Wednesday, March 25, 2009

Are biobanks a good investment?

This week sees a push to establish a Europe-wide biobank. Biobanks are repositories of huge troves of biological samples; DNA, tissues, other kinds of specimens, and the masses of accompanying information, DNA sequence and health data, that researchers hope will make sense of it all. Everyone seems to need a biobank, and many millions of dollars and pounds and euros have been committed to their creation and maintenance, for decades to come.

The claim is that these databases will be revolutionary (our term, but that's the spirit of what's being suggested), by inaugurating an era of 'personalized medicine', in which treatment is tailored to the specific patient. Of course, intelligent and conscientious doctors have always practiced 'personalized' medicine: if a given therapy seems not to be working, they adjust or try something else. But what is being promised now is that the personalized approach will be based on targeting therapies to individual genetic variation (though environments are mentioned, tailored lifestyle advice is nothing new).

Of course, success at this would be a bonanza for pharmaceuticals, but there is a more subtle issue than profit-seeking (largely at public expense) that underlies the idea of making hundreds of thousands, or even millions, of DNAs available, along with extensive biomedical information, to the private sector. At least one thing that would really be profitable compared to combinations of drugs affecting only modest numbers of individuals, is to find a 'blockbuster' drug (like statins, or baldness remedies, or--of course--the next Viagra). A great incentive, but one that private industry should fund on its own.

But one doesn't have to be cynical about private self-interest to raise questions about the scientific merits of biobanking, nor does one have to stress the potential invasion of privacy and other bioethical issues (insurance, employment, psychological impact of genetic disease forecasts, genealogical ancestry, parentage, gene-patenting, etc.).

The idea that a biobank will provide dramatic health improvements is based on a belief in genetic determinism that, despite occasional caveats and denials, is the reason that genotyping could have the high predictive value that's being touted. If stochasticity (randomness), complexity, and environmental effects predominate in the production of important biomedical traits, then genetic risk effects will be small to modest, predictive power too contingent on other unmeasured factors, and statistically unstable over time.

We already know about major disease traits is that they clearly are not due to a few, much less one, genetic mutation, except in a small minority of instances. Here, we're not speaking of the many essentially single-gene diseases, whose causal genes are already identified--there is no need for big biobanks for them.

The problem of understanding genetic causation has been a challenge, and ever-increasing sample sizes have very clearly been showing that we cannot use reductionistic approaches to dissect all (or, probably, even the majority) of genetic risk factors for common complex diseases. We know, in fact, that most are most heavily the result of lifestyle changes.

Many different kinds of evidence support this view. However, it goes against the current belief system--and that's the right term for it--in genetics as the causal basis of almost anything you want to name (except maybe the reason why stoplights are always red when you get to them). Genetic determinism is already spilling over into behavior, which is a potential social tinder box.

As we go to some lengths to try, at least, to explain in our book, simple, and we would argue often simplistic, ideas about genetic causation may fit current theoretical ideas about genomes and their evolution, and are very much in the interests of business-as-usual. But the evidence is that they are not accurate! Biological traits as a rule involve too many different genetic factors for common, serious diseases to be due to common major mutations. The genomic effects on traits are spread widely across genomes because of the nature of the way genomic functional effects build them during development, maintain them during homeostasis, and throughout their evolutionary history.

Biobanking is based on the idea that individual cases of disease are due to individual ('personalized') genotypes to an extent that justifies locking up huge amounts of research resources, essentially permanently, to build, maintain, and hive off large-scale research projects on. This is not a statement that genomes are not involved in traits, normal or otherwise! There is a difference between trying to understand genetic mechanisms in the traits involved (like blood pressure or neural function), but that requires different approaches such as work with model systems, that do not rest on naturally occurring genomic variation. Even there, even working with inbred animal strains (where there is no variation compared to that in human populations), the same levels of complexity are being found.

The issues of large biomedical databases are not wholly simple, because many good things can come of them. The science itself in terms of accuracy of measurements, technology, and so on, will be of high quality, and we'll have legions of data to play with in our academic and government labs. Certainly some health successes will result, too. But after the fact, claims of success rarely ask what other things could have been done with the resources, that might have had much greater impact. Tying up funds in one large, ongoing effort is a priority decision, in a limited world.

There's no doubt that national disease registries have had importance in public health research and policy in many ways. Computerized record systems and the like, and even the storage of tissue samples and so on, can be important. But that's not the same as doing all the analysis, such as whole genome sequencing, as part of the data base. History shows, also, that such data bases don't collect the relevant data needed, often because it's not known what to collect. What's being lobbied for now, and is almost certain to occur, are large, expensive, hypothesis-free fishing ponds.

In addition to what we think is already known about how genomes work and the complexity of genetic causation, we have seen grandiose promises from geneticists before....long before this. It is in the nature of the science establishment (certainly not limited to genetics!) to lobby in public media and to potential funders for our self-interests. Genetics is our business, and we do genetics every day, but this is about public priorities and public good, not fun for geneticists, at least when paid for on such public-sourced rationales. The focus on genes and the promise that genes will cure everything except stop-light frustration, is a misleading and even distracting way to establish such data.

If you disagree, please let us know where you think we're wrong!

No comments: