Soon after people develop a chronic disease such as rheumatoid arthritis (RA), they ask their physicians two searching questions: the first is ‘Why me?’ and the second is ‘What will happen to me now?’ In 1989 the ARC Epidemiology Unit set up the Norfolk Arthritis Register (NOAR) with the aim of providing answers to these two questions. This paper is the first in a series of two reviews which focus on the work of NOAR, a register of a primary care‐based inception cohort of patients with inflammatory polyarthritis (IP). The present paper examines genetic and environmental risk factors for the development of IP and of its subset, RA. The second paper summarizes the natural history and predictors of prognosis [87]. In both papers the NOAR data are presented in the context of a detailed review of the literature which summarizes current knowledge in this challenging area. The cause and prognosis of any disease are best studied by identifying incident cases from the general population. The study of prevalent cases makes it difficult to distinguish factors which lead to the development of disease from those which influence disease outcome. Hospital‐based studies may be subject to selection bias. A major problem in studying rheumatoid arthritis (RA) lies in the definition of the disease. The earlier in the course we try to identify patients, the harder it is to distinguish different forms of inflammatory polyarthritis (IP). Many of the criteria currently used to classify RA [1] [e.g. radiological erosions, rheumatoid nodules and rheumatoid factor (RF)] are features of chronic or severe disease, and may not appear until months or years after disease onset. The Norfolk Arthritis Register (NOAR) is a population‐based register which aims to recruit all patients who potentially have RA by using deliberately broad inclusion criteria. Details of the register have been published elsewhere [2]. General practitioners (GPs) working in the former Norwich Health Authority, which covers a population of almost 0.5 million, predominantly (99%) of European Caucasoid origin, notify NOAR of adults (age 16 yr or older) who have swelling of at least two peripheral joints which has persisted for a minimum of 4 weeks. A parallel notification system operates from the local hospitals. Patients who do not satisfy the 1987 American College of Rheumatology (ACR) criteria for RA are later excluded if they are referred to hospital and are given a diagnosis other than RA, psoriatic arthritis, viral arthritis or undifferentiated IP. Within 2 weeks of notification, patients are visited at home by a research nurse, who takes a history using a structured questionnaire and examines the following joints for tenderness, soft‐tissue swelling and deformity: distal interphalangeal joints, proximal interphalangeal joints (PIP), metacarpophalangeal joints (MCP), wrists, elbows, shoulders, knees, ankles and metatarsophalangeal joints (MTP). Patients are asked to provide a blood sample, which is frozen as serum and EDTA‐anticoagulated specimens. Patients are assessed annually for a minimum of 5 yr. At each visit patients are asked to complete the Stanford Health Assessment Questionnaire [3], which is scored from 0 to 3. The 1987 ACR criteria are applied at each assessment. At the 1‐yr assessment, those who satisfy at least two of the 1987 ACR classification criteria for RA are asked to have X‐rays taken of the hands and feet. The X‐rays are read independently by two observers, and the following joints are graded from 0 to 5 using the grading system of Larsen et al. [4]: PIPs, MCPs, wrists and MTPs II–V. A third observer acts as arbiter if there is disagreement about whether a patient has erosions. At the 2‐yr assessment visit, patients who satisfy at least two of the ACR criteria and who were non‐erosive at the 1‐yr visit are asked to have further X‐rays. During 1994–95 a case–control study was conducted, the results of which form the basis of much of this review paper. The cases were consecutive patients recruited by NOAR between May 1994 and April 1995 who were aged 18–70 yr and who had a symptom duration of less than 12 months at the time of registration with NOAR. The control subjects were selected from the combined registers of all GPs in the Norwich Health Authority. The controls were matched to the case by sex and by date of birth (± 1 day) [5]. If the first control subject approached declined then another was approached until one agreed to participate or until 6 months had elapsed from the time of notification of the case. We tried to recruit two controls for each male case aged less than 45 yr in order to increase the power of the investigation. Cases and controls were mailed identical questionnaires, which asked about marital status, occupation, educational qualifications, current height and weight, previous medical history, current medication, smoking habits and reproductive factors. Controls were assigned a dummy date of disease onset corresponding to that of their matched cases. Odds ratios (OR) were calculated using a conditional logistic regression model which maintained the matching. In total 177 cases were recruited. Controls were found for 165 of the cases, including 13 young males for whom two controls were identified. The analyses are therefore based on 165 cases and 178 controls. Of the cases, 90 (55%) satisfied the 1987 ACR criteria for RA at the baseline assessment, and are referred to in this review as ‘RA’. This paper reviews our current understanding of the range of genetic and environmental factors which may predispose to or protect against the development of RA. The majority of previous studies of the aetiology of inflammatory arthritis have focused specifically on RA. We have therefore examined risk factors in the whole NOAR cohort (‘IP’) and those who satisfied classification criteria for RA (‘RA’). The importance of genetic factors in the aetiology of RA has been recognized for many years. Concordance rates for RA are higher in monozygotic (12–15%) than in dizygotic twins (3–4%), and higher in dizygotic twins than one would expect by chance from population prevalence estimates [6, 7]. These observations support the hypothesis that there is a genetic component to RA. However there is an ongoing debate as to the relative contribution of genetic factors to disease susceptibility as opposed to disease severity. For example, the occurrence of RA in the first‐degree relatives of NOAR RA patients was not increased compared with that of matched controls [8]. Around 2–3% of NOAR RA cases have at least one first‐degree relative with RA compared with around 11% of hospital clinic RA patients [9, 10]. This suggests that genetic factors may be more important in determining disease severity (and thus hospital referral). The association between RA and HLA‐DR4 was first demonstrated more than 20 yr ago [11]. It has since been estimated that up to 50% of the genetic component of RA susceptibility may be accounted for by genes in the HLA region [12]. As more advanced molecular biology techniques have been developed, it has become clear that RA is primarily associated with a specific amino acid sequence on the β1 chain of a number of class II alleles. Studies of clinic‐derived RA patients from a wide range of populations [13] led to the formulation of the ‘shared epitope’ (SE) hypothesis [14]. Variation in the frequency of SE‐positive alleles (DRB1*0101, *0102, *0401, *0404, *0405, *0408, *1001 and *1402) in different populations may account for some of the variation in RA prevalence around the globe. It is unclear how the SE is involved in the causation of RA, but there are a number of possibilities. For example the SE may act as a receptor for an arthritogenic peptide; the SE may model the T‐cell repertoire to favour an arthritogenic response; or there may be molecular mimicry between SE and a pathogen such as the Epstein–Barr virus [15]. As mentioned above, these genes may be more important in determining the clinical course of RA than in determining susceptibility to the disease. It appears, from a number of reports, that the DR4‐positive RA‐associated alleles are more frequent in patients with severe, erosive disease [16], whereas a study of cases ascertained from a population survey did not demonstrate an important association between HLA and RA [17]. We investigated this relationship in the NOAR cohort. Six hundred and eighty consecutive patients with IP, registered between 1991 and 1995, were typed for HLA‐DRB1 alleles. The controls were taken from the case–control study described above or from another population‐based prospective study in Norfolk which is investigating the relationship between diet and the subsequent development of cancer (the EPIC study). We found only a modest association between IP and the presence of any SE [OR 1.8; 95% confidence interval (CI) 1.4, 2.4]. The effect of being homozygous for the SE was only moderately greater (OR 2.1; 95% 1.5, 3.0) than the effect of being heterozygous (OR 1.3; 95% CI 1.1, 1.6) [18]. However the risk of being RF‐positive was substantially higher in those who were homozygous for the SE (OR 3.3; 95% CI 2.1, 5.3) than in those who were heterozygous (OR 1.5; 95% CI 1.0, 2.0). We examined the associations of certain alleles that include the SE. All showed a pattern similar to that described above, with the exception of *0404 (OR 3.5; 95% CI 1.8, 6.8). There was a strong association with IP, RA and seropositivity in cases who were homozygous for *04 and in whom one of the alleles was *0404. For example, individuals who had the genotype *0404/*0404 had a markedly increased risk of having IP (OR 17.9; 95% CI 4.2, 77.8). Similar results were obtained by restricting analysis to the subset of patients with RA. Therefore the ACR criteria did not define a subgroup of patients with a stronger association with HLA. It is important to be aware that, although the majority of patients with IP and RA possess at least one copy of the shared epitope, so do a large proportion of the normal population. According to the NOAR data, in a Caucasian population about 44% of people are SE‐positive but only one in 43 of these individuals will develop RA. The highest risk is conferred by possession of two *0404 genes. This combination occurs in only 0.1% of the Caucasian population, but one in six will develop RA. Among new cases of RA, approximately 5% will be homozygous for *04 and approximately 0.5% will be homozygous for *0404. Until recently, most studies of RA susceptibility have focused on genes in the HLA region. However, it now appears that HLA genes may be more important in determining the clinical course of RA than in conferring susceptibility to the disease. The availability of rapid gene sequencing techniques and highly polymorphic microsatellite markers means that it is now possible to search for other susceptibility genes using both whole‐genome screening and the candidate gene approach. Some preliminary results have been published, and this is likely to be an area of intensive investigation in the next few years. A number of observations point to a role for hormonal and reproductive factors in the development of RA. RA is more common in women than men, especially in the pre‐menopausal age groups [2]. RA is rare before the menarche. Women have a reduced susceptibility to RA during pregnancy. This is followed by a period of increased susceptibility in the immediate post‐partum period [19]. This increased susceptibility does not simply represent a ‘catch‐up’ phase in which those women who might otherwise have developed RA during the 9 months of pregnancy all develop it in the 3 or 4 months post‐partum. Rather, the risk seems to be related to breast‐feeding. A case–control study of 88 women who developed RA within 12 months of delivery of their first child showed that they had a five‐fold increased frequency of breast‐feeding compared with 129 control women [20]. It seems that if a woman ‘survives’ her first exposure to breast‐feeding without developing RA, then her risk after each subsequent pregnancy will be less. Prolactin, the hormone which maintains lactation and which is known to be proinflammatory, may play a role in the onset of post‐partum RA. It is not clear whether pregnancy itself is a risk factor or protective against RA. An early study by Kay and Bach suggested that nulliparous women may have an increased risk of developing RA However, studies including a large study and the NOAR case–control study have to might be a of in the early weeks of pregnancy. In the NOAR case–control women with RA more (OR 95% CI than controls [5]. et al. found a increased of their RA found that of their cases were but not compared with of controls. The was to a higher of the RA In a more study from NOAR, women with RA a higher frequency of of pregnancy than controls (OR 95% CI A number of studies have an increased frequency of It therefore appears that an pregnancy or may be a risk factor for developing RA, many years A number of studies that exposure to the may the risk of developing severe RA or may the onset of RA A protective effect of against the development of RA was first suggested in as an from the College of General study of the effect of the on of women the risk of developing RA. a number of studies have been published which this results are The most for this from a published in which the studies those which were and those which were population‐based The for the studies published up to that time was CI and for the population‐based studies it was CI As in it appears that this between and cases has been maintained in studies published since It suggests that does not protect against the development of RA, but does it from severe to hospital We the between and the development of IP in the context of the NOAR case–control study [5]. We found a protective effect for current of an (OR 95% CI but not for having taken an (OR 95% CI 1.6) The effect was similar for those who satisfied the criteria for RA 95% CI used 95% CI It appear that, in the there is a protective effect from the current of the in a However, there are in these studies the women have whether or not to an and this is likely to be by their and It is possible that is a for another which is the of RA. This is more likely by the that or duration effect has been studies will be difficult the majority of women now an at some point in their This may the NOAR study found protective effect for having used an possible is that is the onset of RA than it The age of onset of RA appears to be higher in women but not in men, which is with the that the of the the onset of RA If hormonal factors are so important in the pre‐menopausal as to the effect of the and the of hormone on the onset of RA. In the NOAR case–control study there was between and the onset of RA, although women with RA did a higher frequency of (OR 95% CI of cases and of controls having used The first study to this found a protective effect in women to but subsequent studies have to this about hormonal factors in with A number of case–control studies have suggested that RA is associated with in and but it is to whether this is a cause or the effect of the disease et al. in before the onset of RA and suggested that this may be involved in the causation of RA than an effect of the disease. However, a more population‐based study from found similar of and in both and women who developed RA, compared with the general population There is some to support the hypothesis that some people who develop RA have a of the which from developing an to this may account for the of some cases of IP, it is of history and the development of RA as pregnancy as history and the development of RA as pregnancy as and the development of RA case–control prevalent incident prospective not and the development of RA case–control prevalent incident prospective not have been as of RA for a However, have been from or and studies for a wide of and have been applied to of patients with RA. In the NOAR patients there was of of RA in or time In the NOAR case–control study the cases did not an increased frequency of before the development of their arthritis [5]. This suggests that RA is not by a It does the that may RA in cases. It has been suggested that Epstein–Barr virus and may play an role in some although results of different studies are There are cases of RA developing after for example, and cases do not up a large proportion of all cases of RA but In the NOAR only of ascertained patients with a history of IP of less than 16 weeks for of the patients satisfied the ACR criteria for RA The be applied to the that may RA There is clear that may an arthritis in women In the arthritis is but in those in whom it it and RA. there is that and may RA in In the NOAR case–control study there was an association between in the 6 weeks symptom onset and the development of RA (OR 95% CI [5]. such individuals only a proportion of all those with RA. from the NOAR case–control study was that previous blood may be a risk factor for RA (OR 95% CI [5]. This did not appear to be to possible such as and In the IP the association with blood was higher for women than for men, and the association was stronger with than were more than yr before symptom onset. is known to have a number of on the including the of and In between HLA on the and blood may lead to the development of and the of specific T‐cell These have been for patients with RA It is possible that there may be a in with a in the that blood is may account for the that there have been cases of RA following blood in years. medical risk factors for RA include other in disease, and is a risk factor for the development of IP There is from NOAR that the of RA is by factors or although both of these may be important in the prognosis of RA. In the NOAR case–control RA was found to be associated with having (OR 95% CI 1.0, and with (OR 95% CI 1.1, [5]. studies have demonstrated a between smoking and the development of RA, in RA, especially in The is less clear in one study a protective effect but a number of large studies have an association between smoking and RA in women smoking is associated with in individuals without RA The role of in is although it is clear that may the development of RA, by many years may be associated with in the of and which influence possible is that factors associated with such as a may be for the development of RA. For example, a study of blood found of in those who later developed RA smoking is associated with other including which be involved in the causation of RA. Of was found to protect against the development of RA in two studies which recruited only women previous studies found an increased of RA in women who were but the Health study found association between and RA The NOAR case–control study was the first to demonstrate an association between and RA in It demonstrated that the association in both was with since being did not an increased There is a that the occurrence of RA is related to There are studies which this The NOAR case–control study suggested that many patients their diet after the onset of arthritis in an to their any study of prevalent cases of RA will be by in habits since disease onset. The EPIC study in which is the diet of a large number of the to the of diet in patients who develop IP and are to NOAR. Many patients that their RA was by or by The most are associated with occupation, and The association between and the onset of RA has been examined in a number of most of which have been than or had a example, they have not a control or have patients after the onset of their in an that of patients with RA with a duration of less than yr had at least one in the before onset the only two studies of RA than 5 more than patients have both association between the occurrence of and the subsequent development of RA A case–control study to NOAR found that or a role in patients who developed RA after the age of 45 yr There is less about the between and RA. In an et al. found that, of patients with RA of less than 1 yr six their disease to and it to In a prospective cohort study of patients with age developed RA in the next 2 yr compared with of controls age studies an association between and and the subsequent development of RA but another two studies to this However, in this the hypothesis related to the effect of than to the associated with many years In the NOAR case–control study there was a association between RA and (OR 95% CI and association with (OR 95% CI [5]. This review may appear to have the aetiology of IP and RA. However, the results from NOAR to any in the associations of these two The as to some patients have disease and on to have RA a does not appear to with the which the It seems more likely that the genetic or and and other which after the onset of disease are more important in determining the outcome. The does not support a of the cause of IP or RA. It seems that a series of or may be before the disease genetic or has been to work is on the by which some of the risk factors above may the development of IP, many years more is known about these it may be possible to a of how IP this had been the next will be in some IP is and in it a ARC Epidemiology of This study was by the Arthritis We are to the GPs and hospital in the Norwich Health for their in patients for In we Norfolk Norwich We the current and previous NOAR and research for their and We of the ARC Epidemiology and
No takes yet. Share an insight, caveat, or question.
Symmons et al. (2000) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: