The pathogenesis of hidradenitis suppurativa: Evolving paradigms in a complex disease

Hidradenitis suppurativa is a complex inflammatory skin disease with the molecular pathogenesis of disease incompletely understood. Recent observational and experimental insights into disease pathogenesis are challenging long‐held beliefs regarding the causes and mechanisms of disease. The most effective treatments to date are anti‐inflammatory in nature suggesting inflammation is the major driver of disease activity. This study critically evaluates the existing literature regarding the mechanisms of disease pathogenesis. Specifically, it questions the role of follicular occlusion as the central driver of disease activity and reframes hidradenitis suppurativa as a complex autoinflammatory and autoimmune disorder. Ongoing efforts to understand the mechanisms of disease will no doubt lead to more efficacious therapeutics to control this burdensome disabling disease.


| INTRODUCTION
Hidradenitis suppurativa (HS) is a complex inflammatory disorder manifesting in chronic, recurrent, painful nodules abscesses, and malodorous draining tunnels with a predilection for flexural areas of skin. 1 The disease has features of both autoinflammation and autoimmunity with a variety of inflammatory pathways and cell types involved. The precise pathogenesis of the disease is incompletely understood. Historically, HS has been considered a disorder of follicular occlusion, however, novel insights have suggested it may come under the spectrum of an autoinflammatory keratinization disease (AiKD). 2 Limitations to our understanding of the pathogenesis of HS include the lack of a reliable animal model of disease and the identification of significant inflammatory heterogeneity.
Continual re-evaluation and integration of current clinical, histological and molecular data into our pathogenic model of HS is essential to advance our understanding of the disease. 3 Challenging existing paradigms 4 through observation, hypothesis, and experimentation (and separating the interpretation of these results from personal selfinterest) 5 is a core component of the scientific process 5 ; and is essential to enable accurate identification of novel therapeutic targets and treatment strategies. 6,7 It is also vital to the understanding of differential treatment response in different individuals and exploring the potential role of variations in inflammatory endotypes in the disease 6 ; in a similar way to how this has been identified in atopic dermatitis. 6 This review aims to synthesize existing knowledge from clinical observation, classical histology, as well as modern molecular biology techniques to evaluate the evidence for HS as either a disorder of follicular occlusion or an AiKD.

| THE EVOLVING PATHOGENIC PARADIGM OF HIDRADENITIS SUPPURATIVA
Historically, HS has been proposed to be a disorder to apocrine glands inflammation, 8,9 although multiple independent histological studies have demonstrated that inflammatory involvement of apocrine glands is a secondary phenomenon, [10][11][12] and that the primary inflammatory drive of the disease exists adjacent to keratinocytes of the follicular infundibulum and interfollicular epidermis. 10,13,14 It is now widely accepted that the primary drive of disease activity centers upon the follicular infundibulum 13,14 (Figure 1). Additionally, other disorders such as pilonidal sinus disease 15 and dissecting cellulitis of the scalp 16 share many clinical, histological, and inflammatory features with HS [15][16][17] and are beginning to be considered as uni-localized variants of disease. 18 Melnik's seminal 2013 paper 19 began to shift the pathogenic paradigm of HS away from an apocrine-gland-based inflammatory or infectious disorder, to a disorder of follicular occlusion and proposed dysregulated Notch signaling 19,20 as the unifying feature of HS pathogenesis.
Emerging evidence as to the role of the inflammasome, [21][22][23][24] complement, [25][26][27][28][29] and IL-1 isoforms [30][31][32][33] has led to the suggestion of HS as an AiKD. 2 Evidence of systemic inflammation, 34,35 activation of F I G U R E 1 (A) Multiple contributing factors including genetic alterations (including documented gamma-secretase-complex polymorphisms), exogenous environmental stimuli, and contributions from a dysregulated microbiome, activate resident dendritic cells via the production or stimulation of various inflammatory factors. These then lead to the stimulation and activation of various T cell subsets which lead to keratinocyte and other inflammatory cell activation via various products including IL-17, TNF-a, IL-1a, and IL-1b. The downstream effects on follicular epithelium and interfollicular epithelium leads to the production of pro-inflammatory mediators by proliferating keratinocytes. This is the likely association with comedone formation in established disease. (B) Pseudo-psoriasiform epidermal hyperplasia in the epidermis of HS lesions is the site of cytokine production including IL-1a, IL-1B, TNF-alpha, IL-17C, and CXCL1/CXCL8 isoforms. This then feeds into crosstalk with dermal inflammatory and stromal (fibroblast) lesions leading to a self-perpetuating influx of B cells and T cells (via CCL20 and CXCL13), macrophages, monocytes, and neutrophils. It is hypothesized that surrounding the follicular infundibulum that keratinocyte-fibroblast crosstalk may be a mechanism by which epithelialized tunnels may develop by outgrowth of keratinocytes from the outer root sheath. (C) Established epithelialized tunnels are a major producer of inflammatory cytokines and chemokines to a higher level than that expressed in the overlying epidermis. Transepithelial neutrophil migration leads to development of the infiltrative proliferative gelatinous mass (IPGM) found in the lumen of such tunnels. The surrounding collection of B cells dendritic cells and neutrophils (producing neutrophil extracellular traps), all contribute to the self-amplifying inflammation seen in severe disease B cells 36,37 and plasma cells 36,38,39 have raised the possibility of HS having an autoimmune or antibody-mediated component. 37 However, follicular occlusion is still considered the "primem movens" of HS 4 preceding the inflammatory drive of disease.

| FOLLICULAR OCCLUSION
3.1 | Comedones are clinically and experimentally a product of inflammation, rather than a cause Histological studies illustrate the prominent role of comedogenesis, follicular hyperkeratosis, and comedogenesis in HS tissues. 10,13,14 However, in each instance, the coexistence of perifollicular inflammation is comparably prominent. 10,13,14 Clinically, comedones (both open and closed) as well as typical double-sided comedones 40 are present in diseased areas, inflamed tissues, and also in scarred, noninflamed tissues. 13,14,[40][41][42] They are also present in areas not exposed to flexural occlusion. 43 Von Laffert et al. 13,14 report comedones as more common in endstage fibrotic and scarred lesions and independent of the follicular unit. 13,14 These comedones are more likely to be those of the doubleended variety which were once pathognomonic of HS. 40 From these clinical observations, we can conclude that comedones are associated with HS, however, the establishment of causation requires mechanistic evidence. Such mechanistic evidence is available thanks to Investigations into comedogenesis in acne research. [44][45][46][47] Recent findings have identified subclinical inflammation as preceding comedogenesis in acne prone skin 44 disrupting the long-standing assumption that follicular occlusion is the primary initiating factor in acne. 46 The molecular mechanisms of comedogenesis involve follicular keratinocytes producing a number of proinflammatory mediators [44][45][46][47][48][49] (including antimicrobial peptides, microbial associated proteins including Lipotechoic acid, CCL20, and IL-1α). Ex vivo studies of the follicular infundibum 50 isolated in vitro are able to recapitulate the formation of comedones with addition of IL-1α and prevent formation with the addition of IL-1RA. 50 While these studies were focused upon highly sebaceous follicular units which are distinctly different from skin-bearing apocrine glands, the Th17 associated mediators which are involved in the HS inflammatory response are similar. 6,51 Therefore, these in vitro studies highlight the theoretical prospect of these immunuological mechanisms being common between different anatomical locations. Reproduction of these experiments using follicular infundibular from apocrine gland bearing regions would hopefully be definitive in confirming or refuting these findings. Ongoing clinical trials using IL-1α antagonists (such as Bermekimab) may contribute to our knowledge in this area. Therefore, we can conclude that molecular and ex-vivo evidence suggests comedone formation is possible secondary to subclinical inflammation, rather than inflammation solely being the result of comedone formation and follicular rupture. 4 The precise mechanisms have been demonstrated in human skin, however, require validation in apocrine gland bearing skin given the unique immunological mileu of these sites.

| Skin fold occlusion is associated with microbiome alterations and subsequent pro-inflammatory keratinocyte responses
On a microscopic level, follicular occlusion in HS refers to occlusion of the follicle at the infundibulum, 4,10,13,14 however, clinically, "follicular occlusion" tends to be associated with specific body sites including the axillae, inguinal flexures, and sub-mammary folds. 4 These are areas of predilection in HS as they are prone to increased frictional trauma, altered pH, increased moisture levels, and microbiological colonization. 54,55 Among obese patients, the posterior neck folds, abdominal pannus, gluteal cleft, and inner thighs, and other anatomical sites can undergo similar microbiological milieu alterations secondary to heat, moisture, and pH changes. 56 The follicular infundibulum is significantly different to other parts of the hair follicle (including the bulb) because of its immunologically active nature and its related role in developing an immune tolerance to commensal micro-organisms. [57][58][59] Healthy infundibular keratinocytes functioning within a normal physiological environment are responsible for producing CCL20 as well as antimicrobial peptides. 57 Due to the increased moisture and reduced pH of the stratum corneum that HS causes, colonization of porphyromonas species. 60,61 Further, the release of pre-formed 1α in keratinocytes 62 can be provoked by bacterial specials other than porphyromonas (including Staph Aureus 63 and Propionibacterium ances 64 ) ( Figure 1). Additionally, there is ancillary evidence suggesting that yeasts play a role in the inflammatory process of HS. 65 Recent observational research has demonstrated that anti-saccharomyces cerevisae antibodies 66 (present in severe HS disease) can cross react with other fungi and bacteria. 65 The specific mechanism of such microbiological specials remain poorly understood, however, their significant role in producing a pro-inflammatory response (via keratinocytes either directly or indirectly) has been demonstrated in observational studies investigating both early and advanced HS. 61 36,38 NETosis, 36 and epithelialized tunnels. 68 HS has been shown to be an inflammatory disease in multiple molecular [69][70][71][72][73] and histological studies. 67,69 Despite this principle being well established, it must be recognized that the majority of biological samples (HS lesional tissue) was taken from patients who have longstanding and severe disease. 36,38,39 Additionally, until very recently, standardized biopsy sites and requirements had not been defined for HS tissue studies. 74 These factors have lead to two limitations in our current understanding of HS as an inflammatory process. First, there is a lack of studies investigating early HS disease and factors which may precede or provoke HS development. Second, because HS has multiple morphologies, it cannot reliably be concluded that a biopsy taken from one site is representative of the various morphologies that exist within the spectrum of HS disease. 74 For these reasons, it is important to review and consider any study results which do not specify such factors (disease severity and the anatomical location of tissue samples) with caution.
The common inflammatory signatures on qRT-PCR identified in HS studies of lesional tissue include TNF-α, IL-1α, IL-1β, IL-6, IL-17A, IL-17F, IL-32, IL-36α, IL-36g, and IL-10. 73 Additional chemokines include CCL3, CCL5, CCL27, and BLC. 70 Nonlesional tissue also demonstrates upregulated levels of many of these cytokines 51,70 although variation does exist due to previous lack of standardized biopsy sites and combination of both partially treated as well as untreated specimens. 70 RNAseq transcriptomic studies demonstrate strong B-cell signatures with IgG1 and IgG3 immunoglobulins and aspects of the complement cascade highly upregulated. 39 Additionally, signals of keratinocyte hyperplasia (Keratin 6, Keratin 16) are also seen with keratinocyte derived factors being elevated in lesional and perilesional tissue compared with unaffected and control skin. 67 Variation in cytokine levels do occur (between lesional, perilesional, and nonlesional tissue) in terms of type and degree of inflammation although reliable characterization of inflammation matched with disease morphology (e.g., nodules vs. tunnels) is yet to be undertaken. Scarred tissue demonstrates decreased inflammatory profiles compared to nonscared areas, 70 and the presence of occult dermal tunnels can also induce highly inflammatory profiles in normal-appearing skin. 70 The use of clinical ultrasound has been suggested as a method of confirming or excluding the presence of tunnels before biopsy. 74 Analysis of serum has identified IL-1β, IL-6, IL-8, IL-10, IL-12p70, IL-17, TNF-α as upregulated in multiple studies, 70 however, conflicting results exist between studies between serum levels of IL-10, IL-17, and IFN-y which may be secondary to the severity of included participants and the methods of cytokine analysis. 70 The majority of data regarding serum inflammation is based upon patients with Hurley stages 2 and 3 disease 70 with the changes in serum inflammatory markers in early and mild disease unclear.
In terms of establishing mechanism-it has been assumed (based on observational studies) that perilesional inflammation is of the same character (albeit less intense) than nearby lesional inflammation. 70,71,75 Therefore, given the known feedback mechanisms between IL-1 and IL-17 76 leading to self-perpetuating feedforward inflammation, it is reasonable to assume that lesional tissue inflammatory characteristics may be replicated by adding proinflammatory cytokines to perilesional tissue. This experiment (conducted by Vossen et al. 75 ) was unable to replicate the lesional HS inflammatory profile suggesting that IL-1α and/or IL-1β are not the sole triggers necessary to induce the development of lesions in HS. 75,77 Other possibilities are that a combination of multiple inflammatory mediators are required; or as-yet-unknown predisposing factors are involved in inducing active inflammatory nodules on a background of perilesional subclinical inflammation. 77 This raises the prospect that the process of inflammation in HS is more complex than initially thought. The underlying assumption thus far in HS research is that perilesional tissue represents the same inflammatory profile as lesional tissue, differing only in the degree, intensity, and more superficial location of inflammation. 77 However, an alternative hypothesis, that the inflammatory characteristics of perilesional tissue is distinct from lesional tissue 77 remains to be thoroughly investigated.

| Disease initiation is associated with systemic subclinical inflammation and dysregulated infundibular keratinocytes
The site of initial inflammation in HS is centered upon the infundibulum of the hair follicle. 13,14 Given the active immunologic role of the follicular infundibulum, 49 a degree of baseline inflammatory activity around the follicle is considered normal. Our understanding, however, of the factors which initiate and perpetuate the immoderate and self-perpetuating peri-follicular inflammation associated with the pathogenesis of HS is not complete. Multiple factors such as hormonal dysregulation, insulin resistance, and obesity have been suggested as possible associations of HS disease which may contribute to a generalized pro-inflammatory state. 78 Furthermore, diseases such as psoriasis, 79 84,85 indicate that there are benefits to reducing systemic inflammation via weight loss and smoking cessation. 82,83 The mechanisms of these pro-inflammatory cascades are complex and incompletely understood. [86][87][88] Smoking, via polycyclic aromatic hydrocarbons can directly alter follicular keratinocyte differentiation resulting in comedogenesis. 89 It can also produce widespread methylation changes and systemic increases in IL-6, CRP, Fibrinogen, and multiple members of the NF-kB family. 88 Adipose tissue can produce proinflammatory signatures including IL-6, IL-1β, TNF-α in the setting of chronic nutrient excess. 86,87 Additionally, adipokines can mediate both inflammation and the development of insulin resistance 90 which is also associated with HS. 78 Keratinocytes in the infra-infundibulum of the follicle express Type 1 and 5 hydroxy-testosterone 91 ; modulating infundibular keratinocyte differentiation programs both directly; as well as via fibroblast activation and fibroblast-keratinocyte interactions, 92 contributing to androgen-induced follicular changes. 92 These associations indicate that systemic inflammation and infundibular keratinocyte dysregulation are likely to be factors which pre-dispose patients to suffer from clinical disease.
There are contradictory reports 93

| TH17 FEED-FORWARD INFLAMMATION IS PROMINENT IN ESTABLISHED DISEASE
While the mechanisms behind Th17's role as a feed-forward selfamplifier in HS remains uncertain, it is well established that the TH17 axis is strongly implicated with clinical disease. 69 It has been presumed that the activation of the TH17 axis in HS is similar to that of it's activation in psoriatic disease. A predisposition of a Th17 immune response at cutaneous locations that are apocrine gland rich (such as the axillary regions) has been experimentally indicated. 51 Literature regarding psoriatice disease presents well-established evidence surrounding the positive feedback loops (or "feed-forward mechanisms") that exists between IL-1β, IL-6, and TNF-α by IL-17 leading to further IL-1β, IL-6, and TNF-α production as well as downstream activation of acute-phase reactants, neutrophilic and complement-mediated inflammatory responses. [95][96][97][98] This is further perpetuated through leukocyte-keratinocyte interactions, 92,99 further amplifying antimicrobial peptide and chemokine production (including CXCL1 and CXCL8) 98 36,38 indicates that a longstanding and severe HS may secondary to autoimmune or antibody mediated dysregulation. No product or cell, however, has been identified as an autoimmune target among HS patients. 37 In other dermatological inflammatory diseases such as eczema and psoriasis, B cells are also present but, similar to HS, no known autoimmune targets have been identified. 37 Researchers have suggested that the B cells are secondary to combined B cell and T cell chemo-attractants (CXCL13 or CCL20) and thus amplify inflammation that is mediated by T cells. 37 A study by Byrd et al. 36  GSC mutations in the pathogenesis of HS is unknown. 102 What is known, however, is that over 70 substrates are involved in the cell cycle and inflammation (such as EGFR, IL-1, TNF-α, and Notch) are cleaved from the GSC complex. 102 Melnik and Plewig proposed that Notch signaling is the unifying element in HS pathogenesis via associations with keratinocyte proliferation, smoking, and sequence variants in GSC. 19,20 The molecular evidence for Notch being associated with keratinocyte hyperproliferation is well established, 108,109 however, dysregulated Notch signaling is also associated with other inflammatory skin disorders including psoriasis, atopic dermatitis, and alopecia areata. 103 Notch dysregulation may be an epiphenomenon secondary to keratinocyte proliferation (as it is present in multiple other inflammatory dermatoses) rather than the primary cause of HS. 103 In silico evidence 102 has identified HS-specific GSC substrates ERbb4 and Tie1 102 as differentially expressed substrates that distinguish the transcriptome of HS from familial Alzheimer's disease and other inflammatory skin diseases. 102 ErbB4 and Tie1 are components of the EGFR pathway (active in the follicular infundibulum) 49 and are associated with SOX9 and Wnt signaling linked with hair cycle progression, IL-17A production 49,110 (through shared downstream Act1 activity), and epithelial cell fate, 110 all mechanisms identified in transcriptomic analysis of HS tissues. 39,72 In vitro studies have demonstrated diverse pro-inflammatory results of Nicastrin knockdown including IL-36a production, 111 alterations in EGFR signaling 112 as well as increased sensitivity to interferon mediated pro inflammatory pathways. 113 Recently, mutations in POFUT1 have been identified in cases of Dowling Degos Disease associated with HS. 114,115 POFUT-1 is a fucosyltransferase which is active upon multiple substrates including Notch and EGFR 116 and is important for post translational modification of receptors. 116 Therefore, abnormal activity of the EGFR pathway is linked with infundibular keratinocyte differentiation 49 and Th17 inflammatory pathways. 110 The link to clinical disease activity is supported by reports of HS associated with the use of EGFR antagonists in oncology. 117 Therefore, dysregulation of EGFR associated pathways secondary to GSC mutations may explain both the infundibular localization of HS, the involvement of the Th17 immune axis and cases of HS-like features in the setting of EGFR antagonism.

| The evidence and proposed mechanisms for follicular rupture
Follicular rupture is proposed as the primary mechanism that follicular occlusion leads to diffuse dermal inflammation in HS. 4 The histological evidence for follicular rupture is largely based upon findings from observational studies which show both dense perifollicular and intrafollicular inflammation which discontinues in the follicular epithelium in affected tissue. 10,13,16 Longstanding disease often has a noticeable absence of follicular and adnexal structures replaced with dense inflammatory infiltrates and scarring 118 consistent with the known profound dermal inflammation but the mechanisms and process of follicular rupture are poorly understood.
Danby et al. 119 documented PAS staining of the basement membrane zone (BMZ) in lesional HS tissues, and identified decreased staining compared to healthy controls. 119 The authors proposed that defects or thinning of the BMZ may predispose the follicle in HS to rupture, with subsequent spillage of intrafollicular contents into the dermis stimulating the inflammatory cascade. 4 The authors point to reduction in PAS staining to support this claim. In examination of the his-  118 and the development of dermal tunnels. 99 These EMT-associated signaling routes have also been identified in malignancy and wound healing processes and are thought to play a role in the metastatic potential of malignancy as well as the pathogenesis of longstanding wounds. 99,125 Follicular rupture, therefore, may be more accurately considered as a process of "follicular disassembly" which is precipitated by chronic inflammatory changes via EMT and abnormal extracellular remodeling wound healing programs. 125

| DERMAL TUNNELS ARE ACTIVE INFLAMMATORY ORGANS AND THEIR DEVELOPMENT IS ORCHESTRATED BY DERMAL INFLAMMATION
Currently no mechanistic explanation exists as to how the dermal tunnels of advanced HS form. The dermal tunnels seen in HS disease are unique structures which are made up of stratified squamous epithelium and reflect the structure of the overlying epidermis. 126 This differs to other tunnel-like structures which can be appreciated in other chronic inflammatory conditions including fistulizing Crohn's disease where the tunnels do not recapitulate mucosal structures in the same way. 127 Comprising of keratinocytes, melanocytes, Langerhans cells 126 these tunnels express pro-inflammatory mediators and are a second source of keratinocyte-derived inflammatory chemokines and cytokines in the dermis. These tunnels have significant association with the systemic inflammatory nature of HS in both tissue and serum. 128 It is thought that these dermal tunnels seen in HS are derived from abnormal keratinocyte overgrowth from the outer root sheath of the follicle, 57,99 although the exact mechanism and pathogenesis of tunnel formation is not known. An association between tunnel development and signaling from the dermis is likely as the tunnels do not extend beyond the dermis into subcutaneous tissue. 99 This aligns with the known development of hair follicles and the process of early anagen downgrowth in the hair cycle. 129,130 These processed are mediated via PDGFα -derived signaling from the dermal condensate. 130 Further, PDGFα mediated signaling has been demonstrated in transcriptomic data from HS-associated fibroblasts. 99 These fibroblast-derived signals are known to be secondary to inflammation-mediated epigenetic modifications and it is thus reasonable to assume that the development of dermal tunnels is an inflammatory process. However, once these tunnels are established, the CXCL1/8 gradient established across the epithelia 98 (including tunnels) results in transepithelial neutrophil trafficking and NET formation in tunnel lumen. 36 This results in development of the infiltrative proliferative gelatinous mass (IPGM) 131 and biofilm formation in HS tunnels. 132 Consequently, this provokes the recruitment of additional inflammatory cells to surround these dermal tunnels, contributing to the ongoing cycle of ongoing and severe inflammation and drainage. Therefore, rather than an end stage phenomenon, dermal tunnels are an inflammation driven process resulting in the development of active inflammatory organs contributing to the perpetuation of dermal inflammation in established disease.

| CONCLUSIONS
The available histological and molecular evidence suggest inflamma-