HLA and disease: haplotype sharing in multiplex families

Recently it has been argued that the distribution of shared haplotypes in multiplex sibships for HLA-associated diseases may be an indicator of the disorder's underlying mode of transmission. Specifically, it has been suggested that the presence of multiple disease susceptibility genes and/or loci may be inferred when an inverse relationship between the amount of haplotype sharing and the number of affected sibs is observed in families where neither parent is affected. This claim is evaluated using extensive computer simulations. It is shown that a variety of haplotype sharing patterns are possible, even for the simplest models, and that for a large segment of the parameter space the actual distribution of shared haplotypes is opposite to that predicted. Accordingly, the inference that more than one locus is involved in the etiology of an HLA-associated non-Mendelian disease, if based only on the distribution of shared haplotypes in multiplex sibships, is unjustified.