Corynebacterium Prosthetic Joint Infection

Abstract

Identification of Corynebacterium species may be challenging. Corynebacterium species are occasional causes of prosthetic joint infection (PJI), but few data are available on the subject. Based on the literature, C. amycolatum, C. aurimucosum, C. jeikeium, and C. striatum are the most common Corynebacterium species that cause PJI. We designed a rapid PCR assay to detect the most common human Corynebacterium species, with a specific focus on PJI. A polyphosphate kinase gene identified using whole-genome sequence was targeted. The assay differentiates the antibiotic-resistant species C. jeikeium and C. urealyticum from other species in a single assay. The assay was applied to a collection of human Corynebacterium isolates from multiple clinical sources, and clinically relevant species were detected. The assay was then tested on Corynebacterium isolates specifically associated with PJI; all were detected. We also describe the first case of C. simulans PJI.

INTRODUCTION

While Gram-positive aerobic cocci cause most cases of prosthetic joint infection (PJI), Corynebacterium species are occasional causes. Since the publication of the few reports on Corynebacterium PJI, several new Corynebacterium species have been described. The genus now includes over 85 species (http://old.dsmz.de/microorganisms/bacterial_nomenclature_info.php?genus=Corynebacterium), not all isolated from humans. As normal human flora, Corynebacterium species are common contaminants in clinical specimens. Because of this, as well as challenges in their identification, they have not received a great deal of attention in clinical practice. They are, however, increasingly recognized as causes of significant human infection, including PJI (1, 16, 20, 22).

Corynebacterium species commonly isolated in the clinical laboratory include C. amycolatum, C. aurimucosum, C. glucuronolyticum, C. jeikeium, C. pseudodiphtheriticum, C. striatum, C. tuberculostearicum, and C. urealyticum (8). C. glucuronolyticum and C. urealyticum are involved predominantly in urinary tract infection and C. pseudodiphtheriticum in respiratory tract infection (8); the other five commensal flora species (10) are potential causes of device-associated infections, including PJI (13, 19). Corynebacteria that cause orthopedic device-associated infections are not usually identified to the species level, and specific Corynebacterium species involved in PJI are not definitively known.

As mentioned, many laboratories do not routinely identify Corynebacterium species, because they are frequently isolated as contaminants and their identification is challenging. In our laboratory, the term “small non-spore-forming Gram-positive bacillus resembling Corynebacterium species” is often used for coryneform organisms, since it can be challenging to differentiate Corynebacterium species from Turicella otidis, Arthrobacter species, Brevibacterium species, Dermabacter hominis, Rothia dentocariosa, Exiguobacterium acetylicum, Helcobacillus species, Oerskovia turbata, Cellulomonas species, Cellulosimicrobium species, Microbacterium species, Curtobacterium species, and Leifsonia aquatica using a simplistic strategy (8). For species-level identification, phenotypic testing such as with API Coryne (bioMérieux, Durham, NC) has been used. Sequencing-based methods targeting the 16S rRNA gene or rpoB give more precise species identification (11). Of these, rpoB sequencing is ideal but not commonly available; none of these methods is rapid.

As an alternative approach for rapid identification of Corynebacterium species, a PCR assay targeting the Corynebacterium species that most frequently cause human infection, with a specific focus on those causing PJI, was designed. In addition to detection of Corynebacterium species, the assay differentiates the antibiotic-resistant species C. jeikeium and C. urealyticum from other species. The assay was first tested on a collection of Corynebacterium isolates representing multiple species. It was then tested on isolates specifically associated with PJI, which were also identified using rpoB or 16S rRNA gene sequencing. Finally, we compared our results on the microbial ecology of PJI to the literature on the topic.

MATERIALS AND METHODS

Control isolates.

Nineteen clinical Corynebacterium species characterized by partial rpoB gene sequencing (2) and C. glutamicum ATTC 13032 were studied (Table 1). The five specifically targeted clinical species are in bold in Table 1. In addition, 338 non-Corynebacterium isolates, identified using phenotypic methods or 16S rRNA gene sequencing, from patients with biofilm-associated infections were used to assess cross-reactivity (Table 2).

Table 1.

Characteristics of Corynebacterium species and their detection by the described real-time PCR assay

Speciesa	Clinical source(s)b	Antimicrobial susceptibilityb	Frequency in clinical specimensb,c	Frequency in PJIc,d	No. of isolates tested	Melting temp (°C)
						Channel 1e	Channel 2f
C. accolens	Eyes, ENTg	Susceptible	−	−	1	−	−
C. afermentans	Skin, blood	β-Lactam susceptible	−	−	4	+61, 67h	−
C. amycolatum	Skin, foreign body, blood	Possible resistance	+++	+++	3	+64	−
C. aurimucosum	Skin, female genitourinary tract		+++	+++	4	+∼56	+50i
C. confusum	Foot, breast		−	−	1	+60	−
C. diphtheriae	Oropharynx	Susceptible	+	−	1	+57
C. durum	Respiratory tract, blood		+	−	1	−	−
Corynebacterium group F1	Urinary tract	Penicillin susceptible, often macrolide resistant	−	−	2	−	−
C. glucuronolyticum	Male genitourinary tract, blood	Often tetracycline resistant	++	−	1	+63	−
C. glutamicum	Animal species		−	−	1	+64
C. imitans	Skin, blood		+	−	1	+56	−
C. jeikeium	Skin, foreign body, endocarditis	Often resistant to multiple agents	+++	++	2	−	+61
C. mucifaciens	Blood, peritoneal fluid	Susceptible	+	−	1	+62	−
C. propinquum	Oropharynx		−	−	1	+55	−
C. pseudodiphtheriticumj	Oropharynx, Respiratory tract, endocarditis	β-Lactam susceptible	++	−	1	+55	−
C. riegelii	Female urinary tract, blood		+	−	1	−	−
C. singulare	Skin		−	−	1	+52	−
C. striatum	Skin, respiratory tract, foreign body	Possible macrolide, fluoroquinolone, tetracycline resistance	+++	+++	2	+50	−
C. tuberculostearicum	Skin, foreign body, endocarditis		+++	−	2	+56, 64h	−
C. urealyticum	Urinary tract, blood, groin	Often resistant to multiple agents	++	−	1	−	+61

^aBoldface indicates species most frequently isolated in the clinical laboratory.

^bBased on data from references 3, 8, and 9.

^c−, exceptionally or never observed; +, rarely observed; ++, frequently observed; +++, very frequently observed.

^dBased on data from references 13 and 19.

^eMeasured using F2/F1 color compensation.

^fMeasured using F3/F2 color compensation.

^gENT, ear, nose, and throat.

^hDouble peak on melting curve.

ⁱNo crossing point on quantification curve.

^jC. pseudodiphtheriticum is closely related to C. propinquum based on rpoB sequence; they also share the same PCR profile (positive only with probes 1 and 2, with a T_m of 55°C).

Table 2.

Non-Corynebacterium isolates (n = 338) from patients with biofilm-associated diseases tested for cross-reactivity

Group, genus, or species	No. of isolates
Staphylococci	201
S. aureus	85
Coagulase-negative staphylococci	116
S. epidermidis	80
S. lugdunensis	16
S. warneri	8
S. capitis	3
S. caprae	3
S. simulans	3
Othera	3
Propionibacterium spp.	41
P. acnes	29
P. avidum	10
P. granulosum	2
Enterobacteriaceae	28
Escherichia coli	5
Enterobacter cloacae	4
Proteus mirabilis	3
Klebsiella pneumoniae	3
K. oxytoca	2
Otherb	11
Streptococcus spp.	15
S. agalactiae	3
S. dysgalactiae	3
S. pyogenes	2
S. pneumoniae	2
S. salivarius	2
Otherc	3
Bacteroides fragilis group	13
B. fragilis	8
B. thetaiotaomicron	2
Otherd	3
Gram-positive anaerobic cocci	12
Finegoldia magna	8
Othere	4
Pseudomonas aeruginosa	10
Enterococcus faecalis	9
Granulicatella adiacens	3
Otherf	6

^a1 S. haemolyticus, 1 S. hominis, 1 S. saprophyticus.

^b1 Citrobacter freundii, 1 C. koseri, 1 E. aerogenes, 1 Morganella morganii, 1 Pantoea agglomerans, 1 P. vulgaris, 1 Providencia rettgeri, 1 Salmonella sp., 1 Serratia liquifaciens, 1 S. marcescens, 1 Shigella flexneri.

^c1 S. anginosus, 1 S. mitis, 1 S. mutans.

^d1 B. caccae, 1 B. distasonis, 1 B. ovatus.

^e1 Peptoniphilus asaccharolyticus, 1 P. harei, 1 Parvimonas micra, 1 Anaerococcus prevotii.

^f1 Candida parapsilosis, 1 Dermabacter hominis, 1 Gordonae terrae, 1 Pseudomonas fluorescens, 1 Rhodococcus equi, 1 Staphylococcus saccharolyticus.

Clinical isolates from the site of PJI.

Nine Corynebacterium species isolated from the site of hip or knee PJI between 1999 and 2008 were studied. They were characterized by partial rpoB (2) or (if rpoB failed) partial 16S rRNA gene (12) sequencing, and their antimicrobial susceptibilities were determined following current guidelines (4).

DNA extraction.

DNA was extracted using the QIAamp DNA Minikit (Qiagen, Valencia, CA).

Real-time PCR assay design.

A potential target corresponding to a polyphosphate kinase gene, pvdS2, of C. jeikeium strain K411 (accession number NC_007164) was identified in the genomes of C. aurimucosum ATCC 700975, C. jeikeium K411, and C. urealyticum DSM7109 (14, 15, 18). A consensus sequence was created using Sequencher (GeneCodes, Ann Arbor, MI) and primers (forward and reverse 1 [Table 3]) designed with Roche LightCycler Primer Design software. Primers (Integrated DNA Technologies, Coralville, IA) were initially evaluated on DNA from control organisms with SYBR green (Roche Applied Science, Indianapolis, IN) detection using the LightCycler 1.0 (Roche Applied Science, Indianapolis, IN). C. jeikeium, C. urealyticum, C. aurimucosum, and C. striatum, but not C. amycolatum or C. tuberculostearicum, were detected. The putative kinase genes of the two nondetected species were amplified with primers Seq Coryne kin F and R (Table 3) and sequenced, as previously described (12). Based on known and new sequence data, a third primer (reverse 2) was designed to enable the detection of the five clinically relevant PJI species with all three primers combined. Three fluorescence resonance energy transfer (FRET) probes, i.e., a donor probe labeled with fluorophore (fluorescein isothiocyanate [FITC]) at the 3′ end (probe 1) and two acceptor probes, one labeled with the fluorophore LC705 at the 5′ end (probe 3, specific for C. jeikeium) and the other with the fluorophore LC640 at the 5′ end (probe 2, detects other Corynebacterium species), were incorporated (Table 3). Primers and probes (TIB Molbiol, Berlin, Germany) were evaluated with DNA from Corynebacterium species (Table 1) and non-Corynebacterium species (Table 2).

Table 3.

Oligonucleotide primers and probes

Primer or probe	Sequence (5′ → 3′)
Forward	CGRTTGTACCARGARCGGT
Reverse 1a	GCACCTSAAYCCSCGT
Reverse 2b	CAACGAGCACCTSAACCC
Seq Coryne kin F	GTRCAGAAKCCCATSACGCGC
Seq Coryne kin R	GCSGGYAAGGGYGGCWCC
Probe 1	TAGCGCTGGAAGTACCASGAGGT-FL
Probe 2	LC640-GACTCRCGCGGCGACGG
Probe 3	LC705-GACTCGCGCTCGGAAGG

^aFragment size with primers forward and reverse 1, 143 bp.

^bFragment size with primers forward and reverse 2, 149 bp.

PCR optimization.

Two microliters of template DNA was added to 18 μl PCR mix (LightCycler FastStart DNA Master HybProbe mixture; Roche Applied Science) for a 20-μl final reaction volume. The cycling parameters initially evaluated were preincubation for 10 min at 95°C followed by 35 cycles of 95°C for 10 s, 55°C for 10 s, and 72°C for 22 s. Melting curve analysis (starting at 45°C) was performed with a temperature transition rate of 0.8°C/s to determine the melting temperature (T_m). The assay was optimized for primer concentration, forward/reverse primer ratio, Mg²⁺ concentration, annealing temperature, and number of cycles. Parameters giving the lowest crossing point for standardized positive controls were selected.

Optimization of the LightCycler PCR assay.

The optimal Mg²⁺, probe, and forward primer and reverse primer concentrations were 4 mM, 0.2 μM, 0.4 μM, and 2.0 μM, respectively. The optimal annealing temperature and cycle number were 57°C and 35, respectively.

LOD.

The limit of detection (LOD) was determined by testing serial 10-fold dilutions of known concentrations of C. tuberculostearicum and C. jeikeium.

RESULTS

Inclusivity and cross-reactivity.

All isolates of the five common human PJI species (C. amycolatum, C. aurimucosum, C. jeikeium, C. striatum, and C. tuberculostearicum) were detected (Table 1). Nine other species, including C. glucuronolyticum, C. pseudodiphtheriticum, and C. urealyticum, were also detected. Four unusual species (C. accolens, C. durum, Corynebacterium group F1, and C. riegelii) and non-Corynebacterium isolates (Table 2) were not detected.

Limit of detection.

The LODs were 200 and 440 CFU/ml for C. tuberculostearicum and C. jeikeium, respectively.

Melting curve analysis.

The LC640 probe is detected in channel 1 of the LightCycler (measured using F2/F1 color compensation) and the LC705 probe in channel 2 (measured using F3/F2 color compensation). In the melting curve analysis (Table 1), most species tested yielded a T_m only in channel 1. Characteristic T_m values were found for some species (e.g., 64°C for C. amycolatum). C. jeikeium and the closely related C. urealyticum yielded a T_m in channel 2 only. Only C. aurimucosum yielded T_m values in both channels, but with a T_m different from that of C. jeikeium in channel 2, allowing differentiation. C. aurimucosum did not yield a crossing point in channel 2 using quantification curve analysis.

Interpretative criteria.

Interpretative criteria were established by assessment of the bacteria shown in Tables 1 and 2. A T_m of >60°C in only channel 2, in conjunction with a quantification curve, was considered positive for C. jeikeium or the closely related C. urealyticum; the absence of a quantification curve or the presence of a quantification curve and the absence of a corresponding T_m of >60°C was considered negative for C. jeikeium and C. urealyticum. A T_m in channel 1, in conjunction with a quantification curve, was considered positive for non-jeikeium/urealyticum Corynebacterium species. Certain species are differentiated by their T_ms. The absence of a T_m in both channels was considered negative for Corynebacterium species (with the caveat that C. accolens, C. durum, Corynebacterium group F1, and C. riegelii are not detected).

Clinical isolates from the site of PJI.

All isolates from the site of clinically defined PJI (5) were detected (Tables 4 and 5). IDRL-6128 yielded a T_m of >60°C in channel 2 only and was identified as C. jeikeium by rpoB sequencing. IDRL-8271 yielded a T_m in both channels (T_m of <60°C in channel 2) and was identified by partial 16S rRNA gene sequencing as C. aurimucosum (rpoB sequencing failed). The remaining isolates had a T_m in channel 1 only. Three (IDRL-6031, -6110, and -6281) had a T_m of 64°C and were identified as C. amycolatum, and one (IDRL-7596) had a T_m of 55°C and was identified as Corynebacterium propinquum. IDRL-7734 yielded an unexpected T_m of 52°C and was identified as Corynebacterium simulans by rpoB sequencing. Finally, IDRL-6330 yielded a T_m of 66°C; rpoB sequencing failed, and partial 16S rRNA gene sequencing showed 100% and 99.7% homology with two unnamed Corynebacterium sp. strains, Corynebacterium sp. strain 96447 (6) and Corynebacterium sp. strain 3301750 (11), respectively (Table 5). All results were therefore as expected (Table 1).

Table 4.

Clinical characteristics and culture results for nine Corynebacterium species isolated from the site of hip and knee PJI

Isolate no.a	Source	Criteria for PJI (acute inflammation/visible purulence/sinus tract)	Antibiotics (MIC, μg/ml)	Tissue culture (no. of culture-positive tissues/no. of tissues cultured, organism detected)	Sonicate fluid cultureb	Synovial fluid culture	Colony morphology
6031	Hip	NDc/+/−	Penicillin (>8), ciprofloxacin (>2), minocycline (1), TMP-SMXd (>2-38)	3/4, Corynebacterium sp.	ND	ND	White, tiny
6110	Knee	+/+/−	Penicillin (1), cefazolin (8), levofloxacin (>4), minocycline (1), TMP-SMX (>2-38), vancomycin (2)	1/2, Corynebacterium sp.	ND	ND	White, tiny
6128	Hip	−/+/−	Penicillin (8), cefazolin (8), levofloxacin (2), minocycline (1)	2/5, C. jeikeium	ND	ND	Translucent, tiny
6281	Knee	+/−/−	Penicillin (2), cefazolin (8), erythromycin (>4), levofloxacin (>4), vancomycin (2)	2/4, Corynebacterium sp.; 1/4, Enterococcus sp.	Corynebacterium sp. (probable contaminant)	ND	White, tiny, dry
6330	Knee	+/+/−	ND	0/3, Corynebacterium sp.; 1/3, SCNe	Corynebacterium sp., S. lugdunensis	SCN	White, tiny, slow growing
7065	Knee	+/−/−	ND	0/4	Corynebacterium sp.	ND	Translucent, tiny, creamy
7596	Knee	ND/+/+	ND	1/5, Corynebacterium sp.; 2/5, yeast	Corynebacterium sp.	−	White, tiny
7734	Knee	+/+/−	Penicillin (1), vancomycin (≤2)	4/6, Corynebacterium sp.	Corynebacterium sp.	Coryne. sp.	White, small, dry
8271	Knee	ND/−/+	Penicillin (1), ceftriaxone (2), vancomycin (≤1)	1/5, Corynebacterium sp., 5/5, Escherichia coli, SCN	Corynebacterium sp., E. coli, SCN	ND	White, large (2 mm)

^aIsolates from monomicrobial Corynebacterium PJI cases are in bold.

^bImplant sonication as described by Trampuz et al. (17).

^cND, not done.

^dTMP-SMX, trimethoprim-sulfamethoxazole.

^eSCN, coagulase-negative Staphylococcus species.

Table 5.

Molecular identification of Corynebacterium species isolated from the site of hip and knee PJI

Isolate no.	Real-time PCR melting temp (°C)		Sequence (% identity)
	Channel 1a	Channel 2b	rpoB	16S rRNA gene
6031	+64	−	C. amycolatum (99)
6110	+64	−	C. amycolatum (99)
6128	−	+61	C. jeikeium (95)
6281	+64	−	C. amycolatum (99)
6330	+66	−	−	Corynebacterium sp. strain 96447 (100) (6), Corynebacterium sp. strain 3301750 (99.7) (11)
7065	+62, 68	−	C. afermentans (97)
7596	+55	−	C. propinquum (98)
7734	+52	−	C. simulans (98)
8271	+56	+54	−	C. aurimucosum (99)

^aMeasured using F2/F1 color compensation.

^bMeasured using F3/F2 color compensation.

Four of the 9 isolates (IDRL-6031, -6128, -6281, and -7734) were isolated from at least two tissue cultures, and these were all monomicrobial infections. For the remaining isolates, whether they were the cause of the PJI with which they were associated is unknown; some may be contaminants (Table 4). Most isolates were susceptible to the antimicrobial agents tested, with some exceptions. C. amycolatum IDRL-6281 was intermediate to penicillin and resistant to cefazolin, erythromycin, clindamycin, and levofloxacin; C. jeikeium IDRL-6128 was resistant to penicillin and cefazolin and intermediate to levofloxacin (Table 4).

DISCUSSION

Corynebacterium genus-specific real-time PCR allows more rapid identification than API Coryne or sequencing-based approaches. We are not aware of another Corynebacterium-specific real-time PCR assay. We developed an assay that detects the most common human Corynebacterium species, with a specific focus on PJI. Based on our review of the literature and clinical laboratory experience, we focused on five species. A novel polyphosphate kinase was targeted, and three probes with two detection fluorophores were incorporated to differentiate C. jeikeium (and C. urealyticum) from other Corynebacterium species.

There are only a few published studies about Corynebacterium and PJI, most of which were performed before the identification of recently described Corynebacterium species. In 2004, Roux et al. published a study of Corynebacterium species isolated from bone and joint infections (13). Of the 31 patients reported, 8 presented with prosthetic joint infection (2 each with C. amycolatum and C. striatum, 3 with C. aurimucosum, and 1 with C. jeikeium). In 1998, von Graevenitz and al. published a study analyzing 60 patients presenting with PJI or open fracture infection (19). Seventy-three coryneform bacteria were identified to the species level; the most frequent species were C. amycolatum, C. striatum, and C. jeikeium. Nine isolates were considered clinically significant as sole agents of PJI (4 C. striatum, 3 C. amycolatum, and 1 each C. jeikeium and Corynebacterium species). At the time, C. aurimucosum and C. tuberculostearicum had not been described (they were described in 2002 [21] and 2004 [7], respectively). In both studies, C. amycolatum was highlighted as a cause of PJI.

A retrospective analysis of Corynebacterium species isolated at our institution from the site of hip or knee PJI was performed. Nine available isolates were analyzed with the described assay, all were detected, and the sole C. jeikeium isolate was correctly identified. This study confirms Corynebacterium species as causes of PJI. Corynebacterium PJI is rare. Four monomicrobial infection cases were found, which is unusual, with five case reports (1, 16, 20, 22) and the cases in the von Graevenitz study (19), previously describing this entity. We describe a case of C. jeikeium PJI, which has been previously reported (16); as expected, this isolate was resistant to penicillin and cefazolin. Two cases of C. amycolatum PJI are also described; this commensal flora organism has been previously described as a cause of PJI (13, 19). Finally, we describe one case of C. simulans PJI which, to best of our knowledge, has not been previously described. The isolate was susceptible to penicillin and vancomycin, and the patient's outcome was favorable following two-stage implant exchange and a course of intravenous ertapenem. Overall, C. amycolatum was identified in three cases. We also identified C. propinquum in one case, with the same T_m (55°C) as the closely related species C. pseudodiphtheriticum (11) (Table 1). C. propinquum and C. pseudodiphtheriticum are commensal flora and/or pathogens of the respiratory tract, especially the oropharynx. Finally, for IDRL-6330, rpoB sequencing failed and 16S rRNA gene sequencing showed 100 and 99.7% homology with two unnamed Corynebacterium sp. strains, Corynebacterium sp. strain 96447 (6) and Corynebacterium sp. strain 3301750 (11), respectively. This isolate may represent a yet-to-be described species.

This study highlights the need for antimicrobial susceptibility testing on clinically significant Corynebacterium species. C. jeikeium is not the only penicillin-resistant species; furthermore, some C. jeikeium strains lack penicillin resistance (9).

Human microbiome studies are now frequent in the literature. Gao et al. analyzed superficial skin bacterial biota of the human forearm and found that the Corynebacterium genus was one of the most frequent genera present, with C. tuberculostearicum being the most frequent Corynebacterium species detected (10).

In conclusion, we designed a Corynebacterium real-time PCR assay which detects Corynebacterium species frequently described in human pathology and specifically in PJI. This assay is rapid and specific and can differentiate C. jeikeium and C. urealyticum, which are usually resistant to multiple antibiotics, from other species. Our analysis of clinical isolates from the site of PJI (all detected by our assay) and review of the literature highlight two points: the need for identification and antimicrobial susceptibility testing of clinically significant Corynebacterium isolates.