Comparison of Detection Methods for Heteroresistant Vancomycin-Intermediate Staphylococcus aureus, with the Population Analysis Profile Method as the Reference Method

Abstract

Staphylococcus aureus clinical isolates with vancomycin MICs of 2 μg/ml have been associated with vancomycin therapeutic failure and the heteroresistant vancomycin-intermediate S. aureus (hVISA) phenotype. A population analysis profile (PAP) with an area under the curve (AUC) ratio of ≥0.9 for the AUC of the clinical isolate versus the AUC for hVISA strain Mu3 is most often used for determining hVISA, but it is time-consuming and labor-intensive. A collection of 140 MRSA blood isolates with vancomycin MICs of 2 μg/ml by reference broth microdilution and screened for hVISA using PAP-AUC (21/140 [15%] hVISA) were tested by additional methods to detect hVISA. The methods included (i) Etest macromethod using vancomycin and teicoplanin test strips, brain heart infusion (BHI) agar, and a 2.0 McFarland inoculum; (ii) Etest glycopeptide resistance detection (GRD) using vancomycin-teicoplanin double-sided gradient test strips on Mueller-Hinton agar (MHA) with 5% sheep blood and a 0.5 McFarland inoculum; and (iii) BHI screen agar plates containing 4 μg/ml vancomycin and 16 g/liter casein using 0.5 and 2.0 McFarland inocula. Each method was evaluated using PAP-AUC as the reference method. The sensitivity of each method for detecting hVISA was higher when the results were read at 48 h. The Etest macromethod was 57% sensitive and 96% specific, Etest GRD was 57% sensitive and 97% specific, and BHI screen agar was 90% sensitive and 95% specific with a 0.5 McFarland inoculum and 100% sensitive and 68% specific with a 2.0 McFarland inoculum. BHI screen agar with 4 μg/ml vancomycin and casein and a 0.5 McFarland inoculum had the best sensitivity and specificity combination, was easy to perform, and may be useful for clinical detection of hVISA.

Vancomycin has been the most reliable therapeutic agent against methicillin-resistant Staphylococcus aureus (MRSA) for the past 3 decades. However, despite its sustained in vitro microbiologic inhibitory activity, clinicians continue to debate its utility for MRSA infections (30, 36). Widespread empirical use of vancomycin to cover Gram-positive organisms, including MRSA, has likely contributed to the growing burden of less susceptible strains, and many health care facilities have reported an upward trend of vancomycin MICs for MRSA isolates over the past 5 years (19, 35, 40). In addition, a number of investigators have reported that MRSA infections caused by isolates with vancomycin MICs of 2 μg/ml (at the upper limits of the Clinical and Laboratory Standards Institute [CLSI] susceptibility range) and/or the heteroresistant vancomycin-intermediate S. aureus (hVISA) phenotype were associated with prolonged bacteremia, greater rates of complications, and vancomycin therapeutic failures (22, 23, 31).

Infections involving hVISA pose a unique problem. Such strains are susceptible in vitro to vancomycin (MIC < 4 μg/ml) and thus are classified as susceptible by standard clinical laboratory methods but contain subpopulations of 1 in 10⁶ cells that can grow in the presence of ≥4 μg/ml of vancomycin (9, 27). Although the true prevalence of hVISA is unknown, estimates from a limited number of studies range from 1.3% to 27% of all MRSA isolates (1, 4, 5, 8, 21). Because of the increasing number of reports of vancomycin treatment failures and reports of poor outcomes for patients infected with hVISA (3, 5, 13, 21, 23, 25), an accurate and practical method for the detection of hVISA among MRSA isolates in the clinical laboratory is of growing importance.

Standardized reference methods for susceptibility testing, such as CLSI broth microdilution, agar dilution, and standard Etest methods, fail to detect hVISA, in part due to the small inoculum, the relatively poor support of growth on Mueller-Hinton agar plates, or a combination of both (38). Inoculum size is critical to detection of the minor subpopulation of resistant cells. Additionally, hVISA strains are notoriously slow growing, with thickened cell walls and unique pleomorphic features, such as small-colony variants (45). Screening for hVISA by the population analysis profile-area under the curve (PAP-AUC) method has been the most reliable and reproducible approach but is labor-intensive, costly, and unsuitable for routine use in clinical laboratories. A variety of alternative methods for detection of the heteroresistant phenotype have been evaluated with varying success, and no single standardized clinical laboratory method has been established to date (17, 36, 38, 43).

In this study, we chose PAP-AUC as the “gold standard” to define hVISA. Using PAP-AUC as the reference method, we evaluated the Etest glycopeptide resistance detection (GRD), the Etest macromethod, and a newly described brain heart infusion (BHI) screen agar containing casein and vancomycin for the detection of hVISA (42) using a collection of 140 invasive MRSA isolates with vancomycin MICs of 2 μg/ml.

MATERIALS AND METHODS

Bacterial strains.

One hundred forty invasive MRSA clinical isolates with vancomycin MICs of 2 μg/ml by reference broth microdilution using standard CLSI methods (6, 7) were collected as part of the Active Bacterial Core surveillance (ABCs) of the Centers for Disease Control and Prevention (CDC) Emerging Infections Program from 2005 to 2007. Isolates were collected in five U.S. metropolitan areas in five states: 40 isolates from California, 30 from Georgia, 13 from Minnesota, 35 from New York, and 22 from Tennessee. Isolates were stored at −70°C in BBL Trypticase soy broth (TSB; Becton Dickinson and Company, Sparks, MD) containing 20% glycerol and were subcultured twice prior to testing on blood agar medium (Trypticase soy agar [TSA] with 5% sheep blood). All testing for each isolate was performed on the same day, using the same McFarland preparation to perform each test when the same inoculum size was required. MRSA strains Mu3 (the archetype hVISA strain), ATCC 29213 (a vancomycin-susceptible S. aureus [VSSA] strain), and Mu50 (VISA) were included as controls with each experiment (11), and results were recorded for each isolate only when the positive and negative controls were confirmed for each method. All strains were independently tested in duplicate to examine the reproducibility of each method.

PAP-AUC.

Heteroresistance was evaluated by PAP-AUC using the microdilution technique previously described by Pfeltz et al. (26) and Satola et al. (32) and modified as described below.

A 0.5 McFarland standard suspension from an overnight culture in BBL TSB (Becton Dickinson and Company) of the isolate to be tested was prepared. Serial dilutions (10⁻¹ to 10⁻⁷) were made in sterile saline, and 100 μl of 10⁻⁶ and 10⁻⁷ dilutions was spiral plated by hand using an inoculating turntable onto two media: Trypticase soy agar with 5% sheep blood to check for purity and Difco BHI agar (Becton Dickinson and Company) to determine viable counts. An aliquot of 100 μl of undiluted 0.5 McFarland standard suspension was spiral plated, and four 10-μl droplets from each of five dilutions (10⁻¹ to 10⁻⁵) were dropped onto BHI agar plates containing 0, 0.25, 0.50, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 4.0, 6.0, and 8.0 μg/ml of vancomycin. Vancomycin analytical powder was commercially purchased (Sigma Chemical Company, St. Louis, MO).

Plates were air dried and incubated for 48 h at 35°C. Colony counts (log₁₀ numbers of CFU/ml) were determined at 24 and 48 h by counting the number of colonies on each plate or in each droplet, averaging the number of colonies in the four droplets for each vancomycin concentration, and converting the number of colonies to number of CFU/ml using the appropriate dilution factor. The log₁₀ number of CFU/ml for each strain was plotted against the vancomycin concentrations, and the AUC was determined using GraphPad Prism software (San Diego, CA). Isolates were defined as hVISA if the AUC of the test isolate divided by the AUC of the corresponding strain, Mu3, was ≥0.90.

Standard Etest.

Vancomycin and teicoplanin MICs were determined by standard Etest methods using a 0.5 McFarland standard inoculum on Mueller-Hinton agar (MHA) plates (Remel, Lenexa, KA), according to the manufacturer's manual (EAS 003; AB Biodisk, Solna, Sweden), using vancomycin and teicoplanin Etest strips (bioMérieux, Durham, NC). MIC endpoints were read according to the manufacturer's recommendations. MIC values were evaluated at the actual endpoint and also after rounding up to the next highest doubling dilution, which is recommended by the manufacturer for results reporting.

Etest GRD.

The Etest GRD was performed according to the manufacturer's instructions (AB Biodisk) using a double-sided predefined gradient of vancomycin and teicoplanin for the detection of glycopeptide-intermediate S. aureus (GISA) or hetero-GISA (hGISA). A 0.5 McFarland standard inoculum was prepared and swabbed onto MHA with 5% sheep blood (BBL, Becton Dickinson and Company). The zone of inhibition was read at 24 and 48 h after incubation at 35°C. The test isolate was considered positive for hVISA if the Etest GRD strip result was ≥8 μg/ml for either vancomycin or teicoplanin.

Etest macromethod.

The Etest macromethod was performed using a 2.0 McFarland inoculum on BHI agar plates (Difco, Becton Dickinson and Company), according to the manufacturer's manual (EAS 003; AB Biodisk), using vancomycin and teicoplanin Etest strips (bioMérieux). Etest macromethod assays were performed on the same day and with an inoculum from the same initial culture as that used for the standard Etest and the PAP-AUC assays. Heteroresistance by the Etest macromethod was defined as MICs for vancomycin and teicoplanin of ≥8 μg/ml or a teicoplanin MIC of ≥12 μg/ml regardless of the vancomycin MIC.

BHI screen agar plates.

BHI agar (Difco, Becton Dickinson and Company) with the addition of 16 g/liter pancreatic digest of casein (Becton Dickinson and Company) and 4 μg/ml vancomycin (Sigma) was prepared in-house at the CDC.

Two inoculum densities prepared in saline were utilized: a 0.5 McFarland standard recommended by CLSI and a 2.0 McFarland standard suspension. Four 10-μl droplets from each McFarland suspension were dropped by a pipette onto the BHI with casein and 4 μg/ml vancomycin screen agar plates, allowed to air dry for approximately 5 min, and incubated at 35°C. Plates were examined at 24 and 48 h, and individual colonies in each droplet were counted. A droplet with >20 colonies was considered too numerous to count accurately and designated TNTC. A droplet with confluent growth was scored as having confluent growth. An isolate was considered hVISA if at least one droplet had two or more colonies.

Statistical analysis.

Sensitivity and specificity analyses were conducted to evaluate the performance of the Etest macromethod, Etest GRD strip, and the BHI screen agar plate technique using PAP-AUC as the reference method. Sensitivity analysis describes the fraction of correctly identified true positives (hVISA) by each test method, while specificity analysis describes the fraction of correctly identified negatives (non-hVISA, VSSA).

RESULTS

hVISA was determined by PAP-AUC for 140 MRSA clinical blood isolates that demonstrated an initial vancomycin MIC by reference broth microdilution of 2 μg/ml. Twenty-one of 140 (15%) isolates were hVISA by PAP-AUC, defined as a value of ≥0.90 for the ratio of the AUC of the isolate to the AUC of hVISA strain Mu3. The PAP-AUC ratios (ratio of the AUC of the isolate to the AUC of Mu3) ranged from 0.57 to 1.48, with a mode of 0.74. The frequency distribution of the PAP-AUC ratios for all 140 isolates is shown in Fig. 1. Almost half (66/140, 47%) fell within the range of 0.69 to 0.79; 10 isolates (7%) had ratios between 0.85 and 0.89, borderline for hVISA (Fig. 1).

FIG. 1.

Frequency distribution of the PAP-AUC ratio of the isolate AUC to the Mu3 AUC for 140 MRSA isolates with vancomycin MICs of 2 μg/ml. PAP-AUC ratios ranged from 0.57 to 1.48, with a modal ratio of 0.74, and 47% fell within the range of 0.69 to 0.79.

Standard Etests were performed on all 140 isolates as a comparison to the reference broth microdilution MICs and not as a method for hVISA detection. MICs ranged from 1.5 to 3 μg/ml (mode = 2 μg/ml) for vancomycin and 0.5 to 6 μg/ml (mode = 1 μg/ml) for teicoplanin. Vancomycin MICs were 1.5 μg/ml for 48 isolates, 2 μg/ml for 84 isolates, and 3 μg/ml for 8 isolates. Since a standard Etest MIC of 1.5 μg/ml could be rounded to the nearest doubling dilution for reporting purposes (i.e., MIC = 2 μg/ml), the MICs for 132/140 (94%) isolates were in agreement with the reference broth microdilution MICs of 2 μg/ml. Six of the eight isolates with standard Etest vancomycin MICs of 3 μg/ml were found to be hVISA by PAP-AUC, and the remaining two had slightly elevated ratios of the AUC of the isolate to the AUC of Mu3 of 0.75 and 0.83 (Table 1). Additionally, all isolates with standard Etest teicoplanin MICs of 3 μg/ml or greater were hVISA by PAP-AUC.

TABLE 1.

Comparison of PAP-AUC ratio, BHI screen agar results, and standard Etest, Etest macromethod, and Etest GRD vancomycin and teicoplanin MICs for isolates with PAP-AUCs ≥ 0.69^a

Isolate	PAP-AUC ratio	MIC (μg/ml)						BHI casein screen agar
		Standard Etestb		Etest macroc		Etest GRDd
		VAf	TPg	VA	TP	VA	TP	48 h, 0.5 McFarland		48 h, 2.0 McFarland
								No. of colonies/droplete	hVISA	No. of colonies/droplet	hVISA
CA-156	0.69	2	0.75	3	2	1.5	1.5	0,0,0,0	−	0,0,0,0	−
GA-6	0.69	2	1.5	3	3	2	4	1,4,0,1	+*	3,3,5,6	+*
GA-772	0.69	1.5	1	8	6	2	4	3,2,0,0	+*	10,c,c,c	+*
TN-196	0.71	2	0.75	3	2	1.5	2	0,0,0,0	−	c,c,c,c	+*
TN-61	0.72	2	2	4	8	2*	8*	0,0,0,0	−	c,c,c,c	+*
GA-165	0.74	1.5	1	3	2	1.5	3	0,0,0,0	−	10,10,6,8	+*
GA-983	0.74	2	1.5	8	2	3	4	5,3,3,2	+*	7,3,5,4	+*
MN-102	0.75	3	1	3	3	2	4	0,0,0,0	−	0,0,0,0	−
CA-488	0.76	2	2	8*	8*	1	6	0,0,0,0	−	c,c,c,c	+*
NY-507	0.78	2	1.5	3	2	1.5	3	2,0,0,0	+*	2,3,0,1	+
CA-108	0.83	2	1.5	6	6	1.5	4	2,1,0,0	+*	t,t,4,10	+*
NY-14	0.83	3	1.5	4	4	1.5*	8*	0,0,0,0	−	0,0,0,0	−
CA-118	0.85	1.5	1.5	3*	12*	1.5*	32*	0,0,0,0	−	t,t,t,t	+*
CA-131A	0.85	2	1.5	4	6	2	6	3,0,0,0	+*	3,3,0,0	+*
NY-240	0.85	2	2	3	3	1.5	3	0,0,0,0	−	3,2,2,2	+*
NY-301	0.85	1.5	2	4	6	1.5	3	0,0,0,0	−	0,0,0,0	−
NY-406	0.85	2	2	8	4	1.5	3	1,0,0,0	−	5,3,t,t	−
GA-27	0.86	2	1.5	4	4	2	2	3,4,2,3	+*	c,c,c,c	+*
NY-18	0.87	2	1	8	4	2	1.5	1,0,0,0	−	1,0,1,0	−
NY-415	0.88	2	1.5	8*	8*	1	3	0,0,0,0	−	6,10,10,10	+*
CA-124	0.89	3	2	8*	8*	1.5	4	0,0,0,0	−	c,c,c,c	+*
NY-328	0.89	2	2	4*	12*	1.5	4	2,2,0,0	+*	2,t,t,10	+*
GA-984	0.90*	2	2	8	3	2*	8*	t,t,t,t	+*	c,c,c,c	+*
NY-593	0.90*	1.5	1	3	2	1	2	2,3,4,5	+*	15,13,19,12	+*
MN-055	0.91*	2	3	8*	8*	2*	8*	1,2,3,2	+*	c,c,c,c	+*
NY-341	0.91*	2	2	4	3	1.5	2	1,0,2,0	+*	5,3,5,7	+*
CA-105	0.92*	2	3	4	8	1.5*	12*	1,0,0,0	−	t,t,t,t	+*
GA-12	0.92*	2	2	8*	8*	1.5*	8*	6,1,4,4	+*	c,c,c,c	+*
GA-15	0.92*	2	4	4*	12*	4*	32*	c,c,c,c	+*	c,c,c,c	+*
GA-231	0.93*	2	1.5	8*	8*	2	4	2,0,0,0	+*	2,4,2,2	+*
CA-39	0.95*	2	1.5	8*	8*	2	4	t,t,t,t	+*	c,c,c,c	+*
NY-439	0.95*	2	2	3	3	1.5	4	2,1,1,0	+*	7,8,8,10	+*
CA-436	0.99*	2	3	8	4	2*	8*	0,0,0,0	−	1,2,1,2	+*
CA-11	1.01*	2	3	8*	8*	2*	12*	2,t,t,4	+*	c,c,c,c	+*
CA-9	1.03*	2	1.5	8*	12*	1.5	4	7,3,4,1	+*	c,c,c,c	+*
CA-545	1.04*	2	2	3	4	1.5	4	12,12,14,10	+*	t,t,t,t	+*
MN-105	1.05*	3	2	8	6	2*	8*	2,3,1,0	+*	c,c,c,c	+*
CA-598	1.06*	1.5	1	4	6	1.5	4	1,2,2,1	+*	2,3,5,6	+*
NY-224	1.27*	3	4	6*	12*	3*	8*	c,c,c,c	+*	c,c,c,c	+*
CA-544	1.34*	2	3	8*	12*	3	6	c,c,c,c	+*	c,c,c,c	+*
NY-356	1.37*	3	6	8*	16*	4*	32*	c,c,c,c	+*	c,c,c,c	+*
NY-360	1.38*	3	3	6*	12*	3*	12*	c,c,c,c	+*	c,c,c,c	+*
NY-417	1.48*	3	2	8*	12*	3*	32*	c,c,c,c	+*	c,c,c,c	+*

^aData are for 43 isolates. An asterisk signifies that hVISA was detected by that method based on boldface values and text.

^bStandard Etest, 0.5 McFarland standard inoculum at 24 h.

^cEtest macro, Etest macromethod, 2.0 McFarland standard inoculum at 48 h; the isolate was hVISA if vancomycin and teicoplanin MICs are ≥8 μg/ml or the teicoplanin MIC is ≥12 μg/ml regardless of the vancomycin MIC.

^dEtest GRD, 0.5 McFarland standard inoculum at 48 h. The isolate is hVISA positive if either the vancomycin or teicoplanin MIC is ≥8 μg/ml.

^eTotal number of colonies in each 10-μl droplet. c, confluent growth; t, too numerous to count.

^fVA, vancomycin.

^gTP, teicoplanin.

Etest macromethod and Etest GRD were performed on all 140 isolates, and results were read at both 24 and 48 h. For each test, only the reading at 48 h was considered for hVISA determination. Vancomycin MICs ranged from 2 to 8 μg/ml (mode = 3 μg/ml), teicoplanin MICs ranged from 1.5 to 16 μg/ml (mode = 3 μg/ml), and 17/140 (12%) isolates met the criteria for hVISA by Etest macromethod. By Etest GRD, vancomycin MICs ranged from 0.75 to 4 μg/ml (mode = 1.5 μg/ml), teicoplanin MICs ranged from 0.75 to 32 μg/ml (mode = 2 μg/ml), and 15/140 (11%) isolates met the criterion for hVISA.

All 140 MRSA isolates were evaluated on BHI screen agar plates (containing additional casein and 4 μg/ml vancomycin) using two different concentrations of inoculum delivered in four 10-μl droplets. The presence of growth was examined at 24 and 48 h. Growth was best evaluated at 48 h for each inoculum. With a 0.5 McFarland inoculum, 27/140 (19%) isolates met the criterion for hVISA. When a 2.0 McFarland inoculum was used, 59/140 (42%) isolates met the criterion for hVISA. To verify reproducibility, 15 isolates with ratios of the isolate AUC compared to the AUC of Mu3 of between 0.65 and 1.38 were selected, blinded, and screened in duplicate for hVISA by two additional laboratory personnel. The BHI screen agar method was 100% reproducible within our laboratory.

A comparison of the PAP-AUC ratios with results obtained using BHI screen agar, standard Etest, Etest macromethod, and Etest GRD are shown in Table 1. Only results for those isolates positive or borderline positive for hVISA by any method or with a vancomycin MIC of 3 μg/ml by standard Etest are shown (n = 43). No isolate with a PAP-AUC ratio of the isolate AUC to the AUC of Mu3 of less than 0.69 was found to be hVISA by any method, and the majority of isolates (102/140, 73%) demonstrated no evidence of hVISA by any method evaluated (data not shown). Only 8/21 (4%) isolates that were hVISA by PAP-AUC were hVISA by all methods evaluated.

The isolates defined as hVISA by each test method and the PAP-AUC range of the ratio of the AUC of the isolate to the AUC of Mu3 is shown in Table 2. With the exception of the BHI screen agar method using the higher 2.0 McFarland inoculum, no false positives were identified by any of the evaluated methods when the PAP-AUC ratio was <0.69 (Table 2). The highest false-positive rate occurred in isolates with borderline PAP-AUC ratios. Eight of 18 (44%) isolates with PAP-AUC ratios between 0.80 and 0.89 were positive for hVISA by one or more of the test methods, whereas only 6/66 (6%) of isolates with PAP-AUC ratios between 0.69 and 0.79 were positive for hVISA.

TABLE 2.

Isolates defined as hVISA by different test methods and range of PAP-AUC ratio of isolate AUC to Mu3 AUC

PAP-AUC ratio range	No. (%) of isolates tested	No. (%) hVISA isolates by method
		Etest macromethoda	Etest GRDa	BHI with casein screen agarb	Totalc
0.57-0.68	35 (25)	0	0	0	0
0.69-0.79	66 (47)	1 (1.5)	1 (1.5)	4 (6)	4 (9)d
0.80-0.89	18 (13)	4 (22)	2 (11)	4 (22)	8 (44)d
0.90-1.48	21 (15)	12 (57)	12 (57)	19 (90)	19 (90)e
Total	140 (100)	17 (12)	15 (11)	27 (19)

^aResults at 48 h.

^bResults at 48 h; 2 or more colonies in one of four 10-μl droplets with a 0.5 McFarland suspension.

^cTotal number of isolates hVISA by one or more test methods not including PAP-AUC.

^dPercentage false positive at the given PAP-AUC range.

^eTrue positives.

The sensitivity and specificity of the Etest macromethod, Etest GRD, and BHI screen agar methods using PAP-AUC as the reference method are shown in Table 3. Although the specificities (i.e., fewer false positives) of the Etest macromethod and the Etest GRD were slightly better at 24 h (100% at 24 h versus 96% at 48 h and 99% at 24 h versus 97% at 48 h, respectively), the sensitivities (i.e., fewer false negatives) significantly improved at 48 h. However, the sensitivities of both the Etest macromethod and the Etest GRD were relatively low at 57%, correctly identifying only 12 of the 21 hVISAs by PAP-AUC. The performance of the BHI screen agar with casein and 4 μg/ml vancomycin varied between the two different inocula and between reading at 24 or 48 h. For both the 0.5 and 2.0 McFarland standard inocula, the sensitivity increased at 48 h, while the specificity decreased: 43% sensitive and 100% specific at 24 h and 91% sensitive and 94% specific at 48 h using a 0.5 McFarland inoculum and 76% sensitive and 89% specific at 24 h and 100% sensitive and 67% specific at 48 h using a 2.0 McFarland inoculum. All methods were performed in duplicate, with no more than a 3% difference in sensitivity and specificity occurring between the two sets of tests.

TABLE 3.

Sensitivity and specificity values for the different methods in determining hVISA

Method	Correct no. of hVISA isolates identified (n = 21)	No. of false positives	No. of false negatives	% sensitivity	% specificity
Etest macromethod, 24 h	3	0	18	14	100
Etest macromethod, 48 h	12	5	9	57	96
Etest GRD, 24 h	9	1	12	43	99
Etest GRD, 48 h	12	3	9	57	97
BHI screen agar, 0.5 McFarland, 24 h	9	0	12	43	100
BHI screen agar, 0.5 McFarland, 48 h	19	8	2	91	94
BHI screen agar, 2 McFarland, 24 h	16	13	5	76	89
BHI screen agar, 2 McFarland, 48 h	17	39	0	100	67

DISCUSSION

The clinical significance of hVISA has been debated since its initial description in 1997, and the uncertainty is likely to continue for the foreseeable future (2, 15, 16). Clinical MRSA infections with hVISA strains have been associated with treatment failure and increased length of hospital stay (5, 41). Since no commonly used clinical laboratory methods, including automated susceptibility testing systems, can identify hVISA, a practical, real-time method for detection of reduced susceptibility to vancomycin is imperative. However, given the continuum of resistance illustrated in Fig. 1, finding the perfect test to detect hVISA will be a challenge. In this survey of clinical MRSA isolates with vancomycin MICs of 2 μg/ml by the reference broth microdilution method, a BHI screen agar method was shown to have the highest sensitivity and specificity in detecting hVISA compared to the PAP-AUC reference method. Heteroresistance has been seen in strains with vancomycin MICs as low as 0.5 μg/ml; however, we chose to focus our study on a collection of clinical MRSA isolates with vancomycin MICs of 2 μg/ml since these strains demonstrate heteroresistance at a greater rate (1), and in a previous study we found no heteroresistance by PAP-AUC among MRSA isolates with vancomycin MICs of 0.5 and 1.0 μg/ml by the reference broth microdilution method (32). All methods evaluated reliably detected heteroresistance among a collection of MRSA isolates with vancomycin MICs of ≥4 μg/ml by standard Etest (data not shown).

The screen agar approach is an inexpensive and efficient way to test multiple clinical isolates on a daily basis. However, several screen agar methods differing in medium composition, inoculum density, and choice of antibiotic and antibiotic concentration have been evaluated for the detection of hVISA and demonstrated poor sensitivity (12). Currently, the only CLSI vancomycin screen agar method in place for clinical isolates for the detection of vancomycin-resistant S. aureus (VRSA) and possibly VISA is BHI agar containing 6 μg/ml vancomycin (BHIA6V), a method originally established for detection of vancomycin resistance in enterococci (39), and several studies showed this method to have a very low sensitivity for the detection of hVISA (44, 45). The CDC recommends this as a supplemental test for VISA detection, with the caveat that strains with vancomycin MICs of ≤4 μg/ml will not be reliably identified (http://www.cdc.gov/ncidod/dhqp/ar_visavrsa_labFAQ.html) and screen agar plates with a lower concentration of 3 μg/ml vancomycin have a very high false-positive rate (20).

In addition to the vancomycin concentration, the base medium of the screen agar appears to be important. The addition of supplements that enhance growth of hVISA could potentially improve the detection of hVISA by screen agar methods. Willey et al. found the addition of pancreatic digest of casein to BHI agar and 4 μg/ml vancomycin an improvement to other screen agars for the detection of VISA, and 97.7% of VISA strains in their study were successfully detected with high specificity within 24 h (42). The addition of 20% horse serum to BHI has also been suggested as a means to differentiate between hVISA and VSSA (14).

Although increasing the inoculum from a 0.5 to a 2.0 McFarland standard improved the sensitivity of the BHI screen agar method at 48 h (91% and 100%, respectively), there were 39 false positives with the higher inoculum when using the criterion for hVISA of two or more colonies growing in at least one of the four 10-μl droplets. Using a higher cutoff for the 2.0 McFarland inoculum (growth of 10 or more colonies from at least one of the four 10-μl droplets), the BHI screen agar method was 62% sensitive and 98% specific at 24 h and 81% sensitive and 90% specific at 48 h. The quadruplicate testing method (i.e., use of four 10-μl droplets and looking for growth in any one of the four droplets) enhanced sensitivity; up to 8/19 (42%) hVISA isolates might have been missed if only one droplet was inoculated on the screen agar plate.

The BHI screen agar method failed to identify only two isolates found to be hVISA by PAP-AUC. However, both isolates had a standard Etest teicoplanin MIC of 3 μg/ml (above the European Committee on Antimicrobial Susceptibility Testing [EUCAST] breakpoint of ≤2 μg/ml; www.eucast.org) and were hVISA by the Etest GRD teicoplanin MIC. Historically, S. aureus acquired resistance to teicoplanin, a glycopeptide that is widely used outside the United States, much earlier than resistance to vancomycin (18) and teicoplanin resistance is frequently accompanied by a small increase in vancomycin resistance (10). Additionally, Shlaes et al. (33) demonstrated that penicillin binding protein 2 (PBP 2) is overproduced in a teicoplanin-resistant S. aureus mutant strain compared with the level of production by its parent strain, and overproduction of PBP 2 is also observed in hVISA strain Mu3. Although MRSA strains that are resistant to teicoplanin can still remain susceptible to vancomycin and cross-resistance does not necessarily occur, all VISA strains identified to date have shown reduced susceptibility to teicoplanin (34), making teicoplanin a useful marker for VISA and, potentially, hVISA. Ultimately, in the absence of molecular determinants specific to hVISA, the use of more than one method may be necessary for optimal screening sensitivity.

The Etest macromethod and the Etest GRD incorporate steps to improve detection of hVISA, such as use of enriched media, prolonged incubation, both vancomycin and teicoplanin, and a higher inoculum (Etest macromethod). Use of these methods as routine clinical laboratory tests for hVISA detection has been increasing, yet no standardized diagnostic methodology exists for either. Additionally, interpreting the results of both Etest methods for hVISA can be subjective and variable, and performing replicates is costly. Initial studies suggested that the Etest macromethod was a potentially effective screening test for detection of hVISA (24, 28, 38); however, the accuracy of this method varies significantly between different studies (12), probably due to the criteria used to define hVISA (29) and the variability of the inoculum size utilized (24, 37, 44). Though we followed the manufacturer's recommended method and inoculum size, we found low sensitivity, similar to results reported by Adam et al. (1).

In conclusion, we found a BHI screen agar method that was easy to perform using standard media and inoculum sizes commonly utilized in clinical microbiology laboratories to be a useful method to identify hVISA. BHI agar plates with the addition of casein and 4 μg/ml vancomycin had the highest sensitivity and specificity in identifying hVISA compared to the two Etest methods when using PAP-AUC as the gold standard method. The sensitivity of detection improved significantly with incubation for 48 h, underscoring the apparent slow growth of hVISA isolates and the importance of longer incubation even at the expense of slightly lower specificity. Although the positive predictive value for this test was only 72%, 4/8 (50%) of the false positives were near the cutoff for hVISA by PAP-AUC (ratios of 0.83, 0.85, 0.86, and 0.89) and might have biologic relevance in the continuum of emerging resistance. Larger-scale studies to further evaluate the value of the BHI screen agar method for identification of hVISA in a clinical setting are warranted.