对大规模人群的关联分析工作的争论

[编者的话]

在大规模人群中对疾病相关的遗传标记的关联分析计划正在英美国家大规模的展开，许多对此表现出乐观的态度，甚至表示在不久也许就在年底就可以拿到所有复杂疾病相关的遗传标记。但是，美国两个遗传学家从统计遗传学的角度衡量，对这样的关联分析计划提出了批评，称其为“只管钓鱼，不动脑子”，认为目前统计方法并不足以对付复杂疾病，这种计划是“去死胡同的一次昂贵旅行”，他们认为还是基于家系的研究方法更合适。到底谁更有道理呢，有兴趣的同学可以看一下这篇来自biomed news的报道。

Two American geneticists, critical of current statistical methods in human genetics, are at the center of international debate about proposed studies in Britain and the US that seek the genetic and environmental determinants of common maladies such asheart disease and cancer. Searching for predisposition genes to complex diseases in large populations is an expensive trip up a blind alley, they say.

The two have attacked the UK Population Biomedical Collection, a new effort to be funded by the Wellcome Trust and the Medical Research Council. The study proposes to collect blood samples from 500,000 people aged 45-65 over the next twenty years, inorder to try to match genetic polymorphisms against common diseases. Two workshops were held in April for preliminary planning to create "a national database of unprecedented size."

"I don't know any statistical geneticist in the UK who supports it", statistical geneticist Joseph Terwilliger of New York's Columbia University told BioMedNet News. Both he and his partner in the debate, geneticist John Blangero of the SouthwestFoundation for Biomedical Research in San Antonio, Texas, say the new British study exemplifies the weaknesses of the latest approaches to gene-hunting in complex diseases.

They also direct criticism at plans in the US to make more use of association studies, which are designed to detect and match single-nucleotide polymorphisms (SNPs) with disease in large populations. Though the cost of using SNP markers across thewhole genome has been prohibitive, a recent new proposal by Eric Lander at the National Institutes of Health to create a haplotype map (of commonly co-inherited SNP markers) means far fewer SNPs are necessary. "In the States, I thought the pendulum hadswung against genome-wide scans," said Blangero. "Now they're retooling."

Many research teams in the past two decades have launched studies that scan large populations (or many pairs of siblings from large numbers of small families), seeking patterns that appear to connect certain regions of the genomes with symptoms ofcomplex and multifactorial diseases. The large-population strategy is "ridiculous," says Terwilliger, who calls such studies "fishing, not thinking." Studies of large collections of siblings are likewise underpowered, says Blangero. "Why should wewaste money on studies with sub-optimal designs?" he asked.

Their message stands in direct contrast to the high-flown optimism of many scientists in the field. For instance, at a presentation in Melbourne in June, geneticist Charles Cantor of Sequenom declared that because of his firm's high-throughputdetection system for SNPs, "by the end of the year we will have found most of the genes that cause complex disease."

Scientists such as Cantor say that common polymorphisms are behind common diseases, and that large population-based studies will amplify their weak signals. They also believe that it will be possible to track such polymorphisms by using enough markersthat are in linkage disequilibrium with them.

Terwilliger argues that the correlations between markers and disease genes are chaotic and difficult to analyze in any given population, and that it is likely that common diseases are influenced by a whole range of factors, not just by commonpolymorphisms.

"The point is: we are complex, disease is complex, and we should be glad about it; if it were simple we'd all be dead," Terwilliger told BioMedNet News. Even in a less complex disease, retinitis pigmentosa, he points out, a range of gene variants areresponsible, which could be identified only by tracking them down individually in large pedigrees.

If Terwilliger is right, then methods that use large populations to amplify weak genetic signals will produce a load of genetic noise. Blangero also points out a source of false information in DNA pooling studies: PCR amplification is not uniform. Oneallele may amplify disproportionately and give a 1-2% apparent enrichment, which is close to the expected variability of the markers being measured.

Because of these difficulties, Blangero advocates a return to family-based studies, as he himself is doing in Nepalese and Mexican American families. "I'm a great advocate for returning to family linkage studies and telling people it's not thatdifficult." Terwilliger has been involved in similar studies in Finland.

Geneticist Nick Martin at Queensland Institute of Medical Research in Australia maintains that false leads are inevitable, and that there is no right way to go about finding genes for complex disease. (His own studies of non-identical twin pairs hintat genes for twinning, endometrosis, and mole counts.) It's up to the scientific community to sort it all out fterwards, he says.

"We won't know till we do it," Martin added. "It's like 1490: We're in an age of exploration." Tom Meade, director of the Epidemiology and Medical Care Unit at St Bartholomew's Hospital in London, and chair of the scientific protocol development group for the UK Population Biomedical Collection, responds to criticisms of the study the same way:"The only way to find out is to do the study."

But is it worth the effort? Proponents of the UK study (sometimes referred to as the Biobank) claim it will be an invaluable resource giving researchers the statistical power to predict disease risk. What's remarkable about the study, says OxfordUniversity epidemiologist Rory Collins, "is that no-one else is doing it." But Terwilliger maintains that at least a large chunk of the proposed data from the British study is already available in Finnish registries.

Collins told BioMedNet News that "because of the huge resources, all sorts of people are fighting over what [the study] will be."

"The British study is planning to spend ￡50-60 million without deciding what it is for," observes statistical geneticist David Clayton at the Cambridge Institute for Medical Research. "There are an awful lot of cavalier things going on in the genomeera."