The Extent of Linkage Disequilibrium in Four Populations with Distinct Demographic Histories

Alison M. Dunning,1,* Francine Durocher,2,* Catherine S. Healey,1 M. Dawn Teare,2 Simon E. McBride,1 Francesca Carlomagno,1, Chun-Fang Xu,4 Elisabeth Dawson,5 Susan Rhodes,5 Saeko Ueda,1 Eric Lai,6 Robert N. Luben,3 Elizabeth J. Van Rensburg,7 Arto Mannermaa,8 Vesa Kataja,9 Gadi Rennart,10 Ian Dunham,5 Ian Purvis,4 Douglas Easton,2 and Bruce A. J. Ponder1

1CRC Department of Oncology, 2CRC Genetic Epidemiology Group, and 3EPIC, University of Cambridge, 4U.K. Molecular Genetics, GlaxoWellcome Medicines Research Centre; and 5Sanger Centre, Cambridge; 6U.S. Discovery Genetics, GlaxoWellcome, Inc., Research Triangle Park, NC; 7Department of Human Genetics, University of Pretoria, Pretoria, South Africa; 8Department of Clinical Genetics and 9Department of Oncology and Radiotherapy, Kuopio University Hospital, Kuopio, Finland; and 10Department of Community Medicine and Epidemiology, Carmel Medical Centre and Technion Faculty of Medicine, Haifa, Israel

Received August 21, 2000; accepted for publication October 18, 2000; electronically published November 14, 2000.

The design and feasibility of whole-genomeassociation studies are critically dependent on the extent of linkage disequilibrium (LD) between markers. Although there has been extensive theoretical discussion of this, few empirical data exist. The authors have determined the extent of LD among 38 biallelic markers with minor allele frequencies >.1, since these are most comparable to the common disease-susceptibility polymorphisms that association studies aim to detect. The markers come from three chromosomal regions1,335 kb on chromosome 13q12-13, 380 kb on chromosome 19q13.2, and 120 kb on chromosome 22q13.3which have been extensively mapped. These markers were examined in 1,600 individuals from four populations, all of European origin but with different demographic histories; Afrikaners, Ashkenazim, Finns, and East Anglian British. There are few differences, either in allele frequencies or in LD, among the populations studied. A similar inverse relationship was found between LD and distance in each genomic region and in each population. Mean D is .68 for marker pairs <5 kb apart and is .24 for pairs separated by 1020 kb, and the level of LD is not different from that seen in unlinked marker pairs separated by >500 kb. However, only 50% of marker pairs at distances <5 kb display sufficient LD ( > .3) to be useful in association studies. Results of the present study, if representative of the whole genome, suggest that a whole-genome scan searching for common disease-susceptibility alleles would require markers spaced 5 kb apart.

*The first two authors contributed equally to this work.
**Present affiliation: Dipartimento di Biologia e Patologia Cellulare e Moleculare, Facolta' di Medicina e Chirurgia, Universita' degli Studi di Napoli "Frederico II", Naples, Italy.