Impact of pregnancy on progression of preclinical autoimmune disorders: a prospective cohort study

Abstract

Objectives

The objective of this study was to evaluate the role of pregnancies in the progression from the preclinical phase of autoimmune disorder to a definite rheumatic disease.

Methods

A cohort study of women with symptoms and laboratory findings suggestive for autoimmune disorder were enrolled during the first trimester of pregnancy and followed-up for 5 years with clinical and laboratory assessment. Multinomial logistic regression was used to compute the risk of progression to definite autoimmune disease correcting for confounders.

Results

At the end of follow-up, out of 208 subjects, 81 (38.9%) were considered negative, 53 (25.5%) had symptoms and abnormalities of autoantibody profile compatible with a non-criteria rheumatic status and 74 (35.6%) had a definite rheumatic disease (43 undifferentiated connective tissue disease, 5 systemic lupus erythematosus, 3 SS, 10 antiphospholipid syndrome, and 12 miscellaneous autoimmune disorders). The median time from enrolment to definite diagnosis was 28 months (interquartile range = 18–42). The rate of progression towards a definite autoimmune disease was 47.1% (48/102) among subjects with one or more subsequent viable pregnancies compared with 24.5% (26/106) of those with no subsequent pregnancies (adjusted odds ratio = 4.9, 95% CI: 2.4, 10). The occurrence of preeclampsia during the index pregnancy or subsequent pregnancy was an additional and independent risk factor for progression to a definite autoimmune disease (adjusted odds ratio = 4.3, 95% CI: 1.2, 14.8).

Conclusions

Among women with suspected autoimmune disease during pregnancy, additional viable pregnancies and diagnosis of preeclampsia were independently associated with an increased rate of progression to definite rheumatic disorder. Hormonal modifications associated with pregnancy could worsen preclinical rheumatic disorders favouring their progression to a defined autoimmune disease.

Introduction

Autoimmune CTDs are a heterogeneous group of chronic disorders characterized by inflammation of various tissues, production of non-organ-specific autoantibodies and clinical manifestations including nephritis, neurological disorders, thrombosis and arthritis. These disorders are more prevalent in women and often occur during childbearing age with negative effects on both maternal and neonatal outcomes [1]. Literature suggests that the diagnosis of autoimmune rheumatic diseases is preceded by a preclinical phase lasting from months to several years and characterized by mild symptoms and the presence of detectable autoantibodies without meeting the diagnostic classification criteria [2].

Systemic autoimmune rheumatic diseases even in their preclinical phase are associated with increased rates of adverse pregnancy outcomes [3, 4] and with increased risks of subclinical atherosclerosis, ischaemic heart disease and lung disease [2, 5]. Several studies have shown that obstetric adverse outcomes such as miscarriages, fetal growth restriction (FGR) and preeclampsia can occur before the established diagnosis of rheumatic diseases [6–8], suggesting a potential causal role of unrecognized autoimmune disorders in the event of pregnancy complications [3]. Since this evidence comes from retrospective studies, it remains unclear whether adverse pregnancy outcomes are the result of a worsening of rheumatic disorder or the adverse pregnancy outcome itself could be the trigger for subsequent progression of autoimmune disease.

Concerning the influence of pregnancy on rheumatic autoimmune disease, literature is rich in studies focusing on the frequency of flares and the evolution of overt CTDs during in pregnancy [9]. However, our experience suggests that screening measures in pregnancy based on a two-step approach with a self-reported questionnaire and autoantibody profile identifies early, incomplete or undifferentiated rheumatic disorders, which can affect pregnancy outcomes [4, 10]. The purpose of this study was to evaluate the potential role of pregnancy in the clinical evolution of early autoimmune disorders. The study was a prospective 5-year follow-up of a cohort of women enrolled during their first pregnancy and with symptoms and signs suggestive for an early or defined rheumatic disorder.

Methods

Subjects were recruited among pregnant women who attended our department during the first trimester of pregnancy. Enrolment criteria included: (i) nulliparity; (ii) singleton pregnancy; (iii) antenatal care and delivery at our department; (iv) no previous diagnosis of rheumatic disease; and (v) absence of fetal malformations or chromosomal anomalies.

As previously described [10], after written informed consent each woman was asked to complete a screening questionnaire including any connective tissue disorder symptoms (Fig. 1). Women who answered positively to one or more of the questions were tested for the presence of circulating autoantibodies, including ANA, anti-double-stranded DNA (anti-dsDNA), anti-extractable nuclear antigen (ENA), antiphospholipid antibodies (aPL), in particular anticardiolipin antibody (aCL), anti-β₂-glycoprotein I antibodies (aβ2GPI) and lupus anticoagulant (LAC), according to standardized methods, as previously described [10]. The ANA test was considered positive at a titre ≥1:80. Anticardiolipin antibodies were considered positive if >40 MPL-GPL/ml and low-titre if 10–40 MPL-GPL/ml; aβ2GPI antibodies were considered positive if >10 U/ml and low-titre if 7–10 U/ml.

Women who tested positive both to with the questionnaire and to the autoantibody profile were referred to the rheumatology unit of our hospital for further clinical evaluation. Rheumatologists were unaware of the results of the questionnaire.

The population for the study was a selected cohort of women with symptoms plus autoantibodies who were prospectively followed up for 5 years by the same obstetric-rheumatological team. The pregnancy ongoing during the enrolment was defined as index pregnancy. The study complied with the Declaration of Helsinki and was approved by the local ethics committee of our foundation (686-rcr2011-bis-23).

During both the index pregnancy and subsequent pregnancies, subjects underwent monthly clinical and ultrasonographic evaluation. In order to monitor the evolution of rheumatic disorders during pregnancy, patients were followed up with clinical-rheumatological and laboratory evaluations every 1–2 months during pregnancy and every 6–12 months after delivery, depending on the severity of symptoms.

Rheumatic diseases were classified according to widely used criteria for UCTD [11], SLE [12], spondyloarthropathies [13], APS [14], SS [15], SSc [16], MCTD [17] and Behçet’s disease [18]. Patients with suspected rheumatic disease (symptoms plus autoantibodies) but not fulfilling the above-mentioned criteria were classified as the non-criteria for diagnosis group. The diagnosis of rheumatic disease was excluded in women with disappearance of symptoms and who tested negative for autoantibodies at subsequent evaluations; these subjects constituted the control group.

Preterm delivery was defined as birth before the 37th week of gestation. The diagnosis of FGR was made according to standard guidelines [19]. Preeclampsia was diagnosed according to standard criteria [20]. Small for gestational age (SGA) infants were diagnosed when birth weight was below the 10th percentile of the Italian population curves [21]. Data on demographic and clinical characteristics, pregnancy, delivery, neonatal outcomes, rheumatological symptoms and the autoantibody profile were collected and stored in a dedicated database.

Continuous and categorical data were compared by Kruskal–Wallis analysis of variance and χ² test, respectively. The Bonferroni correction was used to adjust P-values in case of comparisons of either continuous or categorical data between multiple categories. Time to the progression to a definite diagnosis of a rheumatic disorder was modelled as a survival variable and the role of exposure variables (preeclampsia, FGR) on the speed of progression was evaluated by Kaplan–Meier survival analysis and curves. Logistic regression with multiple outcomes (negative, non-criteria and definite rheumatic disorders) was used to evaluate the role of subsequent pregnancies after inclusion in the study on the occurrence of outcomes adjusting for potential confounders.

Results

During the study period 4108 women attended our department during first trimester and completed the questionnaire. The proportion of subjects who answered positively to the questionnaire was 10.2% (420 of 4108). Of the 420 subjects with symptoms, 286 (68.1%) tested positive for autoantibodies and were sent to the rheumatology unit for evaluation and follow-up. Out of 286 first trimester pregnancies with suspected rheumatic diseases enrolled in the study in the period 2013–2016, 208 were followed up after the index pregnancy for at least 5 years. The 78 subjects excluded from the study were either followed up at other institutions (50 subjects) or declined to participate into the study (28 subjects). At the end of follow-up, women were divided into three groups according to a definite rheumatological clinical and laboratory evaluation: negative, non-criteria for a definite rheumatic disorder and definite rheumatic disease.

Overall, rheumatological symptoms and autoantibody profile at the index pregnancy are reported in Table 1. Arthralgia was one of the most common symptom (89/208). None of the six pregnant women who had a history of thrombosis (all peripheral thrombosis) tested positive for aPL antibodies. Overall, there were only two pregnant subjects who tested simultaneously positive to two aPL (aCL IgG and anti-β2GPI) but without a history of thrombosis. Among antiphospholipid antibodies the most frequent were aCL IgM and aCL IgG (50/208; 24%). ANA positivity had a frequency of 88.5% (184/208) and 53.3% (98/184) had positivity ≥1:160.

Table 2 reports the association between autoantibody profile and symptoms at enrolment and final diagnosis. Subjects with ANA titre ≥1:160 or positive for ENA (mainly anti-Ro) had an increased rate of a specific rheumatic disease diagnosis at follow-up. Compared with asymptomatic women with negative autoantibody status at 5 years of follow-up, subjects who developed a definite rheumatic disease had lower rates of photosensitivity and increased rates of arthralgia and xerostomia.

At the end of 5-year follow-up 81 (38.9%) subjects tested repeatedly negative for ANA or other autoantibodies, were asymptomatic and were considered negative, 53 (25.5%) had mild symptoms and abnormalities of autoantibody profile and were considered as non-criteria rheumatic disease status and, finally, 74 (35.6%) had a definite rheumatic disease. Among patients with a definite diagnosis there were 43 (20.7%) UCTD, 10 (4.8) APS, 6 (2.9%) SLE, 3 (1.4%) SS and 12 (5.8%) miscellaneous rheumatic disorders (8 spondyloarthropathies, 2 Behçet’s disease, 1 MCTD and 1 systemic sclerosis). The median time to a diagnosis of a definite rheumatic disease after the index pregnancy was 28 months (interquartile range [IQR] 18–42), whereas the final diagnosis of negative or non-criteria rheumatic disease status was made after 5 years of follow-up. During the follow-up, the median number of rheumatological visits was 14 (IQR 10–17).

During the index pregnancy, six women were treated with low molecular-weight heparin for previous thrombosis and two with steroids for severe arthropaties in the third trimester. After 5 years, therapy was given to 46 (62.2%) subjects with diagnosis of a definite rheumatic disease: 37 women were treated with HCQ, 11 with steroids and 8 with immunosuppressant.

Out of 208 women enrolled, 102 had subsequent pregnancies (Table 3). Compared with asymptomatic women with negative autoantibody status at 5 years of follow-up, subjects who developed a definite rheumatic disease had increased rates of subsequent viable pregnancies, spontaneous abortions and overall diagnosis of preeclampsia in the index pregnancy. The rates of a definite rheumatic disorder at the end of follow-up were 24.5% (26/106) among subjects with no subsequent pregnancies, 44.4% (40/90) among those with subsequent normotensive pregnancies and 66.7% (8/12) in those with subsequent pregnancies complicated by preeclampsia (P < 0.001 by χ² analysis). On the other hand, the rate of a definite rheumatic disease diagnosis was also associated with the incidence of FGR in the subsequent pregnancies being 24.5% (26/106) in uniparas, 43.5% (37/85) in subsequent pregnancies with adequate fetal growth and 64.7% (11/17) in subsequent pregnancies complicated by FGR. The speed of progression to a definite rheumatic disease was higher among women with subsequent pregnancies complicated by preeclampsia and/or FGR (Fig. 2).

To evaluate the independent role of variables associated with subsequent development of a definite rheumatic disease, we used multinomial logistic regression. The model used and the results are reported in Table 4. After adjustment for the confounder effect of the age of the subjects, one or more viable pregnancies after enrolment and a diagnosis of preeclampsia in index or subsequent pregnancy were independently associated with an increased risk of a definite rheumatic disease in the 5 years following the index pregnancy (OR 4.9, 95% CI: 2.4, 10; P < 0.05). Overall, and after adjustment for the age of the subjects, an increasing number of subsequent viable pregnancies (χ² for trend = 12.9, P < 0.001) or abortions (χ² for trend = 5.3, P = 0.02) after the enrolment were correlated with increased rates of progression to a definite rheumatic disorder. However, when the occurrence of abortion was examined as a categorical variable (yes, no), and after adjustment for confounders, it was uncorrelated to the progression to a definite rheumatic disease (OR = 1.6, 95% CI: 0.39, 6.27).

Discussion

In this study, we prospectively followed up a cohort of women with a suspected rheumatic disease detected in the first trimester of pregnancy through a two-steps screening method including the administration of a questionnaire of symptoms and autoantibody testing. After 5 years, out of the 208 women followed up, 38.9% were negative and asymptomatic, 25.5% had non-criteria rheumatic disorders and 35.6% had a definite diagnosis of a rheumatic disease. Women with a final diagnosis of a rheumatic disease were more likely to have ANA ≥1:160, ENA positivity, arthralgia and xerostomia at enrolment. Compared with negative subjects they had higher rates of preeclampsia in the index or subsequent pregnancies, and increased rates of subsequent viable pregnancies or miscarriages. In multivariate analysis, after correction for confounders, one or more viable pregnancies after enrolment and a diagnosis of preeclampsia either in index or in subsequent pregnancies resulted independently associated with an increase in the risk of progression to a definite rheumatic disease.

The strengths of this study were the use of two-steps screening in a cohort of pregnant women without a previous diagnosis of autoimmune rheumatic disease, the enrolment at first trimester of their first viable pregnancy, a prospective and longitudinal approach, a standardized evaluation of pregnancy outcome and an adequate follow-up to diagnose progression of rheumatic disorders. Main limitations include the relatively small sample size, which makes it difficult to evaluate the effect of pregnancies on different rheumatic disorders. In addition, since this was not a population study and did not include infertile women, the results could not be generalized and the relationship between number of viable pregnancies and risk of progression to a definite rheumatic disease should be interpreted cautiously.

To our knowledge, this was the first cohort study evaluating the clinical evolution from symptoms suggestive of rheumatic disorders to a definite rheumatic disease associated with pregnancy. The rate of rheumatic disorders diagnosed during the index pregnancy at enrolment was similar to that reported in a previous study of 5232 pregnant women screened during the first trimester of pregnancy with a two-steps approach including a questionnaire and autoantibody profile [3]. In this study, arthralgia and xerostomia at enrolment were the most predictive symptoms of a subsequent diagnosis of rheumatic disease. This finding is coherent with previous data in a cohort of pregnant women at first trimester with a suspicion of autoimmune rheumatic disease, where arthralgia/swollen joint and xerostomia were more frequent in patients with an undiagnosed UCTD or major CTD [10, 22]. Data from pregnant women with a diagnosis of UCTD showed a prevalence of joint involvement of 57.9% and of xerostomia of 15% [23]. Recently, even in non-pregnant UCTD patients, joints symptoms were found to be the most frequent manifestations in case of progression to a definite CTD [24].

Concerning laboratory findings, in our cohort the prevalence of ANA ≥1:160, ENA and anti-Ro positivity at enrolment was higher in patients with a subsequent diagnosis of a definite rheumatic disease. Studies investigating the natural history of SLE in the general population showed that high-titre ANA and ENA were detected 3.4 years before the onset of disease [25]. Previous data from pregnant women screened during the first trimester showed an ANA or ENA positivity >90% among the subjects with a subsequent diagnosis of a definite rheumatic disease [10]. Finally, in a group of non-criteria patients at first trimester of pregnancy, ANA positivity and arthralgia/swollen joints were found to be independently associated with an increased likelihood of persistence of symptoms and with the diagnosis of a definite rheumatic disorder (OR 5.1 and 2.8 for ANA positivity and arthralgia respectively) [4].

In our study, in a 5-year follow-up, the median time from a suspicion of rheumatic disorder to a definite disease was 28 months (IQR 18–42). This longitudinal approach allowed the evaluation of clinical evolution from mild rheumatic symptoms and autoantibody positivity to a formal diagnosis of a rheumatic disease. In addition, it allowed detection of false antibody tests and transient symptoms that can be common in the general population should not be considered as preclinical rheumatic disease. Data from literature investigating the influence of pregnancy on the evolution of autoimmune diseases were mostly retrospective and focused on differentiation of established UCTD into CTD and on frequency of flares [23, 26, 27]. Castellino et al. [26], during a 15-month follow-up, reported an increased rate of flare (32%) and evolution into a well-defined CTD (12%) in pregnant UCTD patients compared with non-pregnant UCTD patients (11% flare rate and 2% differentiation to definite CTD). A more recent cohort of 81 UCTD women followed up in pregnancy and postpartum period [27] presented 13% flare rate and 3% differentiation to SLE. Furthermore, a multicentre retrospective study on 133 women with UCTD reported a rate of disease evolution to a definite CTD of 12% in a mean time of 5 years [23].

Whether pregnancy or parity could influence the development of autoimmunity remains the subject of debate: results of studies on the association between pregnancy and SLE and rheumatoid arthritis showed a weak correlation or were contradictory [28–30]. A wide retrospective cohort study from medical registers [31] suggested a slight association between pregnancy and the risk of subsequent maternal autoimmune diseases, finding a higher relative risk (RR) in the first year after delivery (vaginal delivery RR = 1.1 [1–1.2] and caesarean delivery RR = 1.3 [1.1–1.5]) compared with miscarriages. As parity has been already considered as a potential risk factor for autoimmune disease [28–30], in this study we enrolled only women at their first viable pregnancy, in order to avoid the bias of previous multiparity. In our study, one or more viable pregnancies following the index pregnancy was an independent risk factor for the progression to a rheumatic disease, with a strong association (OR 4.9). Furthermore, the risk of development of a definite rheumatic disease was independently associated to a diagnosis of preeclampsia in index or subsequent pregnancies (OR 4.3), suggesting that pregnancy complications precede and lead the progression to rheumatic disorders. This finding was coherent with a previous large retrospective study that found a significant correlation between hypertensive disorders of pregnancy and development of aPL antibodies with subsequent CTD [32]. Moreover, higher rates of preeclampsia, gestational hypertension and FGR were also found in pregnant UCTD women who later developed a definite CTD [23]. Since asymptomatic or paucisymptomatic stages of rheumatic disorders are accompanied by an autoimmune response facilitating both pregnancy failures and the development of a faulty placentation [25, 28, 29] leading to preeclampsia, the association between disorders of pregnancy and subsequent progression to a rheumatic disease seems obvious. Less clear is the independent association between the number of normal pregnancies and subsequent progression from mild symptoms to a definite rheumatic disorder [1]. Although sex hormones are considered to play a main role in the higher prevalence of rheumatic diseases among females compared with males, nulliparity has been classically associated with an increased risk of several rheumatic disorders [33]. This association has been ascribed to the negative effect of autoimmunity on fertility and on the time to conceive [1, 33]. The association between parity and risk of rheumatic disease has been conflicting, although most of the evidence derives from retrospective case–control studies [1, 33]. The main difference of our study compared with others was the inclusion of pregnant subjects in the follow-up, thus excluding subjects with fertility problems eventually associated with autoimmune disorders.

The biological basis of the potential relation between autoimmune disorders and number of pregnancies involves the important immunological and hormonal changes of pregnancy itself [34]. During pregnancy and the post-partum period there is an increased risk of flares and disease activity in SLE subjects and an increase in the progression from UCTD to a definite CTD [1, 23, 26]. The mechanism favouring the progression from mild UCTD to a definite rheumatic disease has been attributed to the predominantly T helper (Th) 2-type cytokine profile typical of pregnancy [23, 26]. In addition, both oestrogens and progesterone during pregnancy could increase the Th2-type and reduce the Th1- and Th17-type response dependent on oestrogen receptors in T cells, favouring the disease progression of several autoimmune diseases such as SLE or SSc [35].

In conclusion, in a cohort of pregnant women with suspected rheumatic disorders we found 35.5% of evolution into a definite disease in a median time of 28 months. The rate of progression from the preclinical phase to a definite diagnosis of rheumatic disease appeared to be positively correlated with the number of subsequent viable pregnancies, and to preeclampsia at enrolment or in subsequent pregnancies. Immune modification associated with pregnancy such as a modified Th2 T cell response could be responsible for the progression of preclinical disorders to a definite CTD and to SLE.

Data availability

All data relevant to the study are included in the article. Additional data are available upon reasonable request.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors have declared no conflicts of interest.

References