Material and methods

PATIENTS AND FAMILIES

Fifty five coeliac disease families containing 84 affected sibships were studied. All were of northern European ancestry. The diagnosis of coeliac disease was established in all cases by demonstration of severe enteropathy with villous atrophy on small bowel biopsy. All made a symptomatic recovery and the majority of patients had a follow up biopsy showing mucosal recovery on a gluten free diet. All affected family members were typed for HLA-DQ alleles by PCR-SSP (details from O Olerup, personal communication).

MARKER TYPING

DNA was salt extracted from EDTA venous blood samples. Subjects were genotyped at the following loci: D6S260, D6S273, and D6S426. A total of 12.5 ng of genomic DNA was amplified using PCR in 10 μl of 75 mmol/l Tris-HCL (pH 9), 0.01% (v/v) Tween, 1.5 mmol/l MgCl₂, 0.25 U Taq polymerase, 20 mmol/l (NH₄)₂SO₄, 0.1% (w/v) BSA, 0.2 mmol/l each dNTP, and each fluorescently labelled primer at 5 ng μl^-1. Thirty cycles of the PCR were carried out at 94°C for one minute, 55-50°C for one minute, and 72°C for one minute with a final extension step at 72°C for 10 minutes. The products were electrophoresed through 6% denaturing polyacrylamide gels and were sized by an automatic ABI 377 (PRE-ABI), using the programs GENESCAN and GENOTYPER to allocate alleles.

Marker allele frequencies were estimated from observed relative frequencies in founder subjects. (These were identical to the relative frequencies seen across all typed subjects.) Distances between markers were obtained from the Genethon (http://www.cephb.fr/ceph-genethon-map.html) database.

STATISTICAL ANALYSIS

Linkage analysis of the data was undertaken in accordance with the maximum likelihood score (MLS) approach implemented in the program MAPMAKER-SIBS.4 This program calculates, for a sample of affected sib pairs, the MLS and the corresponding maximum likelihood estimate of the IBD vector. To assess the contribution of the MHC to sibs with coeliac disease, haplotype sharing among affected sib pairs was assessed with the “all pairs” option. The allele sharing probabilities of affected sib pairs (ASPs) depend on the contribution any gene makes to the genetic variation of the trait. The 2, 1, and 0 sharing probabilities are given by: Z₀=α₀(1/λ_s), Z₁=α₁(λ_o/λ_s), Z₂=α₂(λ_mz /λs),5where Z_i=P (sibs share i marker alleles IBD‖ASP) and α_i=P (sibs share i marker alleles IBD). For i=0, 1, 2, α_i equals 1/4, 1/2, and 1/4 respectively. When a marker is unlinked, Z_i=α_i, λ_s=K_s/K, λ_o=K_o/K, and λ_mz=K_mz/K where K_s, K_o, and K_mz are the sib, offspring, and monozygotic recurrence risks respectively, and K is the population risk of the disease. These formulae hold true irrespective of the mode of inheritance at the disease locus, the number of alleles, and their frequencies, penetrance, and population prevalence.5 Using the allele sharing probabilities among ASPs the relative risks of coeliac disease associated with the MHC can be determined. To compare the haplotype sharing probabilities observed in this study with published data, values of Z₂ and Z₁ were combined. Heterogeneity is then given by: X² _overall-(Σt_i)²/n, where i refers to individual study ASP probabilities, n the number of studies in which ASP probabilities have been determined, and t_i is X_i ².

Results

All of the affected subjects in the families studied had HLA class II genotypes compatible with possession of the DQ(α1*0501,β1*0201) alleles. The marker D6S273 lies within the HLA system and D6S260 and D6S426 lie 15 cM upstream and downstream of D6S273 respectively. Table1 shows the lod scores and allele sharing probabilities for the three markers for all the families studied. As expected, there is evidence for excess sharing at HLA and at the two close flanking markers. Significant linkage was seen across the whole region, with a maximum lod score of 3.06 between markers D6S273 and D6S426. Based upon the allele sharing probabilities at D6S273, the sib relative risk attributable to the MHC region is 3.6.

Estimates of the prevalence of coeliac disease in Europe ranged from a high of 1 in 300 in western Ireland to between 1 in 1000 and 1 in 2000 in other regions.6-8 The true prevalence of coeliac disease is likely to have been underestimated because a substantial number of subjects are asymptomatic or have only mild symptoms. Studies using antibodies for screening have shown that the prevalence of coeliac disease among some symptom free subjects is as high as 1 in 200.9-13 Based on a prevalence of 0.005, a 10% sib recurrence risk equates to a relative risk of 20.

The contribution of the MHC region to this relative risk depends on the model of interaction between the HLA linked and unlinked loci. They could interact either additively (that is, the penetrance of the disease is represented by the sum of the penetrances contributed by two or more loci) or multiplicatively (that is, the penetrance of the disease is the product of the penetrances contributed by two or more loci). The familial risks seen in sibs and monozygotic twins are most parsimonious with a multiplicative model, since a simple additive model would violate the mathematical relationship between sib, parent-offspring, and monozygotic twin relative risks. (The relative risk in monozygotic twins, λmz, is given by: 4λs-2λpo-15.) Under the assumption of a multiplicative model, the contribution of the MHC region to the overall familial risk of coeliac disease (given by: lnλ_MHC/lnλ_s) will be 44%. A number of other studies have also examined the sharing across the HLA region in sibships with coeliac disease.14-25 In all but one of these, sharing between ASPs has been based on determination of HLA status. Table 2 details the allele sharing probabilities and derived sib relative risks associated with the HLA linked locus in the published studies. There is no evidence for heterogeneity between these studies (χ²=11.1, 16 df, p>0.1) so allowing data to be combined for a more precise estimate of the familial risk of coeliac disease associated with the MHC. Using the pooled data, the estimated sib relative risk of coeliac disease associated with the MHC is 3.3 (95% CI 2.3–4.9). Based on this estimate, the relative risk associated with the MHC region does not account for more than 39% of the familial risk of coeliac disease.

Discussion

The mode of inheritance of coeliac disease is unknown. Using family data Pena et al 30proposed the involvement of two distinct unlinked genes in the aetiology of coeliac disease. A prerequisite for developing coeliac disease was homozygosity at the HLA unlinked locus and participation of an independently inherited gene or genes located in or tightly linked to the major histocompatibility system acting in a dominant fashion. This proposal was supported by some but not all other studies.27 31 Most agreed on recessivity at the HLA unlinked locus but differed with respect to dominance or recessivity at the HLA linked disease susceptibility locus. It is clear, however, that since expression of the DQ(α1*0501,β1*0201) heterodimer is responsible for the HLA association, constitutive haplotypes will not behave in a simple mendelian recessive or dominant fashion. It is possible that part of the familial risk may be environmentally determined. The concordance rates between monozygotic twins and HLA identical sibs provide considerable support for the hypothesis that the familial risk of coeliac disease is genetically determined. The only caveat of this is the assumption that the environmental correlation between these two types of sibs is similar.

On the assumption that the familial risk of coeliac disease is genetically determined, we previously estimated the sib relative risk associated with the HLA linked genes to be 1.8 under a dominant model, 4.6 under a recessive model, and 4.0 under a codominant model.2 These estimates are, however, indirect ones derived from the frequency and genotypic risk of the DQ(α1*0501,β1*0201) heterodimer. In this present study the impact of the MHC region on the familial risk of coeliac disease was derived from the haplotype sharing probabilities. Based on these probabilities and those published in other studies, the familial risk associated with the MHC region is less than 4. Given that the risk of coeliac disease in sibs is increased by around 20-fold, the MHC region will not account for more than two fifths of the familial risk and the second gene or genes, unlinked to HLA, are likely to be the stronger determinant of disease susceptibility than the HLA linked locus.