Heterogeneous Vancomycin-Intermediate Susceptibility Phenotype in Bloodstream Methicillin-Resistant Staphylococcus aureus Isolates from an International Cohort of Patients with Infective Endocarditis: Prevalence, Genotype, and Clinical Significance

Abstract

Background

The significance of heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) is unknown. Using a multinational collection of isolates from methicillin-resistant S. aureus (MRSA) infective endocarditis (IE), we characterized IE patients with and without hVISA, and genotyped the infecting strains.

Methods

MRSA bloodstream isolates from 65 patients with definite IE from 8 countries underwent PCR for 31 virulence genes, pulsed-field gel electrophoresis, and multilocus sequence typing. hVISA was defined using population analysis profiling (PAP).

Results

Nineteen (29.2%) of 65 MRSA IE isolates exhibited hVISA by PAP. Isolates from Oceania and Europe were more likely to exhibit hVISA than isolates from the United States (77.8% vs. 35.0% vs. 13.9%; P < .001). The prevalence of hVISA was higher among isolates with a vancomycin minimum inhibitory concentration of 2 mg/L (P = .026). hVISA-infected patients were more likely to have persistent bacteremia (68.4% vs. 37.0%; P = .029) and heart failure (47.4% vs. 19.6%; P = .033). Mortality of hVISA- and non-hVISA-infected patients did not differ (42.1% vs. 34.8%, P = .586). hVISA and non-hVISA isolates were genotypically similar.

Conclusions

In these analyses, hVISA occurred in over one-quarter of MRSA IE isolates, was associated with certain IE complications, and varied in frequency by geographic region.

INTRODUCTION

Infections caused by heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) are poorly understood [1]. Defined as the presence of subpopulations (typically at a rate of 1 organism per 10⁵–10⁶) of MRSA with intermediate vancomycin resistance [2], the hVISA phenotype has now been reported in clinical MRSA isolates from many parts of the world [3–7]. However, fundamental questions involving hVISA - including its prevalence, global distribution, and clinical significance - persist.

Nowhere is the issue of antibiotic efficacy more important than in infective endocarditis (IE) caused by MRSA. Vancomycin - often out of necessity rather than choice - remains the first-line treatment for this condition. However, vancomycin’s efficacy is hindered by its slowly bactericidal mechanism of action and poor penetration into valvular vegetations [8, 9]. Furthermore, several studies have reported an association between the hVISA phenotype and deep-seated endovascular infections such as IE [3, 7, 10, 11]. Because of these observations, we reasoned that patients with MRSA IE would be at particularly high risk for the presence of the hVISA phenotype, and that bloodstream isolates from such patients would be ideal to address many of the unresolved issues regarding the prevalence, global distribution, and clinical significance of hVISA. However, the relative infrequency of MRSA IE in any single center made a study focusing upon this infection impractical until now.

The current study had two major objectives: 1) to define the prevalence and clinical significance of the hVISA phenotype in MRSA IE; and 2) to evaluate the genotypic characteristics of MRSA IE bloodstream isolates exhibiting the hVISA phenotype. To address these issues, we used MRSA bloodstream isolates obtained from a large, contemporary, international cohort of prospectively-identified and clinically well-characterized patients with definite IE.

PATIENTS AND METHODS

Patients and settings

Data from the International Collaboration on Endocarditis-Prospective Cohort Study (ICE-PCS) were used for this study. The ICE-PCS Microbiological Repository contains more than 1,500 bloodstream isolates of bacteria obtained from prospectively-identified patients with definite IE from more than 16 countries between November 13, 1999 and January 20, 2006 [12, 13]. For the current study, bloodstream isolates from all patients with definite MRSA IE were eligible for inclusion.

Definition

Definite IE was defined according to modified Duke criteria [14]. Chronic immunosuppressive therapy was defined as the administration of recognized immunosuppressive agents (including oral corticosteroids or other agents such as those used in solid organ transplantation or rheumatologic disorders) for more than 30 days at the time of IE diagnosis. A cardiac device was defined as a permanent pacemaker, cardioverter-defibrillator, or prosthetic cardiac valve. Intravascular access devices were defined as an arterial-venous fistula or an indwelling vascular catheter. A chronic indwelling central catheter was defined as a tunneled or cuffed catheter, or a subcutaneous port catheter. A stroke was defined as an acute neurological deficit of vascular etiology lasting more than 24 hours [15]. Vascular/immunologic evidence of endocarditis was defined as the presence of physical examination findings of endocarditis present on initial evaluation, including Osler’s nodes, Janeway lesions, Roth spots, conjunctival hemorrhage, splinter hemorrhages, or peripheral vascular embolic events. Congestive heart failure was defined according to the New York Heart Association classification system [16]. Persistent bacteremia was defined as >3 days of bacteremia despite receipt of an antibiotic to which the isolate was susceptible in vitro. Health care associated IE was defined as either nosocomial infection or non-nosocomial health care associated infection. Nosocomial infection was defined as IE developing in a patient hospitalized for more than 48 hours prior to the onset of signs/symptoms consistent with IE. Non-nosocomial health care associated infection was defined as IE diagnosed within 48 hours of admission in an outpatient with extensive health care contact as reflected by any of the following criteria: (1) received intravenous therapy, wound care, or specialized nursing care at home within the 30 days prior to the onset of S aureus IE; (2) attended a hospital or hemodialysis clinic or received intravenous chemotherapy within the 30 days before the onset of S. aureus IE; (3) was hospitalized in an acute care hospital for 2 or more days in the 90 days before the onset of S aureus IE; or (4) resided in a nursing home or long-term care facility [17]. The remaining clinical, echocardiographic, and outcome variables were defined as reported elsewhere [12].

Laboratory methods and susceptibility definition

All S. aureus isolates were identified with use of standard methods. The minimum inhibitory concentration (MIC) of vancomycin was determined by E-test (AB Biodisk, Solna, Sweden) according to the manufacturer’s instructions, and by broth microdilution method according to manufacturer protocol (Siemens, Berlin, Germany) and Clinical and Laboratory Standards Institute recommendations [18]. Except where noted, all vancomycin MIC values in the text were defined according to E-test. Reference strain ATCC 29213 was used as a positive hVISA control strain. The MIC of daptomycin was determined using E-test according to the manufacturer’s instructions. All MICs were interpreted by a single blinded investigator (JFF).

Population Analysis Profiling

All MRSA isolates were assessed by population analysis profile-area under the curve (PAP-AUC) using the technique described by Wootton et al. [19]. Organisms were cultured in tryptone soy agar plate from an overnight growth, adjusted to a 10⁸ CFU/ml density, and spiral plated (Eddy jet spiral platers, NY, USA) onto brain-heart infusion agar (Difco, Detroit, MI, USA) plates containing 0, 0.5, 1, 1.5, 2, 3, 4, and 8 mg/L vancomycin. Colonies were counted after 48 hours incubation at 37°C and the viable count was plotted against vancomycin concentration using GraphPad Prism (GraphPad; San Diego, CA, USA). This was then used to calculate an area under the curve. Isolates were defined as hVISA if they had a PAP-AUC ratio of >0.9 and <1.3 when compared with Mu3 reference strain (ATCC 700698).

Multiplex PCR

Genomic DNA was prepared as previously described [20]. Bacterial determinants including adhesins, toxins, agr group I-IV and other genes were screened by multiplex PCR as described before [20, 21]. The isolates were classified as Staphylococcal chromosomal cassette mec (SCCmec) type I, II, III, or IV by multiplex PCR, as previously described [22].

Pulsed Field Gel Electrophoresis (PFGE)

PFGE with SmaI was performed on all isolates, and the gels analyzed using BioNumerics version 4.0 (Applied Maths, Kortrijk, Belgium) as previously described [23].

Multilocus Sequence Typing (MLST)

MLST was performed as described previously [24, 25]. PCR fragments of 7 housekeeping genes (arcC, aroE, glpF, gmk, pta, tpi, and yqiL) were obtained from chromosomal DNA and were directly sequenced. MLST allele names and sequence type (ST) were derived from http://www.mlst.net. Clonal complexes were assigned to groups of isolates sharing 6 of 7 alleles by use of eBURST (http://eburst.mlst.net).

Statistics

Simple descriptive statistics were used to describe the demographics and other characteristics of the patients and for the genetic and antibiotic sensitivity profile of the bacterial isolates. Statistics are presented as medians with 25^th and 75^th percentiles for continuous variables, and as frequency counts and percentages for categorical factors. The statistical significance of associations between variables was calculated using the Kruskal-Wallis test for continuous measures and with Fisher’s exact test for cross-classifications of categorical data. For all tests, a p-value of < 0.05 was considered statistically significant. All statistical analyses were performed using SAS, version 9.1.2 (SAS Institute, Cary, NC, USA).

RESULTS

Patient characteristics and outcome

Isolates from a total of 65 prospectively-enrolled patients from 13 centers in 8 countries with definite MRSA IE were available in this study. Of these, 19 (29.2%) exhibited the hVISA phenotype using the PAP-AUC method. Patients with IE due to hVISA were older (median age, 73.1 vs. 60.6 years; P = .037) (table 1), were more likely to be from Oceania or Europe (P < .001) (figure 1), and were more likely to have native valve IE (94.4% vs. 53.3%; P = .005). None of the 65 patients were injection drug users. Patients with hVISA had a higher rate of persistent bacteremia (68.4% vs. 37.0%; P = .029) and congestive heart failure (47.4% vs. 19.6%; P = .033). Other complications and inhospital mortality did not differ significantly between groups.

Table 1.

Clinical characteristics and outcome of 65 patients with infective endocarditis due to heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) and vancomycin-susceptible S. aureus (VSSA)

Characteristic	VSSA (n=46)	hVISA (n=19)	P
Patient demographics :
Male sex	26 (56.5)	9 (47.4)	0.589
Age, years, median (25th–75th percentiles)	60.6 (47.6–74.1)	73.1 (59.9–79.6)	0.037
Predisposing conditions :
Diabetes mellitus	20 (43.5)	8/18 (44.4)	1.00
Hemodialysis dependent	12 (26.1)	3 (15.8)	0.522
History of cancer	6 (13.0)	3 (15.8)	0.714
Chronic immunosuppressive therapy	8/45 (17.8)	3 (15.8)	1.00
Other chronic disease	34 (73.9)	10 (52.6)	0.144
Recent Invasive procedures	19 (41.3)	9 (47.4)	0.795
Presumed place of acquisition :			0.26
Nosocomial	18 (39.1)	12 (63.2)
Non-nosocomial health care-associated	18 (39.1)	5 (26.3)
Community-acquired	10 (21.7)	2 (10.5)
Type of infective endocarditis :			0.005
Native	24/45 (53.3)	17/18 (94.4)
Prosthetic	12/45 (26.7)	1/18 (5.6)
Other	9a/45 (20.0)	0/18 (0)
Clinical findings :
Fever, > 38°C	39/42 (92.9)	16/17 (94.1)	1.00
Vascular/ immunologic evidence of endocarditisb	9 (19.6)	7 (36.8)	0.20
New cardiac murmur	8 (17.4)	3 (15.8)	1.00
Septic pulmonary infarcts	2 (4.3)	2 (10.5)	0.574
Echocardiographic findings :
New regurgitation	18 (39.1)	10 (52.6)	0.411
Intracardiac vegetation :	43 (93.5)	14 (73.7)	0.041
Aortic and/or mitral valves	26 (60.5)	11 (78.6)
Tricuspid valve	7 (16.3)	2 (14.3)
Aortic/mitral plus tricuspid valves	1 (2.3)	1 (7.1)
Other	9a (20.9)	0 (0)
Treatment :
Glycopeptide therapyc	43d (93.5)	18 (94.7)	1.00
Cardiac surgery during this hospitalization	14 (30.4)	3 (15.8)	0.353
Complications and Outcome :
Stroke	9 (19.6)	4 (21.1)	1.00
Congestive heart failure	9 (19.6)	9 (47.4)	0.033
Intracardiac abscess	4 (8.7)	3 (15.8)	0.408
Persistent bacteremia	17 (37.0)	13 (68.4)	0.029
New conduction abnormalities	2/44 (4.5)	3 (15.8)	0.156
In-hospital death	16 (34.8)	8 (42.1)	0.586

NOTE. Data are no. (%) of patients, unless otherwise indicated.

^aIncludes infections involving other cardiac devices (7 patients), intracardiac catheter (1 patient), or other myocardial structure (1 patient).

^bphysical examination findings of endocarditis present on initial evaluation, including Osler’s nodes, Janeway lesions, Roth spots, conjunctival hemorrhage, splinter hemorrhages, or peripheral vascular embolic events.

^cTreatment details missing for 1 VSSA-infected patient and 1 hVISA-infected patient.

^dOne patient received fosfomycin.

Figure 1.

Prevalence by geographic region of heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) phenotype among MRSA bloodstream isolates from patients with infective endocarditis. Percentages of patients with hVISA phenotype are specified above each column.

Patient treatment

Rates of glycopeptide therapy did not differ between groups (43 of 46 [93.5%] VSSA-infected patients vs. 18 of 19 [94.7%] hVISA-infected patients, P = 1.00). Of the 43 VSSA-infected patients treated with glycopeptides, two received teicoplanin. All 18 hVISA-infected patients treated with glycopeptides received vancomycin. Twenty-two (51.2%) of 43 VSSA-infected patients treated with a glycopeptide also received combination antimicrobial therapy: vancomycin plus rifampin (7 patients), vancomycin plus rifampin plus aminoglycoside (6 patients), vancomycin plus aminoglycoside (3 patients), vancomycin plus trimethoprim-sulfamethoxazole (3 patients), vancomycin plus linezolid (1 patient), teicoplanin plus quinupristin-dalfopristin (1 patient), or fosfomycin plus aminoglycoside (1 patient). Eleven (61.1%) of 18 hVISA-infected patients receiving vancomycin also received rifampin (5 patients), aminoglycoside (3 patients), fosfomycin plus imipenem (2 patients), or linezolid (1 patient). No statistically significant differences between VSSA and hVISA–infected patients were noted in rates of surgical therapy (30.4% VSSA vs. 15.8% hVISA; P = .353) or mean time to surgical treatment (20.5 days VSSA vs. 6 days hVISA; Kruskal-Wallis P = .130). The frequency of patients transferred from another facility also did not differ between VSSA- and hVISA-infected patients (20/46 [43.5%] vs. 8/19 [42.1%]; P = 1.00).

Genotypic characteristics of hVISA Isolates

Several genes were highly conserved in both hVISA and VSSA isolates (table 2). For example, spa, ebps, hlg, and efb were present in all of the tested isolates. The hVISA isolates were significantly more likely to contain the genes for fnbA (47.4% vs. 15.2%; P = .011) and seh (21.1% vs. 2.1%; P = .023) than VSSA isolates. The frequency of agr type II polymorphism did not differ between the hVISA and VSSA isolates (P = .586). The distribution of SCCmec types of hVISA isolates was significantly different from that of VSSA isolates (P = .012), with SCCmec type III being more common in hVISA isolates (21.1% vs. 0%; P = .006).

Table 2.

Genotypic characteristics of heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) and vancomycin-susceptible S. aureus (VSSA) isolates from 65 patients with MRSA infective endocarditis.

Gene	VSSA (n=46)	hVISA (n=19)	P
Adhesins :
fnbA	7 (15.2)	9 (47.4)	0.011
fnbB	30 (65.2)	13 (68.4)	1.00
clfA	43 (93.5)	17 (89.5)	0.625
clfB	45 (97.8)	18 (94.7)	0.502
cna	21 (45.7)	12 (63.2)	0.277
spa	46 (100.0)	19 (100.0)	1.00
sdrC	34 (73.9)	17 (89.5)	0.203
sdrD	37 (80.4)	17 (89.5)	0.486
sdrE	34 (73.9)	17 (89.5)	0.203
bbp	43 (93.5)	17 (89.5)	0.625
ebps	46 (100.0)	19 (100.0)	1.00
map/eap	35 (76.1)	13 (68.4)	0.547
Toxins :
eta	25 (54.3)	9 (47.4)	0.785
etb	1 (2.2)	1 (5.3)	0.502
tst	30 (65.2)	9 (47.4)	0.266
sea	40 (87.0)	17 (89.5)	1.00
seb	1 (2.2)	3 (15.8)	0.072
sec	2 (4.3)	1 (5.3)	1.00
sed	14 (30.4)	3 (15.8)	0.353
see	33 (71.7)	16 (84.2)	0.357
seg	41 (89.1)	13 (68.4)	0.067
seh	1 (2.2)	4 (21.1)	0.023
sei	44 (95.7)	18 (94.7)	1.00
sej	13 (28.2)	2 (10.5)	0.196
pvl	6 (13.0)	0 (0.0)	0.169
hlg	46 (100.0)	19 (100.0)	1.00
Other putative virulence genes :
Efb	46 (100.0)	19 (100.0)	1.00
icaA	43 (93.5)	16 (84.2)	0.347
chp	34 (73.9)	11 (57.9)	0.244
V8	36 (78.3)	18 (94.7)	0.154
agr genotype :			0.061
Group I	6 (13.0)	7 (36.8)
Group II	30 (65.2)	11 (57.9)
Group III	10 (21.7)	1 (5.3)
SCCmec type :			0.012
Type I	6 (13.0)	3 (15.8)
Type II	29 (63.0)	7 (36.8)
Type III	0 (0.0)	4 (21.1)
Type IV	11 (23.9)	5 (26.3)
PFGE type (USA Nomenclature) :			0.077
USA 100	20 (43.5)	6 (31.6)
USA 200	9 (19.6)	1 (5.3)
USA 300	3 (6.5)	0
USA 400	1 (2.2)	0
USA 500	1 (2.2)	0
USA 600	2 (4.3)	0
USA 800	3 (6.5)	2 (10.5)
Undefined	7 (15.2)	10 (52.6)
Clonal complex :			0.186
CC 5	23 (50.0)	8 (42.1)
CC 8	5 (10.9)	6 (31.6)
CC 30	10 (21.7)	1 (5.3)
ST 228	4 (8.7)	3 (15.8)
Other STsa	4 (8.7)	1 (5.3)

NOTE. Data are no. (%) of isolates, unless otherwise indicated.

^ainclude 2 ST45, 1 ST1, and 1 ST111 in VSSA isolates and ST59 in an hVISA isolate.

Using MLST, all but 4 of the 65 isolates were included within one of four clonal complexes (CCs) or sequence types (STs): CC5 (31, 47.7%), CC8 (11, 16.9%), CC30 (11, 16.9%), and ST228 (7, 10.8%). The distribution of these groups did not differ significantly between the hVISA and non-hVISA groups. Among the 48 MRSA isolates with a typeable PFGE profile by USA typing schema, 26 isolates (40.0%) demonstrated a USA 100 pattern and 10 isolates (15.4%) had a USA 200 pattern (figure 2). Three isolates demonstrated the USA300 genotype; none exhibited the hVISA phenotype. The distribution of PFGE profiles among the hVISA isolates and VSSA isolates did not differ significantly.

Figure 2.

Dendrogram of pulsed-field gel electrophoresis profiles of 65 MRSA bloodstream isolates from patients with infective endocarditis. The clonal complex/sequence type, geographic region of origin, and heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) status of each isolate is also represented. CC = clonal complex; ST = sequence type as defined by multilocus sequence typing; Region = Geographic region of source patient; Group = vancomycin-susceptible S. aureus (VSSA) and hVISA.

hVISA, Vancomycin MIC, and Clinical Outcome

The hVISA phenotype was significantly associated with higher vancomycin MIC values determined by E-test (P = .026), with all but one of the hVISA isolates having a vancomycin MIC of ≥1.0 mg/L. In addition, the hVISA phenotype was present in 5 of the 6 isolates (80%) of isolates with a vancomycin MIC of 2 mg/L (table 3). To confirm this association between vancomycin MIC and hVISA phenotype, we repeated the analyses using MIC values as defined by the broth microdilution method. These repeat analyses again indicated a significant association between increasing vancomycin MIC and the presence of hVISA phenotype (Fisher’s Exact test, P = .0063) (table 3). All 65 MRSA isolates were susceptible to daptomycin, with a daptomycin MIC ≤ 1.0 mg/L by E-test (data not shown).

Table 3.

Percentage of heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) isolates by population analysis profile according to vancomycin minimum inhibitory concentration (MIC), as defined by E-test and Broth Microdilution. Overall p-value of association for presence of hVISA and increasing vancomycin MIC is 0.026 using E-test method and 0.0063 using Broth Microdilution method.

E-test
MIC of vancomycin, mg/L	hVISA/MRSA (%)a	VSSA (N = 46)	hVISA (N = 19)
0.5	0/3 (0)	3 (6.5)	0
0.75	1/11 (9.1)	10 (21.7)	1 (5.3)
1.0	4/14 (28.6)	10 (21.7)	4 (21.1)
1.5	9/31 (29.0)	22 (47.8)	9 (47.4)
2.0	5/6 (83.3)	1 (2.2)	5 (26.3)
Broth Microdilution
MIC of vancomycin, mg/L	hVISA/MRSA (%)a	VSSA (N = 46)	hVISA (N = 19)
<1.0	7/40 (17.5)	33 (71.7)	7 (36.8)
1.0	3/11 (27.3)	8 (17.4)	3 (15.8)
2.0	8/13 (61.5)	5 (10.9)	8 (42.1)
4.0	1/1 (100)	0	1 (5.3)

NOTE. Data are proportion (%) of isolates, unless otherwise indicated.

^aRatio and percentages of hVISA phenotype among MRSA isolates.

We also evaluated potential associations between clinical outcome and higher vancomycin MIC values. As shown in table 4, no significant differences in outcome were identified among patients with IE due to MRSA isolates with varying levels of vancomycin MICs.

Table 4.

Characteristics and outcome of 65 patients with MRSA infective endocarditis according to vancomycin E-test minimum inhibitory concentration (MIC) of the infecting isolate.

Characteristic	MIC of vancomycina, mg/L			P
	≤ 1.0 (n = 28)	1.5 (n = 31)	2.0 (n = 6)
Patient Characteristics :
Male sex	18 (64.3)	15 (48.4)	2 (33.3)	0.269
Age, years, median (25th–75th percentiles)	67.5 (58.4–75.9)	65.6 (45.4–74.8)	67.6 (66.5–76.6)	0.554
Predisposing conditions :
Diabetes mellitus	11 (40.7)	13 (41.9)	4 (66.7)	0.570
Hemodialysis dependent	5 (17.9)	9 (29.0)	1 (16.7)	0.623
HIV positive	0 (0.0)	1 (3.2)	0 (0.0)	1.00
History of cancer	5 (17.9)	3 (9.7)	1 (16.7)	0.569
Chronic immunosuppressive therapy	5 (18.5)	6 (19.3)	0 (0.0)	0.785
History of infective endocarditis	1 (3.5)	4 (12.9)	1 (16.6)	0.328
Other chronic disease	19 (67.9)	22 (71.0)	3 (50)	0.679
Recent Invasive procedures	12 (42.9)	13 (41.9)	3 (50)	0.457
History of congenital heart disease	3 (10.7)	2 (6.7)	1 (16.7)	0.443
Presumed place of acquisition :				0.547
Nosocomial	12 (42.9)	13 (41.9)	5 (83.3)	0.187
Non-nosocomial health care-associated	10 (35.7)	12 (38.7)	1 (16.6)	0.697
Community-acquired	6 (21.4)	6 (19.3)	0 (0.0)	0.717
Patient outcome :
Cardiac surgery during hospitalization	7 (25.0)	8 (25.8)	2 (33.3)	0.845
Stroke	6 (21.4)	4 (12.9)	3 (50.0)	0.124
Congestive heart failure	7 (25.0)	7 (22.6)	4 (66.7)	0.107
Intracardiac abscess	4 (14.3)	2 (6.5)	1 (16.7)	0.398
Persistent bacteremia	12 (42.9)	13 (41.9)	5 (83.3)	0.187
New conduction abnormalities	2/27 (7.4)	2/30 (6.7)	1 (16.7)	0.616
In-hospital death	11 (39.3)	9 (29.0)	4 (66.7)	0.242

NOTE. Data are no. (%) of patients, unless otherwise indicated.

^aMIC values defined according to E-test.

DISCUSSION

Over one-quarter of the isolates in this study exhibited the hVISA phenotype. This rate -approximately three times higher than those reported in two recent reports in patients with primarily non-IE infections [6, 26] - may reflect the fact that IE is a high-inoculum infection, with valvular vegetations serving as “sanctuary sites” for the bacteria. Penetration of glycopeptides into these vegetations is limited [27], leading to the potential for prolonged bacterial exposure to sub-therapeutic glycopeptide levels. The presence of hVISA related to prolonged sub-therapeutic glycopeptide exposure seems particularly possible among MRSA isolates with higher vancomycin MICs [28]. In support of this possibility is the observation that over 80% of the IE isolates in the present investigation with a vancomycin MIC of 2 mg/L exhibited the hVISA phenotype.

hVISA-infected patients had more IE complications - including persistent bacteremia and congestive heart failure - than patients with VSSA IE. Congestive heart failure is a leading cause of death in patients with IE [29, 30], and its presence has the greatest impact on prognosis of all IE complications [31, 32]. The association between persistent bacteremia and hVISA has been previously reported among patients with S. aureus infections of multiple types [7, 10, 11], but was confounded by the issue of whether the persistent bacteremia was simply due to the presence of IE. The current study addresses this issue by demonstrating that, even among patients with IE, persistent bacteremia and hVISA are associated. On the other hand, it should be acknowledged that persistent bacteremia per se does not equal failure, particularly as the mortality difference between VSSA- and hVISA-infected subjects (35% vs. 42%) did not achieve statistical significance. Because this observation is consistent with previous reports [7, 10], it is possible that some infections due to hVISA may eventually be treated successfully with vancomycin. However, the recent finding by Rose et al. that daily vancomycin doses of up to 10 grams were ineffective against an in vitro model of a high inoculum hVISA infection argues against this possibility [33]. Thus, our failure to demonstrate a statistically significant difference in mortality between hVISA- and VSSA-infected patients may simply be due to an insufficient sample size. However, the fact that almost 40% of the entire cohort died underscores the need for alternative therapies for MRSA IE.

The prevalence of hVISA among MRSA IE isolates differed by geographic region in our study. IE isolates from Oceania or Europe were more likely to exhibit the hVISA phenotype than isolates from North America. Although the absolute number of isolates and study sites from each region was relatively small, the findings were striking: over three-quarters of MRSA IE isolates from Australia or New Zealand exhibited the hVISA phenotype on PAP analysis. Previous studies evaluating hVISA prevalence have used a variety of laboratory definitions, testing strategies, and patient populations [6, 10, 34–39] (table 5), thereby limiting comparisons. The current investigation overcomes these limitations to make a key point: even in a “high-risk” infection type such as IE, the prevalence of hVISA varies significantly and is potentially associated with non-clinical factors such as geography. Although the reasons for the apparently higher hVISA prevalence in Oceania and Europe are unknown, it is intriguing to note that teicoplanin, a glycopeptide that has been associated with development of the hVISA phenotype [40, 41] is used in both of these regions, but not the US. Further studies will be required to substantiate this interesting speculation.

Table 5.

Prevalence of heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) in previous studies.

Author (reference)	Study period	Geographic region	Design	MRSA, N	hVISA, N	% hVISA	Study population	Screening tests	Confirmatory test
Hiramatsu et al. [35]	1996~1997	Japan	M	1,149	35	3.05	All clinical isolates	sPAP-BHIA-V4	Broth Microdilution
Kim et al. [36]	1999	South Korea	S	3,363	24	0.71	All clinical isolates	sPAP-BHIA-V4	PAP
Sancak et al. [39]	1998~2001	Turkey	S	256	46	17.97	Blood (41%), pus (39.8%), tracheal aspiration/bronchoalveolar fluid (18.4%)	BHIA-V4, MET	PAP
Charles et al. [10]	2001~2002	Australia	S	53	5	9.4	Bloodstream isolates	None	mPAP
Garnier et al. [34]	2001~2002	France	S	2,300	255	11.1	All clinical isolates	BHIA-T4, MET	mPAP
Maor et al. [38]	2003~2004	Israel	S	246	16	6.5	Bloodstream isolates	None	MET
Kosowska-Shick et al. [37]	2006~2007	US	S	982	2	0.2	All clinical isolates: wound (60%), respiratory tract (20.7%), blood (9.7%), etc.	BHIA-V3, MET	PAP
Rybak et al. [6]	1986~2007	US	M	1,499	112	7.5	Clinical isolates from surveillance studies: blood (60%), lung (21%), skin (14%), etc.	MET	mPAP

NOTE. M = multicenter trial; S = single center trial; sPAP-BHIA-V4 = simplified population analysis profile on brain heart infusion agar with 4 mg/L vancomycin; BHIA-T4 = Brain heart infusion agar with 4 mg/L teicoplanin; BHIA-V3 = Brain heart infusion agar with 3 mg/L vancomycin; MET = Macrodilution E-test; PAP = Population analysis profile by Hiramatsu et al. [35]; mPAP = Modified popultion analysis profile by Wooton et al [19]

Detection of the hVISA phenotype in S. aureus is difficult. PAP-AUC is time-consuming, labor-intensive and unsuitable for clinical laboratories. Several other laboratory methods have been considered [42–44], but also have limitations as simple screening tests. Thus, a clinically directed approach will likely be required before hVISA testing can be routinely considered. One readily available strain characteristic that may identify bacteria at higher risk for hVISA is the presence of a vancomycin MIC of 2.0 mg/L. Although only ~ 10% of the cohort had a vancomycin MIC of 2.0 mg/L on E-test, over 80% of these isolates exhibited the hVISA phenotype on PAP testing. Thus, discovery of a high vancomycin MIC MRSA isolate from an appropriate clinical setting (e.g., MRSA IE with persistent bacteremia) could identify situations in which hVISA testing might be helpful. However, any such screening strategy should be properly validated before widespread adoption into clinical practice.

Using PFGE, MLST, and PCR for a large number of putative virulence genes, hVISA isolates were genotypically similar to VSSA isolates. This finding conflicts with a previous report suggesting an association between agr type II polymorphism and the development of vancomycin heteroresistance [45]. Our failure to identify a single clonotype associated with hVISA also suggests that the emergence of hVISA may be due more to the influence of ubiquitous forces – such as widespread glycopeptide use - acting simultaneously upon multiple different bacterial clones in different regions of the world than the global dissemination of a limited number of bacterial clones exhibiting hVISA. This finding also suggests that bacterial genotype is unlikely to be a useful method to target selected clinical isolates for hVISA testing. However, recent studies linking point mutations in graSR and vraSR to the stepwise development of the VISA phenotype [46] increase the possibility of the future development of molecular methods for detection of hVISA strains.

Using both E-test and broth microdilution techniques, our study found an association between higher vancomycin MIC values and presence of the hVISA phenotype. This finding is consistent with a recent report from Detroit [6]. Several studies have reported higher rates of treatment failure [47–49] and mortality [50] among patients with infections caused by MRSA isolates with higher vancomycin MIC values. In the current study, the mortality of infections caused by isolates with vancomycin MIC of 2 mg/L by E-test was particularly high (66%), although the difference did not achieve statistical significance. This finding agrees with a recent report demonstrating suboptimal cure rates of vancomycin in an animal model of high vancomycin MIC (2 mg/L) MRSA aortic endocarditis [8].

The present study has limitations. Our sample size was relatively small, limiting our ability to detect significant differences betwen VSSA and hVISA infections, and between infections with different vancomycin MICs. Next, we had no information regarding timing of complications, antibiotic dosing, trough serum concentration monitoring, prior antibiotic therapy before the onset of MRSA bacteremia, or the ultimate duration of bacteremia after therapy was initiated. We were also unable to consider why individual patients did or did not undergo cardiac surgery, or why patients with VSSA infection underwent cardiac surgery twice as frequently as did those with hVISA (30.4% vs. 15.8%; P=.353). Standard practice in the ICE-PCS Microbiological Repository is to evaluate the first available bloodstream isolate. However, patients transferred from other facilities may have had bacteremia for extended periods of time prior to arrival at the ICE study center. Finally, because serial isolates from the same patient were not routinely available, we were unable to further evaluate the association between persistent bacteremia and the hVISA phenotype.

In summary, this study makes several key observations. The hVISA phenotype occurred in over one-quarter of MRSA IE isolates, and was associated with a higher frequency of IE complications such as persistent bacteremia and heart failure. hVISA was geographically distributed, with highest rates being encountered in MRSA IE isolates from Europe or Oceania. hVISA and VSSA isolates were genotypically similar in this study. Vancomycin MIC values of 2 mg/L were relatively uncommon, but associated with hVISA. Future studies are required to validate the findings of this investigation, to better understand specific risk factors for the emergence of hVISA, and to identify practical, effective screening tests to detect hVISA in the clinical setting.