1. ¹1st Department of Medicine, Christian-Albrechts-University, Kiel, Germany
2. ²Institute for Molecular Biotechnology, Beutenberg Strasse 11, 07745 Jena, Germany
3. ³Department of Internal Medicine, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea
4. ⁴Institue for Medical Informatics and Statistics, Christian-Albrechts-University, Kiel, Germany
5. ⁵Department of Internal Medicine, Yonsei University College of Medicine, Seoul, Korea

Correspondence: Professor Stefan Schreiber, 1st Department of Medicine, Christian-Albrechts-Universität Kiel, Schittenhelmstrasse 12, D-24105 Kiel, Germany. Tel: +49-431-597-2350; Fax: +49-431-597-1842; E-mail: S.Schreiber@mucosa.de

Received 19 April 2002; Revised 22 August 2002; Accepted 26 August 2002.

Abstract

Current debate focuses on the relevance of linkage disequilibrium (LD), ethnicity and underlying haplotype structure to the search for genes involved in complex disorders. The recently described association between single nucleotide polymorphisms (SNPs) of the CARD15 (NOD2) gene and Crohn's disease (CD) in populations of north-European descent provides a test case that we have subjected to detailed SNP haplotype based analyses. We examined 23 SNPs spanning 290 kb, including CARD15, in large North-European and Korean samples of patients with Crohn's disease and normal controls. In Europeans we confirmed that the three disease-associated SNPs occur independently but share a common background haplotype. This suggests a common origin and the possibility of an undiscovered more strongly predisposing mutation. Korean CD patients present a phenotype identical to the European patients and have not previously been screened for CARD15. The three disease-associated SNPs were absent and there was no evidence of association between CARD15 and CD. Consequently, the disease-associated mutations in the Europeans, which are rare, have arisen recently (after the Asian–European split). Our results highlight important issues relevant to mapping the genes that predispose to complex disorders. First, although ethnically divergent populations may present identical phenotypes they do not necessarily share the same set of predisposing genes. Second, although single-locus tests of association showed consistent association with markers throughout the gene, pair-wise LD between markers (r² and D') yielded very little information about actual disease-association. Third, a population comparative approach allowed refining of the marker set through the examination of shared polymorphisms and common LD-groups. This approach, in conjunction with the examination of the mutational steps in a haplotype network, allows unambiguous identification of the potentially causative mutations.