A systematic literature search pertaining to hidradenitis suppurativa and genetic variation was conducted in PubMed/MEDLINE, Science Direct and Google Scholar databases. A comprehensive search strategy was based on the following combinations of free text keywords and Medical Subject Heading (MeSH) terms: “hidradenitis”, “suppurativa”, “acne”, “inversa”, “genetic”, “familial”, “secretase”, “nicastrin”, “NCSTN”, “PSENEN”, “presenilin”, “PSEN”, “APH”, “NOD2”, “PSTPIP1”, “MEFV”, “Syndrome”, “PASH”, “PAPASH”, “Pyoderma”, “SAPHO”, “PASS”, “Dowling-Degos”, “Mediterranean”. The Boolean operators used were “AND” and “OR”. The search covered articles published between January 1980 and January 2022, and was restricted to articles published in the English, Italian and Spanish languages. Furthermore, handsearching and citation review of relevant studies was also conducted to identify studies that were not captured by the electronic database search. Published original studies investigating the genetics of HS in its isolated or syndromic forms through both targeted and untargeted genomic approaches were eligible for inclusion. We excluded studies with 1) absent genetic data 2) duplicate data pertaining to the same proband in separate publications 3) conference proceedings, reviews, editorial letters or comments and 4) studies not directly investigating the association of specific genes with HS phenotypes.

Articles identified from the literature search were screened for duplicates. All studies deemed to be potentially eligible for inclusion were reviewed and data extracted by the investigators. Any discrepancies were resolved by consensus.

The following information was extracted from each eligible article 1) primary author and year of publication 2) locus, the specific genetic variant identified and segregation data when available 3) age at presentation of first lesion 4) primary anatomic sites affected and 5) data on comorbid risk factors, specifically obesity and smoking status.

Identified variants were categorized as missense, nonsense, frameshift indels, splice-site and non-coding regions according to their effect on translation. VarSeq software (Golden Helix) was used for variant interpretation and annotation (Kopanos et al., 2019). Variant pathogenicity was classified according to guidelines from the American College of Medical Genetics/Association for Molecular Pathology (ACMG/AMP) (Richards et al., 2015). These guidelines standardize variant classification by stratification into five categories (pathogenic, likely pathogenic, uncertain significance, likely benign, benign) based on a combination of computational, population, functional and segregation data. Rare variants were considered to be of uncertain significance (VUS) if there is limited or contradictory clinical or functional evidence for pathogenicity. The Human Gene Mutation Database, (HGMD), dbSNP and ClinVar databases (ncbi.nlm.nih.gov/clinvar/) were also interrogated to identify any respective studies and ontologies. To explore genotype-phenotype associations, variants in the GSC genes were considered as an aggregate category and contrasted against variants in genes not forming part of this complex.

The impact of missense variants on protein structure was evaluated through molecular modelling. The DynaMut webserver was used to predict the effects of amino acid substitutions on protein stability, flexibility and to analyze interactions amongst amino acid residues (Rodrigues et al., 2018). The predicted change in stability between the wild type and variant structures derived from SDM, DUET, and mCSM algorithms is reported as ΔΔG in kcal/mol, with negative values indicative of destabilizing variants. Dynamut also reports vibrational entropy changes (ΔΔS_Vib in kcal mol⁻¹K⁻¹) calculated by ENCoM using normal-mode analysis to depict the substitution’s effect on structure flexibility or rigidity. To further categorize the structural impact of missense variants, the Missense3D and HOPE webservers were used. Missense3D assesses 17 different structural features that are essential for protein conformation and stability, such as stearic clashes and disallowed phi/psi angles (Venselaar et al., 2010; Ittisoponpisan et al., 2019). A comprehensive graphical summary of identified GSC variants was generated using ProteinPaint (Zhou C et al., 2016).

The characteristics of the cohort are summarized using descriptive statistics. Normality of age at first lesion was assessed by the Shapiro–Wilk and Kolmogorov-Smirnoff tests. This exhibited a skewed non-normal distribution; non-parametric statistics with medians and interquartile ranges are presented. To compare differences in quantitative variables between two categories, the independent-samples Mann–Whitney U test was applied. The chi-square test was used to compare categorical variables. A p-value of <0.05 was considered statistically significant.

1. Sporadic HS: HS with no identified genetic variation to date

Genetic variants in the GSC account for the majority of identified variants in fHS and sHS cases, but for only a minority of cases of sporadic HS. Cohort studies showed a low prevalence of GSC variants in sporadic forms of HS. The largest multicenter study in cases of predominantly Caucasian ethnicity identified pathogenic GSC variation in only 12 out of 188 (6.4%) patients with HS, of which 51% had fHS (Duchatelet et al., 2020). Smaller scale cohort studies in the United Kingdom revealed no GSC pathogenic variants in HS patients irrespective of family history of the disease (Ingram et al., 2013).

The human GSC is a multimeric, intramembrane-cleaving proteases composed of four subunit domains namely: Nicastrin (NCSTN), Presenilin Enhancer 2 (PEN2), Presenilin 1 (PSEN1) or PSEN2 and Anterior Pharynx Defective (APH) 1A or B. This gives rise to at least six different GSCs, which assemble fully in the endoplasmic reticulum and are transported to the cell membrane (Capell et al., 2005). The GSCs are dynamic and can exist in three conformational states; extended, intermediate and compact (Elad et al., 2015). Cryo-electron-microscopy single-particle analysis has elucidated the three-dimensional structure of human GCS, which was shown to have a horseshoe-shaped transmembrane domain (spanning 19 transmembrane segments) and a large extracellular domain (Lu et al., 2014).

The most widely reported variants in fHS lie in the nicastrin-coding gene NCSTN (OMIM #142690). Nicastrin is the largest subunit of the GSC (accounting for two-thirds of the molecular mass of the entire complex), and is postulated to be the substrate-recruiting protein of the GSC [specifically at the DYIGS and peptidase-like (DAP) region in the ectodomain (ECD)] (Shah et al., 2005). Nicastrin is essential for the assembly (particularly of the transmembrane domain) (Shah et al., 2005), maturation and stability of the GSC (Elad et al., 2015). The protein has a bilobed ECD and a single transmembrane domain (TMD) at its C-terminus (Xie et al., 2014). The interface between the two lobes is maintained by extensive van der Waals contacts, amongst which those formed between Phe287 (of the large loop) nestled within a pocket of hydrophobic amino acids Phe103, Leu171, Phe176 and Ile180 (of the small loop), is highly conserved (Xie et al., 2014). A loop extending from the core of the small lobe of the nicastrin ECD forms a lid which covers the putative substrate binding site within the large lobe, amongst which the residue Trp164 is vital (Xie et al., 2014). This amino acid, as well as Pro-141, Trp-164, Asn-165 and Gly-168 have been shown to be highly conserved in the NCST “lid” domain, but not essential for regulating GSC activity, including Notch (Zhang X et al., 2016). Fluorescence imaging microscopy of intact cells detected conformational change of the nicastrin ECD, which is brought closer to the membrane core upon binding an inhibitor (Elad et al., 2015). Residues Tyr337 and Glu333 (Shah et al., 2005) are both located within the substrate biding pocket ECD (Xie et al., 2014). The importance of Glu333 in proteolysis has been confirmed by mutagenesis studies, demonstrating abolished GSC cleavage, activity and maturation after substituting the residue (Shah et al., 2005; Dries et al., 2009), thus dispelling assertions that Glu333 is only involved in GSC maturation (Chávez-Gutiérrez et al., 2008). Nicastrin also forms complexes with PSEN1 and PSEN2, establishing a “secretasome” which allows for intramembranous proteolysis of the transmembrane proteins, including Notch (Yu et al., 2000).

The Notch signaling pathway is a highly conserved pathway involved in cell-cell communication. It regulates cellular differentiation and proliferation in continually renewing adult tissues such as skin. In these tissues, the notch receptor is activated by various ligands and cleaved by the GSC, releasing its intracellular domain which translocates to the nucleus to regulate gene expression (Mumm and Kopan, 2000). Because of the important roles played by Notch in epidermal and follicular homeostasis as well as inflammation, Notch dysregulation has been touted to underpin the molecular basis of HS in patients with pathogenic variants in GSC-protein coding genes (Okuyama et al., 2008; Melnik and Plewig, 2013). Interaction of the Notch receptor with its ligand (delta or jagged) results in two subsequent proteolytic cleavages of the receptor, the first of which is catalyzed by ADAM-family metalloproteinases and the second by the GSC. The product of the second cleavage, Notch intracellular domain (NICD), subsequently translocates to the nucleus where it acts as a transcriptional regulator for various genes (Borggrefe and Oswald, 2009) after complexing with CBF1-Suppressor of Hairless-LAG1 (CSL) and the co-activator mastermind (Wilson and Kovall, 2006).

In human models, downregulation of Notch signaling pathway has been shown to perturb keratinocyte differentiation and result in uncontrolled proliferation, disorganization of the suprabasal layers of the epidermis as well as dermal invasion (Thélu et al., 2002). These findings were also replicated in murine models wherein alteration in notch signaling resulted in altered sebaceous gland differentiation and terminal differentiation of the epidermis (Pan et al., 2004; Blanpain et al., 2006; Wang et al., 2008). Loss of ADAMS10 (responsible for notch receptor cleavage) (Groot and Vooijs, 2012) in murine models has been shown to result in impaired epidermal differentiation resulting in various pathologies including epidermal hyperproliferation and cyst formation as a result of altered notch signaling (Weber et al., 2011). Studies have also demonstrated that ADAMS10 is downregulated in HS (Frew and Navrazhina, 2020). Dysregulation of ADAM10-Notch signaling axis has been shown to impair the epithelial barrier and favor cutaneous dysbiosis (favoring Corynebacterium spp) (Sakamoto et al., 2021). Potentially, the altered microbiome promotes chronic inflammation by triggering the innate lymphoid cell population in an IL-17R dependent manner (Sakamoto et al., 2021). These findings may partly underpin the dysregulated cutaneous microbiome that accompanies HS (Mintoff et al., 2021a). Murine models have also demonstrated that disruption of notch nuclear target RBP-J results in cyst formation and epidermal hyperkeratinization (Yamamoto et al., 2003). A study scrutinizing publicly available genomic data revealed significant downregulation of Notch 1–4, and suggests ADAM17 as a key mediator in the pathogenesis of HS (Frew and Navrazhina, 2020).

Evidence for the role of Notch in HS can also be drawn from HS + disease such as HS-DDD. DDD is an autosomal dominant genodermatosis characterized by flexural and reticulated pigmentation. The condition is attributed to heterozygous variants in KRT5 (DDD1, OMIM #179850) at 12q13 and POFUT1 (DDD2, OMIM #615327) at 20q11. (Crovato et al., 1983; Stephan et al., 2021). Dubbed as “clinical collision” diseases, HS-DDD provides avenues for understanding pathophysiology and phenotypes (McGrath, 2018). Both POGLUT1 and POFUT1, causative genes in DDD, are established regulators of Notch pathway activity through their respective protein products namely protein-O glucosyltransferase 1 and protein O-fucosyltransferase 1 (Li et al., 2013; Basmanav et al., 2014). Congruently, patients with HS-DDD having underling POGLUT1 pathogenic variants demonstrate abnormal expression of genes encoding the critical element of the Notch pathway (Pavlovsky et al., 2018). Further molecular evidence is derived from two patients with HS-DDD, where reduced expression of NOTCH1 and NCSTN mRNA was demonstrated in lesional skin when compared to non-lesional skin, (Penha et al., 2020). HS-DDD with underlying POFUT1 pathogenic variants has been suggested to share similar Notch disturbance; however robust serological and in-vitro studies are lacking (González-Villanueva et al., 2018). Indirect evidence supporting the role of Notch downregulation in HS is provided by a case report in which a patient treated with a notch signaling inhibitor AL101 (BMS-906024) developed HS (Wiggins and Chon, 2020).

Conflicting findings have also been reported. Nicastrin, Notch 1–3, PIK3R3 and AKT3 levels were found to be significantly higher in lesional skin of 60 HS patients when compared to healthy controls. In lesional skin, these proteins were significantly higher in patients with mild (Hurley stage 1) disease compared to those with moderate and severe disease (Hurley stage II and III), despite excluding smoking and obesity as confounding factors (Hessam et al., 2021). Nicastrin overexpression (particularly dermal) has been associated with hypertrophic scarring as well as with inhibition of Notch signaling resulting in the suppressed production of fibrotic factors such as collagen 1 and 3 and TGF-β1 (Chen et al., 2021). Functional studies have shown that various HS-inducing nicastrin missense variants are active and sustain Notch signaling, and therefore do not fully support the concept of notch as being the singular pathophysiological processes involved in NCSTN-associated HS (Zhang and Sisodia, 2015). Indirect evidence for upregulation of notch can be drawn from other inflammatory pathologies. In patients with psoriasis, the proinflammatory cytokine serum amyloid A (SAA), known to be highly elevated in sera of HS patients (Witte-Händel et al., 2019), has been shown to upregulate Notch1 activity (Rooney et al., 2014).

PSEN has also been implicated in Notch signaling. The homologs PSEN1/PSEN2 are the catalytic subunit of the GSC. PSEN1/GSC is widely distributed in the cell (including its plasma membrane) whilst PSEN2/GSC is mostly restricted to late endosomes and lysosomes (Sannerud et al., 2016). Auto-compensatory mechanisms maintain PSEN levels in equilibrium (Stanga et al., 2018). In animal models, only zebrafish affected by the familial acne inversa-like indel mutation psen1. ptrp233fs (equivalent to human PSEN1 codon P242) had shown a significant alteration in Notch signaling (as opposed to early onset familial Alzheimer disease mutants). Upregulation of genes involved in inflammation was also observed in these mutants (Barthelson et al., 2021). The apparent upregulation of Notch signaling may possibly be accounted for by transcriptional adaptation, the process by which fragments of mutated mRNA translocate to the nucleus leading to transcriptional modulation of “adapting genes” (Sztal and Stainier, 2020). HS patients with underlying PSEN variants are designated OMIM # 613737.

Variants of Presenilin enhancer 2 (PSENEN), which codes for presenilin enhancer 2 (PEN2) have also been described in both sporadic HS as well as HS + DDD. The role of PEN2 in GSC function and Notch signaling was found to be conserved across species (Francis et al., 2002). PEN2 is the last unit to be incorporated within the GSC, with a PEN2 “retention mechanism” ensuring that only fully assembled GSCs are released from the endoplasmic reticulum to the plasma membrane (Capell et al., 2005; Christoph et al., 2007). The incomplete penetrance of PSENEN pathogenic variants has been described in three families by Riis et al (Theut Riis et al., 2020). Potentially, this can be attributed to the fact that PEN2 is not part of the GSC proteolytic domain. Additionally, it is possible that monoallelic pathogenic variants may not cause disease in the absence of other risk factors. In silico studies further support the concept that heterozygous pathogenic variants in NCSTN and PSENEN are not sufficient to cause disease (Nomura et al., 2014; Theut Riis et al., 2020).

APH-1 is the least well-characterized locus in the setting of HS, with a single APH-1B variant (p.His170Arg) having been associated with HS (Theut Riis et al., 2020). The authors postulate that this variant is unlikely to be causative of HS, citing studies which demonstrate much higher APH-1A expression in skin and fibroblast models in which APH-1A exclusively is involved in Notch signaling (Theut Riis et al., 2020). APH1 is a 7-transmembrane helix protein expressed as two homologous isoforms in humans, encoded by two genes (APH1a on chromosome 1; APH1b on chromosome 15). Both APH-1A and APH-1B adopt a water channel topology and transport water across the plasma membrane (Aguayo-Ortiz and Dominguez, 2019). The conserved His170 residue together with His196 play a major role in water transportation across the lipid bilayer (Aguayo-Ortiz and Dominguez, 2019; Dehury and Kepp, 2021).

The significance of GSC variants and the disruption in Notch signaling pathways and its targets in HS, is still evolving. Further functional and multiomic studies are required to determine the definitive role of the Notch signaling pathway in skin disease, including HS (Brandão et al., 2021). A working model of HS pathophysiology suggests that underlying genetic variants lead to lower protein expression which becomes functionally relevant under cellular stress mediated by friction (obesity), temperature and dysbiosis (Pink et al., 2016).

In a subset of patients, HS develops as part of a constellation of other inflammatory, conditions (sHS). The classical HS syndromes in which genetic variation has been described include PASH (pyoderma gangrenosum, acne and suppurative hidradenitis), PAPASH (pyogenic arthritis + PASH) and SAPHO (synovitis, acne, pustulosis, hyperostosis and osteitis) syndromes (Garcovich et al., 2021a). A PSTPIP1 pathogenic variant was recently identified in a patient with proctitis, pyoderma gangrenosum, HS and fever (dubbed “PPHSF” syndrome) (Hieta et al., 2021). Supplementary Table S2 summarizes the genetic variants associated with sHS.

Braun-Falco et al., had documented the first two families with PASH. No pathogenic variants in PSTPIP1 were detected; however afflicted patients had hemi-allelic increase of the CCTG microsatellite motif (Braun-Falco et al., 2012). Other cases of PASH without PSTPIP1 pathogenic variants have also been reported (Gracia-Cazaña et al., 2015; Niv et al., 2017; Lamiaux et al., 2018). Similarly, no pathogenic variants were described in other forms of sHS namely, PsAPASH (Psoriatic arthritis and PASH) syndrome (Saraceno et al., 2015), PsAPSASH (Psoriasis, arthritis, pyoderma gangrenosum, synovitis, acne, suppurative hidradenitis) (Nikolakis et al., 2021) and PASS (pyoderma gangrenosum, acne, hidradenitis suppurativa and ankylosing spondylitis) (Bruzzese, 2012; Leuenberger et al., 2016) syndromes . Follicular occlusion syndromes such as the follicular occlusion triad [HS, acne conglobata and dissecting cellulitis of the scalp (Perifolliculitis capitis abscedens et suffodiens)] (Chicarilli, 1987) and the follicular tetrad (Vasanth and Chandrashekar, 2014) (follicular triad and pilonidal sinus) are described but likewise, they also lack an identified genetic driver.

PSTPIP1 is the locus most frequently implicated in sHS that features in PASH, PAPASH and PPHSF syndromes. Interestingly, PSTPIP1 variants have not been described in isolated forms of HS. PSTPIP1 codes for Proline-serine-threonine phosphatase-interacting protein 1. It has been demonstrated that PSTPIP1 regulates the transition of macrophage’s podosomes to filopodia-like protrusions and modulates their invasive migration (Starnes et al., 2014). Cellular studies have also demonstrated that pyrin modulates the intracellular distribution of PSTPIP1, and co-localizes at the leading edge of cells mitigating cell migration (Shoham et al., 2003; Waite et al., 2009; Akkaya-Ulum et al., 2015). In fact, PTSPIP1 mutant T-cells were shown to have altered f-actin polymerization (Janssen et al., 2018).

MEFV variants have been shown to be more frequent in patients with HS than healthy controls (Vural et al., 2019). A patient of Turkish origin suffering from comorbid PASH and familial Mediterranean fever (FMF) has been described. Targeted analysis in the MEFV gene identified the two heterozygous pathogenic variants (p.M680I and p. V726A); assessment of the rest of his genome was therefore lacking (Vural et al., 2017). Interestingly, a patient with PAPASH was heterozygous for a microsatellite elongation in the PSTPIP1 promotor region, and homozygous for the pathogenic MEFV p. Met694Val variant. However, he was asymptomatic for FMF. The authors suggest that this could possibly be the result of aberrant mutant Pyrin-PTSPIP1 interaction (Vural et al., 2019). The cause-effect relationship underlying the coexistence of a systemic proinflammatory susceptibility typical of FMF with inflammation of the PSU remains to be established.

Cytokine IL-1β may be the common denominator linking variants described in sHS or HS in combination with pyoderma gangrenosum (Galimberti et al., 2016; Witte-Händel et al., 2019). The protein products encoded by PSTPIP1 and MEFV interact in a multimeric protein complex that regulates assembly and activation of the inflammasome by promoting ASC oligomerization and Caspase-1 activation (Yu et al., 2007). The inflammasome regulates innate immunity and epithelial barrier defenses. On the other hand, gain of function mutations in NLRP3 result in increased IL-1 [a defining feature of Cryopyrin-associated periodic syndrome (CAPS)] (Kuemmerle-Deschner et al., 2017). Conflicting results have been observed with regards to IL-1β levels in patients with sHS. No statistically significant differences in serum Il-1β levels were described between patients with PASH syndrome and healthy controls (Marzano et al., 2014). Conversely, PASS has been characterized as an IL-1-driven autoinflammatory disease which responds to treatment with the Il-1 Receptor antagonist anakinra (Leuenberger et al., 2016). Anakinra has been proven to be ineffective in the treatment of a young female with PASH syndrome (Staub et al., 2015).

HS has also been documented to manifest in the setting of specific chromosomal disorders, where it does not constitute a classical diagnostic feature. It has been speculated that diminished Notch receptor processing and signaling could account for HS is the setting of Trisomy 21; however a definitive functional correlation remains elusive (Gasparic et al., 2017). A case report also describes the occurrence of unilaterally distributed HS, possibly due to constitutional mosaicism in a patient with trisomy 1q (Skroza et al., 2019).

HS has been independently described in the setting of two other heritable conditions namely, HS-DDD and Familial Mediterranean Fever (HS-FMF).

A distinctive subtype of Dowling-Degos with HS (HS-DDD) is defined by heterozygous variants in PSENEN (OMIM # 613736) on 19q13. Patients having underlying PSENEN variants but suffering from DDD exclusively have been described (Ralser et al., 2017; Ren and Zeng, 2020). Interestingly, only obese family members harboring the pathogenic PSENEN c.62-1G > C splice variant manifested HS + DDD, whilst their lean, non-smoking relative who also harbored the same mutation manifested DDD exclusively (Ralser et al., 2017). Non-smoking, lean patients from another family having PSENEN 84_85insT variant manifested DDD without HS (Ralser et al., 2017). On the other hand, the c.216delC PSENEN variant was described in non-smoking, lean patients from two separate families manifesting DDD exclusively but also in an unrelated patient with DDD-HS, whose smoking history and weight were not documented (Ralser et al., 2017; Ren and Zeng, 2020; Theut Riis et al., 2020). This suggests that in the context of HS, PSENEN pathogenic variants exhibit incomplete penetrance and variable expressivity, and possibly only manifest disease in the setting of specific triggers such as obesity. A pathogenic NCSTN nonsense variant p. Arg583Ter (c.1747C > T) has also been described in a lean, non-smoking patient with HS-DDD. The variant segregated with the DDD phenotype, but not HS, in the proband’s daughter (Garcovich et al., 2020; Garcovich et al., 2021b). However, the pathogenicity of this variant and its relevance to HS-DDD has been disputed, mainly because the significance of co-existing KRT5 variants was downplayed and deemed benign (Hermasch et al., 2020).

HS can occur in combination with other inherited autoinflammatory syndromes. Two patients with co-morbid HS - mevalonate kinase deficiency (Benhadou et al., 2021), an autosomal recessive inborn error of metabolism which leads to chronic inflammation, have been described. Various studies have also described the co-existance of an HS phenotype in patients bearing pathogenic variation at the MEFV locus (Abbara et al., 2017; Vural et al., 2017; Jfri et al., 2019; Vural et al., 2019). This gene encodes pyrin, a protein which modulates the activity of the GSC. These studies suggest that HS and FMF are autoinflammatory disorders that may possibly share converging pathophysiologic processes. Despite the systemic proinflammatory state in these conditions, proof of causal associations remains lacking.

Patients with HS exhibit extensive phenotypic heterogeneity making genotype-phenotype correlations difficult to establish (Kent, 2009). The task is further complicated by poor interrater reliability of HS phenotypes (van Straalen et al., 2018), differences in severity scoring (Zouboulis et al., 2019) as well as evidence of pleiotropism (Gratton et al., 2020). Notwithstanding, some general patterns are emerging, and encouraging results are emerging even from relatively low-powered studies (Marzano et al., 2022). HS patients having underlying NCSTN variants appear to have a follicular-type HS, in which comedones, papules and folliculitis predominate. Patients with this subtype of HS are likelier to manifest lesions in the nape and back and have co-morbid pilonidal sinus disease (PND) (Xu et al., 2016), a common co-morbidity and a possible intergluteal localization of disease in HS patients (Benhadou et al., 2019). Similarly, phenotyping of a large cohort of HS patients showed that cases exhibiting a follicular (LC2) type phenotype, typified by epidermal cysts, PND and comedones were more likely to have a family history of HS when compared to the patients with axillary-mammary (LC1) and gluteal (LC3) phenotypes (Canoui-Poitrine et al., 2013). An inverse correlation between LC1 HS phenotypes and NCSTN variants has been described (Frew et al., 2019). Patients with an endotype defined by GSC pathogenic variants and higher levels of serum IL-10 are more likely to be non-obese males with predominantly nodular lesions on the trunk and posterior sites and have a history of PND (referred to as “Cluster 1” HS patients). On the other hand, patients with high serum IL-1, IL-17, IL-16 and CRP are more likely to be obese and have later-onset disease, with tunnels and abscesses predominating (Cluster 2) (González-Manso et al., 2020). With regards to patients with sHS, both promoter and PSTPIP1 variants were found to be significantly associated with syndromic forms of disease (Frew et al., 2019).

In a broader sense, HS patients having an affected first-degree relative develop more severe disease at a younger age than their parents. The gender of the affected parents also influences the resulting phenotype in the offspring, with patients having an affected mother more frequently reporting axillary involvement, and patients with affected fathers being significantly more likely to involve the buttocks and the genitals. The mean number of affected body sites was found to be significantly higher in patients whose father has HS then those with a maternal history of the condition (Plenary Lectures, 2019).

Patients who are obese and smokers and without any underlying genetic variants are more likely to present with a “wet” phenotype characterized by draining abscesses in intertriginous regions. Conversely, lean non-smokers with a known family history are more likely to develop a “dry” phenotype characterized by follicular lesions in atypical regions such as the nuchal area and antecubital fossae (Vossen et al., 2020a; Mintoff et al., 2021b), or with syndromic HS.

The interpretation of genetic findings in HS presents several limitations. When compared to other common complex traits, there is a deficiency of large-scale genomic studies on ethnically diverse cohorts. The relative contribution of common vs. rare polymorphisms remains unascertained. Applying high-throughput sequencing studies to kindreds with early-onset familial or atypical HS phenotypes is a valid approach. However, variant prioritization and pathogenicity scoring can be complicated by pitfalls such as the overreliance on in-silico predictors and the use of inappropriate allele frequency cut-offs. Additionally, limited conclusions about the role of monoallelic variants causing recessive disorders can be drawn. It must be acknowledged that the assessment of allele frequencies in aggregate datasets unselected for disease, such as GnomAD, is a valuable approach. However, the presence of rare variants causative of late-onset disease in genomic databases can confound variant classification (Lek et al., 2016). Attributing causality to variants remains a considerable challenge, particularly for missense substitutions that are not structurally deleterious. Functional evaluation using in-vitro or in-vivo models are required to support pathogenicity and robustly define gene-disease associations for disputed loci. This is reinforced by conflicting interpretations of pathogenicity attributed to some variants in clinical databases. Furthermore, studies based on exome capture and sequencing may fail to identify deep intronic variants that modulate splicing or pathogenic structural variation. In the broader context of HS genomic architecture, it is essential to consider that studies sequencing cases with multigenerational early-onset disease are likely to skew towards the identification of high-penetrance variants. These represent the ‘low-hanging fruit’ of genomic discovery, at the expense of variants that lack adequate penetrance to drive familial segregation of disease. Plausibly, such intermediate penetrance variants predispose to later onset or milder disease.

This review is intrinsically limited by study selection criteria, and it is possible that some studies may not have been included (language other than English, Italian and Spanish, articles missing key data and conference proceedings). In addition, key phenotypic data, such as obesity and smoking status, as well as familial segregation of identified variants was not reported by some investigators.