Atopic dermatitis and autoimmunity: the occurrence of autoantibodies and their association with disease severity

Atopic dermatitis (AD) is a widespread condition that appears to be increasing in prevalence and severity worldwide, yet the underlying mechanisms are not well understood. Recent research has identified various similarities between AD and autoimmune conditions, as well as indicating that there may be an association between AD and autoimmunity. This systematic review evaluates the association between AD and autoimmunity, as well as between severity of disease in AD and autoimmunity, with an emphasis on the associations with autoantibodies. MEDLINE (1946 to December 2017) and Embase (1974 to December 2017) databases were searched. Further relevant articles were retrieved from reference lists. Only studies measuring direct indicators of autoimmunity, in humans, were included. Qualitative analysis was carried out for all studies. In addition, quantitative analysis was used to evaluate prevalence of IgE autoantibodies and anti-nuclear antibodies (ANAs) in AD patients and control subjects. The Mantel–Haenszel method was used with a random-effects model. 28 studies assessed the occurrence of autoantibodies in AD patients and 16 studies were used to evaluate association between disease severity and autoantibodies. Pooled analysis from 14 studies, involving 986 AD patients and 441 control subjects, showed that IgE autoantibodies were significantly more prevalent in patients with AD (P < 0.00001) than control subjects. Similar analysis was carried out for ANAs, with eight studies that involved 1045 AD patients and 1273 control subjects. ANAs were significantly more prevalent in patients with AD (P = 0.003). This quantitative analysis supported an association between AD and IgE autoantibodies, as well as between AD and ANAs. There was insufficient data to make similar conclusions for other indicators of autoimmunity. The weight of evidence also suggests an association between IgE autoantibodies and disease severity. There was insufficient evidence to make this link for other indicators of autoimmunity.


Introduction
The term Atopic Dermatitis (AD) was first used by Wise and Sulzberger in 1933 [1]. The World Allergy Organisation defines eczema (also known as atopic eczema dermatitis syndrome (AEDS) and as AD throughout this review) as 'an inflammatory, chronically relapsing, non-contagious and extremely pruritic skin disease' [2].
Prevalence of AD worldwide is high, with 7.9% of 6-7 years presenting with the condition in the ISAAC Phase 3 study [3]. AD appears to be increasing in prevalence globally, with centres reporting increases in self-reported AD and severe AD [3]. There was significant variance in the prevalence of AD between centres, both within and across regions, which has been attributed, in part, to environmental differences [4]. The increasing prevalence and severity of AD makes it essential to understand the underlying mechanisms of the disease, enabling more effective future treatments.
AD is traditionally thought of as an allergic condition with extrinsic allergens driving the immune response. The discovery of a deficiency in the stratum corneum caused by filaggrin gene mutations supports this, as impaired skin barrier function allows for allergens to infiltrate the body, inducing an immune response [5]. AD patients with these mutations are more likely to have persistent, severe, allergen-driven symptoms, but some patients continue to suffer from AD despite removal of stimuli such as dust mites or 1 3 pollen [6]. Overall, as many as two-thirds of AD patients demonstrate 'no measurable allergen-specific IgE antibody sensitization' [7].
One hypothesis as to why chronic inflammation persists, despite removal of environmental allergens, is infection by Staphylococcus aureus. S. aureus colonization was found in more than 90% of patients and was found to correlate significantly with severity [8]. S. aureus has a contribution to most AD lesions, both by worsening inflammation through direct action of toxins and by increasing production of IgE antibodies [9]. It would follow that other skin based microorganisms, such as the fungus Malassezia sympodialis, could also cause this type of exacerbation, but this does not account for patients that suffer AD lesions without allergic sensitisation.
Despite the traditional understanding of AD, there has been evidence to suggest there might be an autoimmune component to the disease. Disease progression in AD is similar to that in known autoimmune conditions, involving periods of relapse and remission. Furthermore, a significant association has been found between AD and 11 autoimmune conditions, as well as the fact that AD patients are more likely to present with 'multiple autoimmune comorbidities' [10]. Finally, immediate skin reactions to some human proteins indicate that, for some patients, IgE autoantibodies are produced [11] which could contribute to disease progression. Researchers have also recently claimed that they have proven that 'atopic dermatitis … is an immune-driven (autoimmune) disease' [12] by treating AD with Dupilumab. This blocked IL-4 and IL-13 [13], to prevent Th2 cytokines from incorrectly targeting autologous tissues.

Methods
This systematic review investigates the link between AD and autoimmunity (with an emphasis on autoantibodies), providing an update to, and widening of scope of, a systematic review from 6 years earlier [14]. The present review evaluates the strength of the link between AD and autoantibodies, and whether there is an association between AD severity and autoantibodies.

Identification of relevant studies
Two modifications were made to the search strategy employed in 2012 by Tang et al. [14]: 'autologous' was added as a search term and the Embase database was searched in addition to MEDLINE. These alterations ensured the search strategy reflected current terminology and that as many relevant papers as possible were found.
The search terms were as follows: The PRISMA diagram [15] in Fig. 1 gives an overview of the number of papers at each stage of the literature search. Titles and abstracts from this search were checked, yielding 165 papers that warranted full-text analysis. These were scanned in full, with 28 studies providing evidence relating to an association between AD and autoantibodies, and 16  that evaluated severity of AD in relation with the occurrence of autoantibodies. During this full-text analysis, the reference sections of papers were checked to identify relevant papers that had not appeared in the search results. Most papers excluded at this point were conference proceedings or included results not linked to autoimmunity and AD.

Language
All literatures evaluated in this systematic review were in English (including papers that had already been translated in to English).

Types of studies
Due to the wide range of study designs present in the literature, all the types of study were included. The main types of study evaluated were case-control studies, case series, and individual case reports.

Types of participants
This systematic review evaluated all studies involving human participants, irrespective of age or gender. Studies varied in the age ranges of the participants (details in Tables 1, 3, 4, 5, 7): most included adults only or a mixture of children and adults, whereas a few studies included children only. However, there was no apparent effect of participants' age on the relationship between autoreactivity and AD ( Table 2) or severity of AD (Table 6). All the literature was included regardless of whether they had clear criteria for selecting patients with AD. These criteria have been noted for all studies in appendices, but have only been referred to in the main text when these criteria are unclear or missing.

Types of publication
In this systematic review, all papers published in journals were included, with the exception of reviews and editorials. Letters to journals were included, as they were a key source of smaller studies. Grey literature was not included.

Outcome measures
The papers evaluated in the systematic review were heterogeneous in both study design and outcome measures. The only restriction was that studies had to provide direct evidence of autoantibodies. When evaluating the association between autoantibodies and AD, any tests that indicated autoantibodies, for example, ELISA and skin prick tests (SPTs), were included. When evaluating the association between severity of AD and autoantibodies, all measures of severity were included. The most common measures of severity were SCORAD (SCORing Atopic Dermatitis) [44] and Rajka and Langeland [45] systems, although some authors used other assessments of severity.

Statistical analysis
Where appropriate, a quantitative analysis was carried out using Review Manager 5.3 [46]. The Mantel-Haenszel method was used with a random-effects model to account for the variation in study design and resultant difference in effect size.

Association of autoantibodies with atopic dermatitis
28 papers were identified that fulfilled the inclusion criteria and were relevant to this part of the review. The main indicators of autoimmunity investigated were specific IgE and IgG antibodies, as well as IgG ANAs.

Autoantibodies of IgE class
14 studies investigated the presence of specific IgE autoantibodies in patients with AD; the details of these studies are summarised in Table 1 (in alphabetical order of the first authors). Twelve were case-control studies, of which 11 indicated that certain IgE autoantibodies were significantly more common in patients with AD. The study quality was generally good, with well-described protocols, often using western blot, ELISA or SPT techniques, and robustly defined criteria for selection of AD patients in most cases. Most studies evaluated the presence of IgE autoantigens to epithelial cells, epithelial cell extracts, and other proposed autoantigens associated with the skin. Furthermore, the control groups were comprised predominantly of healthy individuals which makes comparison between, and pooling of, individual studies much easier and more useful. This meant an analysis could be carried out on the combined data, which established an odds ratio of 12.97 (with 95% confidence limits of 5.14-32.68) and P < 0.00001, indicating that IgE autoantibodies are significantly more likely to be identified in patients with AD compared to the control groups (Fig. 2). This effect seemed to be uniform across all tested antigens, although the majority of evidence concerns specific antibodies to epithelial cell extracts. A qualitative analysis of the studies also indicated that all but one (13/14) concluded that  IgE class autoantibodies are more common in AD patients than controls ( Table 2).

Autoantibodies of IgG class
Eight studies investigated the presence of specific IgG autoantibodies in AD (Table 3). However, it was not possible to carry out a quantitative analysis: whilst the studies predominantly used ELISA techniques, the study design was too varied for a pooled analysis. Some studies only compared AD patients against patients with other conditions, for example psoriasis, whilst others only provided mean titres. Furthermore, only three of the studies had clear definitions for AD diagnosis. Because of the more variable study design, it was more appropriate to carry out a qualitative review. Levels of IgG autoantibodies specific to SART2 161 , SART2 899 , ART4 13 [22], CCL3 [31] and CD28 [32], as well as APL antibodies [34] were significantly higher than in their respective control groups, but levels of autoantibodies specific to β 2 GPI [35], TNF-α, IFN-α, CCL-5, CCL-2, IL-17 [31], and five other peptides [22] were not significantly higher. Autoantibodies against DFS70 [33] and Fc ε R1 α autoantibodies [16] were more commonly identified in AD patients than the respective control groups, but no statistical analysis was carried out. Finally, AD patients were found not to display an autoreactive response to stratum corneum antigens, but psoriasis patients did (although no statistical analysis was carried out). Overall, there is no clear picture as to the presence of IgG autoantibodies in AD patients ( Table 2). The mixed results, combined with more variable study techniques and less clarity with respect to inclusion criteria, means that more research is required into the relationship between AD and IgG autoantibodies.

Autoantibodies of IgA class
Only one study solely investigated IgA autoantibodies in AD: Ress et al. [37] carried out a large case-control study, comprising 297 patients with AD and 52 healthy control subjects. ELISA techniques were used to determine the presence of IgA-anti-transglutaminase 1 (anti-TG1) and IgA-anti-transglutaminase 3 (anti-TG3) in the control group and AD group. However, this study did not indicate an association between IgA autoantibodies and AD ( Table 2), but there are insufficient studies to draw a clear conclusion.

Anti-nuclear autoantibodies
Eight studies investigated ANAs in AD (Table 4), with two of these including ANAs as part of a wider study. Six of the eight were case-control studies. Most of these papers applied clear, commonly used definitions of AD and most used well-described methods-particularly indirect immunofluorescence-to identify ANAs. Seven of the eight papers indicate that ANAs occurred more commonly in patients with AD than in the control subjects (Table 2). Furthermore, the pooling of data showed that patients with AD have a significantly higher frequency of ANAs than the control subjects with an odds ratio of 2.18 (95% confidence intervals 1.31-3.64) and P = 0.003 (Fig. 3).

Association of autoantibodies with severity of atopic dermatitis
The second part of this systematic review considers whether there is an association between severity of disease in patients with AD and the presence of an autoimmune response. From the papers identified using the search strategy, 16 papers evaluated the presence of autoimmunity in combination with disease severity. 13 papers provide evidence of the relationship between IgE autoantibodies and disease severity (Tables 5, 6), and two provide research relating to ANAs ( Table 6). The final paper investigated ANAs, APL antibodies, and ACL antibodies and their relationship with severity (Table 6). Two main methods of assessing severity were used, SCO-RAD [44] and Rajka and Langeland [45] scores, although some researchers used other measures of severity. SCORAD was developed as a standardised method to assess severity of AD that could be used in outpatient clinics. It relies on three components: intensity (responsible for 60% of the total score), extent and subjective symptoms (each responsible ≈ 20% of the score). Rajka and Langeland also put forward a simple scoring index for AD using three parameters, each scored out of 3: extent, course (how much remission in a year) and intensity (scored according to how much it impacts sleep). The sum of these scores is calculated, with 3-4 being defined as mild, 4.5-7.5 being moderate, and 8-9 being severe.
It is difficult to produce pooled data for the studies assessing a link between IgE autoantibodies and severity of AD due to wide variation in the measures of severity and outcome measures but, as shown in Table 6; the weight of evidence indicates that there is an association between the frequency of IgE autoantibodies and disease severity in AD. Furthermore, two studies [48,50] looked at individual patients over time, assessed levels of IgE autoantibodies and SCORAD throughout treatment, and established a temporal link between SCORAD and IgE autoantibodies. Whilst it is impossible to determine from these studies whether the decrease in IgE autoreactivity as severity of disease declines is a separate response to the cyclosporine A treatment, an unrelated biomarker resulting from a decrease in disease severity, or indeed the cause of the reduced severity, it provides strong support for a link between AD disease severity and autoimmunity. This association did not seem to be present with respect to ANAs, as neither study that looked exclusively at ANAs demonstrated a significant association (Table 6), but more studies are required to reach a definitive conclusion. Szakos et al. [34] also indicated that there was no significant difference in disease severity in patients that displayed ANAs, ACL, APL or a combination of ANAs and APL (Table 6). Overall, the available evidence indicates that severity of AD is associated with autoimmunity, but that this response might be limited to IgE-mediated  Hiragun et al. [20] Case-control study* 63 patients with AD (Patients with AD ranged from 0 to 65 years, with a mean age of 29.9 ± 11.5 years.) Moderately clear 'severity of AD was evaluated using severity index of Japanese guideline for AD' Ilves, Virolainen and Harvima [21] Case-control study* 50 patients with AD (Patients with AD ranged from 21 to 64 years, with a mean age of 38 years. Insufficient data to calculate SD.)

Intracutaneous injection
Of autologous sweat autoreactivity. However, further research is required into other indicators of autoimmunity, such as IgG autoantibodies, ANAs and IgA autoantibodies to confirm this, as current evidence is limited in these areas.

Discussion
The weight of evidence presented in this review indicates that there is an association between autoreactivity and AD, as well as disease severity in AD, particularly with regard to autoantibodies of IgE class. However, this review cannot provide evidence for a causal link between the two, as the exact mechanisms for how autoimmunity occurs in patients with AD are unclear. Some papers hypothesise that cross reactivity between Malassezia spp. and the autoantigen hMnSOD could underlie the pathogenic response seen in AD [21,25], whilst many others have identified cells in the skin as a source of autoantigens, potentially indicating that skin damage could lead to an autoimmune response to components within the skin, giving rise to autoimmunity. However, these hypotheses rest on assumptions that have not been proven. Taking all the available evidence into account, it is not possible to declare that autoimmunity is a pathogenic factor underlying the natural history of AD, but it does seem plausible and certainly worthy of future research. This review points to several areas, where future research into autoimmunity and AD would be most valuable. First, some areas of autoimmunity have been less thoroughly investigated when looking at an association with AD. Whilst IgE autoantibodies and, to some extent, ANAs have been covered extensively by research, other markers of autoimmunity have less evidence available. Studies investigating IgG autoantibodies, for example, were varied in terms of control groups, often comparing against groups that have other diseases, and were less clear about the selection criteria for the patients with AD. More research into IgG autoantibodies in AD, across a wide range of autoantigens and in comparison to healthy control groups, would be useful in drawing more robust conclusions.
Five other studies, not specifically covered by this review, have examined aspects of autoimmunity other than autoantibodies (Table 7). Collectively, these studies do indicate that autoimmunity is related to AD, with three of the five indicating statistically significantly higher rates of autoimmune response in patients with AD compared to their control groups, be it through increased histamine release upon stimulation by semi-purified sweat antigens [51], activation of iNKT cells without the corresponding ligand [52], or significantly higher stimulation index of PBMCs in response to hsp60 [53]. In addition, Hide et al.  [47] show a greater frequency of response to intradermal injection of autologous sweat in AD patients, albeit without statistical analysis. The final study [54] does not seem to indicate an increased autoreactive T-cell response in patients with AD, but does indicate that there may be an alteration to the balance of CLA+/CLA− and CCR4+/ CCR4− cells. Overall, these less traditional indicators of autoimmunity show promising results in terms of a link with AD, but further research is required to come to definitive conclusions.
There are also new indicators of autoimmunity that, whilst not meeting the inclusion criteria of this review, could provide insight into AD. Some interesting papers have investigated the function of regulatory T cells (Tregs) in AD with research indicating that the suppressive effect of Tregs on CD8 + CLA + T-cell proliferation is inhibited in AD [55]. This would be a similar mechanism to other autoimmune/inflammatory diseases such as psoriasis, which shows impaired suppression of CD4 + CD25 − T cells [56]. Zhang et al. [55] note that further experiments are needed to address the theory that the CD8 + CLA + T cells in patients with AD might resist suppression, as opposed to there being dysregulation of Tregs, but further research into this field could provide a valuable insight into the mechanism behind autoimmunity. Later research corroborated this [57], with both papers suggesting that reduced inhibition of Teff cells could cause impaired self-tolerance in patients with AD, but further research would assist in elucidating specific mechanisms underlying the autoimmune response.
Another area that could provide useful research into autoimmunity and AD is analysis of genetic variants and potential associations with AD. For example, copy-number variations of the human histamine H4 receptor (HRH4) gene have been shown to be associated with AD [58], as well as a number of other autoimmune conditions such as SLE [59]. Whilst the exact mechanisms involved with the human histamine H4 receptor gene and AD are unclear, Chen et al. [58] suggest that the expression of HRH4 could be enhanced because of inflammatory stimuli found in autoimmune diseases, leading to a stronger response to histamine. This is an example of another potential component to the autoimmune pathogenesis in AD and is an area that warrants future research.
How autoimmunity is related to AD in animal models also warrants further research. Using animal models could lead to greater understanding of whether autoimmunity causes AD, perpetuates AD, or whether it is just an associated factor. Differences in ethical considerations for humans and animals allows for different studies to be carried out in animals, allowing more flexibility in study design and an ability to more easily identify causation.
Finally, further research into how different phenotypes of AD are associated with autoimmunity could provide a strong basis for future clinical research. Patients with AD can be divided into intrinsic and extrinsic categories, as well as other groups, including age of diagnosis and ethnic origin [60]. These groups have different disease characteristics, so more research into the underlying pathogenic mechanisms within these groups, be it related to autoimmunity or other factors, would be useful to aid understanding and potentially improve future treatments.  Funding There is no funding source.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.