 individuals, in order to identify common genetic variants underlying variation in personality. Participants’ scores on Harm Avoidance, Novelty Seeking, Reward Dependence, and Persistence were tested for association with 1,252,387 genetic markers. We also performed gene-based association tests and biological pathway analyses. No genetic variants that significantly contribute to personality variation were identified, while our sample provides over 90% power to detect variants that explain only 1% of the trait variance. This indicates that individual common genetic variants of this size or greater do not contribute to personality trait variation, which has important implications regarding the genetic architecture of personality and the evolutionary mechanisms by which heritable variation is maintained.

No association with DRD4 and Novelty Seeking.  And no association with DRD4 and Novelty Seeking was found in 2 samples in New Zealand, and while some studies in birds have shown an association with a DRD4 variant and exploratory behavior, it isn't consistently reported.  A study of one of these repeat polymorphism has shown that the effect of the variation depends on your socioeconomic status: negative for low SES, positive for high SES.  Context determines what the genotype determines.

Now, all the usual caveats that we mention over and over again apply here -- the definition of the trait may vary enough between studies that they are comparing apples and oranges, and the association may in fact be real in some populations and not in others.  But GWAS should find genes with large effects, and the Verweij et al. study did not.

What if that's because the trait is due not just to a single gene but to gene-and-environment interaction? Would a GWAS still pick it up?  This would depend in part on the size of the study sample and the amount of environmental variation among those sampled.  GWAS detects net results, basically, especially if the environmental factors are not known or measured.  Large effects found in a GWAS may be because there was some tractably invariant environment among the sampled people, that may not be present in another sample, and the same allele may not have its purported effect in those data.

The association of this gene with political ideology could be real.  Or it could be a chance positive result due to multiple testing or other statistical aspects of the sample.  Or it could be real....but only ephemeral, changing rapidly with societal context.  The notion that we can predict these kinds of things very meaningfully is one we should be very circumspect about.  Even after extensive (and expensive), very technically sophisticated studies (unlike many of the behavior one-gene studies mentioned here), we can't make good predictions of very clearly genetically-affected traits like how tall you are or whether you'll have diabetes (and if so, what its particular characteristics, severity, response to treatment, etc. will be).

Every trait you can name probably has at least some heritability; that is, genetic variation contributes to variation in the trait.  Heritability is usually substantial.  But so are chance and the various aspects of lifestyle and environment.  High heritability means that genes are relative important in the population, but does not mean that an individual genotype predicts that person's trait reliably.  GWAS experience, which is now extensive, shows that high penetrating power of single (or even a few) genetic variants is not common.  Often, the trait definition is crude (like 'intelligence' or 'IQ', or 'asthma' or 'autism' or even obesity), but the definitions are changeable and cultural, the measures sometimes arbitrary, and people can have the same value for different reasons.  People with high intelligence measures can have better ability to memorize, or to learn quickly from one exposure to a fact or challenge, or to integrate repetitive exposure, or to do mental work by visualization or by other means, and so on.  

Change the environment or the measuring criterion and you can change the result.  Different approaches to modifying the achieved trait, like therapy for a disease, can apply differently to different people with similar measurement value.  People with similar glucose levels have different ways in which the 'same' trait is manifest during their respective lives.  This should be obvious, since diabetes and IQ have both increased substantially in recent decades, but the underlying genetic variation hasn't.

This is in rough terms why things can be genetically affected in real and substantial ways but, we would say, not genetically determined.  Again, the problem lies in overstating the latter and putting it into policy.  This is where the idea of 'eugenics', that is, of using ideas about inherited worth to make policy to 'improve' the human species and its genome, become heavily sociopolitical rather than scientific.  Historically this has led to discrimination against individuals and against whole groups.  One might be tempted to say that it's 'right wing' to believe in genetic determination of traits like ability or behavior, or the inherent value of one race vs another.  But this would be vague at best and misleading at worst.

To caricature affiliations, we can say that it's not just rapid right-wing behavior genetics that we should be wary of.  The rabid left has left its own trail.  While the Nazis were gassing people who were inherently 'inferior', the Soviets were starving people because, under the influence of one Trofem Lysenko, they believed life was improvable by experience alone rather than by inheritance (a view generally attributed originally to Lamarck in 1813).  In both cases, categorical beliefs about the role of inheritance led to disaster on a major scale.

That is why, despite the fact that of course genetic variation contributes to all biological variation, we should be very careful about making pronouncements that could be used for societal policy.  Today that policy might be favorable to you....but what about tomorrow?  Even if nothing untoward is being done with genetic data these days, prevention is the better part of valor.

Of course, if you want to make strong statements about the genetic determinants and evolution of ostrich knees, feel free, because only a few 'experts' in the world will care one way or the other.


Holly Dunsworth said...

Excellent read.

And as I've been thinking a lot about variation lately I'd like to add...

We, as observers and wannabe explainers, need to be careful with perspective. I'd guess that we err on the side of overemphasizing differences between people and between organisms.

Holly Dunsworth said...

And there's probably a gene for that.

Ken Weiss said...

Some of these are questions for good observational anthropology in comparative cultures. They would look for interpretable universals that might indicate genetic control (which, even so, need not be simple or uniform worldwide).

Unfortunately, cultural anthropology has developed its sometimes rabid anti-science or pro-Darwinian factions, coloring what is observed in the field. And, the panoply of potentially informative cultures, not too influenced by the outside world, is minimal.

Can we rely on past ethnographies? Did observers make relevant observations?

Since these are open questions, and since clearly both genes and 'environment' contribute to trait variation, the debates won't end.

But to me, at least, they are so tied up with social politics, resource, and power distribution that they are inherently political and since we know there can be abusive implications, I think we should be wary and preventive. Of course, advocates on both sides will not change their views based on 'facts', as we know very well and have posted about.

Anne Buchanan said...

I agree, Holly. One consequence of this is that we tend to overlook the significant overlap in the distributions of a given trait in whatever we're calling different populations. Iq and race of course being the big one, but this would be true for most other traits, behavioral and otherwise, as well.