With an estimated prevalence of 3% in adults and 15–20% in children, atopic eczema (AE) is one of the most common diseases in the Western World. The pestering itch with consecutive loss of sleep and concentration and the social stigmatization result in severely affected quality of life of the affected individuals, but also of their families. Even though numerous studies investigated the underlying pathogenesis of AE, central questions are still under debate. This review discusses three key questions: 1. Is AE a genetic or an acquired disease? 2. Is AE driven by intrinsic alterations or rather by extrinsic factors? and 3. Are allergen-specific and Th2 inhibiting attempts or general anti-inflammatory therapeutic regimens more effective? in a disputation manner. Finally, all aspects of AE pathogenesis are summarized in a unifying mosaic model.
Atopic eczema (AE) is a challenge for modern medicine, because it is prevalent, severely affects quality of life of patients and their families, and causes high socioeconomic costs. The pathogenesis of AE is complex. While initial studies suggested a Th2 deviation as primary reason for the disease, numerous studies addressed a genetically predetermined impaired epidermal barrier as leading cause in a subgroup of patients. Recently, immune changes beyond the initial Th2 concept were defined in AE, with a role for specialized dendritic cells as well as newly identified T helper cell subsets such as Th17 and Th22 cells. Furthermore, trigger factors are expanded beyond classical Th2 allergens such as pollen or house dust mites to microbial products as well as self-antigens. This review pieces together our current understanding of immune as well as barrier abnormalities into the pathogenesis mosaic of AE.
1st disputation: is AE a genetic or an acquired disease?
Progenetics and skin barrier impairment
Without any doubt, an impaired epidermal barrier is a key feature of AE. The skin barrier represents one of the largest front lines of the body where the immune system interacts with the environment. An intact epidermal barrier shields most environmental stimuli from immune cells, but this function is disturbed in AE. In fact, many different components of the epidermal barrier are modified in AE, and some are genetically predetermined. Genetic modifications affect many genes encoding the epidermal differentiation complex including profilaggrin, S100 proteins, small proline-rich proteases, antiproteases such as lymphoepithelial Kazal type-related inhibitor (LEKTI), or tight junction proteins such as claudin-1 . Concordance rates of up to 0.86 in monozygotic and 0.5 in dizygotic twins provide clear evidence for a genetic predisposition in AE [2, 3].
The strongest genetic association with AE so far was shown for loss-of-function mutations in the filaggrin gene (FLG). About one-third of European Caucasian patients carry the most frequent FLG null mutations R501X and 2282del4 . Early onset of the disease and severe chronic courses are characteristic for FLG mutation carriers. Moreover, data obtained from mice with a homozygous frameshift mutation in the FLG gene show a stronger cutaneous cellular infiltrate and allergen priming mirrored by elevated IgE levels in mice with the FLG-frameshift mutation as compared to wild-type mice . Studies on FLG mutations conducted in human and murine model systems within the last few years exemplify in a so far unique way how a genetically predetermined epidermal barrier dysfunction might interact with environmental factors and lead to the manifestation of AE. Furthermore, because filaggrin is not expressed by nasal or esophageal epithelium as well as bronchial mucosa , it is likely that allergic sensitizations relevant for the upper and lower airways might originate from FLG-mutation-related epidermal barrier impairment . Exposure to cat allergens in early life increases the risk of eczema and development of sensitizations in particular in children with FLG loss-of-function mutations (R501X and 2282del4) [8, 9]. These data clearly indicate that hereditary factors predict the risk of the development of a clinical phenotype (eczema), while environmental factors such as allergen exposure represent important cofactors, with secondary effects on disease manifestation.
However, epidermal barrier dysfunction occurs also independently from the FLG genotype indicating that other factors with profound impact on the epidermal barrier exist. Interestingly, it has been demonstrated that the microenvironment directly or indirectly modulates skin barrier integrity. The Th2 cytokines IL-4 and IL-13, factors promoting a Th2 polarization such as thymic stromal lymphopoietin (TSLP), IL-25 and IL-33, as well as histamine modulate skin barrier function by influencing keratinocyte functions [10-12]. Keratinocytes that differentiated in the presence of IL-4 or IL-13 showed a lower filaggrin protein expression . Similarly, IL-4 and IL-13 diminished expression of the calcium-binding protein A11 (S100) in keratinocytes cultured in the presence of calcium . Because increased levels of IL-4 expression are observed in acute AE lesions [15, 16], IL-4 might indirectly impact on the expression of epidermal differentiation proteins. Moreover, it has been demonstrated that IL-4 and IL-13 enhance the expression and function of the serine protease kallikrein 7 of human keratinocytes in vitro, which degrades corneodesmosome proteins and promotes desquamation . This Th2 overbalance is in part genetically predetermined, because genetic variations in the gene regions encoding IL-4, IL-4RA, and IL-13 are associated with AE . Thus, hereditary factors might directly or indirectly impact the epidermal barrier dysfunction and thereby influence the clinical course of AE (Fig. 1).
Furthermore, even nonlesional AE skin displays profound differences as compared to the skin of healthy individuals concerning terminal differentiation gene abnormalities . Gene expression of loricin, filaggrin, involucrin, corneodesmosin, or late cornified envelope protein is significantly downregulated already in nonlesional AE skin as an indicator for abnormal keratinocyte differentiation and skin barrier impairment even in the absence of strong skin inflammation. This underlines the hypothesis that hereditary factors primarily determine the course of the disease.
There is no AE without inflammation, and several arguments support the hypothesis that AE is primarily an immune disease. One of the strongest is the fact that patients with coexisting AE and psoriasis are reported [19, 20]. In those rare patients, distinct T-cell subsets infiltrate eczema lesions and psoriasis plaques simultaneously, with Th2 dominance in AE and more Th1 and Th17 cells in psoriasis . Furthermore, the resulting different cytokine composition in AE and psoriasis largely explains the observed epidermal alterations, as the Th2 cytokines IL-4 and IL-13 impair the epidermal barrier function. Even in close proximity to pre-existing psoriasis plaques, AE reactions could be induced using epicutaneous patch testing [atopy patch test, APT ]; thus, specific stimuli define the course of a cutaneous inflammation rather than epidermal alterations in those patients. This argument is supported by the observation that immune-targeted therapeutic approaches regularly lead to resolution of the epidermal phenotype .
Finally, the concept of genetically fixed epidermal alterations as primary cause of AE lacks explanations for common phenomena of the disease phenotype. Namely, acquisition of different sensitizations is commonly observed in patients exposed to similar environmental exposure. Some patients might develop sensitizations against grass, others against birch pollen, and a third group against both [23, 24]. These diverse sensitization patterns are still unpredictable and not associated with known genetic alterations. A second weakness of the barrier theory is the low association of single genes with development of the disease. Even the best characterized gene FLG is altered in only 30% of AE patients, while on the other hand, around 8% of the healthy population carry identical loss-of-function mutations . Thus, there is evidence for a ‘pure immunology’ type of AE, and genetic alterations explain only a part of the AE pathogenesis (Fig. 1).
2nd disputation: is AE driven by intrinsic alterations or rather by extrinsic factors?
Pro-intrinsic cellular alterations
An impaired epidermal barrier is not limited to AE, but is also observed in other skin diseases such as Netherton syndrome or allergic contact dermatitis. These diseases even share some genetic modifications with subgroups of AE patients such as loss-of-function mutations in the FLG gene or in the SPINK5 gene, which encodes the serine protease inhibitor LEKTI [26, 27]. Thus, an epidermal barrier disruption is not sufficient to fully explain AE.
Beyond skin barrier impairment, a high number of genetic modifications of innate and adaptive immune cells exist in AE. Keratinocytes of patients with AE release higher amounts of several pro-inflammatory cytokines and chemokines than keratinocytes from non-AE skin. Furthermore, in a subgroup of AE patients, keratinocytes as well as macrophages weakly respond to bacterial stimuli [28, 29]. Most importantly, AE keratinocytes secrete high levels of TSLP, which promotes dendritic cell (DC)-driven Th2 cell priming and thereby amplifies Th2 immune responses . In contrast, the production of antimicrobial peptides (AMPs) such as LL-37 or human-beta defensins is lower than in other inflammatory skin diseases such as psoriasis [19, 31, 32]. This deficient AMP production by AE keratinocytes is partially caused by the Th2-dominated microenvironment and in part explains the high susceptibility of AE patients to bacterial and viral skin infections [14, 31, 32].
Besides a strong intrinsic Th2 deviation, Th1 immune responses are attenuated in AE. Lower expression of interferon-γ receptor (IFNγR) II by epidermal DCs of patients with AE and attenuated STAT1 phosphorylation as well as upregulation of IFN-γ-induced pathways might be a reason for this phenomenon . Moreover, genetic variants in IFNG and IFNGR1 as well as interferon regulatory factor (IRF) 2 were significantly associated with AE with a history for a severe disseminated viral skin infection caused by herpes simplex virus, called eczema herpeticatum (EH) [34, 35]. Also, a reduced number of regulatory T cells (Tregs) influenced by both environmental and inherited factors predispose to the development of AE in early life .
Another hallmark of AE is the presence of characteristic inflammatory DC subtypes in both dermis and epidermis. Antigen-presenting cells such as CD1apos epidermal DC, dermal DC, and monocytes in AE typically express the high-affinity receptor for IgE, FcεRI, and its corresponding tetraspanins CD9 and CD81 [37-39]. FcεRI-bearing Langerhans cells, characterized by expression of Langerin, are the main CD1apos DC population in nonlesional AE skin. In lesional AE skin, Langerin-negative, CD1apos DCs with strong pro-inflammatory characteristics and high FcεRI expression are detectable. Inflammatory DCs subtypes in the epidermis were named inflammatory dendritic epidermal cells (IDECs). Data from atopy patch tests (APT) performed in sensitized AE patients revealed a decrease in FcεRIpos Langerinpos DCs but an increase in FcεRIpos Langerinneg DCs in the epidermis 24–72 h after allergen application. In line with those results, the number of FcεRIpos, Langerinneg DCs decreases, while FcεRIpos Langerinpos DCs increase after topical treatment of AE and improvement in the skin lesions [40, 41]. Therefore, the described DC subtypes fulfill distinct functions in the skin, and their chemokine receptor pattern together with the release of chemokines in the skin directs the recruitment of DCs during eczema development . In terms of the capacity of DCs to polarize T cells, it has been demonstrated that unstimulated DCs of AE patients are capable to polarize Th1, Th2, or Th17 T-cell subsets. However, a defect in the ability to induce IL-10 and Th1 cells was recently described in DCs of allergic rhinitis patients . Therefore, the local microenvironment that partially depends upon intrinsic alterations in AE directs the nature of infiltrating memory T-cell subsets .
In summary, intrinsic cellular alterations result in an initial Th2 dominated immune response that triggers both a secondary epidermal barrier dysfunction as well as a non-Th2 immune response (Fig. 2). Those cellular alterations represent important pieces of the pathogenesis mosaic of AE.
Is any individual at risk to develop AE upon distinct environmental exposure? Probably yes. Even though most people tolerate classical allergens such as house dust mites or pollen, increasing evidence suggests the stimulus itself rather than intrinsic cellular alterations determines the course of an immune reaction. It is still unresolved why certain harmless molecules are allergenic, but a key seems to be activation of innate immunity. The house dust mite allergen Der p2 mimics an LPS-binding site  and signals through MyD88/TLR2 , and the potent contact sensitizer nickel activates TLR4 . The exciting concept that eczematous reactions somehow mimic innate infection was developed in recent years and is now broadly accepted in the field of contact dermatitis .
Besides general immune activation, also the type of immune response is influenced by environmental triggers. Food and aeroallergens are extrinsic trigger factors in subgroups of AE patients, in particular in early life. Allergen-specific cutaneous immune responses induced by epicutaneous application of allergens in sensitized individuals in an APT setting are initially almost exclusively dominated by Th2 cytokines [15, 19]. This Th2 response is partially explained by intrinsic alterations in AE patients, but also in part by the trigger itself. For example, nonallergenic pollen-derived substances favor a Th2 response by blocking IL-12 , by recruitment of eosinophil and basophil  granulocytes, and by activation of mast cells independent of IgE . Also TSLP, a central intrinsic mediator of Th2 immunity that favors allergic sensitization, is induced upon environmental stimuli such as wounding  with consecutive release of double-stranded RNA  or directly by staphylococcal-derived products .
However, AE is more than an allergen-specific Th2 immune response. While in early stages of inflammation, spongiosis and a dense cellular infiltrate consisting of mostly Th2 cells are typical [15, 24, 32], more chronic lesions display a distinct T-cell infiltrate with Th1, Th17, and Th22 dominance [15, 24, 32, 55-57] and morphology with acanthosis and less immune infiltrate. Again, this phase of the immune reaction is influenced and driven by environmental stimuli, in particular microbial substances. Staphylococcus aureus colonizes almost all AE lesions, and the skin microbiome closely correlates with disease severity . Staphylococcal products downregulate FcεRI on dendritic cells via binding to TLR2 . Furthermore, staphylococcal enterotoxins such as SEA, SEB, and TSST act as superantigens stimulating T cells independent of their specificity via certain T-cell receptor Vbeta chains . SEB induces the Th17/Th22 cytokines IL-17  and IL-22  as well as the pruritic cytokine IL-31 . Both Th17 and Th22 are critical for epithelial homeostasis and induce acanthosis [63, 64], which helps to understand why chronic AE shares clinical and histological aspects with psoriasis. Also, humoral factors associated with microorganisms may contribute to disease pathogenesis of patients with a long-lasting history of AE. In such patients, specific IgE against Malassezia furfur  and staphylococcal enterotoxin  is regularly detected.
Besides microorganisms, autoimmune reactions and/or microbial mimicry determine the course of chronic AE. IgE auto-antibodies against a broad variety of keratinocyte proteins [67, 68] and against a superoxide dismutase of Aspergillus species with cross-reactivity to the human isoform  were identified in AE patients. Also, autoreactive T cells with molecular mimicry against the enzyme thioredoxin of Malassezia species were described in patients with a long history of AE .
Thus, environmental factors induce a mixed T-cell response of Th1, Th17, and Th22 cells in the ongoing immune cascade of AE. This concept is supported by the observation that Th17 cells infiltrating house dust mite  and nickel -induced eczema are not specific for the eczema-eliciting antigen, but rather for Candida or staphylococci . In summary, microbial and self-antigens activate non-Th2 T cells in the course of AE, and their cytokine signaling elicits a second wave of inflammation unrelated to the initial Th2 response (Fig. 2).
3rd disputation: are allergen-specific attempts or general anti-inflammatory therapeutic regimens more effective?
Pro-allergen-specific and Th2 inhibiting attempts
Virtually, all innovative therapies addressing the immune system were tried for efficacy in AE in recent years (Table 1). Because novel therapeutics target more and more specific pathways, their efficacy in AE teaches us important lessons for disease pathogenesis. Besides restoring the epidermal barrier, basically three therapeutic concepts are investigated: induction of tolerance toward specific allergens, inhibition of Th2 immunity, and targeting proteins involved in acute-phase inflammation and the inflammasome (Table 1).
|Mepolizumab||Anti-IL-5||EASI 50: 0/18 |
|Omalizumab||Anti-IgE||EASI/SCORAD 50: 0/3 , 3/3 , 2/11 , 21/21 (all with asthma) , 0/20 |
|Rituximab||Anti-CD20||6/6 improved with EASI 50 , 0/2 improved long term |
|Immunoglobulin i.v.||No long-term efficacy |
|Apremilast||Phosphodiesterase IV inhibitor||Minimally effective [95, 96]|
|Interferon-gamma||EASI 50: 18/40 ; good response 17/40 |
|Infliximab||Anti-TNF-α||2/9 long-term efficacy |
|Etanercept||TNF-αR antagonist||EASI 50: 0/2 |
|Ustekinumab||Anti-IL-12p40||1/1 improved for short term ; 1/1 improved |
|Tocilizumab||Anti-IL-6||3/3 clearly improved |
|Alefacept (withdrawn in 2011)||LFA-3/IgG1 fusion protein||EASI 50: 2/9 ; 10/10 |
|Efalizumab (withdrawn in 2009)||Anti-CD11a||EASI 50: 2/11 ; 6/10 |
|Ongoing clinical studies|
|Aeroderm/Pitrakinra||Aerovance/Bayer||IL-4Rα binding protein (antagonizes IL-4 and IL-13), phase II |
|REGN668||Regeneron||L-4Rα binding protein (antagonizes IL-4 and IL-13), phase II|
|BMS-981164||Bristol-Myers-Squibb||Anti-IL-31, phase I|
|QGE031||Novartis||Anti-IgE, phase II|
|QAW039||Novartis||CRTH2 inhibitor, phase II|
|AMG157||Amgen||TSLP-R binding protein, phase I completed|
|Anakinra||Amgen||IL-1R antagonist, phase I|
|S-777469||Shionogi||Cannabinoid 2 receptor, phase IIa|
|ASB17061||Asubio||Chymase inhibitor, phase I|
|MEDI-4212||MedImmune||Anti-IgE, phase I|
A central question is the role for single allergens and type 2 immunity in AE. Numerous studies are currently investigating immunotherapy regimes, but until now, data are not overall convincing if not disappointing and long-term follow-up studies are missing . Major problems of published studies were the broad inclusion criteria and poly-sensitized patients, so ongoing studies try and identify subgroups of AE patients that may benefit from allergen-specific immunotherapy. Conflicting evidence exists regarding efficacy of the anti-IgE antibody omalizumab. One study reports good clinical effects in all 21 investigated patients, but all of them suffered also from allergic asthma . Other groups report subgroups of responding patients  or no sufficient response rate at all. The response rate may also be related to the level of total IgE, which is why a new substance indicated for higher IgE levels is currently under investigation (Table 1). A pilot study regarding the effects of immunoadsorption in AE patients revealed a clinical success of at least 50% SCORAD reduction after 13 weeks in 8 of 10 patients; however, long-term improvement was not assessed . Intravenous application of immunoglobulins resulted in one study in a moderate response rate, but long-term effects were not observed . Little data exist on the efficacy of the anti-CD20 molecule rituximab that removes B cells from the body. A pilot study performed in six patients showed a moderate to high beneficial effect in all patients . In contrast, another study found poor efficacy of rituximab in AE patients . Also the anti-IL-5 antibody mepolizumab did not convince, with 0 of 18 patients reaching EASI 50 after two injections, although blood eosinophils markedly decreased . H1-antihistamines decreased histamine blood levels in a study with 15 patients, but this decrease correlated with less pruritus only, not with eczema severity . Currently, substances interfering with the IL-4 and IL-13 signaling pathway are under investigation  (Table 1). Another strategy is to induce an immune shift away from Th2 immunity in early life. Recent reports speculate that orally administered bacterial components  or ingestion of yoghurt in early life  might protect from AE.
In summary, only a limited number of studies systematically investigated available drugs interfering with type 2 immunity, but overall they are not promising. This further supports the concept that (allergen-specific) Th2 immune reactions are just an initial part of AE pathogenesis. The hypothesis of early prevention by shifting the immune system toward Th1 immunity is interesting, but needs to be proofed in follow-up studies.
Progeneral anti-inflammatory therapeutic regimens
A third approach for novel therapies in AE targets general pro-inflammatory cytokines. While TNF-α neutralizing drugs such as infliximab and etanercept are of enormous potency in psoriasis treatment, they do not improve AE at all [84, 85]. This fact allows the speculation that Th17 cells may not be of great importance in AE. However, future studies with new molecules neutralizing IL-17 more specifically might clarify this point. Ustekinumab, targeting both the Th17 and the Th1 pathway, may have more beneficial effects in AE; however, here only two case reports [19, 86] exist so far, and one of the patients  did not show long-term effects (data unpublished). A pilot study reports that neutralizing acute-phase proteins could be a useful strategy. All three reported patients dramatically improved upon injection of tocilizumab, a monoclonal antibody against IL-6, but two of them had to discontinue therapy due to severe infections . A study investigating whether targeting the IL-1 pathway could be a good option is currently ongoing (Table 1).
AE is a heterogeneous clinical phenotype characterized by spongiosis, keratinocyte apoptosis, and dermal immune infiltrate in its acute phase. Because this definition is purely descriptive, AE can be regarded as a cutaneous inflammatory response pattern. However, as many ways may lead to Rome, different predispositions may cause AE. Causing factors can be divided into four groups that may cause AE-like symptoms alone, but in most of the cases, a combination of several will lead to the full clinical phenotype (Fig. 1). Those factors include genetically determined alterations either of the epidermal differentiation complex or of the immune system regarding a type 2 response. But both epidermal damage and a Th2 immune reaction may also occur in response to environmental stimuli regardless of the genetic background. In this acute phase of AE (Fig. 2), a disturbed epidermal barrier and an impaired innate immunity are characteristic features. In the ongoing course of cutaneous inflammation, further environmental stimuli as well as intrinsic alterations trigger AE pathogenesis. Microbial substances such as bacterial superantigens and self-antigens induce a second wave of inflammation not characterized by Th2, but rather by IDECs, Th17, and Th22 cells. Once the single mosaic pieces consisting of genetic factors and environmental stimuli exceed the immune threshold, a cutaneous inflammation is initiated (Fig. 3). A combination of this ‘eczema response pattern’ with dominant Th2 immunity (e.g., increased IgE levels) is currently defined as AE. However, the heterogeneity of the disease with its variable clinical course  is reflected by the mosaic pieces as well as by divergent response rates toward innovative and specific therapies (Fig. 3, Table 1). With this pathogenesis mosaic in mind, the major task in the future of AE research will be to identify subgroups that will ultimately lead to individualized medicine [89-91].
Conflict of interest
The authors declare no conflict of interest.