| Human Molecular Genetics | Pages |
©1999 Oxford University Press |
Combined sib-TDT and TDT provide evidence for linkage of the interleukin-1 gene cluster to erosive rheumatoid arthritis
Introduction
Results
Discussion
Materials And Methods
Subjects
Markers and genotyping
Data analysis
Acknowledgements
References
Combined sib-TDT and TDT provide evidence for linkage of the interleukin-1 gene cluster to erosive rheumatoid arthritis
Received April 6, 1999; Revised and Accepted June 30, 1999
Rheumatoid arthritis (RA) is a common disease of unknown aetiology which usually causes progressive destruction of the joints. Familial aggregation, twin studies and segregation analyses suggest that there is a genetic component to RA and the HLA-DRB1 locus in the major histocompatibility complex on chromosome 6 has been shown to be linked to, and associated with, RA susceptibility. It is likely that other genes with weaker effects are also involved, which may be difficult to detect using conventional parametric and non-parametric linkage methods. We have implemented the combined sib-TDT and TDT, in addition to parametric and non-parametric linkage methods, to investigate the candidate genes of the interleukin-1 (IL-1) gene cluster on chromosome region 2q13, since IL-1 is an important cytokine in the control of the inflammatory response that is central to RA pathology. Several tightly linked IL-1 cluster markers yielded suggestive evidence for linkage in the combined TDT in those families in which affected siblings did not share two HLA-DRB1 alleles identical by descent. The evidence was significant in those with severe disease, as assessed by the presence of bone erosions. In contrast, there was no evidence of linkage using non-parametric linkage analysis, but parametric analysis revealed weak evidence of linkage when marker-trait disequilibrium was incorporated into the analysis. The data provide preliminary evidence for linkage of genes of the IL-1 cluster to RA and suggest a possible role for this region in severe erosive disease.
INTRODUCTION
Many autoimmune and inflammatory disorders are multifactorial diseases in which there is a single major gene (often in the major histocompatibility complex) and in addition many minor genes. In insulin-dependent diabetes mellitus, for example, there are ~15 candidate minor genes which are at present being identified and characterized (1). Since these minor loci contribute little to the overall familial clustering of the disease, they are difficult to detect by sib pair linkage analysis and preliminary positive results in genome-wide scans have proved hard to replicate (2,3). The minor genes may have weak effects individually, but may nevertheless be important in combination with other genes, due to epistasis or other interactions. Alternatively, they may have weak effects in the overall population but may be very important in particular subsets of families. Therefore, identification of these genes makes an important contribution to our understanding of disease aetiology.
Methods that rely on the close physical proximity (or identity) of disease genes and markers (and thus allelic association between them) can be more sensitive than traditional allele sharing linkage analysis methods and are arguably more useful for detecting genes of weak effect. At present, association-based methods can be applied in a candidate gene or candidate region approach, but in the future, when sufficient biallelic markers (SNPs) are available, whole-genome screening may become feasible using this approach (4,5). One of the most robust and powerful of these methods is the transmission/disequilibrium test (TDT), which has recently been extended to families in which parents are unavailable (the sib-TDT) (6). The sib-TDT, like the TDT, is a test for linkage in the presence of an association and avoids the pitfalls of population stratification inherent in population-based association studies (6,7). The sib-TDT and related methods are thus likely to become important with regard to late onset diseases such as rheumatoid arthritis (RA).
RA is an example of a multifactorial inflammatory disease with a relatively low degree of familial clustering ([lambda]s = 2-8) (8,9) and it is well established that genes of the MHC, closely linked to HLA-DRB1 or HLA-DRB1 itself, account for >30% of the genetic component of the disease (10). Preliminary results from genome scans indicate that there is at least one more disease susceptibility locus and it is likely that there will be several minor genes that will contribute to it (11,12). One of the most important candidates in RA (together with tumor necrosis factor) is interleukin-1 (IL-1), which is secreted by activated macrophages in the inflamed synovium and initiates the recruitment of immune cells and the progression of inflammation (reviewed in ref. 13). In some animal models of RA, IL-1 is required for development of the disease (14,15) and IL-1 also stimulates bone resorption (16,17). With this spectrum of activities, it is plausible that IL-1 may be associated with either disease susceptibility and/or severity and there is some evidence that it may be a severity factor in other autoimmune diseases (18). The IL-1 system consists of two agonist molecules, IL-1[alpha] and [beta], and the structurally related IL-1 receptor antagonist; competition between these molecules for receptor binding sites in part determines the overall level of IL-1 activity. The three IL-1 genes are clustered within 430 kb on the long arm of chromosome 2 (19). Thus, disease associations with the IL-1 genes may be complex, resulting from a combination of functional interactions at the protein level and linkage disequilibrium between the genes (20).
We have used the combined sib-TDT and TDT, in conjunction with parametric and non-parametric linkage analysis, to investigate the role of the IL-1 gene cluster in RA and, in particular, in the erosive form of the disease.
RESULTS
We carried out a preliminary scan of the IL-1 region for linkage/association to the RA phenotype, in multiplex RA families from the Arthritis Research Campaign's National Repository. A total of 195 nuclear families were genotyped for five microsatellite and four single nucleotide polymorphic markers in the IL-1 gene cluster; the same families had been previously genotyped for HLA-DRB1 (21). The physical marker map of the cluster is shown in Figure 1. Z scores for the combined TDT were determined for the three most common alleles of each multiallelic marker and the rare alleles of the biallelic markers in the overall dataset and after stratification according to the HLA-DRB1 allele sharing status of the affected sibling pairs (Materials and Methods; Table 1). Whilst no single allele tested yielded a Z score that was significant at the threshold of 0.001 (taking into account correction for multiple tests; see Materials and Methods), there was evidence of linkage to the RA phenotype, as judged by the presence of several nominally significant results distributed across the gene cluster. Seven of the Z scores corresponded to P-values of <0.05 and six were <0.02, (i.e. indicating suggestive linkage evidence; see Materials and Methods). Five of the latter were in the HLA-DRB1 (1,0) dataset. In contrast, there was no evidence of transmission distortion of any of the alleles in the 89 HLA-DRB1 (2) families; in fact in these families there was a tendency in most cases towards bias of the relevant alleles in the opposite direction to that in the HLA-DRB1 (1,0) families (Table 1). We concluded from this preliminary scan of the IL-1 gene cluster that there was suggestive evidence of linkage of the IL-1 region to RA, which was most evident in those families not linked to HLA-DRB1.
Figure 1. Polymorphic markers and genes in the interleukin-1 gene cluster.
Table 1. Z scores and P-values from the combined TDT for the RA phenotype
| Marker | Allele | Overall data set | HLA-DRB1 (1,0) | HLA-DRB1 (2) | ||||||
| Z score | Units | P-value | Z score | Units | P-value | Z score | Units | P-value | ||
| D2S160 | 4 | 2.6348 | 71 | 0.004 | 2.1374 | 34 | 0.016 | 1.4169 | 25 | NS |
| 222/223 | 3 | 1.3100 | 104 | NS | 2.1133 | 42 | 0.017 | -0.4015 | 45 | NS |
| +4845 | 2 | 0.3882 | 97 | NS | 0.7178 | 40 | NS | -0.2713 | 44 | NS |
| gz5/gz6 | 3 | 0.1524 | 88 | NS | 2.2715 | 37 | 0.012 | -0.7501 | 41 | NS |
| +3954 | 1 | 0.7493 | 77 | NS | 0.0205 | 32 | NS | 0.6371 | 36 | NS |
| -511 | 1 | 1.6036 | 110 | 0.054 | 2.7215 | 34 | 0.003 | 0.1509 | 62 | NS |
| gaat | 1 | 2.0454 | 102 | 0.021 | 2.7993 | 39 | 0.002 | -0.0159 | 46 | NS |
| y31 | 6 | 1.0879 | 35 | NS | 1.2685 | 12 | NS | -0.0115 | 16 | NS |
| +2018 | 2 | 0.0014 | 87 | NS | -0.3061 | 31 | NS | 0.8774 | 41 | NS |
We hypothesized from the biological activities of IL-1 that one of the most likely roles of the IL-1 genes in RA would be in bone erosion and thus the apparent linkage to RA in the HLA-DRB1 (1,0) families could be due to those individuals with erosive disease. Table 2 shows that the majority of individuals for whom data on erosions were available had erosive disease in all three data sets. We carried out the combined TDT specifically on the erosive phenotype for the five marker alleles with P-values <0.05 in the HLA-DRB1 (1,0) data set: D2S160 allele 4, 222/223 allele 3, gz5/gz6 allele 3, -511 allele 1 and gaat allele 1. In this analysis, individuals with RA but with no erosions were classed as `unaffected' for the purposes of the sib-TDT and individuals without RA were classed as `unknowns' and thus did not contribute to the analysis. This coding was used in an effort to separate the erosive phenotype from the RA phenotype. Individuals without RA were classed as unknown rather than unaffected since they may possess putative erosion susceptibility alleles that would not be expressed in the absence of RA, thus reducing the power for the analysis of erosions. Four out of five of the alleles previously identified are over-transmitted in this erosive group (P < 0.05), despite the smaller number of units available (Table 3). The result for gaat allele 1 (P < 0.0008) is significant at the study-wide threshold of P = 0.001. Thus these data support linkage of IL-1 to erosive RA, but we cannot exclude linkage to non-erosive RA, since insufficient non-erosive individuals were available in this data set.
Table 2. Affectation status for RA and erosions amongst non-founder individuals
| Phenotypea | No. of individuals | ||
| Overall data set | HLA-DRB1 (1,0) | HLA-DRB1 (2) | |
| RA+ and erosion+ | 251 (83.4%) | 91 (74.6%) | 132 (89.8%) |
| RA+ and erosion- | 50 (16.6%) | 31 (25.4%) | 15 (10.2%) |
| RA+ & erosion 0 | 97 | 45 | 39 |
| RA- | 176 | 68 | 70 |
| RA 0 | 2 | 0 | 1 |
| Total | 576 | 235 | 257 |
Table 3. Linkage/association of IL1 markers with erosive RA in HLA-DRB1 (1,0) families
| Marker | Allele | Z score | P-value | Units |
| D2S160 | 4 | 1.1917 | 0.117 | 13 |
| 222/223 | 3 | 2.3497 | 0.009 | 16 |
| gz5/gz6 | 3 | 2.2792 | 0.011 | 16 |
| -511 | 1 | 1.6064 | 0.048 | 17 |
| gaat | 1 | 3.1614 | 0.0008 | 19 |
We used two methods which take different approaches to try to identify disease haplotypes. These were the 3-marker haplotype method described by Merriman et al. (22) and the TRANSMIT algorithm of Clayton (D. Clayton, manuscript in preparation). Use of the Merriman method (Materials and Methods) identified a disease haplotype consisting of the alleles 3 2 3 2 1 1 at the markers 222/223, IL1A+4845, gz5/gz6, IL1B+3954, IL1B-511 and gaat, respectively. The TRANSMIT program identified both this and another haplotype (4 1 4 1 1 1) as being over-transmitted to diseased individuals (P = 0.08 and P = 0.01, respectively).
We also investigated whether linkage could be detected by use of conventional parametric and non-parametric linkage analysis in the HLA-DRB1 (1,0) families. There was no evidence of linkage using the multipoint non-parametric option of GENEHUNTER (23); the NPL scores for the overall set of families and the two HLA-DRB1 subsets were all negative, despite the high information content of the region (Materials and Methods). Using the two multiallelic markers that had yielded strong signals in the combined TDT analysis (222/223 and gaat), we carried out single locus parametric linkage analysis in the HLA-DRB1 (1,0) families by use of the LINKAGE package (24). Likelihoods were determined under the two assumptions of disease-marker linkage equilibrium and disease-marker linkage disequilibrium, with one dominant and one recessive model in each case. Only the RA phenotype was considered, since there were insufficient data on erosions to test the erosive phenotype. Under equilibrium there was no convincing evidence for linkage to RA in 54 families (Table 4). The incorporation of linkage disequilibrium into the analysis did yield a considerable improvement in the LOD scores under both models. A LOD score of 1.268 was obtained under the dominant model (Table 4). This corresponds to a P-value of 0.009 and could be taken as suggestive evidence for linkage. The weakness of these LOD scores was perhaps not surprising since we found the power of these families to be low when ELODS (expected LODS; Materials and Methods) were considered from simulated data (the average power per family to exceed a LOD score of 1.0 was only 0.5% for the dominant model and 1.96% for the recessive model). In comparison, 40 TDT units would have 80% power to detect a 1.4-fold increase in transmission probability at a significance threshold of 0.05. Thus, these families would have considerably more power for the TDT than for parametric linkage analysis.
Table 4. Parametric linkage analysis of HLA-DRB1 (1,0) families
| Marker | Dominant model | Recessive model | ||
| Equilibrium | Disequilibrium | Equilibrium | Disequilibrium | |
| LOD | LOD | LOD | LOD | |
| 222/223 | 0.103 | 0.496 | 0.0 | 0.660 |
| gaat | 0.364 | 1.268 | 0.088 | 1.174 |
DISCUSSION
In the study presented here, the combined TDT provides evidence for linkage of several markers of the IL-1 gene cluster to erosive RA in those families in which siblings do not share two HLA-DRB1 alleles identical by descent. However, we failed to detect evidence for linkage to RA in this region using non-parametric allele sharing methods, consistent with two previous reports in which these methods were used. John et al. (25) excluded a gene of [lambda]s > 1.7 at IL1A and a recent genome-wide scan in European families did not report evidence for linkage in the region (11). A third more recent preliminary report (26) has indicated some evidence for increased allele sharing of IL-1 markers in 97 of the same European families used by Cornelis and colleagues. With regard to parametric linkage analysis, we did obtain positive LOD scores when marker-trait disequilibrium was incorporated into the analysis, despite the low power of these families, providing some support for the combined TDT results. Overall, our results are consistent with the idea that parametric linkage approaches with two general models (and the incorporation of linkage disequilibrium) can be more powerful in complex diseases than non-parametric methods (27-30) and also that the TDT can be more sensitive in identifying genes of small effect (4). The sib-TDT and similar methods that use unaffected siblings are of particular value in late onset diseases, where parental data are often unavailable, allowing the use of data from such families. However, the power of such methods is not as great as that of the TDT for complex diseases with low penetrance, since unaffected siblings carrying disease-predisposing alleles may develop the disease later in life or fail to develop it due to reduced penetrance (6).
Our data are most consistent with the hypothesis that IL-1 genotype affects susceptibility to erosions in persons with RA. However, with the current data set we cannot exclude that it may affect susceptibility to RA itself, since we have insufficient non-erosive subjects. An alternative use of the data in a case-control analysis also confirmed the same markers to be associated with erosive RA (data not shown). Further studies on erosive and non-erosive subjects will be required to resolve this issue.
The interaction between IL-1 and the MHC is also of interest. It is well established that HLA-DRB1 genes are an important risk factor for erosive disease (31) and our data are consistent with heterogeneity between IL-1 and HLA-DRB1 with regard to erosions. In other words, in those families linked to the MHC, IL-1 seems to have no additional effect on susceptibility to erosions, but in those families unlinked to the MHC, IL-1 does seem to be a risk factor for erosions. Since IL-1 is known to induce metalloproteinase enzymes that destroy the matrix of bone and cartilage and to stimulate bone resorption, then it is quite plausible on biological grounds that IL-1 would be a risk factor for erosive RA, regardless of any possible interaction with HLA-DRB1 (17,32,33).
The analysis of disease haplotypes is problematical in this type of data set since genotype information on one or both parents was only available for a minority of the families (56/195). Thus, in the remaining families assignment of parental haplotypes was more uncertain. We used two approaches to attempt to identify disease haplotypes and although these methods gave some agreement, neither method dealt satisfactorily with this problem and thus the results must be interpreted with caution.
Notwithstanding these difficulties, we identified the 3 2 3 2 1 1 haplotype using the method of Merriman et al. (22) and this haplotype was found to be the second most over-transmitted using the TRANSMIT program, with the 4 1 4 1 1 1 haplotype being the other. These share common alleles at the telomeric end of the gene cluster, namely allele 1 at IL1B-511 and allele 1 at gaat, whilst they diverge centromeric to IL1B-511. This could reflect a single locus at the telomeric end or indicate that due to interaction between the IL-1 loci, there is more than one disease-predisposing combination of alleles. The net biological activity of IL-1 results from the relative levels of IL-1[alpha], IL-1[beta] and IL-1 receptor antagonist binding to the soluble and membrane-bound IL-1 receptors; therefore, it might be expected that several combinations of alleles at the three genes, perhaps associated with high IL-1[alpha] and IL-1[beta] and low IL-1 receptor antagonist for example, will affect the overall predisposition to inflammatory disease in an individual. For example, it has been shown that allele 2 of the +3954 marker of IL1B is associated with high IL-1[beta] production from stimulated peripheral blood monocytes, which would be consistent with the 3 2 3 2 1 1 haplotype predisposing to erosive RA (34,35). It may be best to carry out studies of the combinatorial effects by use of a case-control design, in which the interactions between heterozygous and homozygous genotypes at different loci can be readily considered, in conjunction with careful distinction of the erosive and non-erosive phenotypes.
In conclusion, we have found preliminary evidence to suggest that the IL-1 gene cluster is linked to erosive RA. Follow-up studies are now necessary to confirm these findings and to investigate the interactions between the IL-1 genes themselves and between the IL-1 genes and HLA-DRB1, in particular in erosive disease. Such studies will help elucidate the pathogenesis of this debilitating disease.
MATERIALS AND METHODS
Subjects
The subjects studied were ascertained from the Arthritis Research Campaign's National Repository of multicase rheumatoid arthritis families. Details of family recruitment and case verification are described elsewhere (21). Cases were considered as having RA if they satisfied the ACR criteria modified for use in genetic studies (36). Subjects were classified as erosive if there was a documented presence of radiological erosions in the hands or feet. A total of 187 pedigrees was used in this study, which included 195 nuclear families. In 41 of these families, both parents of the affected proband(s) were available for study. The numbers of affected and unaffected non-founder individuals are shown in Table 2. The average age of onset of RA in the affected individuals was 40.3 (13.6) years. Details of the pedigrees can be found at the ARC National Repository website (http://www.arc.man.ac.uk/natrep.html ).
Markers and genotyping
The positions of the marker loci are shown in Figure 1. Physical distances between the markers have been determined previously (20,37). The genotyping and analyses of the microsatellite markers gaat.p33330 (hereafter abbreviated to gaat), Y31, 222/223 and gz5/gz6 were carried out as described previously (20), except that Amplitaq Gold (PE Biosystems, Foster City, CA) was used for the gz5/gz6 PCRs. Primers for D2S160 were 5[prime]-TGTACCTAAGCCCACCCTTTAGAGC and TGGCCTCCAGAAACCTCCAA (Genome Data Base accession no. 133520; http://gdbwww.gdb.org/ ).
IL1A +4845, IL1B -511 and ILRN +2018 markers were genotyped by 5[prime]-nuclease PCR using the PE Biosystems Taqman allelic discrimination system (38), using primers and fluorescent probes designed by use of Primer Express software (PE Biosystems). IL1B +3954 (formerly +3953) PCRs were carried out by PCR-RFLP as described previously (20) and HLA-DRB1 genotyping of these individuals has been detailed elsewhere (21).
Data analysis
The combined TDT and sib-TDT was carried out as described by Spielman and Ewens (6). As suggested by these authors, the TDT was used in those families for which both parents were available, with at least one of them heterozygous for the marker under consideration, and the sib-TDT was used in the remaining families for which there were one or more unaffected siblings with a genotype different to that of the proband. As described by Spielman and Ewens (6), the results for both sib-TDT and TDT were expressed as a Z score and combined, and P-values were determined by use of the normal distribution approximation. The number of test units, where a unit for the sib-TDT was a family and a unit for the TDT was a heterozygous parent-offspring pair, were also calculated. Since we have used multiplex families in these analyses, the combined TDT is only valid as a test for linkage and not as a test for association, since the siblings are not independent. Nevertheless, there must be an association present in order to detect linkage with this method. Since the HLA-DRB1 locus is a major gene in RA, the families were stratified according to HLA-DRB1 allele sharing status in affected sibling pairs, as has been described by others (11). HLA-DRB1 identity by descent allele sharing probabilities for each affected sib pair were determined by use of MAPMAKER-SIBS (39) and this information was used to stratify the families into two groups; HLA-DRB1 (1,0) and HLA-DRB1 (2). Families were considered to be HLA-DRB1 (1,0) if all or the majority of their affected sib pairs had 0 probability of sharing two alleles identical by descent; otherwise they were classified as HLA-DRB1 (2). Out of 167 nuclear families for which HLA-DRB1 data were available, there were 78 HLA-DRB1 (1,0) families and 89 HLA-DRB1 (2) families.
The three most common alleles of each multiallelic marker were tested and the rare allele of each of the biallelic markers. (Note that although gz5/gz6 has four alleles, only alleles 3 and 4 are observed in this data set, therefore this marker was treated as biallelic.) The same 17 alleles were tested in the HLA-DRB1 (1,0) and HLA-DRB1 (2) data subsets (see Table 1) and then five of these alleles were retested against the erosive phenotype (Table 3). In such a study it is difficult to determine the appropriate value for the significance threshold. The individual markers in this study are closely correlated, as they are all within 1.2 cM of one another and there is considerable linkage disequilibrium between them (20). Therefore, the main issue is the non-independence of the markers, which makes the Bonferroni correction over-conservative. Bearing this in mind, we feel a correction equivalent to 52 independent tests is reasonable. Without any a priori hypothesis regarding the direction of expected transmission distortion, we feel it is appropriate to correct the biallelic markers for two tests and the multiallelic markers for four tests (the three alleles tested +1), making 26 tests in all. When applied to two independent data sets, HLA-DRB1 (1,0) and HLA DRB1 (2), this yields an overall correction factor of 52 on which to base a study-wide significance threshold. We have therefore used a study-wide significance threshold of 0.05/52 = 0.001 and have also considered a threshold of 1/52 = 0.02 as suggestive evidence for linkage. The P-values quoted throughout the paper are uncorrected, to allow the reader to interpret them against whichever threshold they feel is appropriate.
Haplotype analysis was carried out using two methods. For the three-marker haplotype method described by Merriman et al. (22) the haplotype option of GENEHUNTER (23) was used to infer the most likely founder multimarker haplotypes in each pedigree. These were haplotypes of markers only, with no information on disease allele status. This information was used to identify a marker haplotype containing the ancestral disease mutation, as described by Merriman et al. (22). The degree of linkage disequilibrium was determined between a three-marker haplotype and the neighbouring marker amongst RA affected non-founder individuals (n = 398; Table 2). D[prime] values were used as the measure of disequilibrium, as defined by Devlin and Risch (40). This procedure was carried out in both directions starting from the 1 1 IL1B -511 gaat haplotype, since these adjacent markers showed the strongest evidence for linkage in the combined TDT. Thus, a series of overlapping three-marker haplotypes were identified, representing the putative disease-associated haplotype. An alternative haplotype analysis was carried out using the TRANSMIT program, which carries out a TDT after assigning haplotypes for each family (D. Clayton, manuscript in preparation). This program was found to be inefficient for families where no parental genotype information was available and thus could only be applied to 56 of the 195 nuclear families. Alleles of the multiallelic markers with frequencies <5% were aggregated and haplotypes with frequencies of <10% were pooled when computing tests. [chi]2 tests were considered valid if the variances of observed minus expected haplotype transmissions were >5. Thus, [chi]2 tests were performed on six haplotypes. Whilst the GENEHUNTER method used the whole data set, in many cases the haplotype assignment reported by the program would only be one of several or many assignments with equal likelihoods and thus the D[prime] values may be unreliable. The TRANSMIT program, on the other hand, could only be applied to the subset of the data for which the haplotype assignments were less uncertain.
Multipoint non-parametric analysis was carried out by use of GENEHUNTER, using the NPLall statistic, to take into account identity by descent sharing between all affected members of each pedigree. The minimum information content in the multipoint analysis was 0.78.
Two-point parametric linkage analysis (i.e. one disease locus and one marker locus) was carried out by use of the MLINK and ILINK options of the LINKAGE package (24,41). Since it has been shown that the use of two general and complementary models is a powerful approach to the parametric analysis of complex diseases in which the underlying model is unknown (42), one dominant and one recessive general model were used. A population prevalence of 0.01 and a disease penetrance of 0.5 were assumed to calculate the sporadic rates. Under dominance, the disease allele frequency used was 0.002, with penetrances fDD = fDd = 0.5, fdd = 0.008, and for the recessive model, the disease allele frequency used was 0.0632, with penetrances fDD = 0.5, fDd = fdd = 0.008. The power of parametric linkage analysis can be greatly improved when marker-trait disequilibrium is incorporated into the analysis (28-30); therefore, the analyses were carried out separately under the assumptions of linkage equilibrium and linkage disequilibrium between the disease and marker loci. The assumption of linkage disequilibrium is a reasonable one for a candidate gene approach using tightly spaced markers, as is the case here. Under linkage disequilibrium, disease-marker haplotype frequencies were used in the likelihood calculations, which incorporates phase information into phase-unknown matings. The disease-marker haplotype frequencies were determined by use of the EH program (41), under the two models described above. The marker genotype frequencies used to determine disease-marker haplotype frequencies were obtained from a set of unrelated probands from the HLA-DRB1 (1,0) group (n = 77) and 212 unrelated healthy control individuals described previously (20).
To estimate the power of the families to detect linkage, simulation was carried out to allow ELODS to be determined for these families under the assumption of linkage. The simulation assumed a marker with six alleles of equal frequency, at a recombination fraction of 0.05, under the dominant and recessive models described above using 100 replicates for each family. Upper confidence limits (mean SD, i.e. the 84th percentile) for two point ELODS were determined with the same models by use of SLINK (43,44). The simulated marker had a PIC value of 0.8 and the gaat and 222/223 markers used in the analysis of the real data had PIC values of 0.4 and 0.7, respectively. The magnitude of linkage/association that it was possible to detect by use of the combined TDT, given the average number of units available across the markers in the two subsets, was also calculated.
ACKNOWLEDGEMENTS
We thank Hazel Holden for expert technical assistance, Martin Nicklin for helpful discussions and D. Clayton for assistance with the TRANSMIT program. This work was supported by an Arthritis Research Campaign programme grant D0528 awarded to G.W.D., Martin J.H. Nicklin and F.S.d.G.
REFERENCES
+To whom correspondence should be addressed. Tel: +44 114 271 2677; Fax: +44 114 272 1104; Email: a.cox{at}sheffield.ac.uk
§Present address: Genetic Research, IHC, Salt Lake City, UT, USA
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Harrison, J. J. Pointon, K. Chapman, A. Roddam, and B. P. Wordsworth Interleukin-1 promoter region polymorphism role in rheumatoid arthritis: a meta-analysis of IL-1B-511A/G variant reveals association with rheumatoid arthritis Rheumatology, December 1, 2008; 47(12): 1768 - 1770. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. T. Soller, H. M. Escobar, S. Willenbrock, M. Janssen, N. Eberle, J. Bullerdiek, and I. Nolte Comparison of the Human and Canine Cytokines IL-1({alpha}/{beta}) and TNF-{alpha} to Orthologous Other Mammalians J. Hered., August 3, 2007; (2007) esm025v3. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. W Duff Evidence for genetic variation as a factor in maintaining health Am. J. Clinical Nutrition, February 1, 2006; 83(2): 431S - 435S. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Julia, D. Gallardo, F. Vidal, J. J. De Agustin, P. Barcelo, M. Vilardell, and S. Marsal Association study between corticotrophin-releasing hormone genomic region (8q13) and rheumatoid arthritis in the Spanish population Rheumatology, December 1, 2003; 42(12): 1534 - 1538. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. C. Read, C. Cannings, S. C. Naylor, J. M. Timms, R. Maheswaran, R. Borrow, E. B. Kaczmarski, and G. W. Duff Variation within Genes Encoding Interleukin-1 and the Interleukin-1 Receptor Antagonist Influence the Severity of Meningococcal Disease Ann Intern Med, April 1, 2003; 138(7): 534 - 541. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Steer, S. A. Fisher, M. Fife, A. Cuthbert, J. Newton, P. Wordsworth, C. M. Lewis, C. G. Mathew, and J. S. Lanchbury Development of rheumatoid arthritis is not associated with two polymorphisms in the Crohn's disease gene CARD15 Rheumatology, February 1, 2003; 42(2): 304 - 307. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. E. Frevel, T. Bakheet, A. M. Silva, J. G. Hissong, K. S. A. Khabar, and B. R. G. Williams p38 Mitogen-Activated Protein Kinase-Dependent and -Independent Signaling of mRNA Stability of AU-Rich Element-Containing Transcripts Mol. Cell. Biol., January 15, 2003; 23(2): 425 - 436. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Nemetz, M Toth, M A Garcia-Gonzalez, T Zagoni, J Feher, A S Pena, and Z Tulassay Allelic variation at the interleukin 1{beta} gene is associated with decreased bone mass in patients with inflammatory bowel diseases Gut, November 1, 2001; 49(5): 644 - 649. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Buchs, T. Silvestri, F. S. di Giovine, M. Chabaud, E. Vannier, G. W. Duff, and P. Miossec IL-4 VNTR gene polymorphism in chronic polyarthritis. The rare allele is associated with protection against destruction Rheumatology, October 1, 2000; 39(10): 1126 - 1131. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||