Decreased Susceptibility to Teicoplanin and Vancomycin in Coagulase-Negative Staphylococci Isolated from Orthopedic-Device-Associated Infections

Abstract

We studied 315 coagulase-negative Staphylococcus strains recovered prospectively during 240 surgical procedures (206 subjects) from proven or suspected device-associated bone and joint infections. Sixteen strains (5.1%) had decreased susceptibility to glycopeptides: 15 (12 S. epidermidis strains, 2 S. capitis strains, and 1 S. haemolyticus strain) to teicoplanin alone (MIC of 16 mg/liter, n = 9; MIC of 32 mg/liter, n = 6) and one (S. epidermidis) to both teicoplanin and vancomycin (MIC, 16 and 8 mg/liter, respectively). Decreased susceptibility to teicoplanin was more prevalent in “infecting” strains (i.e., strains recovered from ≥2 distinct intraoperative samples) than in “contaminants” (i.e., strains not fulfilling this criterion) (8.1% [12/149] versus 2.4% [4/166], respectively [P = 0.022]). One hundred percent (13/13) of S. epidermidis strains with decreased susceptibility to teicoplanin were resistant to methicillin (versus 112/173 [64.7%] for S. epidermidis strains susceptible to teicoplanin; P = 0.021).

Coagulase-negative staphylococci (CoNS), especially Staphylococcus epidermidis, are major nosocomial pathogens causing a variety of device-related infections in humans (25). In most hospitals, 60 to 70% of CoNS isolates are resistant to methicillin (20). This leads to the frequent use of glycopeptide antibiotics for treatment of CoNS infections (6, 19, 21). The first reports of CoNS with decreased susceptibility to glycopeptides were published in the United States in the late 1970s (13) and shortly afterwards in European countries (2, 11, 21). These strains usually show resistance to methicillin and most other antistaphylococcal agents (19), with the exception of the new agents tigecycline, linezolid, and daptomycin (5). Unlike that of Staphylococcus aureus (14, 18), CoNS resistance to glycopeptides applies almost exclusively to teicoplanin (15, 21, 24). The mechanisms involved are unclear, but CoNS with decreased susceptibility to glycopeptides show cell wall thickening and tend to form cellular aggregates (7, 10, 15).

Several studies have reported an increase in the prevalence of CoNS with decreased susceptibility to glycopeptides (8, 24). However, these studies included mainly strains from patients with CoNS bacteremia, and little information about the initial source of infection was provided. To our knowledge, no prospective study associated with orthopedic surgery has been carried out. Resistance of CoNS to glycopeptides may, however, be a major problem in this context. CoNS are the main cause of prosthetic joint infections and other device-related bone and joint infections (BJIs) (12, 16). Most strains are resistant to methicillin and other drugs commonly used for the treatment of BJIs, leading to an increase in the use of glycopeptides (23). It is thus important to know the current prevalence of CoNS strains with decreased glycopeptide susceptibility in orthopedic surgery. This was the aim of this 3-year surveillance study.

We carried out a prospective study of all CoNS isolates recovered intraoperatively from cases or suspected cases of BJI between January 2003 and December 2005 in the Orthopedic Department of the Raymond Poincaré hospital (Garches, France), a national reference center for the management of BJIs in France. Tissue samples obtained intraoperatively were analyzed microbiologically as previously described (16). We included only CoNS isolates from patients with at least three independent samples collected during the same surgical procedure. Patients could be included several times if the above-mentioned criteria were fulfilled. Mixed infections (e.g., isolation of infecting organisms other than CoNS) were not excluded. The “CoNS group” was identified using Gram staining and the following tests: catalase, Slidex latex agglutination (bioMérieux), and tube coagulase (Bio-Rad, Marnes la Coquette, France). CoNS isolates were then identified to the species level by partial sodA sequencing, as described elsewhere (17). The susceptibilities of strains to 15 antibiotics were determined by the disk diffusion method on Mueller-Hinton agar (Bio-Rad) (4) using the following disks: penicillin (6 μg), oxacillin (5 μg), ofloxacin (5 μg), erythromycin (15 μg), lincomycin (15 μg), pristinamycin (15 μg), rifampin (30 μg), gentamicin (10 IU), kanamycin (30 μg), tobramycin (10 μg), tetracycline (30 μg), trimethoprim-sulfamethoxazole (1.25 and 23.75 μg, respectively), fosfomycin (50 μg), vancomycin (30 μg), and teicoplanin (30 μg). Results were interpreted according to the breakpoints of the Comité de l'Antibiogramme de la Société Française de Microbiologie (CA-SFM) (4). Susceptibility to oxacillin was used for determining methicillin resistance; strains with inconclusive responses were further tested by mecA detection, performed as described previously (9). CoNS isolates recovered from various samples but belonging to the same species (identical partial sodA sequences) were deemed to be the same strain if they had the same colony morphology and identical antibiotic susceptibility patterns (1). A strain was defined as an “infecting strain” if it was recovered from ≥2 distinct intraoperative samples (22); a strain not fulfilling this criterion was defined as a “contaminant.” All CoNS strains were cryopreserved after initial culture.

Strains with decreased susceptibility to glycopeptides were detected by screening with an Etest (teicoplanin and vancomycin Etest strips; AB Biodisk Dalvägen, Solna, Sweden). Mueller-Hinton agar plates were inoculated by swabbing (Difco Laboratories, Cockeysville, MD) with a 0.5 McFarland standard bacterial suspension and incubated at 37°C for 24 h. All strains for which the MIC of vancomycin and/or teicoplanin was ≥4 mg/liter with the Etest were further assessed using the 2-fold agar dilution method, with teicoplanin and vancomycin concentrations ranging from 0.25 to 64 mg/liter. Mueller-Hinton agar plates were inoculated with 2 μl (10⁴ CFU) of each strain by use of a Steers replicator and incubated for 24 h at 37°C. The reference strains S. aureus ATCC 25923 and S. epidermidis CIP 105777 (ATCC 35984) were controls. The MIC value was defined as the lowest concentration of antibiotic that completely inhibited bacterial growth. Vancomycin and teicoplanin MIC values were interpreted according to the 2006 recommendations of the Clinical and Laboratory Standards Institute (CLSI) (3), which classify CoNS strains as “susceptible,” “intermediate,” or “resistant” as follows: (i) for teicoplanin, MIC values of ≤8 mg/liter, 16 mg/liter, and ≥32 mg/liter, respectively, and (ii) for vancomycin, MIC values of ≤4 mg/liter, 8 to 16 mg/liter, and ≥32 mg/liter, respectively. A strain with decreased susceptibility to glycopeptides was any strain considered to be “intermediate” or “resistant” to teicoplanin and/or vancomycin according to the above-described definitions. The chi-square test (with Yates' correction for expected frequencies of <5) was used for comparisons, and P values of <0.05 were considered to be statistically significant.

A total of 240 surgical procedures (206 patients) yielding CoNS were included prospectively. The patients (153 males and 53 females) were between 19 and 96 years old (mean age, 56.6 years). Of these patients, 177 underwent a single procedure, 26 underwent two procedures, 1 underwent three procedures, and 2 underwent four procedures. Clinical data were available for 170 surgical procedures: 103 (60.6%) were revision arthroplasty surgery (hip, 80; knee, 23), and 67 (39.4%) were procedures related to other BJIs. A total of 315 CoNS strains were recovered intraoperatively: 186 S. epidermidis strains (59.0%), 48 S. capitis strains (15.2%), 27 S. warneri strains (8.6%), 15 S. lugdunensis strains (4.8%), 11 S. hominis strains (3.5%), 7 S. haemolyticus strains (2.2%), 7 S. caprae strains (2.2%), 4 S. pasteuri strains (1.3%), 4 S. simulans strains (1.3%), 3 S. pettenkoferi strains (1.0%), 1 S. schleiferi strain (0.3%), 1 S. cohnii strain (0.3%), and 1 S. pseudintermedius strain (0.3%). Using the Etest, we found that 147 strains had a MIC of ≥4 mg/liter for vancomycin and/or teicoplanin. We used the agar dilution method to further assess these 147 strains: 131 were susceptible to both teicoplanin (MIC of 1 mg/liter, n = 2; MIC of 2 mg/liter, n = 11; MIC of 4 mg/liter, n = 54; MIC of 8 mg/liter, n = 64) and vancomycin (MIC of 1 mg/liter, n = 10; MIC of 2 mg/liter, n = 88; MIC of 4 mg/liter, n = 33), 15 had decreased susceptibility to teicoplanin only (MIC of 16 mg/liter, n = 9; MIC of 32 mg/liter, n = 6), and one (S. epidermidis) had decreased susceptibility to both teicoplanin (MIC of 16 mg/liter) and vancomycin (MIC of 8 mg/liter). The MIC of teicoplanin was ≥8 mg/liter according to the Etest for the 16 strains with decreased susceptibility to teicoplanin with the agar dilution method (data not shown).

Thus, 16 (5.1%) of the 315 CoNS strains studied had decreased susceptibility to teicoplanin, i.e., 10 (3.2%) intermediate strains (9 S. epidermidis strains and 1 S. haemolyticus strain) and 6 (1.9%) resistant strains (4 S. epidermidis strains and 2 S. capitis strains) (Table 1). Decreased susceptibility to teicoplanin was more prevalent in infecting strains than in contaminants (12/149 [8.1%] versus 4/166 [2.4%], respectively [P = 0.022]). Among the 13 S. epidermidis strains with decreased susceptibility to teicoplanin, 13 (100%) were resistant to oxacillin (versus 112/173 [64.7%] for S. epidermidis strains susceptible to teicoplanin [P = 0.021]) (Table 2). Other resistance markers significantly associated with decreased susceptibility to teicoplanin in the subset of S. epidermidis strains included resistance to lincomycin, rifampin, and fosfomycin (Table 2). We did not test the new antimicrobial agents tigecycline, linezolid, and daptomycin. However, studies of the susceptibility to these newer agents of CoNS with decreased susceptibility to glycopeptides show a reported MIC₉₀ close to the CLSI breakpoints (5). Given the pharmacodynamic profile of the periprosthetic tissue, the determination of the MIC could be advised prior to the initiation of a treatment using these molecules.

TABLE 1.

Prevalence of strains with decreased susceptibility to teicoplanin

Group of strains (total no.)	No. (%) of strains
	Ib	Rb	I or Rb
All strains (315)	10 (3.2)	6 (1.9)	16 (5.1)
S. epidermidis (186)	9 (4.8)	4 (2.2)	13 (7.0)
Other CoNS (129)	1c (0.8)	2d (1.6)	3 (2.3)
Infecting strainsa (149)	7 (4.7)	5 (3.4)	12 (8.1)e
Contaminants (166)	3 (1.8)	1 (0.6)	4 (2.4)e
Resistant to methicillin (149)	10 (6.7)	6 (4.0)	16 (10.7)
Susceptible to methicillin (166)	0 (0)	0 (0)	0 (0)

^aRecovered from ≥2 distinct positive samples.

^bIntermediate (I), MIC of 16 mg/liter; resistant (R), MIC of ≥32 mg/liter; I or R (decreased susceptibility to teicoplanin), MIC of ≥16 mg/liter.

^cS. haemolyticus.

^dTwo S. capitis strains (two subjects).

^eP = 0.022 for infecting strains versus contaminants.

TABLE 2.

Resistance markers associated with decreased susceptibility to teicoplanin in S. epidermidis strains

Antibiotic	No. (%) of resistant strainsa		P value
	I/R to teicoplanin (n = 13)	S to teicoplanin (n = 173)
Methicillin	13 (100)	112 (64.7)	0.021
Ofloxacin	10 (76.9)	88 (50.9)	0.069
Gentamicin	7 (53.8)	66 (38.1)	0.263
Erythromycin	9 (69.2)	75 (43.3)	0.070
Lincomycin	7 (53.8)	35 (20.2)	0.010
Pristinamycin	2 (15.4)	8 (4.6)	0.307
Rifampin	9 (69.2)	54 (31.2)	0.001
Tetracycline	4 (30.8)	27 (15.6)	0.303
Fosfomycin	7 (53.8)	42 (24.3)	0.044
Fusidic acid	7 (53.8)	91 (52.6)	0.928

^aIntermediate (I) and resistant (R) strains were determined according to the 2009 recommendations of the CA-SFM committee (4). S, susceptible.

The orthopedic surgery department of Garches hospital is a French referral center for the management of patients with BJIs. Patients are mainly from the greater Paris area but also come from other regions of France. Our data are thus a good indication of the current situation regarding sepsis in orthopedics in France. We found a relatively low prevalence of resistance to glycopeptides in the CoNS strains studied, with 5.1% of strains showing decreased susceptibility to teicoplanin and 1.9% being resistant; only one strain had decreased susceptibility to both teicoplanin and vancomycin (MIC, 16 and 8 mg/liter, respectively). These data are similar to those from an Italian study of bacteremia caused by CoNS, which reported that the percentage of strains resistant to glycopeptides was 2% in surgical wards (21). However, our figures are considerably lower than those reported in recent European studies, which demonstrated an alarming increase in resistance to teicoplanin in CoNS strains since the beginning of this century. In a study of 1,337 CoNS isolates recovered from 2001 to 2004 from patients with bacteremia in Greece, the prevalence of teicoplanin-resistant strains increased from 0% in 2001 and 2002 to 6.4% in 2003 and 2004 (8). In a recent retrospective French study of 1,039 CoNS isolates recovered in routine practice from 2000 to 2004, the prevalence of decreased susceptibility to teicoplanin was 7.2% in 2000, was 17.2% in 2001, and exceeded 30% in 2002, 2003, and 2004 (24). These discrepancies may be explained by differences in nature of specimens, clinical context of included patients, and infectiousness of the isolated strains. For example, CoNS resistance to glycopeptides has been shown to be four times higher in intensive care units than in surgical wards (21).

The results of our study should not lead to a false sense of security and less-vigilant surveillance of the prevalence of resistance to glycopeptides, in particular to teicoplanin, in CoNS associated with orthopedic surgery. On the one hand, our data show that the resistance to teicoplanin is 3.4 times more frequent in infecting strains than in contaminants. The prevalence of teicoplanin resistance thus seems much higher in truly pathogenic strains, similar to that reported by Tacconelli et al. to occur in intensive care units (21). On the other hand, the CoNS population is clearly shifting toward greater resistance to glycopeptides, probably as a result of pressure due to the increase in use of these molecules in recent years (19). This shift is particularly marked with S. epidermidis, the main cause of device-related BJIs worldwide (12). We found that 100% of S. epidermidis strains with decreased susceptibility to teicoplanin were resistant to methicillin and that this resistance was frequently associated with other markers, such as resistance to lincomycin, rifampin, or fosfomycin. These results are of particular concern, as rifampin and lincomycin-clindamycin are used widely for treatment of CoNS BJIs. Moreover, they suggest the selection of subclones less susceptible to glycopeptides than the multiresistant S. epidermidis clones currently circulating in hospitals. Genotyping studies using pulsed-field gel electrophoresis (15) or ribotyping (21) have shown a broad diversity of S. epidermidis strains with decreased susceptibility to glycopeptides. These findings need to be confirmed in the particular context of device-associated BJIs by using techniques, such as multilocus sequence typing, that allow better analysis of the relationship between the bacterial populations involved.