Heteroresistant vancomycin-intermediate Staphylococcus aureus (hVISA) has been detected in clinical S. aureus strains throughout the world [1–3] and has been associated with vancomycin treatment failure for S. aureus [4]. The prevalence of hVISA has been reported ranging from 0 to 74% [2, 5]. The large variance of the hVISA prevalence is due to the lack of a prevalence study. On top of that, standardized methodologies and criteria to detect hVISA in the clinical microbiology laboratory are not established. The prevalence study with high quality cost effective methodologies are necessary to be conducted with uniformly to identify the true hVISA prevalence. Owing to the unstable characteristics of vancomycin heteroresistant phenotype, establishment of standardized methods to detect hVISA has been challenging [5].

hVISA was first characterized in 1997 [6]. hVISA is defined as isolates which are classified by the Clinical and Laboratory Standard Institute (CLSI) as having a vancomycin broth microdilution minimum inhibitory concentration (MIC) in the susceptible range where a notable subpopulation of cells (approximately one organism in every 10⁵-10⁶ or more) with MICs over the susceptibility breakpoint [6]. A number of current methodologies for detection of hVISA exist such as the Etest Macromethod, Etest Glycopeptide Resistance Detection (Etest GRD: Standard Etest) method and the population analysis profile-area under the curve (PAP-AUC) method [7]. Multiple studies demonstrated high sensitivity, high specificity, consistent and low variation between laboratories in detecting hVISA by the Macro Etest Method [8–10]. Although PAP-AUC is labor intensive, it has long been considered the gold standard for hVISA detection [2]. Furthermore, as it relates to genetic factors which can be markers for hVISA, no single resistance gene has been correlated with hVISA development [11, 12]; however, it has been demonstrated that the expression of hVISA phenotype is well associated with dysfunction in the quorum sensing cluster of genes, the accessory gene regulator group (agr), controlling colonization and virulence, as well as a number of other regulators [13].

In this study, the objectives were to: 1) determine the prevalence of hVISA in clinical S. aureus bloodstream isolates utilizing Macro Etest, Standard Etest, modified PAP-AUC method and Brain Heart Infusion Agar containing vancomycin 4 mg/L (BHIAV4) method. 2) determine the relationship between hVISA, vancomycin MIC, and agr dysfunction among clinical S. aureus bloodstream isolates for the purpose of establishing the standardized method to facilitate the hVISA detection in the clinical setting.

The proportion of methicillin susceptible S. aureus (MSSA) and MRSA isolates in tested S. aureus isolates were 55 and 45%, respectively (Table 1). The CLSI broth microdilution MIC range for S. aureus was 0.25 to 2 mg/L, and for hVISA was 1 to 2 mg/L. The distribution of MICs 0.25, 0.5, 1 and 2 mg/L measured by the broth microdilution method were 6 (2.7%), 21 (9.5%), 159 (72.3%) and 34 (15.5%), respectively (Figure 1). MIC₅₀ of S. aureus (n = 220), MSSA (n = 121), MRSA (n = 99) and hVISA (determined by Macro Etest, n = 12) were 1, 1, 1 and 1.5 mg/L, respectively. MIC_50/90 of hVISA analyzed by broth microdilution, Standard Etest, and Macro Etest were 1.5/2, 4/4, and 6/8 mg/L, respectively.

	S. aureus 220	MSSA 55% (121/220)	MRSA 45% (99/220)	hVISA 5.5% (12/220)
MIC50/90	1/2	1/2 (p = 0.51)a	1/2 (p = 0.46)a	1.5/2 (p = 0.03)a
agr-dysfunction	15% (33/220)	10% (12/121)	21% (21/99) (p = 0.02)b	58% (7/12) (p< 0.0005)c

MIC_50/90 determined by CLSI broth microdilution and the prevalence of agr-dysfunctional

^a MIC distribution of total clinical S.aureus versus MIC distribution of each phenotype (Mann-Whitney U)

^b Prevalence of agr dysfunction in MRSA isolates versus MSSA isolates (Chi-square test)

^c Prevalence of agr dysfunction in hVISA isolates versus VSSA isolates (Chi-square test)

All tested S. aureus isolates displayed CLSI broth microdilution MIC values in the susceptible range, however, the distribution of MICs was significantly different between the total clinical S. aureus and hVISA strains (p = 0.02). Standard Etest applying the CLSI breakpoints identified 13 clinical S. aureus strains as hVISA, and Macro Etest classified 1 out of those 13 hVISA strains as a vancomcyin susceptible S. aureus (VSSA) which also displayed the low modified PAP-AUC ratio (0.54), as shown in PAP profiles in Figure 2. The criterion for Macro Etest for detecting hVISA was the appearance of one or more subpopulation colonies in MIC ≥4 mg/L. hVISA prevalence was higher in MRSA (9.1%) compared to MSSA (2.5%) (p = 0.03) (Table 2). The Etest MIC results displayed high reproducibility (SD < 10%).

With ten hVISA strains screened by Macro Etest and confirmed by modified PAP-AUC with the ratio cutoff value > 0.8, the mean modified PAP-AUC ratio (± SD) was (0.91 ± 0.06). Additionally, the mean modified PAP-AUC ratio (± SD) of seven hVISA isolates with MRSA phenotype was (0.94 ± 0.04), which was higher compared to the mean ratio of three hVISA isolates with MSSA phenotype (0.85 ± 0.02) (p = 0.001). With a higher endpoint criterion modified PAP-AUC ratio > 0.9, five hVISA strains were misclassified as susceptible. The percentage of the sensitivity modified PAP-AUC ratio > 0.8 and ratio > 0.9 in discriminating hVISA were 83.3% and 58.3%, respectively, when the sensitivity of broth microdilution, Standard Etest, BHIAV4 with lower inoculum (10⁸ CUF/mL) and BHIAV4 with higher inoculum (10¹⁰ CFU/mL) were 0%, 100%, 8.3% and 100%, respectively.

Macro Etest MICs were plotted against modified PAP-AUC ratios (Figure 3). The regression line for the data of thirteen isolates identified as hVISA by Standard Etest is displaying the strong relationship between Macro Etest MIC shift and modified PAP-AUC ratio (r² = 0.78). Standard Etest also displayed the relationship between MIC values and modified PAP-AUC ratio (r² = 0.24). In contrast, the CLSI broth microdilusiton MIC values did not have clear correlation with modified PAP-AUC ratios (r² = 0.026). According to the linear regression equation for these two variables, Macro Etest MICs and modified PAP-AUC ratio, the estimated modified PAP-AUC ratio at MIC 4 and 6 mg/L were 0.62 and 0.85, respectively.

The percentages of agr dysfunctional phenotype in total tested S.aureus, MSSA, MRSA and hVISA were 15%, 10%, 21% and 58% (Table 1). The prevalence of agr dysfunctional phenotype was significantly higher in hVISA compared with VSSA (p = 0.02). Additionally, the prevalence of agr dysfunctional phenotype was higher in MRSA compared with MSSA (p = 0.02). BHIAV4 method analysis using lower inoculum (10⁸ CFU/mL) misclassified 11 out of 12 hVISA strains as VSSA. On the contrary, BHIAV4 analysis with higher inoculum (10¹⁰ CFU/mL) identified all 12 hVISA strains as hVISA.

Due to the heteroresistant nature of vancomycin intermediate resistance, a number of clinical S. aureus may exhibit resistant subpopulations not expressed by traditional methods of MIC detection. Therefore, it is necessary to investigate a variety of methods to increase sensitivity of hVISA detection in clinical laboratories. A recent multicenter, multi-national comparison by Wootton et. al. determined that the Macro Etest method displayed high sensitivity (85.9%) in detection of hVISA which is rationale of using this method as an initial screen for strains which display hVISA [9]. In the current study we utilized this methodology to determine that the hVISA phenotype was nearly four times higher among MRSA vs. MSSA. Quantitatively, these results were also confirmed by population analysis profiling methods whereby MRSA displayed the higher modified PAP-AUC ratios compared to hVISA MSSA.

This perhaps is intuitive due to vancomycin being commonly utilized for MRSA; however, hVISA was also detected among MSSA, which is in agreement with the recent findings of a significant increase in vancomycin MICs among MSSA clinical isolates by Wang [17]. These findings may suggest that the use of aggressive vancomycin regimens, in MRSA infections are important, so that the development hVISA phenotype which is driven by suboptimal vancomycin therapeutic exposure can be avoided. Certainly, alternative treatment should be used for MSSA infections in lieu of vancomycin treatment.

The BHIAV4 method described by Hiramatsu [6] has been evaluated in a number of studies. One concern with this methodology is that, the heteroresistant subpopulations arise in very low frequency, which results inability of the BHIAV4 method to detect heteroresistant phenotypes when the inoculum is low. To study the impact of inoculums on detection and mutation frequency, we utilized a high inoculum of S. aureus (10¹⁰ CFU/ml) which is often observed in the severe S. aureus infection including infective endocarditis. As a result, the inoculum level displayed the significant impact on the sensitivity: the modified BHIAV4 method using the higher inoculum 10¹⁰ CUF/mL 100% isolates additional hVISA correctly. Furthermore, at high bacterial density, stationary-phase growth foster biofilm formation via quorum-sensing mechanisms involving agr in S. aureus [13]. Coupled with the finding that proportion of agr dysfunctional strains were nearly five times higher among hVISA vs. VSSA, these findings may provide additional potential mechanistic insights into the development vancomycin resistance. Furthermore, agr dysfunctional S. aureus strains tend to display low activity in metabolic pathway which is growth phase dependant. Therefore, in the context of high bacterial density, it has been hypothesized that stationary phase growth, defective autolysis profiles, biofilim production and the production of thicker cell walls, may be facilitated by quorum sensing mechanisms [11–13]. Finally, from a clinical microbiology standpoint, the evaluation of delta-hemolysin production may be of potential utility to detect staphylococci which display an increase proclivity to develop heterogeneous resistance.

This study had potential limitations. First, we acknowledge the lack of geographically diverse sites for S. aureus, as isolates were from a single institution and limited to nosocomial bacteremia in a defined geographic area. Additional multicenter studies including both nosocomial and community-onset bloodstream infections are necessary to define the true prevalence of hVISA. Second, we utilized PAP using a high initial inoculum of 10¹⁰ CFU/mL to confirm all hVISA isolates determined by the Etest Macro Method. Although, this method may decrease specificity, our VSSA control strain, ATCC 29213 did not display any vancomycin intermediate populations. Third, although we utilized delta-hemolysin expression as a surrogate marker measure of agr function, more quantitative measure of RNAIII expression may be necessary to further define the relationship agr and heteroresistance. Overall, the prevalence of hVISA phenotype in 220 clinical S. aureus isolates was observed to be 5.5% in S. aureus and 9.1% in MRSA strains using the Etest Macro Method, which was well correlated with modified PAP-AUC ratio values. agr dysfunction may add further utility to hVISA detection in the clinical microbiology setting.

hVISA was detected among S. aureus bloodstream isolates, which were classified as susceptible among clinical microbiology laboratories. Evaluating the correlation between Etest MICs and modified PAP-AUC ratio values will add further improvement of discriminating hVISA, and agr dysfunction may be predictive of strains which display a greater predilection to display the hVISA phenotype.