The influence of early life factors on the risk of developing rheumatoid arthritis

Abstract

Rheumatoid arthritis (RA) is a chronic inflammatory disease that develops as a result of the interaction between genetic and environmental risk factors. Although increasing evidence shows the importance of genes in determining the risk of RA, it is clear that environmental factors also have a vital role. Studies to date have tended to concentrate on environmental influences around the time of disease onset. However, a number of pieces of evidence, including the fact that autoantibodies, such as rheumatoid factor (RF), can develop several years before the onset of clinical disease, suggest that environmental factors may influence disease susceptibility during early life. Several recent studies lend weight to this possibility, with an increased risk of RA in the offspring of mothers who smoked during pregnancy and in those with higher birth weight. There has also been a suggestion that the risk of RA is reduced in breast-fed infants. We describe the evidence surrounding the effect of early life factors on the risk of developing RA and possible mechanisms by which they may act.

Introduction

The strongest genetic association with RA is the human leucocyte antigen (HLA), particularly the HLA-DRB1 ‘shared epitope’[1], with two copies associated with a relative risk of developing RA of 3·8–6 in Caucasian patients [2]. Despite the clear role of genetic factors in determining susceptibility to RA, the concordance of the disease in monozygotic twins is only 15% [3], and twin studies have estimated that approximately 50% of the variation in susceptibility of RA can be accounted for by genes [4].

Various environmental factors have also been implicated in the aetiology of RA. These include exposures to smoking, ultraviolet light, climatic differences, exposure to pollutants and silica and life-style factors, such as diet and exercise. In particular, silica dust exposure has been linked to the risk of developing seropositive RA in several epidemiological studies [5–7]. Various infectious organisms have also been suggested as important. These include viruses such as Epstein–Barr virus (EBV) [8] and Parvovirus B19 [9], and bacteria including Streptococcus, Mycoplasma, Proteus and Escherichia coli[10]. Approximately 10–20% of patients with early RA have serological evidence of recent infection that could trigger RA. However, no single infectious agent appears to be predominant. Perhaps this is because the total burden of infectious exposure is more important than any single organism [11].

In recent years, the effect of smoking on the increased risk and severity of RA has emerged as a major story [12–15]. The presence of environmental cigarette smoke and personal history of cigarette smoking has also been recognized to increase the probability of positivity for rheumatoid factor (RF) [13,14,16]. These effects seem to be strong, consistent and dose-dependent. The effect of smoking is particularly pronounced in individuals with the shared epitope. There is also interesting work suggesting that smoking may induce citrullination of peptides that leads in turn to the formation of anti-citrullinated peptide antibodies (ACPA) [17]: a clear example of a gene–environment interaction.

Early life environmental factors

The genetically determined risk of developing RA is set and at present unalterable. This is not true for environmental factors, which may vary over life in type and intensity. For example, during fetal life and infancy, sufficient nutrition is vital to ensure healthy growth and development of the immune system. During infancy and early childhood, infectious exposures have a major effect on shaping the immune repertoire, whereas exposure to smoking and occupational exposures are likely to come later in life. An understanding of the variation in the exposures experienced during life encourages greater sophistication in our studies of environmental risk factors. One result of this has been closer scrutiny of the effect of environmental factors experienced throughout life. In particular, it has stimulated an interest in the effects of early life environmental factors on the development of RA.

A large body of evidence has been accrued over the last 20 years that suggests that early life events may have long-lasting effects on the likelihood of developing disease in adult life. David Barker's work at the University of Southampton led to a hypothesis that explained how events in early life could influence disease in adult life [18]. Examples include the association of hypertension [19], cardiovascular disease [20,21] and non-insulin-dependent diabetes [22,23] with low birth weight. A ‘developmental origin of adult disease’ is well described for other conditions, such as osteoporosis [24]. This work suggests that certain systems can be ‘programmed’ by early experiences in a permanent way, with life-long effects.

In a similar fashion, several pieces of evidence support the importance of early exposure to environmental factors in RA. The first is that juvenile idiopathic arthritis (JIA), including the RA-like phenotype, can be seen in children. This suggests that environmental factors of importance are present in childhood. In addition, markers of the immunopathology of RA, including RF, ACPA and raised high-sensitivity C-reactive protein (hs-CRP), are present many years before the onset of clinically apparent disease [25,26].

There are also suggestions that early life exposure may be important from studies of populations with very high levels of RA. The prevalence of RA in some native American populations has been reported to be exceptionally high: 5·3% in Pima Indians [27] and 6·8% in Chippewa Indians [28]. Interestingly, there has been a decline in the rate of RA and occurrence of RF in consecutive birth cohorts of Pima Indians, implicating the involvement of early life environmental factors [29,30].

Studies of new immigrants can be useful in trying to separate out the relative importance of genes and environmental factors in a disease. Higher incidence rates of RA have been reported in northern than in southern Europe [31]. Li et al. investigated whether country of birth affected the risk of RA in first-generation immigrants to Sweden, and whether such an effect remained after adjustment in second-generation immigrants [32]. The largest protective effect was seen in first-generation immigrants from southern Europe, in particular those from Greece [standardized incidence ratio (SIR) 0·31, 95% confidence interval (CI) 0·17–0·52]. Interestingly, this effect was reduced in second-generation southern Europeans (SIR 0·52, 95% CI 0·21–1·09). The key here is that the second generation of immigrants from Greece have an increased risk of RA, but the risk of RA for those born in Greece then emigrating to Sweden appears to remain the same as that seen in their country of origin. This suggests that environmental factors experienced in the country of origin can have a long-lasting effect on the risk of RA.

Studies of early life environmental factors and RA

Perinatal characteristics

Descriptive studies have suggested associations between the risk of RA in offspring and several perinatal factors, including maternal smoking during pregnancy, high birth weight and breast feeding.

Maternal smoking in pregnancy

In a Finnish study, an increased risk of RA and JIA was observed in girls whose mothers smoked during pregnancy. The likelihood of developing RA, in girls with heavy smoking exposure of 10 or more cigarettes per day, was increased [odds ratio (OR) 2·57, 95% CI 1·13–5·89]. The chance of developing JIA was also increased (OR 2·98, 95% CI 0·95–8·78) [33].

In an identical way to that seen in adults, it appears that cigarette smoking may increase the risk of RA following neonatal exposure. This could occur through stimulating the immune system, in particular, through the citrullination of proteins in the target tissues leading to the production of ACPA [34,35]. The effect of maternal smoking in pregnancy suggests that exposure in utero causes a similar effect to the fetal immune system.

Birth weight

A large cohort study of women from the Nurses' Health Study (NHS) cohort identified higher birth weight (>4·54 kg versus 3·2–3·85 kg) to be associated with a twofold increased risk of developing RA [relative risk (RR) 2·1, 95% CI 1·4–3·3][36]. Adjusting for additional confounding factors did not change this relationship, and findings were similar when limiting to those cases with seropositivity for RF (RR 2·1, 95% CI 1·2–3·6). A further, small study demonstrated high birth weight (≥4000 g versus 3000–3999 g) to be associated positively with RA (OR 3·3, 95% CI 1·4–7·4) [37]. Other studies have also shown a non-significant increased risk of developing RA associated with being born large for gestational age (OR 1·6, 95% CI 0·7–3·3) [38].

Conversely, low birth weight (<3000 g) has been associated with a small reduction in the risk of RA (OR 0·7, 95% CI 0·5–1·0) in some, but not all, studies [37–39].

Breast feeding

The findings for birth weight and future risk of RA seem to be fairly consistent, with high birth weight associated with development of RA. However, the effect of breast feeding on the risk of RA in the offspring is more uncertain. The early initiation of breast feeding (within the period of hospital admission after delivery) has been demonstrated to reduce the risk of RA (mean length of stay 7 days; OR 0·2, 95% CI 0·1–0·70) [37], although this association was not reproduced in a recent, large cohort study [40]. In HLA-DR4-negative children, RF-positive infants were less likely to have been breast fed for >3 months (OR 0·18, 95% CI 0·04–0·99) compared with RF-negative children. This effect was not seen in HLA-DR4-positive children [16].

Other perinatal factors

A case–control study has shown no association between the risk of later life RA and maternal age, maternal marital status, Apgar score at 5 min, multiple birth, congenital malformation, delivery by Caesarean section, season of birth or maternal–child blood group incompatibility (OR 0·6–1·3) [38]. No association has been found between the risk of RA and preterm birth (2 or more weeks premature) [38], with comparable results for RF-positive and RF-negative RA [40].

Prenatal exposure to hormones, in particular diethylstilbestrol (DES), has also been associated with an increased risk of RA [41], although these studies have included very small numbers and results have not always been consistent [42,43].

The infant life

Early life infections

Infections have been proposed as a possible trigger for the onset of RA. It is also shown that infections alter immune system maturation and alter the predisposition to onset of RA in later life. Our group has previously described a significant association between infant hygiene and RF status in adult women [39]. This study observed a lower risk of seropositivity for RF in those sharing a bedroom in childhood (OR 0·48, 95% CI 0·30–0·78, P = 0·003), low birth order (second–fifth birth, and upwards), as well as with lower social class (IIIN–V).

Hospitalization for any infection during the first year of life has, however, been associated with a non-significantly increased risk of RA (OR 1·4, 95% CI 0·8–2·5) [38]. The association between infection during the first year of life and the risk of RA at age 16 years or later has been shown to have a stronger association for seronegative RA (OR any infection 2·6, 95% CI 1·0–7·0) than for seropositive RA (OR any infection 1·2, 95% CI 0·5–2·9), although these associations were based on small numbers. Maternal infections during pregnancy were not demonstrated to be associated with the risk of RA in this study.

Other childhood living conditions

A positive association has been shown with paternal occupation and risk of RA (manual versus non-manual worker; OR 2·8, 95% CI 1·3–5·7) [37]. High weight at 1 year has also been demonstrated to be associated with seropositivity for RF in adulthood [44].

Discussion

A number of studies have investigated the role played by various environmental and early life factors in the onset of RA. To date, studies have shown that maternal smoking is associated with an increased risk of developing RA in girls [33]. In addition, the risk of developing RA has been shown to be increased with high birth weight [36,37]. Reassuringly, low birth weight has been seen to decrease the risk of RA in later life [38]. No association has been shown with preterm birth [38].

Exposure to breast feeding has not been associated consistently with RA, but it may be protective [37]. There are suggestions that exposure to infections during infancy may also influence the development of RA, but these results have been inconsistent and of borderline significance [38].

However, the detail of how multiple factors interact to affect the overall risk of developing RA is still lacking. Such interactions have been described for the effects of environmental exposure during adult life; for example, the independent interaction between smoking, genes and the shared epitope in ACPA and RF-positive and -negative RA [45]. Further studies may also provide information on whether there is a hierarchy of importance among these factors, or if different genes, ethnic and sociocultural backgrounds may have an influence. None the less, it seems likely that environmental factors will interact during early life. Some will influence growth and produce effects on the hypothalamic–pituitary–adrenal (HPA) axis and the development of immune system organs, such as the thymus. Others will stimulate the immune system, such as maternal smoking. In addition, the microbial environment during infant life will shape the immune repertoire. The lack of consistent results in different studies may be due to the fact that more than one environmental ‘hit’ is needed in the development of RA.

The most persuasive association between early life events and the risk of developing RA appears to be seen with maternal smoking in pregnancy and high birth weight. The effects of maternal smoking may well be through the same mechanism of citrullination of proteins seen in adults. Interestingly, the association between RA and birth weight may be through effects on the HPA axis.

Possible mechanisms for high birth weight to influence the risk of developing RA

The finding that high birth weight may be associated with later development of RA has been replicated in a number of studies. If this association is real, what is the probable biological mechanism by which this acts?

One potential mechanism is through effects on the HPA axis, dysregulation of which has been implicated previously in the pathogenesis of RA [46]. The evidence for the importance of the HPA in the development of RA comes from a number of sources. First, the Lewis rat model of adjuvant-induced arthritis has proved a useful method to investigate potential causes of inflammatory arthritis. The hypothalamus of Lewis rats mounts a poor corticotrophin-releasing hormone (CRH) response following injection of the arthritis-inducing adjuvant. Failure to secrete adequate levels of cortisol subsequently appears to allow arthritis to develop, and this can be prevented by the administration of glucocorticoid at the appropriate time [47]. In human studies, polymorphisms in the CRH gene have been identified in individuals with RA [48,49]. Patients with RA have also been demonstrated to have a significantly reduced cortisol response to stress, including a failure to increase cortisol secretion following surgery, compared with chronic osteomyelitis control patients [50]. CRH stimulation tests in these subjects were normal, suggesting hypothalamic, rather than pituitary or adrenal, dysfunction [51].

All these pieces of evidence suggest that the HPA is underactive in RA and that this may be an important factor in the development of disease. A number of studies have shown that birth weight can set the level of functioning of the HPA axis. Plasma cortisol concentrations have been demonstrated to fall progressively with increasing birth weight [52]. Perhaps high birth weight ‘programmes’ a reduced level of cortisol and this leads to a greater propensity to develop RA.

Conclusion

It appears that exposure to environmental factors during early life is important in the aetiology of RA. In particular, high birth weight and maternal smoking in pregnancy seem to have greatest effect on the increased the risk of RA. However, early initiation of breast feeding may be protective against RA (Table 1).

Table 1.

Summary of the effect of early life factors on the risk of rheumatoid arthritis (RA).

Increased risk of RA	Reduced risk of RA
High birth weight	(Low birth weight)
Maternal smoking in pregnancy	Breast feeding
(Hospitalization for any infection during the first year of life)	Sharing a bedroom in childhood (reduced risk of seropositivity for RF)
	(Preterm birth)

RF: rheumatoid factor; statements in brackets ( ), results conflicting or non-significant.

These findings, summarized in Table 2, could be explained by several factors, including stimulation of the immune system and production of ACPA, exposure to infectious organisms and dysregulation of the HPA axis. Evidence for the role of the HPA axis is most persuasive: high birth weight has been shown to increase the risk of RA, and birth weight is related inversely to fasting plasma cortisol concentrations. It is likely, however, that a combination of many non-genetic factors, including those described here, lead to development of RA in those with an existing genetic predisposition.

Table 2.

Summary of the significant results from the studies described.

	RR, OR or SIR (95% CI)	Authors, journal, year [ref. no.]
Maternal smoking in pregnancy (≥10 cigarettes/day)	Effect seen in girls only: OR 2·57 (1·13–5·89)	Jaakkola & Gissler, Int J Epidemiol 2005 [33]
Preterm birth (gestational age <258–266 days)	OR 0·6 (0·7–1·0)	Carlens et al., Ann Rheum Dis 2009 [38]
	RR 1·1 (0·8–1·5)	Simard et al., J Rheumatol 2010 [40]
High birth weight (≥4000–4540 g)	RR 2·1 (1·4–3·3)	Mandl et al., Ann Rheum Dis 2009 [36]
	OR 3·3 (1·4–7·4)	Jacobsson et al., BMJ 2003 [37]
	OR 1·2 (0·6–2·4)	Carlens et al., Ann Rheum Dis 2009 [38]
Low birth weight (<3000 g)	OR 1·4 (0·7–3·0)	Jacobsson et al., BMJ 2003 [37]
	OR 0·7 (0·5–1·0)	Carlens et al., Ann Rheum Dis 2009 [38]
Breast feeding	OR 0·2 (0·1–0·70)	Jacobsson et al., BMJ 2003 [37]
	RR 1·0 (0·7–1·4)	Simard et al., J Rheumatol 2010 [40]
	HLA-DR4-negative; RF-positive versus RF-negative	Young et al., Ann Rheum Dis 2007 [16]
	OR 0·18 (0·04–0·99)
Early life infections	Hospitalization for any infection during the first year of life: OR 1·4 (0·8–2·5)	Carlens et al., Ann Rheum Dis 2009 [38]
	Risk of seropositivity for RF in those sharing a bedroom in childhood: OR 0·48 (0·30–0·78)	Edwards et al., Ann Rheum Dis 2006 [39]
Place of birth	Iraq versus Sweden	Li et al., Arthritis Rheum 2009 [32]
	SIR 1·55 (1·05–2·21)
	Greece versus Sweden	Li et al., Arthritis Rheum 2009 [32]
	SIR 0·31 (0·17–0·52)

HLA-DR: human leucocyte antigen D-related; RR: relative risk; OR: odds ratio; SIR: standardized incidence ratio; CI: confidence interval; RF: rheumatoid factor.

Disclosure

Dr Chris Edwards received advisory fees, speaker fees and unrestricted educational grants from Abbott, Roche, Wyeth, Schering Plough and UCB-Pharma. Dr Alexandra Colebatch received educational support from Abbott and undertakes clinical trials with Schering-Plough.