Genotypic and phenotypic relationships among methicillin-resistant Staphylococcus aureus from three multicentre bacteraemia studies

Abstract

Background

At a time when the molecular epidemiology of methicillin-resistant Staphylococcus aureus (MRSA) was changing, we sought to characterize several genotypic markers and glycopeptide susceptibility features of clinical isolates from patients with bacteraemia.

Methods

One hundred and sixty-eight MRSA bloodstream isolates obtained from three multicentre clinical trials were microbiologically and genotypically characterized.

Results

All isolates were susceptible to vancomycin (MIC ≤ 2 mg/L); 38% belonged to accessory gene regulator (agr) group I, 52% belonged to group II and 10% belonged to group III. Typing of the staphylococcal cassette chromosome mec (SCCmec) showed that 67% were type II and 33% were type IV. The agr group II polymorphism was associated with SCCmec II (P < 0.001). Fifty-three percent of SCCmec II and 27% of SCCmec IV isolates had vancomycin MICs ≥1 mg/L (P = 0.001). One hundred percent of agr II strains were predicted to be members of clonal complex 5. SCCmec II was the genetic marker most predictive of vancomycin MICs of ≥1 mg/L. SCCmec IV isolates were more likely to have vancomycin MICs ≤0.5 mg/L.

Conclusions

Given that SCCmec IV is a marker for a community-based organism for which less prior vancomycin exposure is predicted, we conclude that prior antibiotic exposure in agr group II organisms may account for their increased vancomycin MICs.

Introduction

Vancomycin has served as the cornerstone of therapy for serious methicillin-resistant Staphylococcus aureus (MRSA) infections for 50 years.¹ Despite the fact that microbiological resistance to vancomycin in S. aureus remains very rare, recent years have seen a shift upwards in vancomycin MICs (i.e. the ‘MIC creep’) within the susceptible range,^2–4 with consequential effects on vancomycin efficacy in MRSA bacteraemia and pneumonia.^5–8 In addition to microbiological susceptibility phenotype, certain genotypic markers may also serve as a predictor of vancomycin treatment failure in MRSA bacteraemia.^5,9 However, the relationship between the MRSA genotype and glycopeptide susceptibility in vitro has not been extensively studied. We evaluated a multicentre collection of 168 MRSA bloodstream isolates compiled from three prior clinical trials to further evaluate the relationship between the MRSA agr type, spa type, and SCCmec type and vancomycin susceptibility and to evaluate for MRSA strain differences between the trials.

Methods

Isolates were obtained as part of three multicentre clinical trials evaluating MRSA bacteraemia.^10–12 PCR was used to characterize mec cassettes and to characterize accessory gene regulator (agr) types as described previously.^13–15 spa X-repeat polymorphisms were determined by nucleotide sequencing, as described previously.^15,16 A semi-quantitative delta-haemolysin functional assay was performed to assess agr function.¹⁷ Vancomycin susceptibility testing was performed by CLSI microdilution methods and was evaluated for differences based on agr and SCCmec type (MIC ≤ 0.5 versus ≥1 mg/L).¹⁸ Ordinal data were compared using Kruskal–Wallis analysis of variance. Categorical data were compared using χ² or Fisher's exact test where appropriate. All statistical procedures were performed with Systat 11 (Systat Software Inc., Point Richmond, CA, USA).

Results

One hundred and sixty-eight S. aureus isolates from 168 unique patients were studied. All isolates were susceptible to vancomycin, with MICs of 0.25 mg/L (n = 1), 0.5 mg/L (n = 92), 1.0 mg/L (n = 68) and 2.0 mg/L (n = 7). The isolates consisted of 64 (38%) agr group I, 86 (52%) group II and 18 (10%) group III MRSA. One hundred and thirteen (67%) were staphylococcal cassette chromosome mec (SCCmec) II and 55 (33%) were SCCmec IV. Of the 164 isolates that were tested by spa typing, 83 (50%) were predictive of clonal complex 5.

The agr group II polymorphism was associated with the presence of SCCmec II (P < 0.001) (Figure 1); 73% (83/113) of SCCmec II were agr group II. Eighty-seven percent (48/55) of SCCmec IV were agr group I. All (100%) of the agr group II isolates were predicted to be of clonal complex 5.

Figure 1.

Relationship between accessory gene regulator (agr) type and SCCmec type in 163 MRSA bloodstream isolates. agr group I versus II, P < 0.001; agr group I versus III, P < 0.001; agr group II versus III, P = 0.004.

Vancomycin MICs were significantly higher among SCCmec II MRSA (P = 0.001) (Figure 2). Fifty-three percent of SCCmec II and 27% of SCCmec IV isolates had vancomycin MICs ≥1 mg/L. Statistically significant differences in vancomycin MICs were not noted between agr groups or spa types. Among the isolates showing vancomycin MICs of ≥1 mg/L, 49% were agr group II, 42% were agr group I and 33% were agr group III (P = 0.24). Thirty-nine of the 83 (47%) isolates whose spa typing predicted clonal complex 5 and 33 of 81 (41%) other clonal complex types had vancomycin MICs ≥1 mg/L (P = 0.661).

Figure 2.

Relationship between SCCmec type and vancomycin MIC values in 163 MRSA bloodstream isolates (P = 0.001).

The function at the agr locus was significantly more reduced among MRSA with higher MICs within the susceptible range (Figure 3). Decreased delta-haemolysin production was noted in 86%, 66% and 36% of MRSA with vancomycin MICs of 2, 1 and ≤0.5 mg/L, respectively (P < 0.001).

Figure 3.

Relationship between delta-haemolysin production (absent or present) and vancomycin MIC values for 163 MRSA bloodstream isolates (P < 0.001).

Analysis of the microbiological and genotypic properties of MRSA was segregated by the clinical studies from which they were obtained (Table 1). The earlier two studies enrolling patients in Phase III/IV protocols from 1998 to 2003 highly weighted towards persistent MRSA bacteraemia (>5 days) and inclusive of patients in renal failure showed a different spectrum of organisms from the randomized Phase III trial enrolling from 2002 to 2005 evaluating daptomycin versus comparator and exclusive of renal failure patients. When compared with the later daptomycin S. aureus bacteraemia trial, the earlier trials had MRSA with a lower percentage of agr group I strains (25% versus 51%) and SCCmec IV strains (88% versus 46%), and a significantly higher percentage of MRSA with vancomycin MICs ≥1 mg/L (77% versus 16%) and agr dysfunction (75% versus 24%).

Table 1.

Shifting molecular epidemiology of MRSA bacteraemia

Characteristic	Sample from studies 1 and 2 (n = 81)	Sample from study 3 (n = 87)	P value
Year MRSA isolated	1998–2003	2002–05
agr group, n (%)
I	20 (25%)	44 (51%)	0.001
II	49 (60%)	38 (44%)	0.029
III	12 (15%)	4 (5%)	0.024
SCCmec II, n (%)	71 (88%)	40 (46%)	<0.001
Vancomycin MIC ≥1 mg/L, n (%)	62 (77%)	14 (16%)	<0.001
δ-lysin 0/1, n (%)a	61 (75%)	21 (24%)	<0.001

Study 1: Moise-Broder et al., 2002. Study continued through year 2003.

Study 2: Moise et al., 2002.

Study 3: Fowler et al., 2006.

^aDelta-lysin scoring system: 0, absent; 1, diminished; 2, comparable to agr wild-type strain RN6607 (see reference 17); 3, increased; and 4, markedly increased.

Discussion

It is well understood that there is a differential response to the treatment of infection based on host, pathogen and antimicrobial selection. While appearing straightforward, antibiotic susceptibility as measured in vitro via the determination of an MIC in a clinical laboratory may be complicated by phenotypes too subtle to be detected by standard methods. For example, the heterogeneous nature of susceptibility to glycopeptides has resulted in a recently appreciated discordance between microbiological and clinical vancomycin resistance among serious S. aureus infections such as bacteraemia and pneumonia.¹⁹

In addition to vancomycin MIC, agr group II MRSA have also been linked to vancomycin treatment failure in one study,⁵ but not in another.⁹ With the understanding that different MRSA clones with specific genotypic characteristics may predominate in specific epidemiological settings with consequential differences in antimicrobial selection driving the development of reduced susceptibility, we sought to determine the relationships between the MRSA agr type, SCCmec type and vancomycin susceptibility.

We found that MRSA harbouring SCCmec II were more likely to have vancomycin MICs of ≥1 mg/L. SCCmec IV isolates were more likely to have vancomycin MICs ≤0.5 mg/L. SCCmec II predominated among agr group II strains and SCCmec IV among agr group I strains.

We noted a strong association between increased vancomycin MIC within the susceptible range and agr dysfunction shown by reduced delta-haemolysin activity. Loss of agr has been associated with the glycopeptide-intermediate S. aureus phenotype,¹⁷ glycopeptide tolerance^17,19 and prolonged bacteraemia.⁹ Since a vancomycin MIC of 2 mg/L has been associated with prolonged bacteraemia, this new observation is to be expected, based on prior data.

Given that SCCmec II has been a marker for a healthcare-associated organism with consequential vancomycin selection pressure, these data suggest that the prior finding of agr group II being associated with vancomycin treatment failure may reflect the fact that these clones have predominated in settings of antecedent vancomycin selection pressure. Consistent with this premise is that the agr group II, clonal complex 5 USA 100 MRSA clone includes the first US and Japanese VISAs and vancomycin-resistant S. aureus.^20,21 This inference is strengthened by the finding of another single-centre study, where agr group III MRSA clones predominate, that the agr group III genotype was associated with vancomycin treatment failure.⁹ Thus, differences in vancomycin response at the genetic level likely reflect microbiological differences rather than intrinsic differences in antibiotic susceptibility. However, the reasons for the establishment of different specific MRSA clones in different settings or even different hospitals are unknown and warrant further study.

In addition to spatial differences in MRSA susceptibility, an evaluation of the microbiological and phenotypic properties of the 168 bacteraemia isolates as they sort out by the previous studies from which they were derived shows temporal shifts in MRSA as well. Collectively, the majority (52%) of MRSA in this study belonged to clonal complex 5 (agr group II). However, this genotype was not evenly distributed between the three studies. In the first two studies, where patients were derived from Phase III/IV linezolid and quinupristin/dalfopristin protocols and enriched for patients with prolonged bacteraemia on vancomycin, isolates were heavily weighted towards agr group II (60%), contain SCCmec II, have vancomycin MICs of at least 1 mg/L and have reduced agr function. In the third study, where isolates were more contemporary and excluded patients with renal dysfunction, MRSA were much more likely to be agr group I, contain SCCmec IV and have lower vancomycin MICs and preserved agr function. Only 44% of MRSA from the third study are agr group II. These latter attributes are features of MRSA that have generally been seen in community-onset infections. Nevertheless, the ‘hospital’ and ‘community’ labels associated with specific MRSA genotypes are certain to break down over time, likely in a fashion similar to that of S. aureus with penicillin resistance over the past decades.

The findings from this multicentre study are in agreement with the recent single-centre study from Detroit by Chua et al.,²⁰ showing agr group II MRSA to be associated with hospital-onset infection, SCCmec II and higher vancomycin MICs, and showing agr group I strains to be associated with community-onset infection, SCCmec IV and lower vancomycin MICs. Many agr group I SCCmec IV MRSA have recently been seen in healthcare-onset infections.

In summary, these findings highlight important genotypic and phenotypic characteristics of MRSA bloodstream isolates as they relate to one another, with SCCmec II being the strongest predictor of vancomycin MIC at the upper limit of the susceptibility range. Prior vancomycin exposure in CC5 agr II SCCmec II organisms may account for their higher vancomycin MICs. Currently, the molecular epidemiology of MRSA is a moving target, both geographically and temporally, and therefore, considerable strain heterogeneity may be found among different clinical studies. Therefore, generalizability of findings from clinical trials on MRSA bacteraemia to individual clinical centres needs to be done with caution.

Funding

This study was not funded by external sources.

Transparency declarations

P. A. M. is currently employed by Cubist. G. S. has received research funding from Cubist and Pfizer, has been a consultant for Cubist, Pfizer, Targanta and Ortho McNeil and has been on the speakers' bureau for Cubist, Pfizer and Wyeth. All other authors (D. S. S., D. A. R., N. E.-F. and C. M.) have no conflicts to disclose.