Finegoldia magna Isolated from Orthopedic Joint Implant-Associated Infections

ABSTRACT

The anaerobic Gram-positive coccus Finegoldia magna is a rare cause of infections of bone and joints. The aim of this study was to describe the microbiological and clinical characteristics of orthopedic implant-associated infections caused by F. magna. We retrospectively analyzed samples consisting of anaerobic Gram-positive cocci and samples already identified as F. magna from patients with orthopedic infections. The isolates found were determined to the species level using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS). The antibiotic susceptibility pattern was determined by Etest. Whole-genome sequencing (WGS) was performed. Clinical data were extracted from each patient's journal. In nine patients, orthopedic joint implant-associated infections were identified as being caused by F. magna. The isolates were susceptible to most of the antibiotics tested, with the exception of rifampin and moxifloxacin in a few cases. Five of the nine infections were monomicrobial. The most common antibiotic used to treat the infection was penicillin V, but five of the nine patients received a combination of antibiotics. Eight patients underwent surgical treatment, with extraction of the implant performed in seven cases and reimplantation in only two cases. The WGS showed a relatively small core genome, with 126,647 single nucleotide polymorphisms identified within the core genome. A phylogenomic analysis revealed that the isolates clustered into two distinct clades. Orthopedic implant-associated infections caused by F. magna are rare, but the bacteria are generally susceptible to antibiotics. Despite this, surgical treatment combined with long-term antibiotics is often necessary. The WGS analysis revealed a high heterogeneity and suggested the existence of at least two different Finegoldia species.

INTRODUCTION

Foreign materials, such as nails, screws, and plates, as well as prosthetic devices, are regularly used in orthopedic surgery. The two most common indications for orthopedic implants are fractures and osteoarthritis. The general objectives are to restore stability and function and to reduce the patient's pain. Infection is a rare but very unwelcome complication. The rate of infection is 1 to 2% following prosthetic joint surgery (1) and is higher for other orthopedic implant procedures.

The most common pathogens that cause orthopedic implant infections are coagulase-negative staphylococci (CoNS), Staphylococcus aureus, streptococci, and Enterobacteriaceae (1). More rarely, anaerobic bacteria are found, predominantly Propionibacterium acnes (2), especially following shoulder surgery (3).

Another anaerobic pathogen that has recently attracted attention as a rare cause of infections in bone and joints is the anaerobic Gram-positive coccus Finegoldia magna (4, 5). This species was previously named Peptostreptococcus magnus but was reclassified in 1999 and is currently the only member of its genus (6–8). The involvement of this bacterium in anaerobic soft tissue, bone, and joint infections, especially in the presence of implants, may be more common than previously recognized (4). F. magna is part of the human normal microbiota and is present on the skin and the gastrointestinal mucosa (9). It exhibits adherence to abiotic surfaces, and biofilms can develop with matrix formation (10). The bacterium displays a variety of virulence factors and is considered one of the most pathogenic species among anaerobic Gram-positive cocci (11).

Conventional microbiological methods have revealed heterogeneity in F. magna. Most strains grow slowly; after 2 to 5 days of strict anaerobic incubation, colonies with a diameter of 1 to 2 mm will be obtained. Colony morphology varies from white and slightly elevated to transparent and flat. Colonies with different types of appearances can also be found on the same plate (6, 7). In addition, matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has shown F. magna to be a heterogeneous group where not all strains exhibit the same mass spectrum (12).

The clinical presentation of bone and joint infections involving F. magna is represented both by early postoperative infections following prosthetic joint surgery accompanied by evident symptoms, and late infections often with mild and insidious symptoms (4–6). F. magna isolates are generally sensitive to the antibiotics routinely used to treat infections caused by anaerobic bacteria, but resistance to various antimicrobial agents, such as β-lactam antibiotics, clindamycin, fluoroquinolones, and metronidazole, has been reported (13, 14).

The aim of the present study was to describe the microbiological and clinical characteristics of orthopedic implant-associated joint infections due to F. magna.

RESULTS

Identification of Finegoldia magna.

F. magna was identified in 16 out of 30 samples (Fig. 1). Nine of these were from patients with an orthopedic implant-associated infection of a joint. The remaining patients' infections did not involve a joint or lacked an implant and so were excluded. Two of the isolates represented duplicate samples. The other bacterial species identified were Staphylococcus saccharolyticus, Staphylococcus epidermidis, Peptoniphilus harei, Eggerthella lenta, Parvimonas micra, and Veillonella parvula.

FIG 1.

Flow chart showing the process of identification of Finegoldia magna isolates in orthopedic joint implant-associated infections.

Antibiotic susceptibility testing.

The results of the antibiotic susceptibility tests are given in Table 1. All F. magna isolates were susceptible to benzylpenicillin, amoxicillin, clindamycin, metronidazole, and vancomycin. For three of the antibiotics tested (moxifloxacin, rifampin, and linezolid), EUCAST breakpoints are not available for anaerobic Gram-positive cocci. One of the isolates displayed high MIC values (>32 μg/ml) for both rifampin and moxifloxacin. This isolate, 12T306, was regarded as resistant, since no inhibiting effects of these antibiotics were detected. Three strain-specific nucleotide polymorphisms in the rpoB gene were found in the resistant isolate, leading to the following amino acid substitutions: Gln135Arg, Ile629Val, and Gly1179Asp (Fig. S1). Two additional isolates were found to be resistant to moxifloxacin, displaying MIC values of >32 and 4 μg/ml, respectively, but otherwise were susceptible to all other tested antibiotics. One isolate was resistant to linezolid and showed an MIC value of 6 μg/ml. The largest range of MIC values was noted for rifampin (0.002 to >32 μg/ml).

TABLE 1.

Distribution of MIC values for eight antibiotics, using Finegoldia magna isolated from patients with orthopedic implant-associated infections (n = 9)

^aPG, benzylpenicillin; AC, amoxicillin; CM, clindamycin; MZ, metronidazole; LZ, linezolid; RI, rifampin; VA, vancomycin; MXF, moxifloxacin.

Clinical data.

The clinical data extracted from the patients' records are presented in Table 2. A majority (7/9) of the patients were women, and the mean age was 65 years. Six of the nine patients were diagnosed with a prosthetic joint infection (PJI), and three had an osteosynthesis-related infection, with two in a shoulder joint and one in an ankle. Seven of the nine patients were diagnosed with an implant-associated infection more than 12 months after the primary surgical procedure. Monomicrobial and polymicrobial infections were equally distributed, with five monomicrobial infections and four polymicrobial infections. Only two patients retained their primary implant at follow-up. The implant or prosthetic device was extracted in the majority of cases, and a two-stage exchange surgical procedure was performed in only two cases.

TABLE 2.

Clinical data for nine patients with orthopedic implant-associated infections due to Finegoldia magna

^aF, female; M, male.

^bRA, rheumatoid arthritis.

Genome sequencing.

Since very little genomic knowledge about F. magna is currently available, we decided to sequence the clinical isolates in order to get a full picture of the genetic repertoire of the species. Moreover, since some F. magna isolates were identified with low MALDI-TOF MS scores, we wanted to investigate the population structure and sequence diversity of this species.

Ten strains (the nine clinical isolates and the F. magna reference strain CCUG 54800) were selected for genome sequencing with high coverage, resulting in draft genomes with 20 to 67 contigs. The GC content varied only slightly (31.8 to 32.1%), but the genome sizes varied substantially (1,679 to 1,990 kb), indicating heterogeneity between the isolates. The seven previously sequenced genomes of F. magna varied in size between 1,828 and 2,032 kb.

In order to investigate the genomic relationships between the F. magna isolates, a core genome analysis was performed. The core genome was revealed as being relatively small, constituting only 43% of the total average genome. A total of 126,647 single nucleotide polymorphisms (SNPs) were identified within the core genome. This phylogenomic analysis revealed that the 17 strains clustered into two distinct clades (Fig. 2). Clade A comprised three isolates of the present study and most of the previously sequenced strains (6 out of 7), including the reference strain ATCC 29328 and the ALB8 strain, which has already been studied in detail (15, 16). Clade B included six of our strains and the reference strain CCUG 54800 (isolated from synovial fluid), as well as the previously sequenced strain ATCC 53516 (isolated from the urogenital tract, vaginal). Clade A, with a core genome of 67% and a total of 89,557 SNPs, was more homogeneous than clade B, which had a reduced core genome of 65% and a total of 134,531 SNPs. A further core genome analysis was carried out to investigate the distance between these two clades. The average nucleotide identity (ANI) between strains of clade A was 96.6%, and that between strains of clade B was 94.6% (Table S2). Interestingly, the ANI between the genomes of clade A and clade B was only 90.7% on average. This indicates that clade A and B strains represent two independent subspecies or even two different species, given that an ANI below 94% generally separates two species (17). However, very few clade A-specific or clade B-specific genes could be identified. Instead, the flexible genome was mainly located on strain-specific genomic regions and putative plasmids (data not shown). The type strain ATCC 29328 is atypical, since it is the only strain among the 17 strains analyzed to date that possesses a large 189-kb plasmid.

FIG 2.

Phylogenomic comparison of all currently available genome sequences of Finegoldia magna. Left, the strains can be separated into two distinct clades. Right, the color depicts the SNP density in comparison to the reference (strain ATCC 29328); all strains of clade B harbor more SNPs in the core genome than the clade A strains. The overall core genome is 43% of the total genome size of strain ATCC 29328. The programs Parsnp and Gingr were used for core genome alignment, SNP calling, and visualization.

DISCUSSION

Infection related to orthopedic implants is a rare complication of this type of surgical procedure. F. magna is an unusual opportunistic pathogen. Of the 32 samples found in our study during 2004 to 2016, nine isolates of F. magna could be related to patients with orthopedic joint implant infections and hence were included in the study. As a comparison, a French study which retrospectively investigated bone and joint infections at three university hospitals during 2009 to 2012 found seven cases where F. magna was the etiology of orthopedic implant infections (4).

Antibiotic susceptibility testing showed that all strains were sensitive to benzylpenicillin, amoxicillin, clindamycin, metronidazole, and vancomycin, according to the specific clinical breakpoints for anaerobic Gram-positive cocci set by EUCAST. This confirms previous studies (13, 14) demonstrating an overall sensitivity to many of the antibiotics used to treat anaerobic infections. For two other antibiotics, moxifloxacin and linezolid, we had to apply non-species-specific breakpoints, since specific breakpoints were lacking. Three of the isolates were classified as resistant to moxifloxacin and one isolate as resistant to linezolid. A previous study has also reported resistance to various fluoroquinolones for F. magna (14). No clinical breakpoints are available for rifampin, but one isolate displayed high MIC values (>32 mg/ml) for both rifampin and moxifloxacin. This patient had a history of previous treatment with the combination of rifampin and a fluoroquinolone, which may explain the high MIC value for these two antibiotics, and thus the risk of emergence of resistance during treatment. Resistance to rifampin in S. aureus and P. acnes develops easily as a result of single point mutations in the rpoB gene (18, 19).

The joints involved in our patients' infections included the knee, hip, shoulder, and foot. The majority (7/9) of the infections were diagnosed 12 months or later after the primary operation, regardless of whether they were monomicrobial or polymicrobial. Levy et al. (6), who included only monomicrobial infections with F. magna in their study, showed that a large proportion of these infections had a symptom-free interval of less than 4 months. Their study (6) also divided the infections into posttraumatic open fractures and nosocomial infections. No posttraumatic open fractures were included in our study; all patients had either elective prosthetic devices or implant surgery. Furthermore, Levy et al. identified posttraumatic infections as polymicrobial and nosocomial infections as monomicrobial, while no such pattern could be noted in our study.

The antibiotic most commonly used for long-term treatment was penicillin V, followed by amoxicillin, rifampin, ciprofloxacin, and clindamycin. Combination therapy was used in five of the nine cases. Where rifampin was used, it was combined with ciprofloxacin or fusidic acid. Since F. magna in general is susceptible to benzylpenicillin and amoxicillin, these should be the drugs of choice unless no polymicrobial infection is present. There is currently insufficient evidence to recommend rifampin combination therapy for anaerobic prosthetic joint infections, such as those caused by P. acnes (20), although there are promising data from an experimental foreign-body model (21).

The foreign material was extracted in seven of nine patients at follow-up; only one patient did not undergo surgical intervention at all. In three of five patients where the infections were assessed as being monomicrobial, the prosthetic device was extracted without later reimplantation. Of all nine cases, five had no reimplantation. Only one of the nine patients underwent a two-stage operation without reinfection. The two-stage exchange strategy is otherwise the most common approach to surgical treatment. The reasons for implant removal without reimplantation may be compromised local or general condition of the patient or the absence of additional functional benefit of an arthroplasty (1).

This study has several limitations. When searching the local laboratory database for isolates kept in the freezer but not identified as F. magna before the introduction of MALDI-TOF MS in January 2014, we selected samples from patients where three or more samples were taken at the same time. Isolates may have been missed if insufficient numbers of samples or tissue biopsy specimens were obtained during debridement or revision surgery. However, the time period investigated was 2004 to 2016, and national guidelines since 2004 (revised in 2008) have recommended obtaining ≥5 tissue samples during debridement or revision surgery. These national guidelines also include treatment algorithms with recommendations about antimicrobial therapy and surgical interventions.

Most strains of F. magna grow slowly, with tiny colonies, and so can easily be missed by inadequate incubation. Previous studies (6, 22) have highlighted the importance of prolonged incubation time and careful examination of samples. In many cases, a polymicrobial infection is present (4, 6), and in cases where more rapidly growing pathogens are present, F. magna can be missed or detected less frequently.

Normally, species-level identification by MALDI-TOF MS requires a score of >2.0, according to the manufacturer's instructions. In our study, however, a score of >1.7 was considered to indicate an identification of F. magna. According to the manufacturer, this indicates identification at the genus level, and F. magna is currently the only species in its genus (8). Previous studies have shown that F. magna is a very heterogeneous species, in which some strains may have a mass spectrum that cannot be matched in the Biotyper database (6, 7, 12). Moreover, until recently, only a few spectra were included in the database, which made it more difficult to correctly identify all strains as F. magna (23). Additional methods of identification may be 16S rRNA gene sequencing or WGS (22).

In general, there are only a few studies exploring F. magna and orthopedic implant infections (4–6). Prior to this study, genomic data on F. magna were restricted to 7 strains of diverse origins, including two ATCC strains (ATCC 29328 and ATCC 53516). With the 10 genomes sequenced here, 9 of which were obtained from joint implant-associated F. magna isolates, we can now perform a meaningful phylogenomic analysis. We noted a high degree of heterogeneity among the 17 sequenced isolates but found that the Finegoldia strains could be clustered into two clades. There was a striking genomic difference between the two clades, with an overall average nucleotide identity (ANI) of only 90.7%. This suggests that these two clades represent distinct species within the genus Finegoldia. Strains of F. magna are represented by the clade containing ATCC 29328 and ALB8. The other clade comprised the majority of the sequenced strains isolated from implant infections. The high heterogeneity is in agreement with previous observations, such as the low MALDI-TOF MS score for identifying F. magna isolates and the distinct morphological appearance (6, 7, 12).

The question remains of whether strains from these two clades have a different pathogenic potential. Clade-specific genes were rare, but subtler differences between the two clades might be present, in addition to the high number of strain-specific variations.

Conclusion.

Orthopedic joint implant infections caused by F. magna are rare. We retrospectively identified F. magna isolates from nine patients from 2004 to 2016. All nine isolates were assessed as being susceptible to most of the antibiotics tested, with a few exceptions, including linezolid, moxifloxacin, and rifampin. Both monomicrobial and polymicrobial infections were found. At follow-up, the implants were extracted in most cases, and reimplantation with a two-stage approach was performed in only two patients. Finally, although most of the isolates were highly sensitive to antibiotics, implant-associated infections cannot be successfully treated without extraction of the implants. The results of the WGS analysis revealed a high heterogeneity among the isolates, suggesting that at least two different Finegoldia species exist and that both can act as opportunistic pathogens.

MATERIALS AND METHODS

Bacterial isolates.

Bacterial isolates from orthopedic implant infections, including prosthetic joint infections, are routinely frozen at −70°C at the Department of Laboratory Medicine, Clinical Microbiology, Örebro University Hospital. Since 2004, national guidelines have recommended that ≥5 perioperative samples be taken when performing a surgical procedure of a suspected or verified deep orthopedic infection, such as during debridement and soft tissue revision. We therefore searched our database at the Department of Laboratory Medicine for all samples where ≥3 samples had been retrieved at the same occasion. All isolates of anaerobic Gram-positive cocci, anaerobic streptococci, or F. magna were chosen for further analysis; these were all from 2004 to 2016. CCUG 54800, originally isolated from human synovial fluid, was used as a reference strain.

Identification of F. magna.

All isolates had been stored in preservation medium (Trypticase soy broth with 0.3% yeast extract and 29% horse serum) at −70°C. The identified isolates were subcultured on FAA plates (4.6% LAB 90 fastidious anaerobe agar; LAB M, Heywood, United Kingdom) supplemented with 5% (vol/vol) horse blood and incubated at 36°C under anaerobic conditions for 2 to 5 days. The isolates were then identified to the species level by using MALDI-TOF MS (microflex LT and Biotyper 3.1; Bruker Daltonics, Bremen, Germany).

Antibiotic susceptibility testing.

MICs were determined via Etest according to EUCAST guidelines (www.eucast.org). Eight antibiotics were tested: benzylpenicillin, amoxicillin, clindamycin, metronidazole, linezolid, rifampin, vancomycin (all from bioMérieux, Marcy l'Etoile, France), and moxifloxacin (Liofilchem, Roseto degli Abruzzi, Italy). Antibiotic susceptibility testing was performed on FAA plates with 0.5 McFarland suspensions of bacteria in NaCl and incubation at 36°C under anaerobic conditions for 5 days.

Clinical data.

All isolates that could be identified as F. magna by MALDI-TOF MS with a score of ≥1.7 were linked to the patient's identification (ID; their unique Swedish personal number). The patients' medical records were reviewed according to a predefined protocol in order to extract relevant clinical information. Patients with an infection site other than joints or with an infected joint without prosthetic devices or implants were excluded from the study.

The study was approved by the Regional Ethical Review Board of Uppsala, Sweden (reference 2016/457).

Genome sequencing of F. magna isolates.

Genomic DNA (gDNA) from 10 F. magna strains was isolated using the MasterPure Gram-positive DNA purification kit (catalog no. MGP04100; Epicentre), according to the manufacturer's instructions. The purity and quality of the gDNA were assessed on a 1% agarose gel with a NanoDrop apparatus (Thermo Scientific, Wilmington, DE, USA). Extracted DNA was used to prepare Nextera XT shotgun libraries for the Genome Analyzer II (Illumina, San Diego, CA, USA) with a 112-bp paired-end sequencing run. Libraries were prepared according to the manufacturer's protocol at the Göttingen Genomics Laboratory, Germany. Raw reads were quality controlled with FastQC version 0.11.2 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and subsequently trimmed using Trimmomatic 0.32 (http://www.usadellab.org/cms/?page=trimmomatic) to remove sequences with quality scores lower than 20 (Illumina 1.9 encoding) and remaining adaptor sequences. De novo assembly was performed using SPAdes version 3.5 (24). Automated genome annotation was carried out using Prokka (http://www.vicbioinformatics.com/software.prokka.shtml) and RAST (25). Information about the number of contigs and genome sizes can be found in Table S1.

Bioinformatics and comparative and phylogenetic analyses.

The BRIG application (26) was used for comparative analyses. In addition to the 10 newly sequenced strains, we used seven genomes of F. magna strains stored in GenBank: ACS-171-V-Col3 (accession number AECM01000000), ATCC 29328 (AP008971 and AP008972), ATCC 53516 (CM000955.1), GED7760A (LRPW01000000), BVS033A4 (AEDP01000000), SY403409CC001050417 (AFUI01000000), and ALB8 (JDVC01000000).

The Harvest software package (Parsnp and Gingr programs) was used for core genome alignment, single nucleotide polymorphism (SNP) calling, and visualization (27). Phylogenetic trees were built in MEGA version 6 (28). JSpecies was used to calculate the DNA similarities between all genomes (29).

Accession number(s).

The GenBank accession numbers of the draft genome sequences are NDYJ00000000 (strain 07T609), NDYI00000000 (strain 08T492), NDYH00000000 (strain 09T408), NDYG00000000 (strain 09T494), NDYF00000000 (strain 12T272), NDYE00000000 (strain 12T273), NDYD00000000 (strain 12T306), NDYC00000000 (strain CCUG54800), NDYB00000000 (strain T151023), and NDYA00000000 (strain T160124).