Identification of Biomarkers and Critical Evaluation of Biomarker Validation in Hidradenitis Suppurativa

Abstract

IMPORTANCE

The identification and validation of biomarkers in hidradenitis suppurativa (HS) has potential to improve the understanding and management of this chronic, burdensome disease.

OBJECTIVE

To systematically identify all known HS biomarkers, categorize them by biomarker type, and critically evaluate their validity according to established criteria.

EVIDENCE REVIEW

Eligibility criteria for this review (PROSPERO Registration 230830) included randomized clinical trials, uncontrolled clinical trials, cohort studies, case-control studies, and other observational studies with no restrictions of patient age, sex, race or ethnicity, or language of publication up until December 31, 2020. All articles were categorized into biomarker type, defined using the US Food and Drug Administration Biomarkers, Endpoints, and other Tools (BEST) glossary. Assessment of each identified biomarker was undertaken in line with the US Food and Drug Administration and European Medicines Agency guidelines for the validation of proposed biomarkers. Assessment of the strength of overall data regarding individual biomarkers was undertaken using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) approach.

FINDINGS

A total of 3953 nonduplicate articles were screened, of which 1429 articles were retrieved based on the include/exclusion criteria applied. After full-text screen and data extraction, 106 articles were included in this review. The evidence of strength of 6 categories of biomarkers (susceptibility/risk, diagnostic, monitoring, predictive, prognostic, and pharmacodynamic/response biomarkers) was assessed using GRADE criteria. A total of 48 biomarkers were identified with a minimum GRADE rating of moderate.

Only 1 diagnostic (serum IL-2R), 1 monitoring (dermal Doppler vascularity), and 2 predictive biomarkers (epithelialized tunnels and positive family history of HS) achieved a GRADE rating of high. None of the identified biomarkers had sufficient clinical validity to be recommended for routine use in the clinical setting.

CONCLUSIONS AND RELEVANCE

Major barriers to the identification, validation, and introduction of routine biomarkers in the management of HS include lack of independent biomarker validation studies (especially assumption-free “omics”-based techniques); insufficient assessment of collinearity between identified or proposed biomarkers; and a lack of routine integration of biomarkers into the structure of clinical trials. International consensus among researchers, clinicians, and pharmaceutical stakeholders is required to standardize goals and methods and encourage biomarker integration into future HS clinical trials. This systematic review presents a number of priorities for near-term future research to overcome such barriers and limitations of biomarkers in HS.

Hidradenitis suppurativa (HS), also known as acne inversa, is a chronic autoinflammatory skin disease manifesting in painful nodules and abscesses, comedones, and draining malodorous tunnels.^1,2 These lesions have a predilection to flexural areas of skin, namely the axillae, groin, and submammary regions, but can occur at any site.^1,2 The disease is thought to occur owing to a combination of genetic and environmental factors (including bacterial dysbiosis) and, in a minority of individuals, has strong associations with inherited sequence variants in the gamma secretase complex.^1-3 Hidradenitis suppurativa is associated with multiple inflammatory comorbidities, including obesity, metabolic syndrome, diabetes mellitus, and inflammatory bowel disease, among others.^1-3 It is a heterogeneous disease with multiple phenotypes, clinical presentations, and disease trajectories. It has been proposed that the development of biomarkers may aid in the diagnosis, understanding, and management of this disease.^4,5

A biomarker is a “defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention.”^6(p4) There are several different types of biomarkers defined by the US Food and Drug Administration (FDA) Biomarkers, Endpoints, and other Tools (BEST) glossary (eMethods in the Supplement). The development, validation, and clinical use of biomarkers in HS would reap multiple benefits in the clinical assessment, diagnosis, and management of disease,^6,7 as recently illustrated by the identification of various inflammatory endotypes in the context of atopic dermatitis and their subsequent therapeutic relevance.⁸ Susceptibility/risk and diagnostic biomarkers may aid in reversing the well-documented diagnostic delay in HS, while also identifying high-risk individuals who would benefit from closer monitoring for active disease. Diagnostic biomarkers may also complement the modified Dessau criteria as an objective measure to differentiate HS from other commonly misdiagnosed conditions, such as recurrent folliculitis, acne, or infection. Additionally, monitoring biomarkers (molecular biomarkers of disease activity) would allow for a deeper understanding of the associations between observed molecular findings in HS tissue and clinical disease activity over time. This may help define disease heterogeneity, refine drug repurposing, and identify novel therapeutic targets. Importantly, predictive biomarkers would allow for personalized medicine approaches with targeted selection of therapy to patients with greater reliability of response, rather than the relatively common trial-and-error approach currently used in HS.⁴

The aim of this review was to systematically identify all known HS biomarkers, categorize them by biomarker type, and critically evaluate their validity according to established criteria.^6-8 Additionally, the strength of evidence supporting each biomarker was assessed using Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) certainty criteria.⁹

Methods

This study was registered with PROSPERO (CRD42021230830) and conducted in line with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline. Eligibility criteria for this review included randomized clinical trials, uncontrolled clinical trials, cohort studies, case-control studies, and other observational studies with no restrictions of patient age, sex, race or ethnicity, or language of publication up until December 31, 2020. Data collection was performed independently by 2 authors (S.D.S. and J.W.F.), with any disagreements regarding inclusion of citations being referred to a third author for mediation. All articles were categorized into biomarker type (eMethods in the Supplement). Biomarkers were defined using the FDA BEST glossary (https://www.ncbi.nlm.nih.gov/books/NBK338448/). Assessment of each identified biomarker was undertaken in line with the FDA and European Medicines Agency guidelines for the validation of proposed biomarkers. Assessment of the strength of overall data regarding individual biomarkers was undertaken using the GRADE approach. Detailed methods are presented in eMethods and eTable 1 in the Supplement.

Results

A total of 1429 articles were identified using the documented search strategy (eFigure in the Supplement). A total of 11 susceptibility/risk biomarkers, 85 diagnostic biomarkers, 39 monitoring biomarkers, and 20 predictive biomarkers were identified. Visual presentation of identified biomarkers with GRADE rating of moderate or high and their validity status is presented in the Figure. Biomarkers that met “moderate” or “high” GRADE criteria are listed in Tables 1, 2, 3, and 4; a complete list of all identified biomarkers with gradings (including those with GRADE ratings of very low or low) is presented in eTables 2-5 in the Supplement.

Figure. Biomarkers With GRADE Rating of Moderate or High and Degree of Biomarker Validation.

Multiple susceptibility/risk, diagnostic, and monitoring biomarkers were identified in this review, with lesser numbers of markers being independently validated in external cohorts. Predictive markers examining response to therapy were the only biomarkers that had undergone clinical validation in the setting of a clinical trial, and no biomarkers had assessment of clinical utility to recommend them for routine clinical use. The vast majority of these identified biomarkers met “moderate” GRADE criteria, and the only biomarkers that reached “high” GRADE criteria were serum IL-2R (diagnostic), dermal Doppler vascularity (monitoring), and epithelialized tunnels and positive family history of HS (predictive). Items were assessed based on criteria in line with the FDA biomarker definitions, GRADE criteria, and FDA/European Medicines Agency guidelines for the validation of proposed biomarkers as reported in eMethods in the Supplement. FDA indicates US Food and Drug Administration; GRADE, Grading of Recommendations, Assessment, Development, and Evaluations; GSC, gamma secretase complex; SAA, serum amyloid A.

Table 1.

Susceptibility/Risk Biomarkers (GRADE Moderate/High)

Biomarker	Biomarker level			Study type	Interpretation	Critical evaluation				GRADE evidence profile	References
	HS	OR	P value			External validation	Analytical validation	Clinical validation	Clinical utility
Smoking	Increased	14.87	.001	Case-control, regression modeling	Smoking is associated with a diagnosis of HS	Yes	No	No	No	Moderate	Akdogan et al,10 2018
		1.90	.01								Garg et al,11 2018
FHX	Increased	NR	NR	Twin study/cross-sectional/heritability calculations	Narrow-sense heritability calculated at 77% (van Straalen et al,12 2020)	Yes	No	No	No	Moderate	van Straalen et al,12 2020; Schrader et al,13 2014; Molina-Leyva and Cuenca-Barrales,14 2019
Fasting serum insulin	Increased	1.09	.03	Case-control, regression modeling	Increased fasting serum insulin is associated with a diagnosis of HS	Yes	No	No	No	Moderate	Akdogan et al,10 2018; Vilanova et al,15 2018
Genetic sequence variants in gamma secretase complex, POFUT1, PTSPIP1, etc	Present	NR	NR	Linkage analysis	Association but no analysis of predictive power or potential	Yes	No	No	No	Moderate	Reviewed in Frew et al,16 2017; Jfri et al,172019

Abbreviations: GRADE, Grading of Recommendations, Assessment, Development, and Evaluations; HS, hidradenitis suppurativa; NR, not reported; OR, odds ratio.

Table 2.

Diagnostic Biomarkers (GRADE Moderate/High)

Biomarker	Statistical association		Study type	Interpretation	Critical evaluation of biomarkers				GRADE evidence profile	References
	Population of comparison	Significancea			External validation	Analytical validation	Clinical validation	Clinical utility
Serum IL-2R	HC	PPV, 0.68; P < .001	Observational case-control	Elevation associated with HS	Yes	Yes (age, race, and sex matched in Matusiak et al,182015)	No	No	High	Wieland et al,19 2013
	HC	P < .001								Matusiak et al,20 2009
	HC	PPV, 0.8; P = .001								Matusiak et al,182015
Serum hepcidin	HC	P < .01	Observational case-control	Decreased levels associated with HS	Yes	No, insufficient info on controls in Ponikowska et al,21 2020	No	No	Moderate	Ponikowska et al,21 2020
	HC	P = .03								Ghias et al,22 2019
Combination of serum markers (E-selectin + VEGF + hBD2)	HC/IBD/staph infection	PPV, 85.3%	Multivariate analysis	Combination of markers identifies individuals with HS	Yes	Yes	No	No	Moderateb	Argyropoulou et al,23 2019b
Serum YLK-40/chitinase 3	HC	PPV, 0.8; P < .001	Observational case-control	Elevation associated with HS	No	Yes, matched for age, race, and sex	No	No	Moderate	Matusiak et al,18 2015
Serum amyloid A	HC	P < .001	Observational case-control	Elevation associated with HS	Yes	No, no matching for age or sex in Witte-Handel et al,24 2019 Matched for age and sex in Akdogan et al,25 2020	No	No	Moderate	Witte-Händel et al,24 2019
		P = .008								Akdogan et al,25 2020
Serum CRP	HC	P < .001	Observational case-control	Elevation associated with HS	Yes	No, matched for age but not sex or BMI in Jiménez-Gallo et al,26 2017 No matching for age, sex or BMI in Jiménez-Gallo et al,27 2018 Matched for age and sex in Akdogan et al,25 2020	No	No	Moderate	Jiménez-Gallo et al,26 2017
		P < .05								Jiménez-Gallo et al,27 2018
		P = .01								Akdogan et al,25 2020
Serum IL-8	HC	P < .01	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Jiménez-Gallo et al,27 2018
		Upregulated								Witte-Händel et al,24 2019
Serum IL-6	HC	P < .01	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Witte-Händel et al,24 2019
		P < .001								Jiménez-Gallo et al,27 2018
Serum adiponectin	HC	P < .001	Observational case-control	Decreased levels associated with HS	Yes	No	No	No	Moderate	González-López et al,28 2020; Özkur et al,29 2020
Serum resistin	HC	OR, 1.02; P = .02		Elevation associated with HS	Yes	No	No	No	Moderate
Serum visfatin	HC	OR, 2.21; P < .001		Elevation associated with HS	Yes	No	No	No	Moderate
Serum IL-17	HC	P < .001	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Reviewed in Frew et al,4 2021
		P < .001								Matusiak et al,30 2017
Serum S100A8/A9	HC	PPV, 0.81; P < .001	Observational case-control	Elevation associated with HS	No	Yes	No	No	Moderate	Wieland et al,19 2013
Tissue and serum miRNAs	HC	miRNA-155-5p (P = .005); miRNA-223-p (P < .001); miRNA-31-5p (P = .04); miRNA-21–5p(P < .001); miRNA-146a-5p (P = .01)	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Radhakrishna et al,312019; Hessam et al,32 2017
	Nonlesional	miRNA-155-5p (P = .01); miRNA-223-5p (P = .01); miRNA-31-5p (P = .048); miRNA-21-5p (P = .01); miRNA-146a-5p (P = .04); miRNA-125b-5p (P = .04)		Decrease associated with HS
miRNA 155, 223, and 31	miRNA-155-5p	P = .01 (L vs NL)	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Radhakrishna et al,31 2019; Hessam et al,32 2017
	miRNA-223-5p	P = .01
	miRNA-31-5p	P = .048
	miRNA-21-5p	P = .01
	miRNA-146a-5p	P = .04
	miRNA-125b-5p	P = .04		Decrease associated with HS
Bacterial species	Corynebacterium (HC)	P < .001	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Ring et al,33 2017; Guet-Revillet et al,34 2017; Naik et al,35 2019; Ring et al,36 2019; Schneider et al,37 2020
	Acinetobacter/Moraxella (HC)	P = .10			Yes	No	No	No	Moderate
	Staphylococcus epidermidis	Not given			Yes	No	No	No	Moderate
	Porphyromonas/Peptoniphilus (HC)	P = .02			Yes	No	No	No	Moderate
	Propionibacterium acnes (NLT)	P < .001			Yes	No	No	No	Moderate
Tissue IL-17	HC	P < .01	Observational case-control	Elevation associated with HS	Yes	No, no matching for Witte-Händel et al,24 2019	No	No	Moderate	Kelly et al,38 2015
		P < .001								Wolk et al,39 2011
		P < .01								Hessam et al,40 2018
		P < .001								Witte-Händel et al,24 2019
Tissue IL-17A	NLT	P = .004	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Wolk et al,39 2011
		P = .05								Kelly et al,38 2015
		P < .001								Witte-Händel et al,24 2019
		P < .01								Hotz et al,41 2016
		P < .01								Navrazhina et al,42 2020
Tissue IL-17F	NLT	P < .01	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Witte-Händel et al,24 2019; Navrazhina et al,42 2020
Tissue IL-1B	HC	P < .01	Observational case-control	Elevation associated with HS	Yes	No, no matching for age or sex in Witte-Händel et al,24 2019	No	No	Moderate	Kelly et al,38 2015
	HC	P < .001								Witte-Händel et al,24 2019
	HC	P < .001								Wolk et al,39 2011
Tissue IL-1B	NLT	31-Fold increase, P = .003	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Wolk et al,39 2011
		P < .001								Kelly et al,38 2015
		P < .001								van der Zee et al,43 2011
Tissue S100A7	HC	P < .001	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Wolk et al,39 2011
		P = .002								Batycka-Baran et al,44 2021
Tissue HBD3	NLT	P < .05	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Wolk et al,39 2011
		P < .05								Emelianov et al,45 2012
		P < .01	Multivariate logistic regression							Akdogan et al,25 2020
Tissue IFN-γ	NLT	P < .001	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Wolk et al,39 2011
		P < .01								Hotz et al,412016
Tissue citrullinated H3 protein	HC	P = .03	Observational case-control	Correlation with disease severity	Yes	No	No	No	Moderate	Byrd et al,46 2019
		P < .01								Lowe et al,47 2020
Fecal calprotectin	Inactive disease	P < .001	Observational case-control	Elevation associated with HS	Yes	No	No	No	Moderate	Eşer et al,48 2020
		Descriptive data only								Lloyd-McLennan et al,49 2021

Abbreviations: BMI, body mass index; CRP, C-reactive protein; GRADE, Grading of Recommendations. Assessment, Development, and Evaluations; HC, healthy controls; HS, hidradenitis suppurativa; IBD, inflammatory bowel disease; L, lesional; miRNA, microRNA; NL, nonlesional; NLT, nonlesional tissue; OR, odds ratio; PPV, positive predictive value; staph, staphylococcus.

^aP values are reported as they were in the original sources.

^bAbstract only.

Table 3.

Monitoring Biomarkers (GRADE Moderate/High)

Biomarker	Disease severity association		Study type	Interpretation	Critical evaluation				GRADE evidence profile	Reference
	Disease severity index	Significance			External validation	Analytical validation	Clinical validation	Clinical utility
Doppler dermal vascularity	Pain NRS	R, 0.98; P value, NS	Univariate and multivariate analysis	Correlation with disease severity	Yes	Yes	No	No	High	Nazzaro et al,50 2019
	Pain VAS	R, 0.67; P < .001								Grand et al,51 2021
	Fistula size	Size ≥25.6 mm associated with presence of Doppler signal: OR, 11.51; P = .01 and mixed vasculature distribution: OR, 3.84; P = .002								Caposiena Caro et al,52 2018
	Sartorius score	Score >19 associated with mixed vasculature distribution: OR, 6.65; P < .001								Caposiena Caro et al,52 2018
Sonographic dermal tunnel diameter	HS-PGA	R, 0.75; P < .001	Univariate	Correlation with disease severity	No	Yes	No	No	Moderate	Grand et al,51 2021
BMI	Hurley stage	P < .001	Univariate and multivariate regression modeling	Correlation with disease severity	Yes	No	No	No	Moderate	Theut Riis et al,53 2018
	PGA score	P < .001
	Number of areas affected	P < .001
	Patient-reported severity	P < .001
	mHSS	R, 0.36; P = .002 for severe (mHSS ≥70) vs moderate and mild disease: OR, 1.12; P = .03 for severe and moderate (mHSS ≥40) vs mild disease: OR, 1.21; P < .001								Hessam et al,54 2015
	Hurley stage	Hurley III and II vs I: OR, 1.03; P = .01								Schrader et al,13 2014
	Hurley stage	Hurley III vs I and II: OR, 1.25; P = .03								Alatas et al,55 2020
Serum ESR	Hurley stage	Kruskal-Wallis: P = .02; R, 0.30; P = .046	Multivariate analysis and regression modeling	Correlation with disease severity	Yes	No	No	No	Moderate	Akdogan et al,25 2020
	mHSS	R, 0.6; P < .006								Jiménez-Gallo et al,27 2018
	HS-PGA	R, 0.6; P < .001								Jiménez-Gallo et al,26 2017
	Hurley stage	Kruskal-Wallis: P < .001								Jiménez-Gallo et al,26 2017
	Hurley stage	Only in male patients with HS: Hurley I vs III: P = .03; Hurley II vs III: P = .02	Observational case-control							Matusiak et al,20 2009
	Number of skin areas involved by HS lesions	Only in male patients with HS: R, 0.41; P = .04								Matusiak et al,20 2009
Male sex	Hurley stage	Hurley III and II vs I: OR, 2.11; P < .001	Multivariate analysis and regression modeling	Correlation with disease severity	Yes	No	No	No	Moderate	Schrader et al,13 2014
	Hurley stage	Hurley III vs I and II: OR, 2.56; P = .02								Alatas et al,55 2020
Serum SAA	Hurley stage	Kruskal-Wallis: P = .03; R, 0.31; P = .04	Observational case-control	Correlation with disease severity	Yes	No	No	No	Moderate	Akdogan et al,25 2020
	Sartorius score	R, 0.32; P = .03								Witte-Händel et al,24 2019
CRP	Hurley stage	Kruskal-Wallis: P = .003; R, 0.44; P = .003	Observational case-control and multivariate regression	Correlation with disease severity	Yes	No	No	No	Moderate	Akdogan et al,25 2020
	mHSS	P < .001
	Hurley stage	Kruskal-Wallis: P < .001								Hessam et al,54 2015
	Hurley stage	Kruskal-Wallis: P < .001								Jiménez-Gallo et al,26 2017
	Hurley stage	Hurley I vs III: P = .02								Matusiak et al,20 2009
	Hurley stage	Hurley I and II vs III: OR, 1.35; P = .001								Alatas et al,55 2020
	HS-PGA	R, 0.54; P < .001								Jiménez-Gallo et al,26 2017
	mHSS	R, 0.50; P < .001								Hessam et al,54 2015
Serum IL-8	mHSS	P = .02	Observational case-control	Correlation with disease severity	Yes	No	No	No	Moderate	Jiménez-Gallo et al,27 2018
	HS-PGA	R, 0.41; P < .001								Jiménez-Gallo et al,26 2017
	Hurley stage	Kruskal-Wallis: P = .02								Jiménez-Gallo et al,26 2017
Serum IL-17	Hurley stage	Hurley I vs II: P = .01; Hurley I vs III: P = .005; R, 0.35; P < .001	Observational case-control	Correlation with disease severity	Yes	No	No	No	Moderate	Matusiak et al,30 2017
	HS-PGA	R, 0.37; P = .001	Observational case-control	Correlation with disease severity						Jiménez-Gallo et al,26 2017

Abbreviations: BMI, body mass index; GRADE, Grading of Recommendations, Assessment, Development, and Evaluations; HS, hidradenitis suppurativa; mHSS, Modified Hidradenitis Suppurativa Score; NRS, numeric rating scale; NS, not stated; OR, odds ratio; PGA, Physician Global Assessment; VAS, visual analog scale.

Table 4.

Predictive Biomarkers (GRADE Moderate/High)

Biomarker	Response to therapy		Study type	Interpretation	Critical evaluation				GRADE evidence profile	Reference
	Treatment	Outcome measure (significance)			External validation	Analytical validation	Clinical validation	Clinical utility
Epithelialized tunnels	Adalimumab	HiSCR (OR, 0.47; P = .01)	Multivariate analysis and regression modeling retrospective cohort	Tunnels decrease the odds of achieving a clinical response and increase the length of time to achieve a clinical response	Yes	No	Yes	No	High	Frew et al,56 2020
		Time to HiSCR (HR, 0.70; P = .03)								Frew et al,57 2021
Family history	Adalimumab	Time to HiSCR (HR, 2.01; P < .001)	Multivariate analysis and regression modeling; retrospective cohort	Family history increased time to loss of HiSCR	Yes	No	Yes	No	High	Frew et al,57 2021
Sonographic vascularization and fibrosis	Adalimumab	Descriptive results only	Multivariate analysis and regression modeling; prospective cohort	Vascularization and fibrosis decrease with clinical response	Yes	No	No	No	Moderate	Nazzaro et al,502019
rs59532114 BCL2 gene	Adalimumab	HiSCR (P < .001)	Genome-wide association study	BCL2 gene variant (increasing BCL2 mRNA) reduced response to adalimumab	No	No	Yes	No	Moderate	Liu et al,58 2020

Abbreviations: GRADE, Grading of Recommendations, Assessment, Development, and Evaluations; HiSCR, Hidradenitis Suppurativa Clinical Response; HR, hazard ratio; mRNA, messenger RNA; OR, odds ratio.

Susceptibility/Risk Biomarkers

Serum biomarkers, including fasting serum insulin,^10,15 have been independently validated as associated with a diagnosis of HS (Table 1). Demographic variables, such as smoking status^10,11 and positive family history,^12-14 are also associated with disease (Table 1). Linkage analysis has demonstrated significant association of sequence variants in various components of the gamma secretase complex¹⁶ and other non-gamma secretase complex genes, including POFUT1¹⁷; however, there is a lower prevalence of such variants in White populations with HS compared with East Asian populations, suggesting possible unidentified risk variants.^59,60

Other serum risk biomarkers that had a low GRADE score and require future independent validation include serum RBP4,⁶¹ ghrelin,⁶¹ visfatin,¹⁰ TLR10 sequence variants,⁶² birthweight,⁶³ childhood body mass index,⁶³ and preceding conditions such as type 1 diabetes⁶⁴ (eTable 2 in the Supplement). Prospective longitudinal studies are required to assess the clinical validity of such biomarkers and avoid selection and recall bias in existing retrospective studies.⁶⁵

Diagnostic Biomarkers

Currently, the diagnosis of HS is based on clinical criteria defined by the modified Dessau definitions.^40,66,67 A plethora of proposed biomarkers have been identified in tissue and serum (Figure) that differentiate between HS and healthy controls (Table 2). However, limited assessment of the analytical validity of such markers has been undertaken.

Only 1 diagnostic biomarker (serum IL-2R) achieved a GRADE rating of high based on multiple independent validation studies^18-20 and preanalytical (with age-, sex-, and race-matched controls) and postanalytical validation (calculation of a positive predictive value). One combined panel of serum biomarkers²³ (E-selectin/vascular endothelial growth factor [VEGF]/hBD2) initially achieved a GRADE rating of high based on independent validation and analytical validity as well as the ability to differentiate HS from inflammatory bowel disease and staphylococcal skin infection.²³ However, this rating was downgraded to moderate because the publication was in abstract form and the study had not undergone peer review.²³

Diagnostic biomarkers with validation in more than 1 study include tissue IL-17,^24,38-40 IL-1B,^24,38,39 serum hepcidin,^21,22 E-selectin/VEGF/hBD2 combination,^23,67 YLK-40,¹⁸ serum amyloid A (SAA),^10,24 C-reactive protein (CRP),^10,24,26,27 S100A7,^{39,41,44,45,68} IL-8,^24,27 and IL-6.^24,27 Additional markers include fecal calprotectin,^48,49 serum visfatin,²⁵ serum IL-17,^30,41,69 microRNA,^31,32 tissue bacterial species,^33-37 IL-1B,^24,39,43 IL-17A,^24,27,39,⁷⁰ IL-17F,^24,42 hBD3,^39,45 and IFN-γ.^39,41,71,72 A wide range of other markers (eTable 3 in the Supplement) without external validation were also identified. The low GRADE assessments of these markers are a reflection of the lack of independent validation and observational univariate correlation (Table 2; eTable 3 in the Supplement).

Monitoring Biomarkers

The validation of proposed biomarkers to HS disease activity has typically been undertaken using validated clinical outcomes, such as the Sartorius score, HiSCR (Hidradenitis Suppurativa Clinical Response), and IHS4 (International Hidradenitis Suppurativa Severity Score System)^11-13 (Table 3). However, validation is often undertaken at 1 time point rather than longitudinally. Monitoring biomarkers with moderate GRADE assessment include serum IL-17,³⁰,^38,69 SAA,^24,25 CRP,^24,25,27 IL-8,^26,27 and sonographic measures including tunnel diameter^51,73-75 (Table 3). Sonographic dermal vascularity was the only monitoring biomarker to achieve a GRADE rating of high owing to independent validation and assessment of analytical validity.^50-52,73-75 Serum erythrocyte sedimentation rate,^25-27 tissue citrullinated H3,^46,47 and body mass index^53,54 are also associated with disease severity using various measures (Table 3) and validated in independent studies. Other monitoring biomarkers identified are listed in eTable 4 in the Supplement.

Significant overlap between diagnostic and monitoring biomarkers was identified (eTables 3 and 4 in the Supplement) with a number of low GRADE-assessed diagnostic markers also having univariate association with disease severity. None of the proposed monitoring biomarkers have undergone analytical or clinical validation to demonstrate validity in a longitudinal setting. This is a particularly important aspect given the significant natural variability of clinical disease activity in HS⁷⁶ and should be an aspect of future monitoring biomarker studies.

Predictive Biomarkers

Two predictive biomarkers (presence of epithelialized tunnels^56,57 and positive family history⁵⁷) achieved a GRADE assessment of high based on their independent validation and integration into the PIONEER 1 and PIONEER 2 phase 3 clinical trials of adalimumab in HS (Table 4). This indicates that particular morphological and clinical characteristics may be predictive of a response to adalimumab therapy as measured by HiSCR (Table 4). Additionally, BCL2 sequence variants (rs59532114)⁵⁸ and elevated serum IL-6 levels have been associated with a decreased odds of clinical response to adalimumab⁵⁸ and infliximab,⁷⁷ respectively. It should be noted, however, that these results require external validation in future studies (eTable 5 in the Supplement). Given the recent insights into B cells being the site of action of adalimumab in HS,^47,78 tissue cell markers such as BAFF and IL-1a^47,78 are potential novel predictive markers that require validation in external studies (eTable 5 in the Supplement).

Prognostic, Pharmacodynamic/Response, and Safety Biomarkers

To our knowledge, no longitudinal studies have undertaken assessment of the risk of disease progression. Currently, to our knowledge, no studies have identified pharmacodynamic or safety markers in the setting of HS therapies.

Discussion

The identification, development, and validation of biomarkers for HS are vital aspects to improving clinical management of patients with this chronic, burdensome disease.⁴ Current biomarkers (Figure, Tables 1, 2, 3, and 4; eTables 2-5 in the Supplement) are primarily diagnostic biomarkers based on small patient cohorts, with few biomarkers integrated into large-scale clinical trials.^79,80 Of 155 identified biomarkers, 44 achieved a GRADE rating of moderate, and only 4 biomarkers were rated high. Elevated serum IL-2R was associated with a diagnosis of HS,^18-20 likely reflecting the systemic nature of HS inflammation. The monitoring biomarker dermal Doppler vascularity was associated with disease severity,^50-52 which may be useful in future clinical and research studies. The presence of epithelialized tunnels in the setting of adalimumab treatment decreased the odds of achieving a clinical response^56,57 and increases the length of time to achieve a clinical response.⁵⁷ Similarly, a positive family history of HS increased time to loss of clinical response in adalimumab studies.⁵⁷

Barriers and Limitations to Biomarker Identification and Validation

Limitations and proposed future directions of biomarker research in HS are summarized in eTable 6 in the Supplement. A major limitation to existing biomarker studies is the broad number of studies (including clinical/phenotypic markers, imaging-based markers, proteomics, transcriptomics, and genetic-based markers) with insufficient depth of biomarker validation. Additionally, given that (to our knowledge) there are no extant longitudinal studies evaluating the risk of developing HS, all susceptibility/risk biomarkers are based on cross-sectional studies of cases and controls. The identified susceptibility/risk biomarkers would require validation in longitudinal case-control studies to truly reflect the risk of developing HS. The majority of low GRADE-rated biomarkers were identified via univariate association with clinical outcomes (HiSCR, Sartorius score) and patient-reported outcomes (pain, Dermatology Life Quality Index) compared with healthy control participants. Such univariate associations require robust validation in large independent data sets. This should preferentially be undertaken via assumption-free (or -omics) methods to address known issues of selection bias and collinearity,⁸¹ rather than via reductive methods, such as examination of only a preselected range of markers.

Currently, as the majority of studies are based on reductive methods, collinearity is likely to exist within the compiled list of biomarkers with collinearity describing the linear association in a statistical regression model between 2 independent variables.⁸¹ Many examined markers have been identified based on previous studies in psoriasis vulgaris. Owing to this approach, many identified diagnostic biomarkers target mediators shared by psoriasis vulgaris and HS inflammatory pathways (including S100A7, IL-17A/F, and CXCL1/8). The lack of assumption-free methods leads to selection bias, which is only just beginning to be addressed by omics approaches.^47,78 Identification of collinearity in an assumption-free data set enables the identification of the most specific or unique marker, and this may address the issues with identified markers not differentiating between HS and other inflammatory disorders. Addressing the issues of assumption-free biomarker identification and collinearity is important for future HS biomarker research.

Specific barriers to the identification and validation of susceptibility/risk biomarkers include the long-standing issue of diagnostic delay in HS¹⁶ and an incomplete understanding of the pathogenesis of disease.^1,2,82 Particularly in the context of genomic risk biomarkers, large genome-wide association studies are required before reliable identification of novel risk biomarkers is likely to occur. Additionally, further basic research is needed to elucidate how such identified genetic loci translate to the observed molecular inflammatory pathways in HS, given the conflicting data regarding Notch, PI3K, and AKT signaling in HS.^60,83,84 This will be important to understand how such risk loci translate into clinical disease.

Clinical outcomes used in the validation of monitoring biomarkers were largely confined to Hurley staging and the Modified Hidradenitis Suppurativa Score, with the HS Physician Global Assessment also used (Table 3). There is known variability and issues with reproducibility in Hurley staging⁸⁵ and lesional counts.⁷⁶ Further validation needs to be undertaken with other outcome measures, such as the IHS4⁸⁶ and HS Area and Severity Index.⁸⁷

Additionally, recent mechanistic insights suggest that some tissue and serum monitoring biomarkers may be reflective of the different immunological pathways that are seen in individuals with and without tunnels.^88,89 This is supported by the negative association of epithelialized tunnels as a predictive biomarker with clinical response to adalimumab.^56,57 Further work is needed to clarify how such structures should be defined and identified, whether clinically, sonographically, or histologically, for such biomarkers to be relevant to the practicing physician. The continued integration of biomarkers into the design of future clinical trials is a vital step forward in translating biomarker research directly to the clinic. Currently, post hoc analyses of 2 clinical trials using complement inhibitors (vilobelimab⁹⁰ and avacopan⁹¹) have identified epithelialized tunnels as a significant predictor of clinical response to this class of agents. Consideration of patient stratification by a validated predictive biomarker would aid the identification of more targeted therapeutics for subpopulations of individuals with HS (eg, those with and without epithelialized tunnels) as is done for fistulizing and nonfistulizing Crohn disease in the gastroenterology field. These data, combined with recent evidence suggesting that neutrophil- and B-cell–associated pathways are associated with these tunnels,^78,88 further emphasize the need for evaluation of collinearity among predictive biomarkers in particular and validation with assumption-free methods.

The role of bacterial-based biomarkers requires further investigation given the established role of bacterial dysbiosis in HS.^33-37 It is acknowledged that the inflammatory markers identified in multiple studies may be influenced by the presence or absence of specific microbionts,³⁷ and hence, future studies correlating tissue and serum inflammatory profiles with microbiome studies are needed to examine the relevance and collinearity of bacterial dysbiosis to identified immunological alterations.

Tissue-based biomarker identification is prone to variability and heterogeneity in findings owing to natural immunological variability across cutaneous sites,^92,93 as well as inconsistency in methods in identifying of lesional, perilesional, and nonlesional tissue.⁹⁴ While recommendations for lesion definitions and biopsy sites have been published,^94,95 implementation and further consensus regarding such definitions is required to standardize tissue-based biomarker investigations. Although outside the bounds of our search strategy, recent publications have identified serum IL-8, CCL-19, and CXCL9 as predictive biomarkers of adalimumab response in the PIONEER phase 3 randomized clinical trials.⁹⁶ This study would have raised serum IL-8 level to a moderate GRADE rating and also provides independent validation of the importance of B-cell–associated chemokines in response to TNF-α blockade.⁴⁷

The identification of prognostic biomarkers, particularly regarding progression to scarring/advanced disease or association with other complications such as squamous cell carcinoma,^97-99 is an important area of need for future biomarker research. Biomarker research in the context of longitudinal registry efforts to understand the clinical course of HS, such as the multi-institutional Hidradenitis Suppurativa Prospective Observational Registry and Biospecimen Repository (HS PROGRESS), will be critical to developing robust prognostic biomarkers.¹⁰⁰

Use of HS Biomarkers in Trials and the Clinic

The results of this systematic review suggest that in HS, no single biomarker has yet been adequately validated to recommend for routine clinical use. Identified biomarkers with the highest GRADE recommendations included serum IL-2R as a diagnostic biomarker,^18-20 dermal doppler vascularity as a monitoring biomarker, and epithelialized tunnels and family history as predictive biomarkers.^56,57

Identification and validation of high-quality predictive biomarkers are especially relevant given the number of mid-phase clinical trials that have failed to achieve primary outcome(s) owing to suspected disease heterogeneity.^90,91 At present, the main focus of clinical trials is on clinical outcomes, while less emphasis is placed on analysis of serum or tissue-based biomarkers. Aside from the well-documented issues with outcome measures in HS clinical trials,^86,87 the consideration of patient stratification by a validated predictive biomarker would aid the identification of more targeted therapeutics for subpopulations of individuals with HS (eg, those with and without epithelialized tunnels) as is done for fistulizing and nonfistulizing Crohn disease in the gastroenterology field. The first step to realize this would be the examination of collinearity between predictive biomarkers, given the theoretical underpinnings linking epithelialized tunnels, neutrophil activity, and B cells in HS.^56-58

Limitations

Priorities for future biomarker studies should address the described limitations in the current literature as outlined previously (eTable 6 in the Supplement). More specifically, identified biomarkers require independent validation in an assumption-free omics setting in both tissue and whole blood/serum.⁸¹ Independent studies looking at the collinearity of identified clinical and inflammatory markers would complement feature selection algorithms⁸¹ to identify associations between biomarkers and the most appropriate biomarker for a given task (eg, diagnostic, monitoring). Current work in development and validation of clinical outcome measures (including the IHS4,⁸⁶ Severity and Area Score for Hidradenitis,⁴⁰ and HS Area and Severity Index⁸⁷) will likely provide benefits in biomarker validation given that newer outcome measures take into account the variable morphologic characteristics (nodules, abscesses, and tunnels) that can contribute to disease activity in HS. This has particular relevance in the setting of predictive and monitoring biomarkers, where differential response may be seen in different disease morphologic characteristics (eg, nodules vs tunnels).

Conclusions

Studies in the extant literature provide an important step in the development of biomarkers for daily clinical use in HS, although no biomarker is yet at the point of established clinical utility. Biomarkers hold great potential to advance our understanding of disease pathophysiology as well as the clinical management of HS via identification of novel treatment targets. However, the process of biomarker validation requires multicenter and independent validation of findings and assumption-free methods to identify the most appropriate marker for clinical validation. Additionally, unique issues in the setting of HS include the need for assessment of biomarker stability over time as well as acknowledgment of the deficiencies in existing clinical outcome measures. Overall, to advance this field in HS and more robustly validate identified biomarkers, stakeholder consensus is required to outline and propose standardized methods for the identification, investigation, and validation of biomarkers in HS.

Key Points.

Question

What is the role of biomarkers in hidradenitis suppurativa (HS), and how are they validated?

Findings

In this systematic review, a total of 48 biomarkers were identified with a minimum Grading of Recommendations, Assessment, Development, and Evaluations rating of moderate; only 1 diagnostic (serum IL-2R), 1 monitoring (dermal Doppler vascularity), and 2 predictive biomarkers (epithelialized tunnels and positive family history of HS) achieved a high rating. None of the identified biomarkers had sufficient clinical validity to be recommended for routine use in the clinical setting; priorities were presented for near-term future research to overcome barriers and limitations of biomarkers in HS.

Meaning

The identification and validation of biomarkers in HS has potential to improve the understanding and management of this chronic, burdensome disease.