Patients and Lesions

In the Centre Médical de l’Institut Pasteur, a referral center for HS patients in France, microbiological samples are routinely obtained before initiation of antimicrobial treatment of HS patients. In this study, we performed a microbiological analysis of all HS lesions sampled from patients who consulted for the first time in our center for active HS during June 2007–February 2011 (Table 1). We excluded patients who received systemic or topical antibiotic drugs during the month before sampling. The clinical severity of lesions was assessed and designated by the same physician using the clinical severity staging of Hurley (12). Briefly, according to Hurley, stage 1 lesions correspond to nodules or abscesses, single or multiple, without sinus tracts or hypertrophic scars. Stage 2 lesions are single or multiple but nonconfluent lesions with sinus tracts and formation of scarring. Stage 3 lesions correspond to diffuse or nearly diffuse involvement of multiple interconnected sinus tracts or abscesses across an entire area.

Lesion Samples

We collected 2 types of lesion samples: 1) transcutaneous samples (from punch biopsies, ultrasonography guided biopsies, and needle aspirations) performed under strict asepsis using 5% povidone-iodine solution, and 2) swab specimens of superficial purulent drainage collected by using the Portagerm system (bioMérieux, Marcy l’Etoile, France) without aseptic preparation. Transcutaneous samples were collected only from patients who gave informed consent. Such samples were obtained from all closed abscesses or nodules and were also recommended for suppurative lesions. For suppurative lesions, we also suggested collecting purulent drainage by swab and collecting an additional control specimen at a 5-cm distance from the lesion, considering that biopsy may fail to reach the infectious site. No transport medium was used for punch biopsy specimens and purulent drainage collected by puncture. Samples were sent to the laboratory within 1 hour after sampling.

Bacterial Cultures and Identification Methods

To grow anaerobic bacteria, we homogenized biopsy samples using a sterile porcelain mortar in 0.5 mL of Schaedler broth (bioMérieux, Marcy l’Etoile, France). Purulent drainage and swab specimens were directly discharged in 0.5 mL of Schaedler broth; 50 μL of the suspension was seeded on agar plates, including an Uriselect4 agar plate (Bio-Rad, Marnes-la-Coquette, France), a colistin-nalidixic acid (CNA) blood agar plate, and a Columbia blood agar plate (bioMérieux). Uriselect4 and CNA agar plates were incubated at 37°C under 5% CO₂ for the isolation of aerobic and microaerophilic bacteria. Columbia agar plates were incubated anaerobically for 2 weeks. Cultures were analyzed at days 2, 7, and 15 by the same physician throughout the study. Anaerobic cultures were considered positive when the abundance or diversity of the bacterial culture was increased under anaerobic conditions. Plates were streaked by using the 4-phase pattern for isolation of predominant colonies. When the number of bacterial colonies was <200, colonies were counted; when >200 colonies, we assigned a colony count of 500 or 1,000 to bacterial colonies that reached the third or fourth quarter of the plate, respectively.

A maximum of 10 colonies per sample was identified by matrix-assisted laser desorption–time-of-flight mass spectrometry by using the Andromas system (13). When no identification was obtained, the 16S ribosomal RNA gene was sequenced. Altogether, 417 bacterial isolates were identified from the 162 culture-positive samples. Bacterial species were grouped in 12 categories (Technical Appendix Figures 1–5).

Bacterial Metagenomics

To investigate more precisely the anaerobic microflora and to decipher whether nonculturable species could be associated with HS lesions, we performed a metagenomic study on 6 consecutive samples including 1 Hurley stage 1 abscess and 5 chronic suppurating lesions. We used sterile dry swabs to collect samples that were immediately frozen at −80°C. We extracted DNA extraction using the MagNa Pure technology (Roche Pharma, Boulogne-Billancourt, France). We amplified each sample using the eubacterial universal 16S primer set 27F/338R described by Fierer et al. that targets the V1 and V2 hypervariable regions of the small subunit of the ribosomal RNA gene (14). We used Platinium PCR SuperMix (Invitrogen, Carlsbad, CA, USA) to elicit amplification reactions. PCR products were purified and concentrated by using the UltraClean PCR Clean-up Kit (MoBio, Carlsbad, CA, USA). Samples were sent to GATC Biotech AG (Konstanz, Germany) to be pyrosequenced by using 454 Life Sciences sequencing (454 Life Sciences, a Roche Company, Branford, Connecticut, USA).

An average of 4,407 quality sequences (2,564–7,008 sequences) were obtained from each sample. We analyzed data using the QIIME software (15). Similar sequences were clustered into operational taxonomic units by using Uclust (16), with a minimum identity of 0.97. We assigned taxonomy using the RDP Classifier (17).

Biostatistics

Statistical analyses were performed with R software version 3.01 (http://www.r-project.org). Hierarchical ascending classifications were performed to characterize sample profiles according to their bacterial composition. We computed the Euclidean distances between samples or species according to the bacterial abundance and then built dendrograms using the Ward aggregation criterion (18). Clustering results were figured in a heatmap chart (pheatmap R package, http://cran.r-project.org/web/packages/pheatmap/index.html) where samples and species are reordered according to the dendrograms.

We used the Fisher exact test to compare categorical variables. All tests were 2-sided; p values <5% were considered to be significant. We performed statistical analyses in R software.

Identification of Microbiological Profiles Associated with HS Lesions

Culture results of the 183 samples are presented in Technical Appendix Figure 1. Of the 125 lesional samples, 106 yielded a positive culture. To identify single or multiple bacterial species specifically associated with HS lesions, hierarchical clustering was performed on these samples (Technical Appendix Figure 2). This strategy identified 2 microbiological profiles (Technical Appendix Figures 3–5).

The profile A group identified Staphylococcus lugdunensis as a unique or predominant isolate (Technical Appendix Figure 3). This group of samples comprised 24 lesion samples: 22/38 lesions were at Hurley stage 1 and 2/45 were at Hurley stage 2.

The profile B group was characterized by a mixed anaerobic flora composed of strict anaerobes and/or anaerobic actinomycetes and/or streptococci of the milleri group (Technical Appendix Figure 4). Various other bacteria, such as S. aureus, coagulase negative staphylococci, corynebacteria, Enterobacteriaceae, Propionibacterium spp., and Enterococcus spp., were inconstantly and in smaller amounts associated with the mixed anaerobic flora, especially when lesions were sampled by swabbing. Profile B organisms were found in 77 samples corresponding to 65 lesions.

Five samples from lesions did not correspond to the 2 identified microbiological profiles (Technical Appendix Figure 5). These samples were 2 Hurley stage 1 samples yielding pure culture of Propionibacterium acnes and 2 samples yielding corynebacteria, Enterobacteriaceae, and coagulase negative staphylococci. A pure culture of Streptococcus pyogenes was recovered from the fifth sample. A review of the medical file of the patient revealed that he was admitted for an acute infectious syndrome that is unusual in HS patients and probably corresponded to an acute superinfection.

Considering the polymicrobial nature of profile B, including bacterial species known to be nonpathogenic skin commensals, we aimed at determining the relevance of these bacteria in the pathogenic process. To achieve this goal, we first analyzed the culture results of 45 purulent drainage on swabs for which a perilesional control sample was obtained (Technical Appendix Figure 6). S. aureus, coagulase negative staphylococci, micrococcaceae, corynebacteria, Propionibacterium spp., Enterobacteriaceae, and Enterococcus spp. were isolated from both purulent drainage and perilesional controls. However, these organisms were isolated less frequently and in lower quantity from purulent drainage than from controls. Conversely, strict anaerobes, actinomycetes, and streptococci of the milleri group were recovered from almost all purulent drainage samples (42/45) and rarely from controls (6/45).

We next compared the culture results of 23 open suppurating lesions that were sampled by biopsy, needle aspiration under strict asepsis, or swabbing, for each of which a control perilesional swab was obtained (Technical Appendix Figure 7). Anaerobes, actinomycetes, and streptococci of the milleri group were isolated from purulent drainage, biopsies, and aspirations, but not from perilesional swabs. By contrast, S. aureus, non-lugdunensis coagulase-negative staphylococci, corynebacteria, Enterobacteriaceae, and Propionibacterium spp. were commonly isolated from purulent drainage and swabs but very rarely from biopsies.

These data demonstrate that anaerobes, actinomycetes, and streptococci of the milleri group are specifically associated with HS lesions. Other species isolated from purulent drainage samples were part of the normal skin flora and were likely to be present as contaminants.

Correlation of Culture Results with Disease Severity and Lesion Topography

Our next task was to analyze the association of microbiological profiles with clinical severity, topography of the lesions, and gender (Table 2). Profile A was almost exclusively associated with Hurley stage 1 lesions, whereas profile B was predominantly associated with Hurley stage 2 and stage 3 lesions. Additionally, profile A tended to be associated with lesions of the breasts and buttocks.

Twelve patients had 2 or more lesions that had positive culture results (Table 3). Microbiological profiles were the same in 2 lesions of the same patient in 8 cases and different in 4 cases, indicating that microbiological profiles are not specific to a given individual.

Composition of Polymicrobial Anaerobic Flora as Assessed by Culture

The predominant anaerobic flora (1–7 different bacterial colonies per lesion) was studied in 36/62 profile B lesions (Table 4). A total of 95 anaerobic isolates were identified, including gram-positive cocci (predominantly Anaerococcus spp., Peptoniphilus spp., and Finegoldia spp.) and gram-negative rods (predominantly Prevotella spp., Porphyromonas spp., Bacteroides spp., and Fusobacterium spp.).

Predominant Actinomyces species were A. turicensis (30% of isolates), A. radingae (23% of isolates), A. neuii (14% of isolates), and Actinobaculum schaali (21% of isolates). Less frequently recovered species were A. massiliae (3 isolates), A. europaeus (2 isolates), A. funkei (1 isolate), and A. urogenitalis (1 isolate). In 14 lesions, >1 Actinomyces species (2 to 4) was identified.

Among streptococci of the milleri group, Streptococcus anginosus and Streptococcus constellatus, were recovered from 18 and 12 lesions, respectively. Both species were identified in the same samples in 2 cases.

Composition of the Anaerobic Flora as Assessed by Metagenomics

Figure

Figure. Microbial diversity of hidradenitis suppurativa (HS) lesions as assessed by high-throughput 454 sequencing. The bacterial diversity of 6 consecutive and representative HS lesions was assessed by high-throughput sequencing. Sample 1 corresponded...

We investigated the microbial diversity of 6 HS lesions by high-throughput sequencing. These lesions comprised 5 chronic suppurating lesions sampled by swabbing and 1 S. lugdunensis abscess sampled by needle aspiration (Figure). High-throughput sequencing confirmed that Staphylococcus was the far predominant taxon (99%) within the Hurley stage 1 abscess. In chronic suppurating lesions, anaerobic species (Prevotella, Porphyromonas, Anaerococcus, and Mobiluncus spp.) were the predominant taxa. Three bacterial orders were present in all samples: Bacteroidales (mainly Prevotella and Porphyromonas); Clostridiales (mainly Peptoniphilus, Anaerococcus, Parvimonas, Dialister, and Finegoldia); and Actinomycetales (Actinomyces, Actinobaculum, and Mobiluncus).

Thus metagenomic data were consistent with culture results, facilitating a more exhaustive description of the anaerobic flora. It should be pointed out that metagenomics did not identify noncultivable bacteria associated with HS lesions.

Altogether, we identified 2 microbiological profiles specifically associated with HS lesions. Profile A was characterized by pure or predominant culture of S. lugdunensis that was mostly associated with Hurley stage 1 lesions. Profile B was represented by a mixed flora composed of gram-negative and gram-positive strict anaerobes, anaerobic actinomycetes, and streptococci of the milleri group. This profile was mainly associated with open suppurating lesions observed in Hurley stages 2 and 3, but also with 24% of Hurley stage 1 lesions. No microbiological differences could be identified between Hurley stage 2 and 3 lesions by culture methods.