Article Text
PDF
Abstract
Background The susceptibility to rheumatoid arthritis (RA) is partly heritable, but whether the severity of RA is also influenced by genetics has not been determined. The evaluation of the heritability of the severity of RA is basic to further studies on genetic factors. A study was undertaken to determine whether joint destruction is heritable.
Methods Iceland has an unique comprehensive genealogy database covering today's population and stretching back to ≥1000 years ago, as well as genome-wide single nucleotide polymorphism data for a large part of the population. Hand and feet x-rays of 325 Icelandic patients with RA were scored according to the Sharp-van der Heijde method. The degree of relatedness between patients was estimated in two ways: (1) kinship coefficients (KC) on the genealogical data were expressed; and (2) the identical-by-descent (IBD) was estimated applying long-range phasing of the genetic profile of the patients. The degree of relatedness was tested against the similarity in joint destruction rates by linear regression analysis and the heritability of joint destruction was calculated.
Results Significant associations between degree of relatedness and similarity in joint destruction rates were observed for both methods of determining relatedness (pKC=0.018, pIBD=0.003). The estimated heritability was 45% using KC and 58% using the estimated IBD data.
Conclusions The severity of joint destruction in RA is influenced by genetic factors.
Statistics from Altmetric.com
Introduction
Rheumatoid arthritis (RA) is a chronic inflammatory disease that is characterised by destruction of joints. It affects about 0.5–1% of the population and is associated with considerable disability and costs for healthcare services and society. Radiographic joint destruction reflects the cumulative burden of inflammation and is conceived as an objective measure of RA severity. The severity of joint damage varies markedly between patients and the processes behind this difference are incompletely understood. Inflammatory markers and autoantibodies are potent risk factors for joint destruction, but explain about 30% of the variance in joint destruction.1 In order to increase the understanding of the processes driving interindividual differences in joint destruction, genetic studies on the outcome of RA have been performed. Nevertheless, it is unknown to what extent genetic factors account for the difference in joint destruction.
Taking advantage of the unique data collection available in Iceland, we therefore investigated whether the rate of joint destruction in RA is heritable. Iceland has an extensive nationwide genealogical database with data available for the last 10 centuries. Moreover, genome-wide genetic data are present for a substantial proportion of the population. This created a unique opportunity to study the relation between the level of relatedness and the similarity in rate of joint destruction between patients in a population-based study.
Methods
Study population
Iceland had 317 630 habitants in 2010, most of whom live in the area of Reykjavik. The research cohort of Landspitali, the National University Hospital of Iceland, included 800 patients with RA diagnosed in the second half of the 20th century. Since RA mainly affects the hands and feet, sets of hand and feet were required and were available in 325 patients. Joint destruction was measured using the Sharp-van der Heijde scoring method (SHS).2 The x-rays were scored by two readers (AWV and RK) To test the consistency of scoring between and within the scorers, intercorrelation and intracorrelation coefficients (ICC) were calculated; 12% of the x-rays were scored by both scorers and each scorer rescored 15% of their own scored x-rays. The ICCs between and within readers were all >0.95. For each patient, the yearly progression rate of joint destruction was determined by dividing the total SHS by the number of disease years at the time of the x-ray.3
Genealogy database
The genealogy department of deCODE has an ongoing project which involves electronic registration of available genealogy data for the last 10 centuries in Iceland, compiled from a variety of sources including genealogy manuscripts, censuses and church books.4 5 Examination of mitochondrial sequences of maternally related individuals indicates that the accuracy of the maternal connection exceeds 99.3%.6 Moreover, after having genotyped 25 000 Icelanders, it has been found that the sum of false paternity and labelling error is 1.5%. The computerised genealogy database is stored and maintained within a relational database. Each record in the database consists of a blinded personal identifier, identifier to parents, sex, and dates of birth and death rounded off to the nearest decade or half decade.
Genome-wide single nucleotide polymorphism (SNP) data
Genome-wide data are available for most of the Icelandic population including approximately 80% of the patients with RA. Genome-wide data for 267 of 325 patients (82.2%) was available.
Relatedness
The relatedness between patients was calculated in two ways. First, relatedness was measured with the kinship coefficient (KC) using the Icelandic genealogy. The KC is the probability that an allele selected randomly from one individual of a pair is inherited from the same ancestor with an allele at the same locus selected randomly from the second individual. Second, using genome-wide SNP data, relatedness was determined using identical-by-descent (IBD) sharing. IBD is a measure of ‘realised’ genetic sharing. To estimate whether the genetic resemblance is based on an identical ancestor, ancestor-specific haplotypes were used which were previously identified in 35 528 Icelandic people using long-range phasing. Given the presence of two DNA strands, the allelic composition of a haplotype is challenging to estimate. The long-range phasing method defines the composition of haplotypes by matching patients based on their haplotypes to two groups of surrogate parents. If one of these surrogates is homozygous for the major allele, the major allele thus belongs to the haplotype of that group of surrogate parents and the other haplotype has the minor allele. As such, this method estimated the IBD with high certainty. More detailed information on the possibilities of this method is provided elsewhere.7 These haplotypes were used to estimate IBD sharing in the current set of patients.
Statistical analyses
Each patient was paired with all other patients, creating patient-pairs to analyse the difference in rate of joint destruction (phenotypic similarity) versus the level of relatedness (genotypic similarity).
The association between genetic and phenotypic similarity was analysed using a linear regression model. The heritability (h2, the amount of variance explained by heredity) was estimated using a standard model in which total variance was modelled as environmental plus genetic variance. Maximum likelihood estimates of the parameters were obtained using R version 2.12.2. (R-project, Vienna, Austria). The dependent variable in each analysis was the difference in the yearly progression rate for each pair transformed on a log scale to fit normality.
Results
Measures of relatedness
The characteristics of the patients are presented in table 1. From the 325 patients with KC data, 52 650 unique pairs of patients could be made. From the 267 patients with IBD data, 35 511 pairs were made. The mean (SD) KC and IBD were, respectively, 0.00043 (0.005) and 0.001 (0.010). The correlation between KC and IBD was 0.98 (p<0.001).
Patient characteristics (n=325)
Association of relatedness with phenotype
Genealogical relatedness (KC) was significantly associated with the similarity in joint destruction rates (p=0.018, figure 1A). Likewise, the relatedness of patient-pairs measured by genome-wide SNP data was significantly associated with similarity in radiological progression (p=0.003, figure 1B).
Relation of patient-pairs between degree of relatedness: (A) kinship coefficients (KC), (B) identical-by-descent (IBD) and similarity in rate of joint destruction. The KC is the probability of IBD, and the IBD determined using genome-wide SNP data presents the ‘realised’ IBD. KC is therefore on average half the IBD. For instance, a KC of 0.25 and IBD estimate of 0.5 concern full siblings and parent-child relations. Similarly, a KC of 0.125 and IBD estimate of 0.25 concern half-siblings and aunt-nephew relations.
Effects of age and gender
In some cohorts, age at diagnosis and gender are associated with joint destruction.8 9 Patients with a family history of RA have a lower age of disease onset than patients with sporadic RA and no differences in gender. To evaluate whether similarity in age of disease onset could underlie the association between relatedness and similarity in joint destruction, the associations of relatedness with similarity in age of onset were tested. No significant associations were found (p=0.21 and p=0.17 for KC and IBD, respectively), making it unlikely that age influenced the analyses.
Heritability
Heritability estimations were performed to approximate the proportion of variance in joint destruction rated that is explained by genetic factors. Based on the KC as a measure for relatedness, the estimated heritability was 45%. Using the IBD estimate, the estimated heritability was 58%.
Discussion
The present study investigated whether heritable factors affect the rate of joint destruction in RA. Classic studies on heritability evaluated twins. With regard to the outcome of RA, a large number of twins with long-term radiological data would be required; such datasets are not widely available. Although it was recently observed that the rate of joint destruction in monozygotic twins was more similar then in dizygotic twins, the number of twins was too low to calculate the heritability.10 However, analyses on the influence of genetic factors can be performed on any patients as long as the degree of relatedness between patients is accurately determined. The Icelandic population is unique in this respect since extensive genealogical data are available and the majority of the population is genome-wide genotyped. We observed that the level of relatedness was significantly associated with similarity in the progression rate of joint destruction. The amount of variance in joint destruction explained by the genetic factors was estimated to be 45–58%. Since only a few genetic variants have been identified and replicated to associate with joint destruction to date,11 12 the findings of our study may promote further research to identify genetic severity factors in RA.
Some of the RA cohort did not have complete sets of hand and feet x-rays and were therefore excluded from the analyses which resulted in a smaller number of patients. Studying the whole RA cohort would have resulted in increased power compared with that obtained in the current analyses.
The percentage of women was similar to previous studies,4 and the anti-cyclic citrullinated peptide (CCP) autoantibody status and distribution of joint destruction rates were comparable to other cohorts. We therefore assume that the patients studied are representative of the population of Iceland with RA.
We studied potential influencing covariates. Age was not associated with genetic similarity. In theory, treatment may have affected the variance in the rate of joint destruction. We did not adjust for treatment effects since >75% of the patients were diagnosed before the modern tradition of aggressive treatment in RA started and we assumed that the effect of eventual treatment is limited. Common environmental factors between Icelandic people could have led to an overestimation of the heritability as estimated by the KC, but not the analyses on genetic IBD.13
A further limitation to the study is that x-rays were performed at different time points with regard to disease duration and progression rates of joint destruction were expressed by estimated yearly progression rates. This method presumes a linear development of joint destruction over time, resulting in a possible underestimation of the progression rate in patients with disease of long duration at the time of the x-ray. Having repeated measurements over time in all patients would overcome this limitation but, to our knowledge, no such cohorts who also have genealogical data exist internationally. However, the correlation between disease duration and yearly progression rate was low (r=0.02), indicating that this possible underestimation was limited.
Unfortunately, the present study was underpowered to evaluate whether the results are different for joint destruction in anti-CCP-positive and anti-CCP-negative patients with RA. Although the heritability of joint destruction in anti-CCP-positive patients was estimated to be substantial (h2=75%), the degree of relatedness and similarity in joint destruction rate were not significantly associated (p=0.30). This probably results from the low number of CCP-positive patients with RA (n=160). The results of these subanalyses should therefore be interpreted with much caution. Estimations on the heritability in anti-CCP-negative patients were not possible owing to the non-normal distribution of the estimated yearly progression rates in the anti-CCP antibody patient-pairs (also after log transformation).
In conclusion, based on extensive genealogical and genetic data, the current study shows that genetic predisposition plays a role in the pathogenesis of progression of joint destruction in RA. These results provide a basis for further research into genetic factors determining the severity of joint destruction in RA. Identification of new risk factors is relevant to increase the understanding of the processes leading to joint damage and to develop targeted therapies.
Acknowledgments
The authors acknowledge deCODE Genetics for providing the genetic and genealogical data and Ari Kárason and Stacy Steinberg for their assistance in the logistics and analyses.
References
Footnotes
-
Funding The work of RK was supported by the Dutch Arthritis Foundation, Reumafonds. AHMvdHvM was supported by Netherlands Organisation of Health Research and Development (ZonMw).
-
Ethical approval Permission was obtained from the National Ethical Committee and all patients gave their informed consent.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.