Information sources for this review included PubMed (1946-July 1 2018), Scopus (2004- July 1 2018) and Web of Science (1990-July 1 2018) as shown in Figure 1. Search strategy is presented in Table 1

Eligibility criteria for this review included cohort studies, case-control studies and other observational studies with no restrictions of patient age, sex, ethnicity or language of publication. Eligible studies included:

Studies deemed not eligible included those which:

Data collection was performed independently by 2 authors (JWF & JEH), with any disagreements regarding inclusion of citations being referred to a third author (JGK) for mediation. Information was collected using a standardized data collection form (available as Extended data ²² ) with the principal outcomes of interest being the cytokine of interest, measured level of cytokine in lesional HS skin or serum. Comparison data against either peri-lesional, unaffected or control skin or serum was also collated. If data from individual patients was not available then the aggregate data including average change and statistical analyses of the significance of change was collected.

For each individual cytokine, where more than one study reported results, heterogeneity was assessed using the chi-squared tests for homogeneity. Homogeneity was defined as a chi squared value >0.05. All statistical analysis was undertaken using R (version 3.5.1)

Potential sources of bias in the identified studies are acknowledged including the small size of patient cohorts, the variability in sampling, laboratory techniques and the inclusion of patients being treated with a wide-variety of medications including immunosuppressants. Bias was also assessed using the NIH quality assessment tool for observational studies ²³ .

The summarized demographic data of the patients and controls comprising this review are included in Table 2. The 564 reported cases comprised of 231 males (40.9% reported cases) and 333 females (59.0%). 24 cases were unreported (4.1%). The average age was 38.5 years (n=560, 18 cases unreported). 141 individuals were current smokers (82.4% reported cases), 8 ex-smokers (4.7% reported cases), 22 non-smokers (12.8% reported cases) and 407 unreported. Obesity (BMI>30) was reported in 85 individuals (42.5% reported cases), with 115 (57.5%) individuals non-obese (BMI<30) and unreported in 378 cases. 8 cases reported diabetes mellitus out of 24 reports (33% of reported cases). 12/38 cases reported a positive family history of HS (31.6% reported cases). Hurley Stage was reported as stage 1 in 68 individuals (17.4% reported), stage 2 in 199 individuals (51% reported cases) and stage 3 in 123 individuals (31.6% reported cases) with 188 cases going unreported. The average mHSS (modified hidradenitis suppurativa score) was 78.1 (n=247 cases). Biopsies were largely taken from the axillae (n=32, 43.8%) and groin (n=35, 48.0%), with a minority of samples being taken from the genital and perianal region (n=6, 8.2%). At the time of sampling patients were on treatment including Clindamycin+ Rifampicin (n=18); adalimumab (n=26); Metformin (n=2); levothyroxine (n=1); MABp1 (n=10); tetracyclines (n=12) Infliximab (n=2); other antibiotics (n=4). Treatment was not specified in 74 cases, with no treatment in 86 individuals and treatment withheld in 85 patients.

Only 5/19 (26.3%) studies analysed both lesional tissue and serum levels of cytokines, enabling direct comparison between these two compartments. 8/19 (42.1%) studies provided age and sex matched controls, 5/15 (33.3%) studies stratified by disease severity and no studies stratified by lesion site or comorbidities. 8/19 (42.1%) studies stratified or accounted for treatment or reported discontinuing treatment up to 3 weeks prior to sample collection ( Table 3).

A total of 81 discrete cytokines were analysed over the 19 studies (presented in Table 4). 6 studies provided a total of 78 outcomes from tissue of lesional or peri-lesional biopsies, 4 studies provided a total of 30 results from serum analysis and 1 study provided 15 results from exudate analysis. The remaining 8 studies did not provide quantification of cytokine levels but did provide analysis of the change and significance between lesion and control samples. The degree of change between lesional and control samples varied widely from 1.5 times the control level (IL-1RA p=0.0112) to 149 times the control level (IL-17 p<0.05). 33 cytokines were evaluated in more than one study. Only IL-1β, IL-6, IL-8, IL-17A and TNF-α had data from 5 or more separate studies.

Cytokines and inflammatory proteins which were elevated in more than one study in lesional tissue included IL-1β, IL-6R, IL-10, IL-17A, IL-36α, IL-36β, IL-36 γ, IL-36RA, TNF-α, sTNFR2, hBD1, hBD2, hBD3, s100A7, LL37/Cathelicidin, CCL3, CCL5, CCL27 and BLC. Cytokines and inflammatory proteins elevated in peri-lesional tissue included IL-1β, IL-17, IL-36β, IL-36RA, IL-37, IL-38 and TNF-α. IL-37 was the only cytokine identified which showed significant differences between lesional and peri-lesional tissue, with a 1.81 times elevation in lesional compared to peri-lesional tissue (p=0.0002) ³ . IL-17 was elevated in unaffected HS tissue compared to control patient tissue (p<0.05) in one study ³¹ . In HS tissue, S100A9, hBD1 and hBD2 were reduced but this data did not meet statistical significance. Two studies measuring IL-1β levels showed no statistically significant difference between lesional and control skin ^{7, 25} . No significant elevation of IL-6 was seen in lesional tissue compared to control with the exception of 1 study ²⁵ . IL-8 levels only just made significance in two studies ^{5, 7} , with one study showing significant elevation of IL-8 in lesional compared to control tissue ²⁴ . Two additional studies showed no significant difference ^{4, 8} . TNF-α levels were significantly elevated compared to control tissue in two studies ^{7, 31} but not significantly in 2 additional studies ^{4, 24} . sTNFR1 was significantly elevated in one study ²⁶ whilst showing a non-significant difference in a second study ²⁵ . CCL5 was significant in 2 studies in lesional tissue compared with controls ^{4, 26} . One methodology using muramyl dipeptide (MDP) did not reach statistical significance compared to stimulation with Pam2CSK4 Lipopeptide, and non-treated (NT) cells. IFN- γ was elevated in lesional tissue with no significance in one study ²⁸ and significance in another ⁴ .

Elevated cytokines and inflammatory proteins in HS serum included IL-1β, IL-6, IL-8, IL-10, IL-12p70, IL-17, TNF-α, sTNFR1, CRP, ESR, LC2, and MMP2. TNF-β, and IFN-γ were elevated in wound exudate from active HS lesions. IFN-γ was noted to be decreased in HS patient serum compared to healthy control serum, despite the elevation in wound exudate. Conflicting results were seen in serum findings in IL-10, IL-17 and IFN-γ. One study demonstrated elevated serum IL-10 levels compared to control ⁵ whereas two other studies ^{8, 27} showed no significant difference. Whilst two studies ^{4, 5} illustrated elevated IL-17 Serum levels in HS patients, one study ⁷ showed no significant difference between patients and controls. IFN-γ showed no statistically significant decrease in the serum of HS patients compared to control in one study ⁹ but a significant difference in a larger, higher powered study ⁴ .

Because adalimumab improves HS through TNF antagonism ^{1, 2} , this cytokine must be classified as pathogenic. TNF mediates inflammation in a classic “sepsis” cascade in tissues—in this pathway LPS from gram negative bacteria activates TNF release from cells, and then TNF stimulates production of IL-1b, IL-6, and IL-8, leading to neutrophil attraction into sites of infection ^{2, 4} . Increases in IL-1β and IL-8 measured in HS, as well as neutrophil accumulation, could result from this pathway. Alternatively, in psoriasis, TNF is a major cytokine that acts on the IL-23/Type 17 T-cell pathway at two points. First TNF induces IL-23 synthesis in myeloid (CD11c+) dendritic cells in the skin ³⁴ . Second, TNF (as well as other cytokines that also activate NF-kB) act synergistically with IL-17A or IL-17F to increase synthesis of many other cytokines, chemokines, and inflammatory molecules in keratinocytes and other cell types. There are several clues that an IL-23/Type17 T-cell pathway may be active in HS which include detection of T _h17 T-cells in skin infiltrates, increased production of IL-17A, and increased production of LL-37/cathlecidin, S100A7, S100A8, S100A9, LCN2, IL-8, beta-defensins and IL-36; which are all molecules induced by IL-17 in keratinocytes, as also the presence of psoriasis-like epidermal hyperplasia in some reports. The increased production of CCL20 ⁴ , would be predicted to increase tissue infiltration of both T _h17 T-cells and CD11c+ DCs, which have both been observed in HS, and increased production of TGF-β could increase differentiation of T _h17 T-cells from precursors and/or influence scarring in skin lesions. If IL-17 is driving inflammation in HS, one would expect to see increased production of additional chemokines that regulate neutrophil chemoattraction (CXCL1, CXCL2, CXCL3). Epidermal hyperplasia is not presently explained in HS, but this could be related potentially to increased expression of IL-19, IL-20 or IL-22, which are associated with the IL-23/Type 17 T-cell axis. If IL-22 is produced in HS lesions, this would implicate T _h22 T-cells as a T-cell type also associated with the IL-23/Type 17 T-cell axis. There is an uncertain role for other T-cell subsets in HS. Increased production of CXCL9 and IP-10 (CXCL10) are often linked to production of IFN-γ from T _h1 T-cells in inflammatory sites, but IL-26 or IL-29, which are also cytokines produced by T _h17 T-cells are alternative activators of STAT1 and CXCL9 production. IL-32 production in HS may also be linked to a T-cell subset that produces this cytokine. Low production of T _h2 associated cytokines (IL-4, IL-5, or IL-13) has been measured in HS, suggesting an unlikely role of this T-cell subset. Likewise, the presence and function of T regulatory cells (Tregs) in HS lesions needs further study. IL-10 which is elevated in HS could be produced by either Tregs or the cDC1 (BDCA3+) DC subset, but levels may be inadequate to control tissue inflammation. At present, dendritic cell subsets are also incompletely characterized in HS. Potential sources of IL-12 or IL-23 are CD11c+ DCs, which includes the tissue resident BDCA-1+ (cDC2) subset and less mature inflammatory DCs, which are abundant cells in inflammatory lesions of psoriasis or atopic dermatitis but have not been investigated in HS. Cytokine contributions by other cell types such as innate lymphoid cells, macrophages, mast cells, and other leukocytes also remains to be determined.

The methodologies of cytokine analysis varied widely ( Table 5). 92 results were produced using electrochemical luminescence (ECL) procedures from three separate systems and manufacturers. 62 results were produced using ELISA. 18 results ⁴ were performed with either ELISA or ECL but not further specified. 15 results were produced using polymerase chain reaction (PCR) with three separate systems from three manufacturers. Four discrete cytokines (IL-10, IL-17, TNF-α and IFN-γ) were analysed using all three techniques (ECL, ELISA and PCR), whilst 15 discrete cytokines (IL-6, IL-8, IL12p40, IL-17A, IL-22, IL-23, S100A7, S100A8, S100A9, RNAse7, IP-10, CCL5, CCL20, CCL27) were analysed using ELISA and ECL only. We note IL-17 levels may well be below the lower limit of quantification with ELC and ELISA based approaches, with only the Singulex platform having the ability to quantify levels of IL-17 present in blood and serum of normal subjects.

Assessment of bias is presented in Table 6. Two of the 14 questions regarding participation rate and loss to follow up were considered not applicable. All included studies identified clear objectives and a clearly defined study population. No clear inclusion or exclusion criteria were specified for 17 of the 19 studies. Power estimation was made for one study ³³ , and recording of all exposures (disease activity, comorbidities etc) were made prior to assessment of the outcomes (cytokine levels). The timeframe of analysis was sufficient to identify an association, but only 10 of the 19 studies (52.6%) documented different levels of exposures (disease severity, metabolic comorbidities, family history etc). There were no serial measures of cytokine levels in the majority of studies. Only three studies ^{5, 25, 33} , examining cytokine levels after monoclonal antibody administration has measurements at two distinct time points. Outcomes of interest (cytokine levels) were measured consistently within studies, however there was great variance in the methods of measurement and analysis between studies ( Table 5). No studies took into account known confounding variables into analysis of their results by stratification or regression analyses.

36 of the 81 identified cytokines or inflammatory proteins were assessed by more than 1 study. 23 of those cytokines had raw data available. No studies had sufficient measures of spread in order to calculate I ²measure of heterogeneity and so chi-squared statistic was used as an alternate marker of heterogeneity ( Table 7) along with a funnel plot ( Figure 3). In total, 18 individual cytokines (78.2%) were found to demonstrate heterogeneity. Only eight cytokines (Serum IL-10, Lesional IL-1α, IL-12p70, hBD1, hBD2, hBD3, S100A9 and GMCSF) illustrated homogeneity. Due to this high level of heterogeneity and concerns regarding the methodological quality of included studies, meta-analysis was not deemed appropriate to perform.

Regarding methods of cytokine analysis, a number of authors have identified variability in cytokine levels measured with different forms of multiplex assays as well as traditional ELISA methods ^{35– 39} . Different methods of cytokine analysis are known to be prone to variability, with some cytokines more sensitive than others. For example, IFN- γ and IL-1β were overestimated compared with ELISA methods ³⁷ , whilst IL-6 levels were underestimated ³⁷ . IL-6 levels when compared across four different multiplex assays showed significant variation in detectable range, accuracy and responsiveness ³⁶ . The correlation of TNF-α between ELISA and Multiplex assays was also poor (r=0.31) ³⁶ . Issues also exist with minimum detectable levels of cytokines with specific bead-based arrays ³⁶ As an example, minimal detectable dose readings reported for IL-12p70 using some multiplex arrays ³⁹ are higher than the levels reported in lesional HS samples ⁶ . Therefore, whilst the general trends in the level of consistently elevated or suppressed cytokines in HS are reliable, the quantification of individual cytokines as well as the relationship between comorbidities and cytokine levels requires further research with consistent, reliable and accurate methodologies in order to further dissect the inflammatory cascade in this disease.

The majority of elevated cytokines and inflammatory proteins identified in lesional skin of HS (TNF-α, IL-1β, IL-6, IL-8, IL-11, IL-23, IL-17A, IL-33, IL-36, LL-37, S100A7, S100A8, S100A9, GM-CSF, TGF-β, hBD2, hBD3, CCL3, CXCL9, CXCL11, PDGF, CCL5, CCL-20, MIF, GM-CSF and LCN2) are those known to be produced by keratinocytes, as well as perpetuating a self-amplification pathway ³⁴ ( Figure 2). Additionally T-cells produce IL-17A, IL-17F, IL-26, IL-29, and IFN-γ; dendritic cells produce IL-12, IL-23 and possibly IL-39; neutrophils produce S100A8 and S100A9 (calgranulin); and innate lymphoid cells also contribute IFN-γ, IL-17A and IL-17F. This inflammatory model has been well documented and explored in both psoriasis and atopic dermatitis ^{34, 40} . The psoriasiform epidermal hyperplasia seen in HS (mediated by IL-17 and maintained by IL-23-mediated T _h17 stimulation) ³⁴ reflects this common inflammatory pathway.

The other elevated non-keratinocyte produced cytokines in HS (IL-4, IL-5, IL-10, IL-16, IL-17A, IL-22, IL-32, IL-36, hBD1), are produced by a combination of dendritic cells, monocytes, neutrophils and CD4+ T cells. IL-4 and IL-5 as key cytokines in the T _h2 axis are consistent with the findings of Mast cells in HS ⁴¹ , as well as the pruritus, which is frequently reported by patients. IL-10 in HS is produced by Treg cells ² (although dendritic cells may also be a source), and whilst quantitatively the IL-10 signal appears paradoxically elevated, it can be explained by the up-regulation of T cells including Treg cells, which although significantly elevated from baseline, are not elevated enough in comparison to T _H17/IL-17/IL-22 signal to counteract this strong pro-inflammatory cascade ² . Further exploration of these cytokines may reveal the initial trigger(s) of the inflammatory cascade in HS, or correlations with known pro-inflammatory comorbidities.

In light of investigations in psoriasis and atopic dermatitis, the role of dendritic cells in HS needs to be clarified, as dendritic cell influx has been reported in histological studies ^{41, 42} , and they may contribute to the high IL-10 and IL-15 levels reported. IL-32 is a second cytokine produced by dendritic cells, but has only been reported in one study ²⁹ . Further research into the functional role of IL-32 in the activity of dendritic cells in HS would be of value. The role of IL-20, IL-22, IL-24 and IL-26 needs further clarification. IL-19, TSLP and CCL17 (TARC) have not yet been examined in HS and this is required in order to further explore the role of dendritic cell, monocyte and T cell activation and migration in this disease.

It is well established that smoking, obesity and diabetes are strongly associated with HS ^{13– 19, 42, 43} . The immunological effects of smoking include increase in number and responsiveness of dendritic cells, altered function of Treg cells and activation of Th17 pathways ⁴⁴ , whilst obesity and diabetes can result in production of IL-1β, IL-6 and TNF-α through activated macrophages in adipose tissue ^{45, 46} . These potential mechanistic pathways (which may prime or contribute towards inflammation in HS) require validation in functional studies. However, if they are a significant contributor to inflammation, the presence or absence of these comorbidities need to be considered in future cytokine studies as confounding variables in order to identify significant biochemical markers independent of these other pro-inflammatory states that reflect the pathogenesis of HS.

The role of the microbiome ^{42, 43} in stimulating chronic inflammation has parallels in diabetes ⁴⁷ and colonic inflammation ⁴⁸ and the presence of Porphyromonas and Peptoniphilus species has been associated with a subpopulation of patients with HS ⁴² . Porphyromonas has been associated with systemic inflammation and atherosclerosis through aberrant toll-like-receptor 4 signalling ⁴⁸ and is not part of the natural cutaneous flora ⁴³ . Altered cutaneous and gastrointestinal microbiome can also act via microbiome metabolites (including lipopolysaccharides, short chain fatty acids and bile salts) ⁴⁹ through stimulation of myeloid dendritic cells via G Protein Coupled Receptors (including GPR41, GPR43 and GPR109A) ^{49, 50} . The microbiome may be implicated as a trigger factor for the initial inflammatory cascade in HS in a proportion of patients. Similarly, the presence of genetic polymorphisms as reported in HS ⁵¹ have the potential to up-regulate inflammatory activity through shedding of IL-6R, IL-15R, TNF-α ⁵² as well as up-regulating the response of dendritic cells to LPS stimulation via ADAM17 (which has been demonstrated to be elevated in a published gene expression study of HS) ⁵³ . These pathways may be involved prior to the activation of keratinocyte-mediated inflammation, and hence, may reveal novel targets for new interventions to control the disease prior to the onset of destructive inflammation.

The limitations to this study include the high degree of methodological variability ( Table 5) and high impact of bias ( Table 6) within the included studies. The lack of individual patient data has also prevented any further analysis into the contribution of comorbidities such as smoking and obesity to variable levels of cytokines in lesional tissue and/or serum. This, along with the high level of heterogeneity in many cytokines ( Table 7), has resulted in analyses of the collated data being limited to descriptive analyses only and limited the generalisability of results.

OSF: Extend data. Data Collection Sheet Cytokine. Review HS. https://doi.org/10.17605/OSF.IO/N2E7A ²²

License: CC0 1.0 Universal

OSF: PRISMA checklist for ‘A systematic review and critical evaluation of inflammatory cytokine associations in hidradenitis suppurativa’. https://doi.org/10.17605/OSF.IO/N2E7A ²²

License: CC0 1.0 Universal