Association between parental autoimmune disease and childhood atopic dermatitis varied by sex: a nationwide case–control study

Atopic dermatitis (AD) is a common inflammatory skin disorder induced by dysfunction of immune suppression sharing similar pathogenesis to autoimmune diseases. To explore the association between autoimmune diseases and AD in children, we linked the birth data from National Birth Registry with National Health Insurance Research Database. There were 1,174,941 children obtained from 2006 to 2012 birth cohort. A total of 312,329 children diagnosed with AD before 5 years old were compared to 862,612 children without AD in the control group. Conditional logistic regression was utilized to calculate adjusted odds ratio (OR) and Bonferroni-corrected confidence interval (CI) for overall significance level of 0.05. In 2006–2012 birth cohort, the prevalence rate of AD was 26.6% (95% CI 26.5, 26.7) before 5 years of age. Having parental autoimmune disease (including rheumatoid arthritis, systemic lupus erythematosus, Sjogren’s syndrome, ankylosing spondylitis, and psoriasis) was associated with a significant higher risk of children AD development. The other associated factors were maternal obstetric complications (including gestational diabetes mellitus and cervical incompetence), parental systemic diseases (including anemia, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, hyperthyroidism, and obstructive sleep apnea), and parental allergic disease (including asthma and AD). The subgroup analysis showed similar results between children’s sexes. Moreover, maternal autoimmune disease had higher impact on the risk of developing AD in the child compared with paternal autoimmune disease. In conclusion, parental autoimmune diseases were found to be related to their children’s AD before 5 years old.


Introduction
Atopic dermatitis (AD), also called atopic eczema, is one of the most common chronic inflammatory skin disease. It affects 15-20% of the world's population, including 20% of children and up to 3% of adults [1]. Moreover, AD can induce subsequent allergic diseases, such as allergic rhinitis and asthma within children, which is often referred to as the concept of "allergic march" [2]. The intense itching and skin manifestations in AD often result in insomnia and poor school or work performance. AD affects not only child development but also the caregiver's quality of life.
AD is a multifactorial disease presenting with different endophenotypes [3]. These variable risk factors include parental systemic diseases, perinatal factors, family allergic history and parental autoimmune diseases [4][5][6]. Previous studies have proved that more mothers of children with AD had received therapy for autoimmune diseases compared to mothers of normal children [6]. However, both maternal and paternal autoimmune diseases may be associated with the occurrence of allergic diseases in their children [6][7][8].
Genes play an important role in linking AD and autoimmune diseases. The commonality of these two diseases include dysfunction of immune suppression, overlapping genetic loci, similar molecular pathways, and disease-implicated cell type [9]. Further study was conducted to figure out the genetic effect on specific autoimmune diseases and allergy. It has been discovered that there are many overlapping regions in the loci of AD and psoriasis, such as chromosome 1q12, 17q25, and 20p12 [10]. And the variant genes on 5p13.2 encoding the IL7 receptor, a key mediator of T-cell-driven autoimmunity, were noted within patients with either AD or multiple sclerosis [11].
All of the aforementioned research suggested that parental autoimmune diseases are associated with allergic disease in their offspring; however, they did not show the common effect on their offspring when both paternal and maternal autoimmune diseases were present. We also wonder whether parental autoimmune diseases will lead to different outcomes between male and female children. Our study hopefully will help clinicians raise awareness of the children at risk of AD and provide them with early treatment, which will improve both the children and their parents' quality of life. The present study is a population-based case-control study using data from the Taiwan National Health Insurance Research Database (NHIRD) and perinatal data in the National Birth Registry.

Data source
In this study, we linked perinatal data from the National Birth Registry (2006)(2007)(2008)(2009)(2010)(2011)(2012) with NHIRD (2003NHIRD ( -2017 to obtain the records of family health history for multiple generations, the relationship between children with their birth parents, and associated demographic data. The Taiwan's National Health Insurance (NHI) was established in 1995 and covered up to 99.99% of Taiwan's population in 2010s. The NHIRD was the subset of NHI insurance claim record and released for research purposes to provide populationbased evidence and support clinical decisions. The NHIRD comprised comprehensive demographic characteristics, clinic visits, hospitalization dates, disease diagnosis codes, procedure codes, prescription codes, and medical costs for reimbursement. This study was approved by the Institutional Review Board (IRB approved number: CS19009) of The Institutional Review Board of Chung Shan Medical University Hospital in Taiwan. Need for informed consent was waived because the data we used comprised a de-identified secondary data set which had been released for research purposes and analyzed anonymously.

Study population
The selection process of the study population is shown in Fig. 1. There were 1,337,660 infants born from 2006 to 2012 in the record of National Birth Registry. In this dataset, it is difficult to identify the correct child identification number among multiple births, so we excluded multiple birth infants (n = 38,295) from analysis. We also excluded stillbirth (n = 438), missing of demographic data (n = 40,246), those with less than 5 years of data (n = 3,445), and their mothers were foreign nationalities (n = 80,295). Lastly, the children who received a diagnosis of AD (ICD-9-CM code 691) before 5 years old would be identified because the prevalence of AD would reach the peak within the interval of 0-5 years old in the progression of allergic march [2,12]. Subjects with AD were defined as having at least three ambulatory visits or one hospital admission with a diagnosis of ICD-9-CM code 691. The validation of ICD-9-CM code 691.8 for AD was proven to have excellent positive predictive values in the inpatient setting [13]. To avoid miscoding for AD, though the code of unspecified contact dermatitis (ICD-9-CM code 692.9) is often misused for eczema, we only identified the patients with code 691 [14]. A total of 1,174,941 singleton births were selected for analysis, and 312,329 (26.58%) of children developed AD prior to the age of 5 years.

Statistical analysis
For all baseline characteristics, the counts and the proportions were calculated for categorical variables among children with AD and children without AD. In the 2006-2012 birth cohort, the prevalence and absolute prevalence difference of child AD was calculated and stratified by perinatal characteristic and parental comorbidities, the variance and 95% confidence interval (CI) of prevalence were estimated by the method of Taylor series approximation. A conditional logistic regression model was conducted to estimate the adjusted odds ratio (aOR) and confidence interval (CI) after adjusting for relevant covariates. The multiple comparison inflated the type 1 error, and we used the Bonferroni correction method to adjust the confidence interval for each statistical hypothesis testing. All statistical analyses were conducted using Statistical Analysis Software (SAS) version 9.4 (SAS System for Windows). A significance level of 0.05 was considered in this study.

Demographic characteristic and percentage of comorbidities
In 2006-2012 birth cohort, there were 862,612 children without AD and 312,329 children with AD, the prevalence rate of AD was 26.6% (95% CI 26.5, 26.7) before 5 years of age. Table 1 gives the prevalence and absolute prevalence difference of child AD stratified by perinatal characteristic and parental disease in the 2006-2012 birth cohort. The absolute prevalence difference was less than 1% for children's sex, birth weight, gestational age, mode of delivery, and maternal preeclampsia. The maternal age at delivery, maternal gestational diabetes mellitus, maternal cervical incompetence, parental autoimmune disease (that including rheumatoid arthritis, systemic lupus erythematosus, Sjogren's syndrome, and ankylosing spondylitis), parental systemic diseases (that including anemia, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, hyperthyroidism, and obstructive sleep apnea), and parental allergic diseases (that including asthma and atopic dermatitis) contribute > 1% absolute prevalence difference of child AD. An absolute prevalence difference greater than 10% was identified in the children whose both parents had rheumatoid arthritis ( Table 3 demonstrated the subgroups analysis stratified by children's sex. In male children subgroup, the maternal GDM, parental autoimmune disease (including RA, SLE, Sjogren's syndrome, AS, and psoriasis), parental systemic diseases (including anemia, hypertension, DM, COPD, Hyperthyroidism, and OSA), and parental allergic disease (including asthma and AD) were associated with a significant higher risk of child AD development. The results of Supplemental Table 1 showed the effect of maternal disease was significantly higher than paternal disease. In female children subgroup, the results were similar with the male children subgroup, and the Supplemental Table 2 also showed the effect of maternal disease was significantly higher than paternal disease.

Discussion
The present study explored the association between parental autoimmune diseases and AD in their children. Our novel findings illustrated two main points. First, parental AD, asthma, RA, and psoriasis increased the risk of AD development in children before 5 years old. Furthermore, the subgroup analysis presented different odds ratios of parental diseases between the sexes of children. Parental RA was more prevalent in male children with AD, while parental psoriasis was more prevalent in female children with AD. The pathophysiologic mechanism of maternal diseases may be explained by the exposure of IgE autoantibodies during pregnancy. One Swedish study mentioned that higher prevalence of asthma was noted in children born to mothers with SLE during pregnancy [4]. Another study found that maternal dermatologic and digestive autoimmune diseases were closely associated with subsequent AD in their offspring [6]. Though the mechanism of relationship between AD and autoimmune diseases remains unclear, it has been found that 65% of the patients with SLE had higher level of IgE autoantibodies [20], which was also prevalent in 50-60% of patients with RA [21]. Studies suggested that elevation of IgE level and self-reactive T cell during pregnancy could induce autoreactivity in early infancy and subsequently increase the risk of developing AD [22,23]. Additionally, our study also found that both maternal and paternal allergic diseases could be risk factors, which implied that genetic effect also played an important role in AD. Filaggrin is a fundamental protein of skin barrier, and mutation of filaggrin (gene R501X and 2282del4) could be perceived as one of strongest predictors for AD [24]. In other GWAS(genome-wide association studies), filaggrin was not only listed as a characterized risk locus of AD, but its effect on genomic printing was prove to be equal between maternal and paternal alleles [25]. However, filaggrin mutations were more frequent in women [25], which would lead to closer relationship between maternal atopic diseases and their children developing AD. The adjusted odds ratios (p value) of paternal diseases exposure compared with maternal disease exposure (as reference group) were showed in Supplemental Table 1 (boys) and Supplemental Table 2 (girls). When compared with maternal AD exposure, the adjusted odds ratio of paternal AD exposure was 0.903 (p < 0.0001) and 0.898 (p < 0.0001) in boys and girls with AD, respectively. Our result showed significant difference and was consistent with the hypothesis.
The distinguishing feature of our study is the subgroup analysis of children's sexes, which showed the different risk factors for boys and girls, respectively. Studies have presented that early-onset AD was more prevalent in boys, while more females would develop AD after puberty [26]. Though there are many possible mechanisms, sex hormones may explain the difference of onset and progression between sexes. Studies have mentioned that level of dehydroepiandrosterone (DHEA), precursor of testosterone, was lower in boys because it was controlled by an X-linked gene. Moreover, in the study by Zhang et al., the AD-driven cytokines, such as interleukins 4 and 11, were found to promote the expression of enzyme for DHEA. This would increase androgen synthesis in sebocytes and decrease total amounts of triglyceride, which could lead to disruption of epidermal permeability barrier [27,28]. Adversely, higher level of estrogen was perceived as a protective factor for AD because of its function of reconstitution in skin barrier. Though our study mainly focused on AD development before 5 years old, it is also noted that increased estrogen and progesterone levels after puberty stimulated the activity of Th2 cells, which Table 2 Adjusted odds ratios of risk factors for child AD using multiple logistic regression analysis The multiple comparison included 16 tests for the association between parental disease and risk of child AD, we used the Bonferroni correction method to adjusted the confidence interval of (1-(0.05/16))×100% = 99.69% for each statistical test when the overall alpha error of 5% was considered a Because the prevalence of SLE was largely higher in female, (female: male = 14:1 in Taiwan, 2011), and the case of paternal SLE was too few to provide the sufficient statistical power, we only calculated the odds ratio of maternal SLE rather than taken the few paternal patients into consideration b The ORs of Obstetric complications was compared between maternal exposure and non-exposure could lead to exacerbation of AD during premenstrual phase and pregnancy, and impact the development of immune system of the fetus [22,28].
Lastly, both RA and psoriasis were found to be associated with AD in several studies. RA and AD shared a similar immune mechanism, including the increased activity of Table 3 Odds ratios of child AD for their parental diseases stratified by children's sex The child-sex stratified multiple comparison included 32 tests for the association between parental disease and risk of child AD, we used the Bonferroni correction method to adjusted the confidence interval of (1-(0.05/32))×100% = 99.84% for each statistical test when the overall alpha error of 5% was considered Th17 cells and dysfunction of regulatory T cells [29]. While most of the studies considered AD as a Th2-predominating disease in acute phase, the switch to Th1 was noted during the chronic period, which was also found in AD patient who developed RA [30,31]. On the other hand, although several studies have identified the common susceptibility loci of AD and psoriasis, the coexistence of these diseases still remained controversial because of their difference in immune pathway [10,32].As a result, studies suggested that subtypes of AD should also be taken into consideration because there are difference in prominent cytokines among these subtypes [33]. The study has a number of strengths. The link between perinatal data in the National Birth Registry and NHIRD enabled us to obtain a real-world data across generation. The participants were drawn from a population-based and highly representative computerized database of medical records. This allowed us to perform our analysis with a real-life setting in an unselected patient population. We took advantage of the detailed medical record in NHIRD and used the strict definitions at least three outpatient visits or one hospital admission with ICD-9 691 [34]. Additionally, we took a variety of autoimmune diseases into consideration and divided the confounding factors into paternal, maternal and parental (both) influences to compare the effects between parents. Lastly, the subgroup analysis showed different risk factors for AD in boys and girls.
Some limitations should be taken into consideration in this study. First, although the AD patients with ICD-9 code was physician-diagnosed, the lack of laboratory skin tests or prescription codes in combination with diagnostic codes could still result in potential misclassification of AD [35]. Hence, we added "at least three outpatient visits or one hospital admission" to the requirement of being diagnosed with AD. Second, autoimmune diseases are usually companied with long-term use of immunosuppressive drugs, such as corticosteroid, hydroxychloroquine, and TNFi (tumor necrosis factor inhibitor). Perinatal exposures to them may give rise to the change in maternal immune system [36,37], and their effects on subsequent AD development in patients' offspring should be well discussed in further studies. Third, previous studies showed that AD was associated with autoimmune disorders in children [38], which implied that past history of childhood AD in these parents could act a confounding factor. However, the time interval of our database was confined to 2003-2017 and that limited us to retrospectively identify the history of childhood AD or asthma in parents. Nevertheless, we have considered the perinatal exposures and parental diseases in this study, which might reduce the potential confounding effect by the childhood AD and asthma from their parents. Lastly, data on parental smoking history and socioeconomic status were not recorded in the NHIRD, despite the fact that these parental exposures could be important risk factors for childhood atopic diseases [39,40]. Consequently, we replaced smoking history as chronic obstructive pulmonary disease and listed it as one of the confounding factors.
In conclusion, this case-control study from 2006 to 2012 birth cohort demonstrated that both paternal and maternal autoimmune diseases are risk factors for atopic dermatitis before 5 years of age, especially seen with parental RA and psoriasis. Therefore, further studies are needed to investigate the underlying mechanisms in genes and immunology. Our findings remind the clinicians to be aware of early recognition of AD in these children, while patients with autoimmune diseases are suggested to receive adequate medication treatment to avoid perinatal exposures.