Juvenile idiopathic arthritis in relation to perinatal and maternal characteristics: a case control study

Background Existing data on associations between maternal and early childhood exposures and juvenile idiopathic arthritis (JIA) risk is scant and inconsistent with previous studies showing potential role for prematurity, number of siblings and infections. We explored JIA and International League of Associations for Rheumatology (ILAR) JIA categories in relation to selected infant (birthweight, size-for-gestational-age, gestational age), and maternal (parity, delivery type, prior fetal loss) characteristics that may be markers for exposures related to two pathways (hygiene hypothesis, microchimerism) potentially associated with autoimmune disorder occurrence. Methods A case–control analysis with 1,234 JIA cases and 5,993 birth year-matched controls was conducted. Exposure information was obtained from WA state birth certificates. Multivariable logistic regression was used to estimate adjusted odds ratios (OR) and 95% confidence intervals (CI) for associations with maternal and early life exposures for JIA and JIA categories. Results Greater maternal parity was associated with a decreased OR for JIA (most marked for persistent oligoarticular JIA, OR 0.32, 95% CI 0.15; 0.71, p for trend = 0.0001). Prior fetal loss (except for oligoarticular JIA) was associated with an increased OR for JIA. Prematurity was associated with increased risk of enthesitis related arthritis (OR 1.9, 95% CI: 1.3–2.9) and rheumatoid factor positive polyarticular JIA (OR 2.2, 95% CI: 1.0–4.8). Conclusions We observed associations of selected maternal factors with JIA, some of which varied across JIA categories. The findings of decreased ORs for JIA in relation to greater maternal parity may be consistent with the hygiene and microchimerism hypotheses. Future studies with biomarkers relevant to these hypotheses will help elucidate any associations.


Background
Juvenile Idiopathic Arthritis (JIA) is a heterogeneous group of seven categories of chronic inflammatory arthritides in children with onset before 16 years of age [1]. JIA likely has a complex etiology, with multiple genetic and environmental factors contributing to its development [2]. Concordance rates of JIA among monozygotic twins are estimated at 25-40%; suggesting that environmental factors may influence JIA risk [3].
Evidence suggests that adult rheumatoid arthritis (RA) [4], JIA [5,6], and other autoimmune diseases such as multiple sclerosis [7,8] may be associated with maternal or early childhood exposures such as maternal parity, birthweight, or gestational age.
Two hypotheses have been suggested relating perinatal and maternal reproductive characteristics to JIA. The "Hygiene hypothesis" suggests that early exposure to infection results in decreased risk of childhood autoimmune disease [9] due to modulation of the developing immune system. Proxy measures often used to assess this include maternal parity (a greater number of prior births being consistent with a greater number of older daycare-or school-aged siblings at home and greater possibility of early infection); infant birthweight, small size-for-gestational age, or gestational age (small, or preterm infants may be less resistant to infection), and mode of delivery (vaginal deliveries may more likely elicit a normal immune response due to early exposure to vaginal flora). The microchimerism hypothesis relates to maternal and/or sibling origin microchimerism (harboring of cells or DNA from another genetically distinct individual) from exchange of cells/DNA in utero from the mother or older siblings during gestation [10]. Microchimerism may potentially increase or decrease disease risk based on the specificity of the acquired genetic material from another individual [11], and some studies suggest that its presence may alter rheumatoid arthritis (RA) risk in women [12][13][14][15][16][17] although we could locate no studies of this in JIA. It has been suggested that the decreased RA risk in parous women may be due to microchimerism among women with prior births [12]. Proxy measures potentially associated with increased microchimerism may include greater maternal parity (greater likelihood that infant has acquired genetic material from older sibling(s)); plurality (twins and higher-order multiples may have acquired genetic material from their sibling in utero); and a maternal history of prior fetal loss (maternally-retained genetic material from a fetus may be passed to subsequent offspring).
Lacking direct indicators of early prior infection relevant to the hygiene hypothesis, or of the presence of microchimerism, we explored associations of JIA and its individual categories in relation to selected maternal reproductive and infant characteristics as proxies for these, as assessed by birth certificate data among children with, and without JIA. Elucidation of these relationships will enhance understanding of possible JIA risk factors.

Study design
This case-control study identified JIA cases at a regional children's hospital in the Seattle-Puget Sound region of Washington State using methods described previously [18]. Appropriate Institutional Review Board approvals were obtained. Potential JIA cases were identified though International Classification of Diseases (ICD-9), Ninth Revision diagnosis codes (720.0, 696.0, 714.32, 714.31, 714.30, 720.89, 714.33, 714.3) from 1997-2010 (N = 2,274) and linked to WA State birth records to identify children born in-state (N = 1,518). Controls were randomly selected in a 4:1 control: case ratio from remaining birth records, matched on year of birth (N = 6,072).
Chart review further refined JIA identification and categorized cases into ILAR categories including: polyarticular rheumatoid factor (RF) positive, polyarticular RF negative, oligoarticular arthritis (persistent and extended), psoriatic JIA, enthesitis related arthritis, systemic JIA, and undifferentiated JIA. We excluded 228 cases who did not meet ILAR criteria [1], and 38 cases with disease onset at > 16 years, resulting in 1,252 cases for analyses. After examining the OR for JIA in relation to plurality status, all remaining analyses were restricted to singleton infants only (1,234 cases; 5,993 controls) because characteristics (for example, birthweight and gestational age) of infants and pregnancies differ for singleton and multiple gestation pregnancies.

Statistical analysis
Multivariable logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CIs) for associations, adjusted for maternal age, and the matching variable birth year. Other factors evaluated for possible effects on the relationships of interest included paternal age; maternal race/ethnicity, educational level, prenatal smoking, and marital status; medical insurance at birth (Medicaid vs. other); sex; and trimester prenatal care began. We adjusted for factors that altered the OR by greater than 10% in univariate analyses. Potential effect modifiers were evaluated by inspection of stratum-specific ORs and the Breslow-Day test for homogeneity. Trends were evaluated using Wilcoxon rank sum test. Analyses were initially conducted to examine JIA overall, and repeated for JIA categories (restricted to factors with at least 5 subjects in a cell). Analyses were conducted using Stata (version 11; StataCorp, College Station, Texas).

Infant and maternal characteristics in relation to JIA
Multiple gestation pregnancy was not associated with JIA (Table 1); OR 1.04, 95% CI: 0.62-1.74). Among singletons, birthweight >4000 g was associated with a decreased OR for JIA (OR 0.81, 95% CI: 0.67-0.98) ( Table 2). The OR for JIA decreased with increasing number of prior maternal deliveries, with the lowest OR observed for children of women with two prior deliveries (0.68, 95% CI: 0.56-0.83 for 2 deliveries, p for trend =0.0007). No associations of JIA were observed with gestational age, size-for-gestational age, C-section delivery or maternal history of prior fetal losses.
The pattern we observed for overall JIA of decreasing ORs in relation to increasing number of maternal prior deliveries was marked for persistent oligoarthritis, with a monotonically decreasing trend ranging from 0.88 (95% CI: 0.69-1.12) for one prior birth, to 0.32 (95% CI: 0.15-0.71) for 4+ prior deliveries (p for trend =0.0001), and a suggestion of trend for extended oligoarthritis. Non-statistically significant decreased ORs were observed for greater numbers of prior deliveries for most other JIA categories, except for RF-(OR 1.24, 95% CI: 0.70-2.18 for 4+ prior deliveries) and RF+ (OR 1.77, 95% CI: 0.75-4.14 for 3+ prior deliveries) polyarthritis.

Discussion
We observed a decreased risk of JIA overall in relation to greater maternal parity, with a significant monotonic trend for persistent oligoarthritis, and suggestion of decreased ORs in relation to increasing maternal parity for most other JIA categories. This finding may be consistent with the hygiene hypotheses as children with older siblings likely have greater opportunity for infection exposure, both in utero and during post-natal life. Additionally a modest, non-statistically significant increased risk of JIA with Cesarean delivery similar to a Swedish registry report [5] provides weak support for the hygiene hypothesis, although this has not been supported in all studies [4]. Consistent with our data, a previous Australian clinicbased study showed that having any siblings (vs. none) was protective for JIA in (OR 0.46, 95% CI 0.28; 0.74, using clinic-based controls); a result slightly attenuated (OR 0.50, 95% CI 0.24; 1.02) when community controls were used [10]. An OR of 1.6 for JIA was observed in relation to being an only child in a Danish registry-based study [19]. In contrast, no risk reduction was observed in relation to having increased numbers of older siblings in a Swedish registry-based study [5]. Birth order was not found to be associated with JIA in two previous studies [6,20].
The presence of microchimerism might increase or decrease JIA risk based on the nature of the acquired genetic material. An inverse association of JIA with greater maternal parity may be related to the presence of    No. prior fetal losses ((≥20 w) c, d None In our study, most JIA category-specific risk estimates for maternal fetal loss were increased; a finding that could be consistent with the microchimerism hypothesis only if presence of increased levels of acquired microchimerism from a mother's prior pregnancy carried JIA-associated risk HLA alleles [11]. However, it has been demonstrated that mothers of children with JIA have impaired reproductive capacity [21], so history of maternal fetal loss may also simply be an indicator of this observed association. Ultimately, we had no information about specific gene traits of children or their families, and this is speculation.
In contrast with previous studies from Sweden and United States [5,6], high birthweight in our study was associated with a significantly decreased risk of JIA overall. Reasons for this difference are not clear.
Previous studies related to a gestational age effect vary. Although Swedish registry data suggest a borderline decreased OR in relation to preterm delivery [5], Shenoi et al [6] noted a significantly increased risk of JIA in relation to prematurity. While our results suggest no association with gestational age overall, enthesitis-related arthritis and RF+ polyarticular JIA had significantly increased ORs in relation to preterm gestation. Mothers of children with JIA have higher rates of preterm delivery [21] and the association of prematurity with JIA needs to be further explored in future studies.
The use of outpatient and inpatient medical records maintained by board-certified pediatric rheumatologists at a large regional children's medical center, with data linkage to state birth records, are major strengths of this study enabling identification of a relatively large number of cases. Exposure information was assessed from birth records prior to onset and diagnosis of JIA, and was not subject to recall bias. Because of the known polygenic genetic predispositions and heterogeneity of the pathogenesis of differing JIA categories, it is important to examine these relationships separately for each category; to our knowledge, this is the first study to do so. Medical chart review allowed us to confirm and ascertain JIA categories, thereby reducing misclassification.
Some limitations of our study include possible inaccuracies in reporting of mother's reproductive history; in particular, inaccurate reporting of fetal loss on the birth certificates may have lead to nondifferential misclassification, biasing the OR estimates towards the null. Birthweight and number of prior live births, however, have been assessed as reliable in comparison to medical records [22]. There is potential for unmeasured confounding due to unrecognized pregnancy loss, psychosocial support factors, and lack of information on fertility, or residual confounding due to the necessity of reliance on birth records for exposure information. The inability to measure these could lead to residual confounding in our risk estimates and could bias the estimates either way. Although we attempted to identify all children with JIA at the sole facility with boardcertified pediatric rheumatologists in Washington State during the study period, and free care is available for families without resources, it is possible that our JIA cases may not be representative of all children with JIA in Washington State. Some cases may have been treated by adult rheumatologists or other practitioners outside this facility, or they may differ from the underlying population in ways that we could not measure. We attempted to evaluate any related bias due to this, although adjustment for socio-economic and other factors potentially related to care access had no effect on the results. Our inability to examine the presence of younger siblings is also a limitation.

Conclusions
We observed decreased JIA risk with increasing parity, which potentially supports the hygiene or microchimerism hypotheses, with consistent results demonstrated in many JIA categories. Of note, our results indicate the need to consider JIA categories as separate entities; indeed our results were often inconsistent across categories, suggesting potentially different etiologies. Given the current lack of preventative or curative treatments for JIA, further studies, particularly those capable of examining category-specific JIA associations, are warranted. A better understanding of these associations will help inform prevention programs and hopefully reduce the burden of this debilitating disease.