Immunologic, Microbial and Epithelial Interactions in Atopic Dermatitis

INTRODUCTION

Atopic dermatitis (AD) is the most common chronic inflammatory skin disease.¹ It usually starts in early infancy, and often remains a life-long therapeutic challenge, especially in its moderate-to-severe form.² Its etiology remains only partly understood but is believed to be multifactorial, most likely rooted in a complex combination of immune deviation, impaired barrier function, and environmental risk factors.^3–5 Skin sensitization to allergens seems to be a phenomenon that happens secondarily to a barrier defect, but it is an ongoing debate whether this is a result of genetic mutations affecting the epidermal barrier (outside-in-model) or due to an inflammatory process inhibiting epidermal differentiation (inside-out-model).⁶ As AD has been identified as a heterogeneous disease with activation of more than one inflammatory pathway rather than a uniform condition with a single pathogenic immune axis such as psoriasis, both models are likely to be relevant, with varying nuances depending on the AD population being evaluated.^1,7 AD heterogeneity has not only been shown for intrinsic vs. extrinsic,⁸ pediatric vs. adult,⁹ and Caucasian vs. Asian AD.¹⁰ Also, the age of onset can be quite different, with a minority of patients showing the advent of late-onset AD during adulthood, a subset that can also have some diagnostic challenges due to atypical clinical presentation.¹¹ Elucidating this complex disease is an ongoing endeavor, but previous work has already led to its increased understanding. Importantly, these data have led to the development of new therapeutics which have already been either approved, or which show promise in clinical trials.¹² An important recent concept is that the immune, epithelial and microbial abnormalities in AD extend beyond the inflamed skin to also involve the “clinically normal” AD skin, possibly reflecting the systemic aspects of this disease (Figure 1).

Figure 1. “The roof is on fire” –

Immune cell subsets mediating local as well as potentially systemic inflammation.

AD is more than skin deep

AD is often associated with a variety of atopic/allergic comorbidities, including food allergy, allergic rhinoconjunctivitis, and asthma.¹³ Circulating skin-homing T-cells in severe patients show activation markers at levels higher than in healthy controls and even psoriasis,¹⁴ as well as alterations in polarizing cytokines, with a primary Th2/Th22 skewing.^15,16 Abnormalities such as immune activation and barrier abnormalities are not only found in lesional, but also in non-lesional AD skin, both in longstanding as well as early onset pediatric disease.^10,17–19 There are also increases in serum biomarkers, with CCL17 (a Th2-associated chemokine) being a robust marker of disease severity.^20–22

Recently, AD has now also been linked to other, non-allergic conditions.^23,24 Similar to what has been shown in psoriasis, adult AD patients showed increases in cardiovascular risk factors.²⁵ Those included increases in body mass index (BMI), higher odds of heavy smoking, sedentary lifestyle, arterial hypertension, and lifetime pre-diabetes, increased alcohol intake, and decreased rates of vigorous physical activity compared to non-AD individuals.^26–30 Recently, severe AD patients without any history of cardiovascular events were reported to show an increased prevalence of coronary artery disease, with coronary plaques in 48.1% of AD patients, compared to only 21.2% in healthy control subjects, as assessed by coronary computed tomography angiography.³¹ Whether AD is itself an independent risk factor for cardiovascular disease is an ongoing debate, and there might be heterogeneity among AD populations.³²

AD is also associated with higher rates of neuropsychiatric disorders such as depression, anxiety, suicidal ideation, attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD).^32–41 The exact mechanisms are still obscure but may be due to the harmful effects of AD on quality of life,^42,43 such as debilitating chronic itch accompanied by sleeplessness.⁴⁴ However, psychiatric disorders including depression, anxiety and autism have been linked to increased levels of pro-inflammatory cytokines,^45,46 that are capable of penetrating the blood-brain barrier⁴⁷ and thus could potentially influence and modulate behavior and emotions.^{32,39,48–50} Future studies will hopefully show whether AD skin disease itself is capable of shaping systemic disease, and influencing systemic immune abnormalities, or vice versa.

Elucidating the AD immune map via clinical investigation

The fact that AD is primarily a T-cell driven disease^51,52 has initially been proven by the therapeutic efficacy of T-cell targeting agents such as cyclosporine, efalizumab, and alefacept.^53,54,55 As no single animal model reflects the complex phenotype of AD, clinical trials using targeted therapeutics in humans are mandatory to fully understand this disease.^56–58

Broadly acting therapies

Broadly immunosuppressive therapies such as cyclosporine A (CsA), systemic corticosteroids, and NB-UVB have been used for decades to treat moderate-to-severe AD, but their advent is more based on serendipity than on a detailed understanding of their mode of action.⁵⁹ It has lateron been demonstrated that CsA and NB-UVB improve measures of epidermal hyperplasia and immune infiltrates, including the decrease of dendritic cell and T-cell infiltrates, but with less effects on terminal differentiation genes such as loricrin and filaggrin.^59–61 Overall, CsA resulted in more rapid and stronger improvement in genomic and cellular dysregulation, and had stronger effects on non-lesional skin.⁶⁰ Also, “inert” moisturizers such as petrolatum can robustly modulate antimicrobials and epidermal differentiation, in line with its beneficial response in AD patients.⁶² While such studies can help to identify disease biomarkers, the disease mediating cytokines need to be assessed using targeted therapies.

Th2 cytokines

It has taken a considerable amount of time since the first identification of IL-4 and IL-13 up-regulation in AD in 1994, until the approval of dupilumab, a humanized monoclonal antibody that inhibits signaling of these two key Th2 cytokines via IL-4 receptor (IL-4R) blockade. Dupilumab was recently FDA-approved for treatment of moderate-to-severe AD patients not well controlled with topical treatments.^63–66 It has been evaluated in three phase III trials, with a total of 2119 adult patients, not adequately controlled by topical medications.^67,68 Patients with an IGA score (investigator global assessment; ranging from 0 to 4, with higher numbers showing more severe disease) above 3 were included. The primary endpoint was the proportion of patients reaching an IGA of 0 (clear) or 1 (almost clear) and at least a 2-point improvement. Additional endpoints included EASI-75 (improvement of at least 75% in Eczema Area and Severity Index/EASI from baseline). Endpoints were assessed at week 16 of treatment. All studies showed significantly higher clinical improvements with dupilumab compared with placebo. The percentage of drug-treated patients reaching the primary endpoint with the now approved dosing (300mg s.c. q2w) were 38%, 36% and 39% in the three trials, as compared to placebo with 10%, 8% and 12%, respectively. EASI-75 responses were 51%, 44% and 69% with dupilumab, but only 15%, 12% and 23% with placebo, respectively. Dupilumab was generally well tolerated. Of note, patients on dupilumab had overall higher rates of conjunctivitis than placebo.^67,68 However, the mechanism of this adverse event remains to be elucidated.

The dupilumab trial data prove that Th2 cytokines are drivers of AD in adult patients, and clinical trials in pediatric populations are ongoing (NCT02407756, NCT02612454, NCT03054428).

Whether the blockade of single Th2 cytokines (such as IL-4 or IL-13) alone is able to control the disease similarly to IL-4R antagonism remains to be determined. While there are no clinical data on IL-4 blockade, studies with anti-IL-13 antibodies tralokinumab and lebrikizumab showed clinical efficacy in phase II studies.⁶⁹ However, the high placebo responses in these trials, possibly due to topical corticosteroid use, does not allow comparison with dupilumab monotherapy data.⁷⁰ The concurrent application of topical corticosteroids might have masked drug effects, as demonstrated by previous publications of trials in AD.^71,72 IL-5, another Th2-associated cytokine, has been implicated in eosinophil recruitment, and elevated blood eosinophil levels are known to occur in AD patients. Mepolizumab, a humanized antibody directed against free IL-5, was assessed in moderate-to-severe AD, but only led to modest improvements in clinical severity, albeit in a very short (2 week study),⁷³ and a phase II study is currently being conducted (NCT03055195). Another Th2 centered treatment approach is blockade of IL-31 responses, a cytokine that has been implicated as an itch mediator.⁷⁴ Nemolizumab, a humanized antibody blocking the IL-31 receptor A component, showed significant reductions in pruritus compared to placebo in a phase II trial, with concomitant reductions in clinical scores.⁷⁵

Therefore, IL-31 blockade seems to be a promising strategy for AD, and possibly also for other pruritic skin diseases. Another Th2-centered treatment approach is blockade of the TSLP-OX40 (OX40L) pathway, which has been suggested to be involved in Th2 immune activation and tolerance induction.^76–78 Blockers of OX40 (NCT02683928) and related mechanisms, including TSLP or its receptor TSLPR (NCT00757042, NCT01732510) are currently being investigated in early phase clinical trials.

Cytokine targeting beyond Th2 – IL-17 and IL-22

Despite the fact that blockade of Th2-centered inflammation shows promising results, the immune map in AD is more complex and heterogeneous than mere Type 2 cytokine-centric inflammation. While EASI-75 responses were achieved in approximately half of patients receiving dupilumab in phase III studies, the other half did not attain adequate responses,⁶⁷ but the cause remains to be elucidated. While Th2 pathway activation is common to all studied AD subsets, including adult European American and Asian, as well as infant and adolescent patients,^9,10,51 other T helper cell axes are detected variably in different AD populations, and their relative roles need to be still clarified. The Th1 axis seems to increase in chronic, adult AD, with low activation levels found in early-onset disease in adults and children.^9,79 Asian AD populations showed additional, strong Th17 activation, with some overlap to psoriasis.¹⁰ Early-onset pediatric AD also showed increased Th17-skewing compared with adults.¹⁰ Ongoing and future trials (NCT02594098) should demonstrate whether Th17 blockade can control AD skin disease, and whether these results differ depending on ethnicity or age. Another distinct T helper cell subset that is strongly upregulated in AD is the Th22 pathway.^80,81 IL-22 receptors are present primarily in epithelial cells, including the epidermis.⁸² In vitro studies suggested that IL-22 inhibits epidermal differentiation and promotes barrier defects,^83,84 a concept that is currently being tested in a phase II clinical trial with the anti-IL-22 antibody fezakinumab (NCT01941537).

IL-12/IL-23p40

Ustekinumab, a fully human monoclonal antibody blocking the subunit of IL-12 and IL-23, thereby inhibiting Th1, Th17 as well as Th22 responses, is effective for psoriasis.⁸⁵ A phase II placebo-controlled trial conducted in the United States using the approved psoriasis dosing showed clear trends of efficacy in AD but effects were likely diminished by concomitant topical glucocorticosteroid use.⁷² Also, treatment intervals seemed too long, with disease recurrence between the dosing periods.⁷² Similarly, there was no difference in ustekinumab and placebo effects in a Japanese cohort, but again, topical steroid use might have masked possible treatment effects.⁸⁶ Consequently, trials with higher dosing and shorter treatment intervals will be needed to fully assess the potential of this therapeutic agent.

IgE

A considerable proportion of AD patients show increases in total serum IgE, and are often sensitized to multiple environmental allergens. Dupilumab treatment indeed suppressed blood levels of specific IgEs for a wide range of allergens in serum from AD patients, after 16 weeks of treatment.⁸⁷ IgE has long been a logical candidate for AD treatment development. Randomized trials, however, failed to show clinical effects of IgE blockade using omalizumab,^88,89 a humanized antibody that is highly efficacious for treating chronic spontaneous urticaria.⁹⁰ These results, together with higher immune activation in patients with intrinsic than extrinsic AD,⁸ suggest that IgE is likely an epiphenomenon rather than a driver of AD, mediating primarily atopic/allergic comorbidities, but not directly impacting the atopic skin inflammation.

IL-6

IL-6 is increased in skin and blood of AD patients, but its role in disease pathogenesis is unclear.^91,92 Tocilizumab, a humanized monoclonal antibody blocking the IL-6 receptor, is currently approved by the FDA for rheumatoid arthritis treatment. In a case series of three AD patients suffering from severe disease that were treated with tocilizumab, all patients had substantial decreases in pruritus, and all obtained an EASI-50 response. However, the study was not placebo-controlled, and two patients developed signs of bacterial infection,⁹² consistent with increased infectious complications of this agent in RA patients.⁹³ This makes IL-6 blockade less attractive for AD treatment, especially in light of other potentially safer therapeutic approaches.

TNF-α

TNF-α, which has been linked to Th1/Th17 responses, is central to several inflammatory diseases, including psoriasis.⁹⁴ While TNF-α can be increased in skin and blood of AD patients,^95,96 there is no convincing evidence that TNF-blockade alleviates skin inflammation in these patients.^97,98 Conversely, there are several reports showing that TNF-blockade can lead to AD exacerbation and blood eosinophilia,^99–102 indicating that TNF-α might even have a protective effect in AD.

Small molecules

Several small molecules have also been assessed for treatment in AD. Crisaborole is a topical phosphodiesterase 4 (PDE4) inhibitor recently approved for the treatment of mild-to-moderate AD in 2 years of age or older patients.¹⁰³ PDE4 activity is increased in AD skin, resulting in decreased intracellular levels of cyclic adenosine monophosphate, which in turn leads to increased production of proinflammatory cytokines.¹⁰⁴ Crisaborole decreases levels of TNF-alpha, IL-12 and IL-23 and other inflammatory mediators¹⁰⁵ that are increased in AD. Other small molecules with potentially promising results include the oral PDE4 inhibitor apremilast,^106,107 the janus kinase blockers including tofacitinib, baricitinib and PF-04965842 (NCT02001181, NCT02576938, NCT02780167),^108–110 and the histamine H4 receptor (H4R) antagonist ZPL389,^51,111 but their efficacy needs to be demonstrated in larger clinical trials.

In sum, it has been demonstrated that T-cells are essential drivers of AD, but a considerable contribution by a specific T-cell immune axis has so far only been consistently proven in humans for the Th2 axis (IL-4, IL-13, IL-31). Other immune T-cell axes (Th1, Th17/IL-23, Th22) might also contribute to the complex and heterogeneous phenotype of AD (Figure 2), and their blockade (Table 1) could possibly close the therapeutic gap of patients not sufficiently responding to various Th2-centered treatment approaches.

Figure 2. Potential “driver” cytokines in AD.

Immune mediators that have been shown to contribute to AD skin inflammation (IL-4, IL-13, IL-31, IL-12/IL-23), and that are potentially involved in shaping the disease phenotype, that are currently being investigated in clinical trials (IL-22, IL-17).

Table 1:

Biologics approved or currently being investigated in clinical trials for the treatment of AD.

Agent	Target	Phase	Manufacturer	ClinicalTrials.gov
Dupilumab	IL-4Rα	Phase III published	Regeneron	NCT01949311
Nemolizumab	IL-31R	Phase II published	Chugai	NCT01986933
Ustekinumab	IL-12/23p40	Phase II published	Janssen	NCT01806662
Tralokinumab	IL-13	Phase II completed	MedImmune	NCT02347176
Lebrikizumab	IL-13	Phase II completed	Hoffmann-La Roche	NCT02340234
QGE031	IgE	Phase II completed	Novartis	NCT01552629
Fezakinumab	IL-22	Phase II completed	Pfizer	NCT01941537
GBR830	OX40	In Phase II	Glenmark	NCT02683928
Secukinumab	IL-17	In Phase II	Novartis	NCT02594098
BMS-981164	IL-31	Phase I completed	BMS	NCT01614756
Tezepelumab	TSLP	Phase I completed	Amgen	NCT00757042
MK-8226	TSLPR	In Phase I	Merck	NCT01732510

^*TSLP thymic stromal lymphopoietin; TSLPR thymic stromal lymphopoietin receptor.

Epidermis – victim and perpetrator

While a breach in the epidermal barrier is a characteristic component of AD, its pathogenic role for disease development is less clear. Loss-of-function mutations of the FLG gene results in a defective expression of the barrier protein filaggrin, which is currently the strongest risk factor for AD.^112–114 Filaggrin is a key component of terminal differentiation and skin barrier function that also involves pH regulation and epidermal hydration.¹¹⁵ FLG loss-of-function mutations have been linked to more severe and persistent AD, leading to a higher degree of immune dysregulation with type 1 interferon-mediated stress responses and increased IL-1 cytokine levels, as well as elevated rates of allergic sensitization and skin infections.^116–124 Nevertheless, this mutation is found only in a fraction of patients, and it is rarely present in African-American patients.¹²⁴ Conversely, patients with FLG mutations can outgrow their disease.¹²² Also, dupilumab has been demonstrated to work equally well both in filaggrin mutated and wild type patients.⁶⁴ Importantly, filaggrin is downregulated in all adult AD patients irrespective of FLG mutation status, most likely due to Th2/Th22-skewed inflammation. IL-4 and IL-13 significantly reduce filaggrin gene expression of keratinocytes in vitro,¹²⁵ an effect that was also observed with IL-22 treatment of full thickness skin equivalents that have been designed to reflect cytokine effects in vivo.⁸³ These data suggest that Th2/Th22 responses in adult patients with AD directly compromise the skin barrier. Surprisingly, though, early-onset pediatric AD did not show such a decrease in filaggrin expression despite the presence of Th2 and Th22-associated inflammatory mediators,⁹ mandating further research on barrier regulation in this age group.

Another epidermal feature of AD is an impaired innate immune response, that results in increased microbial colonization and, consequently, recurrent skin infections.^126,127 This process is further pronounced in FLG mutated skin,¹²⁸ consistent with the observation that filaggrin components do also have antimicrobial activity.¹²⁹ But irrespective of filaggrin expression, antimicrobial peptides, a key component of skin host defense, show a relative deficiency in AD patients,¹³⁰ most likely due to down-regulatory effects of IL-4 and IL-13.^131–135 IL-22, especially in combination with IL-17, may upregulate antimicrobial peptides,¹³⁶ which suggests that Th2 activation is primarily involved in antimicrobial peptide deficiency in AD.

The microbiome in shaping the immune system

During the last decade, the interaction between the environmental microflora and the immune system has been demonstrated to regulate the immune response.^137–139 Whole metagenome profiling suggests that AD-associated microbiomes can elevate the risk to develop AD flares by influencing the microenvironment of the skin surface as well as through interactions with the host immune system.¹³⁹ S. aureus seems to play an eminent role in this regard, and AD skin is frequently superinfected with this pathogen.¹ Indeed, antibiotics or dilute bleach baths that have antimicrobial activity have become a treatment option.¹⁴⁰ In general, AD flares are associated with a decreased diversity of the skin microbiota accompanied by increased abundance of S. aureus.^141,142 Species-level investigation of AD flares showed that S. aureus has greater predominance in patients with more severe AD, while S. epidermidis predominance was found in patients with less severe disease.¹⁴³ Consistently, S. aureus upregulates IL-4, IL-13 and IL-22 expression in skin.^144–146 Interestingly, S. aureus strains from more severe AD patients induced epidermal thickening, as well as expansion of Th2 and Th17 cells in a mouse model, that was stronger than S. aureus from less severe AD and healthy controls.¹⁴³ Importantly, skin commensal bacteria such as S. epidermidis and S. hominis are protective against S. aureus via production of antimicrobial peptides.¹⁴⁷ These strain-specific, highly potent antimicrobial peptides that can selectively kill S. aureus were commonly present in healthy control skin, but only rarely found on AD skin.¹⁴⁷ Conversely, reintroduction of commensal strains harboring antimicrobial activity to AD patients decreased colonization by S. aureus in vivo.¹⁴⁷ This demonstrates that commensal skin bacteria can protect against pathogens, and that dysbiosis can drive disease.¹⁴⁷

Recently, bacterial community structures and diversity was shown to shift over time in a birth cohort evaluating skin samples on day 2, month 2, and month 6 of life.¹⁴⁸ In this cohort, increased commensal staphylococci early in life reduced the risk of developing AD at month 12, while in infantile onset AD, there was no difference between the AD population and healthy control subjects,¹⁴⁸ further adding to the notion that AD is quite a heterogeneous disease.

Given the fact that AD is characterized by a high level of skin barrier defects, accompanied by frequent bacterial and viral superinfection,¹⁴⁹ blocking immune pathways might raise the concern of increasing the risk of skin infections. However, dupilumab was not only well tolerated, but also reduced S. aureus abundance on lesional and non-lesional AD skin in a phase II trial.⁸⁷ Indeed, dupilumab treatment resulted in lower rates of infectious adverse events than in placebo in clinical trials, demonstrating that immune activation directly leads to skin dysbiosis and infection, as suggested by preclinical data.^134,135,150

CONCLUSION

The interaction of immune responses in AD with epithelial dysfunction and microbial dysbiosis is complex, but clinical trials are helping to elucidate key steps in this process (Table 2). Future clinical trials with targeted therapies will likely further contribute to the elucidation of AD skin biology. Results might not only optimize treatment approaches, but will most likely also increase our understanding of AD disease heterogeneity.

Table 2:

Key messages on the immunologic, microbial and epithelial interactions in AD.

AD has a complex pathogenesis and considerable phenotypic heterogeneity. AD is Th2 centered, with variable activation of other immune axes depending on the population. FLG loss-of-function mutation is the strongest risk factor for AD, but only present in a minority of patients. AD flares are associated with a decreased diversity of the skin microbiota together with increased abundance of S. aureus. Skin commensal bacteria are protective against S. aureus via production of antimicrobial peptides. Increases in blood inflammatory and cardiovascular mediators, abnormalities in non-lesional skin, and associated allergic and non-allergic comorbidties suggest its systemic nature.