Genetic Analysis of Complex Diseases

Neil Risch, in a series of seminal papers in 1990 (1, 2), demonstrated the utility of sibpair linkage analysis in identifying genes for complex genetic traits. In doing so, he defined what is the current paradigm for the genetic dissection of common complex genetic diseases. The recent Perspective by Risch and Kathleen Merikangas (3) again shapes the future of human disease gene mapping by defining what will undoubtedly become the statistical “state of the art.” Risch and Merikangas advocate conducting genomic screens based on association studies of candidate genes using the transmission disequilibrium test (TDT). It is important, however, not to infer from their arguments that current linkage analysis methods cannot detect most genes underlying complex disease. Risch and Merikangas extend the current paradigm to include anticipated technological advances. However, the magnitude of g (defined as the relative risk in the heterozygote) in complex traits and the application of this approach using current molecular technology must be considered. In their formulation, Risch and Merikangas show that a TDT approach is more powerful than a sib pair approach, particularly for disease alleles with small genetic effects. This conclusion is based on the sample size required to detect a gene with a g # 4. They show that, while sib pair analysis requires a practical (for example, 100 to 400) number of sib pairs to detect a gene with g 5 4 and a disease allele frequency p, between 0.1 and 0.5, the number of sib pairs required becomes impractical (for example, more than 1000) when g # 2. Previously, Risch (1, 2) established the use of a sibling recurrence risk ratio (ls) to estimate the power of a sib pair design to detect linkage. Estimable from epidemiologic data, ls is calculated as the ratio of the recurrence risk in siblings of an affected individual and the population prevalence of the disorder. This ls represents the overall recurrence risk ratio, which may result from the actions of a single gene or multiple genes acting additively or epistatically. If a number of genes are hypothesized, the genespecific ls (referred to here as lgs) may be estimated by assuming a model (additive or epistatic), the number of genes, and partitioning ls accordingly. Many researchers are accustomed to evaluating the magnitude of genetic effects by using ls rather than g. We calculated lgs corresponding to g # 2 and p 5 0.01, 0.10, 0.5, and 0.8, respectively. The results indicated that for g # 2, lgs , 1.3. It has previously been shown that genes with lgs , 1.3 would be difficult to detect using sib pair methods (2, 4). The curve comparing lgs with g shows that even genes with moderate effect (for example, lgs , 2) may produce g that can be detected by linkage analysis in reasonably sized samples of sib pairs. These results indicate that linkage analysis of complex disease based on genomic screens using current microsatellite markers can be a fruitful enterprise in complex genetic diseases. An excellent example is the discovery of the late onset Alzheimer’s disease susceptibility gene APOE. Using Risch and Merikangas’s formulas, we calculated the number of sib pairs that would have been necessary to detect the effect of APOE on the risk of AD. The g in individuals heterozygous for APOE-4 is 4.5 (5) and the frequency of APOE-4 in the general population is about 15% (6); the resulting probability of allele sharing is Y 5 0.625, and the minimum number of affected sib pairs required to detect linkage is 164. Alzheimer’s disease has an overall ls of 5, with a lgs of only 2 for APOE (7). Other complex diseases, such as multiple sclerosis (ls 5 30) (8) and autism (ls 5 150) (9) have substantial genetic components. Even if there are 10 epistatic genes of equal multiplicative effect underlying multiple sclerosis (ls 5 30; lgs 5 1.4), linkage analysis should be able to detect them. Because it is difficult to determine a priori which disease alleles have minor or moderate genetic effects, linkage analysis should not be arbitrarily abandoned. Risch and Merikangas point out that genomic screening of candidate genes is several years from becoming reality. When the molecular resources become available, the advantages of genomic screening using TDT, such as increasing power to detect minor genetic effects, allowing the use of singleton cases, and testing effects of functional polymorphisms in genes, will make this the method of choice. Until then, well-designed linkage studies of complex traits will still be able to detect genes of major or moderate effect. William K. Scott Margaret A. Pericak-Vance Section of Medical Genetics, Department of Medicine, Duke University Medical Center, Durham, NC 27710, USA Jonathan L. Haines Molecular Neurogenetics Unit, Massachusetts General Hospital, Charlestown, MA 02129, USA REFERENCES AND NOTES ___________________________