Relation of risk of systemic lupus erythematosus to west African admixture in a Caribbean population

Abstract

Risk of systemic lupus erythematosus (SLE) is higher in people of west African descent than in Europeans. The objective of this study was to distinguish between genetic and environmental explanations for this ethnic difference by examining the relationship of disease risk to individual admixture (defined as the proportion of the genome that is of west African ancestry); 124 cases of SLE and 219 matched controls resident in Trinidad were studied. Analysis of admixture was restricted to 52 cases and 107 controls who reported no Indian or Chinese ancestry. These individuals were typed with a panel of 26 single-nucleotide polymorphisms and five insertion/deletion polymorphisms chosen to have large allele frequency differentials between west African, European and Native American populations. A Bayesian model for population admixture, individual admixture and locus ancestry was fitted by Markov chain simulation. Mean west African admixture (M) was 0.81 in cases and 0.74 in controls (P=0.01). The risk ratio for SLE associated with unit change in M was estimated as 32.5 with a 95% confidence interval (CI) of 2.0–518. Adjustment for measures of socioeconomic status (household amenities in childhood and years of education) altered this risk ratio only slightly (adjusted risk ratio: 28.4, 95% CI 1.7–485). These results support an additive genetic model for the ethnic difference in risk of SLE between west Africans and Europeans, rather than an environmental explanation or an "overdominant" model in which risk is higher in heterozygous than in homozygous individuals. This conclusion lays a basis for localizing the genes underlying this ethnic difference in risk of SLE by admixture mapping.

Appendix

Marker information content for ancestry (f)

We shall consider a haploid population formed by equal admixture between two populations X and Y. The ancestry at a marker locus in a randomly chosen haploid genome can be modelled as a Bernoulli random variable with parameter ½. The prior variance of locus ancestry (before typing) is then ¼. If the marker is informative for ancestry, the posterior variance of locus ancestry (after typing the locus) will be less than the prior variance. The marker information content for ancestry f is defined as

1 − (the expected posterior variance of locus ancestry/the prior variance of locus ancestry).

For a locus with k alleles and allele frequencies p_iX and p_iY

$$f = 1 - 2\sum\limits_{i = 1}^k {{{p_{iX} p_{iY} } \over {p_{iX} + p_{iY} }}} $$

For a biallelic locus, f is equal to Wahlund's standardized variance of allele frequencies, defined as

$${{{\rm{H}}_{\rm{T}} - {\rm{H}}_{\rm{S}} } \over {{\rm{H}}_{\rm{T}} }}$$

where H_T is the heterozygosity of a total population formed by pooling equal numbers from the two subpopulations, and H_S is the mean heterozygosity of the two subpopulations.