Info

aIn 187 multicase and 362 single-case families (containing 808 affected and 1574 unaffected subjects); DRX = not DR15.

bIn 937 sporadic cases and 739 controls; DRX = not DR15 or DR17. Abbreviations: OR, odds ratio; CI, confidence interval.

aIn 187 multicase and 362 single-case families (containing 808 affected and 1574 unaffected subjects); DRX = not DR15.

bIn 937 sporadic cases and 739 controls; DRX = not DR15 or DR17. Abbreviations: OR, odds ratio; CI, confidence interval.

Japan, of the DPB1*0501 allele with the opticospinal form of MS (38) implicates HLA in the determination of the anatomical distribution of demyelinating lesions. Whatever the model, it has been estimated that the HLA region accounts for no more than 15% to 50% of the total genetic risk in MS (39).

Association

Association studies are hampered, from the outset, by the difficulty in selecting appropriate candidates from a genome comprised of over 30,000 genes. In general, the genes that have been studied by association analysis in MS have been either functional candidates, chosen on account of the presumed role of the molecules they encode in, e.g., the sequence of immune-cell interactions that culminates in inflammatory demyelinaton; or positional ones, chosen on account of their location in regions positive for linkage in genome-wide screens. Among the dozens of non-HLA candidates, both functional and positional, studied thus far in case-control datasets in MS (40), none has been consistently shown to be associated with the risk of developing MS. Typically, an initial report of association, published in a high-impact journal, is followed by a less publicized train of negative reports, as has been the case with studies of the genes encoding myelin basic protein (41-44) and CD45 (45-48), as well as a score of others (49). This phenomenon has also plagued association studies in other genetically complex disorders (50).

In a recent review, Colhoun et al. (51) identified three main reasons for this pervasive inability to replicate genetic associations: the failure to attribute findings to chance, publication bias, and inadequate sample size. The authors propose the following remedies for these problems: a more stringent threshold for the declaration of statistical evidence of association—specifically, a reduction of the probability value indicating significance from the traditional 5 x 10~2 to the more Bayesian 5 x 10~5; internet-based reporting of the results of negative studies; and, for replication studies, sample sizes large enough to detect or exclude effects somewhat smaller than those reported in previous positive studies. They also remark that the prior probability of association will increase when candidate genes are chosen based on functional and positional information, and when the frequencies of common haplotypes are preferentially investigated.

The last two recommendations have been taken to heart in two recent studies of candidate genes in MS. Barcellos et al. (52) investigated, by family-based association analysis, 47 single-nucleotide polymorphisms (SNPs) located in 34 genes encoding proteins involved in inflammatory pathways in two independent datasets of American MS patients and controls. Association in an initial dataset between an SNP in exon 10 of the NOS2A gene on chromosome 17q11 and the risk of MS was confirmed in a second dataset, and subsequent analysis of common haplotypes containing this SNP further corroborated the association. NOS2A was selected as a plausible functional candidate in this study; it encodes the inducible isoform of nitric-oxide synthase, an enzyme that might contribute to MS pathogenesis through the production of neurotoxic oxygen radicals (53,54). In a second study, with a similar multi-stage design, Zhang et al. (55) genotyped 123 SNPs in 66 genes selected on the basis either of their location in regions linked to MS or other autoimmune diseases, or of the potential functional role of their protein products in MS. They ultimately identified one predispositional and one protective five-SNP haplotype spanning the IL7R gene on chromosome 5p13. 1L7R encodes the interleukin 7 receptor; signaling via the receptor induces somatic recombination of the T cell-receptor and immunoglobulin genes, which in turn promotes the proliferation and survival of T and B lymphocytes (56).

Despite the appeal of the Bayesian approaches—whereby the careful selection of candidates increases the prior probability of a true association—recent advances in molecular biology, including the sequencing of the human genome and the development of high-throughput genotyping assays, have made possible a novel, anti-Bayesian approach to complex-disorder gene mapping: the genome-wide association study.

0 0

Post a comment