Article Text

Download PDFPDF

The art of replication
  1. Michael Kabesch
  1. Professor Michael Kabesch, Hannover Medical School, Center for Pediatrics, Clinic for Pneumology and Neonatology, Carl-Neuberg Strasse 1, D-30625 Hannover, Germany; Kabesch.Michael{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Hot topics in science, like fashion, come and go. Asthma genetics is no exception to this rule. Asthma candidate genes come and very few stay for good, changing from “candidate” gene to “disease” gene. Only an exclusive few survive a thorough replication process.

The asthma candidate genes studied in the replication approach by Blakey and colleagues1 (pages 381–7) were originally discovered by positional cloning, a technique based on linkage studies using microsatellite markers and subsequent fine mapping of these linkage signal loci in families. Until recently, when genome-wide association studies (GWAS) entered the field, this was the method of choice to detect novel asthma candidate genes. In contrast to candidate gene studies, which are hypothesis driven and always only as good as the underlying hypothesis, positionally cloned genes had the aura of being the better asthma candidate genes, as they were detected by a systematic genomic approach not potentially biased by the researcher’s own belief in a hypothesis. Seven years after the first positionally cloned asthma candidate genes were published in 2002/2003, positional cloning has lost its nimbus. It became clear that these genes are no better candidates than those detected by hypothesis-driven approaches. Bias is not excluded by positional cloning. What counts at the end of the day in complex disease studies is replication and verification.

Interestingly, the interpretation of replication has become a matter of debate itself. Some argue that larger populations may not necessarily be better for replication of genetic signals in complex diseases. Disease definitions may be broader in large studies as phenotype assessment may be limited in larger studies. Environmental conditions in the replication study may be very different from the population where the initial result was acquired. As gene–environment interactions are an important factor in complex diseases and as asthma could be a syndrome with different underlying mechanisms, this diversity may influence the prospect for replication.

Another important issue in replication is that different but related phenotypes are sometimes used synonymously to argue for positive replication. This approach has been called “loose replication”. For example, instead of replicating the association of single nucleotide polymorphism (SNP) A in gene 1 with asthma, an association of SNP B in gene 1 with elevated immunoglobulin E (IgE) in the replication study is observed. As elevated IgE may be related to asthma, this is counted as a loose replication of the initial finding. However, this seems rather like an extension of the original finding, which needs further replication itself to increase the likelihood that the reported association is real. Nevertheless, these studies, as well as failures to replicate, should be published to evolve a full picture of a gene’s possible effects. Much can be learned from these findings and, in the end, they may be even more important for our understanding of genetics and disease mechanisms in complex diseases than straightforward replication.

Blakey et al studied five positionally cloned genes, initially found in linkage and association studies in children. They now aimed to replicate associations with asthma, wheezing and IgE in their large British population sample. They found small but significant effects for polymorphisms in DPP10 and ADAM33 (and borderline for GPR154), with odds ratios of around 1.1 per allele.

Is this the final verdict on the role of these genes in asthma? Rather not. Although this is a large and well phenotyped cross-sectional and probably unbiased population sample, it is a snap-shot in time. The individuals studied were born in 1 week of 1958 in Britain. It may be assumed that environmental factors were very different at the time these individuals developed their asthma compared with children growing up nowadays. Looking back on the second half of the 20th century we have witnessed rising prevalence rates for asthma, and fundamental changes in lifestyle, diet, and personal and meta-environments. It could very well be that asthma of today’s children and youth is the result of mechanisms different from those relevant for asthma development 50 years ago. However, when associations are found consistently across time, this could point to very basic and timeless asthma mechanisms. This is nothing but a hypothesis—but one to consider when indulging in replication studies such as this that look simple at first glance but may hold more food for thought than expected.



  • Competing interests: None.

Linked Articles