Small fish in a big pond.

I has been known for some time that many rheumatic and autoimmune conditions are complex in nature, i.e. they have both environmental and multiple genetic aetiological components. The dissection of complex traits has previously been difficult as existing analytical approaches, designed for single-gene disorders, are largely inappropriate. It is now possible to apply new techniques to unravel complex genetic conditions and define them at the DNA sequence level. This has largely resulted from the application of advances in molecular biology, information technology and the construction of detailed genetic linkage maps for the human genome. The rewards for success are potentially great and could include the identification of novel therapies, the ability to refine diagnosis and to target treatment more effectively, and the prediction and quantification of disease risk. The new approach, often called Whole Genome Screening (WGS), examines the human genome for areas linked with disease susceptibility, starting at one end and finishing at the other. By using 0350 highly polymorphic markers, the genome can be divided into sections of around 10 cM. Each can be successively tested for linkage with disease, in nuclear families with at least two affected siblings. Under Mendelian inheritance, for any marker locus with multiple alleles, siblings are expected to share a given parental allele 50% of the time. A significant distortion towards greater sharing of alleles in affected siblings is indicative of genetic linkage with the disease. Large numbers of affected sibling pair (ASP) families are required to detect genes making a relatively weak contribution and this strategy has been dependent on the development of automated high-throughput genotyping [1]. It has already been highly successful in identifying regions of linkage in insulin-dependent diabetes (IDDM) [2, 3] and other conditions [4, 5]. Less enlightened critics have described such ‘exclusion mapping’ as being a ‘fishing expedition’, inferring that a level of serendipity is involved. In one respect, this is completely unfounded as the hypothesis driving this highly systematic approach is that multiple susceptibility genes exist within the genome for a given disease. However, in another respect, this is fishing for genes as, at present, no alternative strategy exists for identifying genes unless there is an a priori rationale for their implication in disease aetiopathogenesis. ASP-based approaches offer a number of advantages over traditional linkage analysis. Firstly, they are not based on fitting a genetic model; thus, not knowing the number of genes or mode of inheritance does not present a problem. Secondly, analysis is restricted to affected individuals; this means that only penetrant genes are being examined and any misassignment of unaffected family members who are carrying the disease allele can be ignored. Rheumatology provides a rich area for those studying complex conditions and many arthritic diseases have a significant oligogenic contribution to susceptibility. Major ASP WGS analyses are under way to dissect the genetics of rheumatoid arthritis (RA), osteoarthritis, ankylosing spondylitis, juvenile chronic arthritis and systemic lupus erythematosus. Preliminary results have been reported for both the French and UK RA WGS studies [6, 7]. In both countries over 200 ASP families have been genotyped with markers at an average genomic spacing of 11 cM. Although analysis of the first 100 families in both studies gave some suggestions of linkage, with the exception of the HLA locus on chromosome 6, these were lost when data from the second 100 families were incorporated. Why should this be and what lessons can be learned? At the 5% level of significance, approximately five false-positive linkage results would be expected for every 100 comparisons made. Thus, over 15 false positives would be predicted for the 300 or so markers used in a WGS. Analysis of a second data set of the same size will only rarely give the same false positive again and linkages can thus be critically re-evaluated. Absolute confirmation, however, requires a larger data set, the size of which depends on the number of hints of linkage detected and number of ASPs in the initial screen [8]. A further point that these studies demonstrate is the close relationship between the number of families analysed and the size of genetic effect which can be detected. Geneticists have found it useful to quantify the genetic component of disease susceptibility by calculating the coefficient of familial clustering (ls). This is calculated as a ratio of the disease risk for siblings of affected cases to the population prevalence. Estimates of ls for RA range from 5 to 10. As individual susceptibility loci are identified, their contribution can also be estimated and this can be expressed as a proportion of the total familial component. In RA, the l for HLA is calculated as being 1.8 [9] and this suggests that HLA accounts for approximately one-third of the genetic susceptibility. It has been predicted that there are at least two loci in addition to HLA and it is, therefore, highly likely that the remaining RA susceptibility loci each contribute a small effect. Added to this is the problem that complex arthritic conditions such as RA exhibit considerable clinical heterogeneity. This heterogeneity will undoubtedly