Multicenter Clinical Evaluation of VRESelect Agar for Identification of Vancomycin-Resistant Enterococcus faecalis and Enterococcus faecium

Abstract

A chromogenic medium for identification of vancomycin-resistant Enterococcus faecalis and Enterococcus faecium, VRESelect, was compared to bile esculin azide agar with 6 μg/ml vancomycin (BEAV) for the isolation of vancomycin-resistant enterococci (VRE) from stool specimens. At 24 to 28 h, VRESelect demonstrated 98.7% (confidence interval [CI], 96.1 to 99.7%) sensitivity and 99.0% (CI, 98.0 to 99.6%) specificity versus 85.1% (CI, 79.8 to 89.5%) and 90.1% (CI, 79.8 to 89.5%) sensitivity and specificity, respectively, for BEAV.

TEXT

Antibiotic resistance among nosocomial pathogens is an increasing problem within health care institutions. Vancomycin resistance within Enterococcus faecalis and E. faecium has proven to be particularly problematic (1). Acquisition of the vanA or vanB resistance cassette(s) confers resistance to vancomycin through altered cell wall biosynthesis (2). Rapid identification and isolation of patients who are colonized with vancomycin-resistant enterococci (VRE) have been shown to be essential for the prevention of hospital outbreaks (1, 3). Chromogenic agar has emerged as a rapid and cost-effective tool for identification of VRE from stool and rectal swab specimens (4, 5). In this study, we evaluated the utility of a novel chromogenic media, VRESelect (Bio-Rad Laboratories, Hercules, CA), which is able to distinguish vancomycin-resistant E. faecium (VREfm) from vancomycin-resistant E. faecalis (VREfs) based on the presence of specific chromogenic enzyme substrates within the agar. We compared VRESelect agar to bile esculin azide agar with 6 μg/ml vancomycin (BEAV; Thermo Fisher Scientific, Lenexa, KS) for the identification of VRE in stool specimens.

Stool specimens (n = 945) were collected in sterile containers from three geographically distinct clinical sites. Specimens were stored at 4°C and inoculated to VRESelect and BEAV within 3 days of collection using indirect inoculation into saline and then plating onto media using a separate swab or inoculating loop for each of the media. Plates were incubated at 35°C in ambient air and examined for growth at 24 to 28 h. BEAV plates were reexamined for growth at 48 h. Pink colonies on VRESelect were presumptively identified as VREfm (Fig. 1). Blue colonies on VRESelect were presumptively identified as VREfs if they were catalase negative (Fig. 1). Determination of catalase activity of blue colonies was necessary given the occasional growth of staphylococcal species with blue colony morphology. Colonies from BEAV that had a black halo were presumptively identified as VRE. Identification of VRE from both media was confirmed using Gram staining (Gram-positive cocci in chains [GPCCh]), catalase (negative), and pyrrolidonyl arylamidase (PYR) (positive). Further species level identification was performed using Vitek 2 (bioMérieux, Marcy l'Etoile, France), and vancomycin resistance was confirmed using a vancomycin Etest (MIC > 32 μg/ml) (bioMérieux) (2). The gold standard was defined as the presence of culture-confirmed VRE growing on either medium at 48 h.

Fig 1.

VRESelect medium plated with E. faecalis and E. faecium. This figure demonstrates the ability of VRESelect to identify E. faecium and E. faecalis based on the color of the colony; on VRESelect, colonies of E. faecalis appear blue and colonies of E. faecium appear pink. For the purposes of this study, fecal specimens were plated directly to VRESelect and BEAV and observed for growth at 24 and 48 h. A mixed culture containing vancomycin-resistant E. faecium (VREfm) and vancomycin-resistant E. faecalis (VREfs) was grown.

The performance of VRESelect and BEAV was evaluated from all three sites (Table 1). Following 24 to 28 h of incubation, the sensitivity and specificity of VRESelect were 98.7% (confidence interval [CI], 96.1 to 99.6%) and 99.0% (CI, 98.0 to 99.6%), respectively. The sensitivity and specificity of BEAV at 24 to 28 h were 85.1% (CI, 79.8 to 89.5%) and 90.1% (CI, 87.8 to 92.3%), respectively. Increasing the duration of BEAV incubation to 48 h increased sensitivity to 95.1% (CI, 91.3 to 97.5%), although specificity decreased to 84.0% (CI, 81.1 to 86.6%). Results from each test site were similar to the aggregate results (Table 2).

Table 1.

Analysis of VRESelect and BEAV for the identification of VRE from 945 samples (combined data)^a

Incubation period (h)	Medium	No. of samples				% sensitivity (CI)	% specificity (CI)	% PPV (CI)	% NPV (CI)
		TP	FP	TN	FN
24–28	VRESelect	219	7	716	3	98.7 (96.1–99.7)	99.0 (98.0–99.6)	96.9 (93.7–98.7)	99.6 (98.8–99.9)
	BEAV	189	71	652	33	85.1 (79.8–89.5)	90.1 (87.8–92.3)	72.7 (66.9–78.0)	95.2 (93.3–96.7)
48	BEAV	211	116	607	11	95.1 (91.3–97.5)	84.0 (81.1–86.6)	64.5 (59.1–69.7)	98.2 (96.8–99.1)

^aTP, true positives; FP, false positives; TN, true negatives; FN, false negatives; CI, 95% confidence interval; PPV, positive predictive value; NPV, negative predictive value.

Table 2.

Analysis of VRESelect and BEAV for the identification of VRE from 945 samples (individual trial site data)^a

Clinical site and no. of specimens sampled	Medium and incubation time (h)	No. of samples				% sensitivity (CI)	% specificity (CI)	% PPV (CI)	% NPV (CI)
		TP	FP	TN	FN
Clinical site A (nb = 270)	VRESelect (24–28)	72	2	194	2	97.3 (90.6–99.6)	99.0 (96.4–99.9)	97.3 (90.6–99.6)	99.0 (96.4–99.9)
	BEAV (24–28)	67	19	177	7	90.5 (81.5–96.1)	90.3 (85.3–94.1)	77.9 (67.7–86.1)	96.2 (92.3–98.5)
	BEAV (48)	73	38	158	1	98.7 (92.7–99.8)	80.6 (74.4–85.9)	65.8 (56.2–74.5)	99.4 (96.5–99.9)
Clinical site B (n = 351)	VRESelect (24–28)	88	3	259	1	98.9 (93.9–99.8)	98.9 (96.7–99.8)	96.7 (90.7–99.3)	99.6 (97.9–99.9)
	BEAV (24–28)	74	27	235	15	83.2 (73.7–90.2)	89.7 (85.4–93.1)	73.3 (63.5–81.6)	94.0 (90.3–96.6)
	BEAV (48)	81	43	219	8	91.0 (83.1–96.0)	83.6 (78.5–87.9)	65.3 (56.3–73.6)	96.5 (93.2–98.5)
Clinical site C (n = 324)	VRESelect (24–28)	59	2	263	0	100.0 (93.9–100.0)	99.3 (97.3–99.9)	96.7 (88.6–99.5)	100.0 (98.6–100.0)
	BEAV (24–28)	48	25	240	11	81.4 (69.1–90.3)	90.6 (86.4–93.8)	65.8 (53.7–76.5)	95.6 (92.3–97.8)
	BEAV (48)	57	35	230	2	96.6 (88.3–99.5)	86.8 (82.1–90.6)	62.0 (51.2–71.9)	99.1 (96.9–99.9)

^aTP, true positives; FP, false positives; TN, true negatives; FN, false negatives; CI, 95% confidence interval; PPV, positive predictive value; NPV, negative predictive value.

^bNumber of specimens tested at site.

After 24 to 28 h of incubation, false positives were observed on both VRESelect and BEAV (Table 3). A total of 7 of 945 isolates were confirmed as false positives from VRESelect; 3 were E. faecalis or E. faecium with vancomycin MIC's lower than 32 μg/ml, 3 were PYR-positive GPCCh other than E. faecalis or E. faecium, and 1 had a Gram stain morphology other than GPCCh. A total of 71 of 945 colonies were identified as false positives from BEAV; 2 were E. faecalis or E. faecium with MIC's lower than 32 μg/ml, 55 were PYR-positive GPCCh other than E. faecalis or E. faecium, 11 were PYR-negative GPCCh, and 3 had Gram stain morphology other than GPCCh. Increasing the incubation time of the BEAV agar to 48 h increased the false-positive rate from 7.51% to 12.28% (116/945 samples).

Table 3.

Analysis of false positives at 24 h for VRESelect and BEAV

Medium	No. of samplesa
	False positive at 24 h	Gram stain morphology other than GPCCh	PYR-negative GPCCh	PYR-positive GPPCH other than E. faecium or E. faecalis	E. faecium or E. faecalis with MIC < 4 μg/ml
VRESelect	7	1	0	3	3
BEAV	71	3	11	55	2

^aGPCCh, Gram-positive cocci in chains; PYR, pyrrolidonyl arylamidase.

False negatives were observed on both media after 24 to 28 h of incubation. Three isolates failed to produce typical morphology on VRESelect, while typical VRE colonies were present on BEAV agar and were confirmed as VREfm. Thirty-three isolates failed to produce typical morphology on BEAV after 24 to 28 h of incubation, while typical VRE colonies were present on VRESelect agar. Of the 33 false negatives from BEAV, 10 were VREfm strains and 1 was a VREfs strain that grew only on VRESelect agar. The remaining 22 isolates were confirmed as VREfm after 48 h of incubation on BEAV.

This multicenter prospective study demonstrated higher sensitivity (98.7% versus 85.1%) and specificity (99.0% versus 90.1%) of VRESelect agar than of BEAV agar after 24 to 28 h of incubation. Increasing the duration of incubation to 48 h increased the sensitivity of BEAV, though decreased specificity was also noted. Given that VRESelect agar had a high sensitivity after 24 to 28 h of incubation, additional incubation was not required for this medium.

While false positives were present on both media, they differed in type and number. Of interest, only 3 false positives from VRESelect were PYR-positive GPCCh compared to 55 from BEAV. This group of false positives consisted primarily of Enterococcus gallinarum and Enterococcus raffinosis. Other studies have also demonstrated the ability of these organisms to grow on chromogenic VRE media, particularly on BEAV (4–7). They are identified as false positives only when tests beyond the routine Gram stain and PYR disk are performed. They are of concern since, in routine practice, they are likely erroneously reported as VRE.

Another potential reason for false positives from VRESelect is misinterpretation of colonies of staphylococcal species with blue colony morphology as VRE. It is essential that catalase testing be performed on these colonies to confirm that they are truly VRE. Occasionally, however, testing was unable to be performed due to the growth of only a single colony. Without catalase testing, these colonies would be erroneously identified as VRE.

Also of note, three isolates with typical VRE morphology on both VRESelect and BEAV failed to confirm as VRE when ancillary testing (catalase, PYR, etc.) was performed using BEAV. These isolates were confirmed to be VRE when ancillary testing was performed using VRESelect. A possible explanation is that the BEAV plates were mixed, containing VRE as well as organisms with similar colony morphologies. While these patients were correctly identified as VRE carriers, the results of antibiotic susceptibility testing might have been misleading.

While rapid and cost-effective, the utility of tests using chromogenic media is dependent on the reliability of organism identification. VRESelect medium provides sensitive and specific results within 24 to 28 h, allowing the rapid detection of VRE carriers and the institution of appropriate infection prevention measures in hospitalized and nursing home patients.