Development of a Real-Time PCR Protocol Requiring Minimal Handling for Detection of Vancomycin-Resistant Enterococci with the Fully Automated BD Max System

Abstract

Vancomycin-resistant enterococci (VRE) are an important cause of health care-associated infections, resulting in significant mortality and a significant economic burden in hospitals. Active surveillance for at-risk populations contributes to the prevention of infections with VRE. The availability of a combination of automation and molecular detection procedures for rapid screening would be beneficial. Here, we report on the development of a laboratory-developed PCR for detection of VRE which runs on the fully automated Becton Dickinson (BD) Max platform, which combines DNA extraction, PCR setup, and real-time PCR amplification. We evaluated two protocols: one using a liquid master mix and the other employing commercially ordered dry-down reagents. The BD Max VRE PCR was evaluated in two rounds with 86 and 61 rectal elution swab (eSwab) samples, and the results were compared to the culture results. The sensitivities of the different PCR formats were 84 to 100% for vanA and 83.7 to 100% for vanB; specificities were 96.8 to 100% for vanA and 81.8 to 97% for vanB. The use of dry-down reagents and the ExK DNA-2 kit for extraction showed that the samples were less inhibited (3.3%) than they were by the use of the liquid master mix (14.8%). Adoption of a cutoff threshold cycle of 35 for discrimination of vanB-positive samples allowed an increase of specificity to 87.9%. The performance of the BD Max VRE assay equaled that of the BD GeneOhm VanR assay, which was run in parallel. The use of dry-down reagents simplifies the assay and omits any need to handle liquid PCR reagents.

INTRODUCTION

Although enterococci are generally considered to be of low virulence, they can cause significant infections, including bloodstream infections, especially in the hospital setting and in severely diseased patients (1). Enterococcus faecalis and Enterococcus faecium together represent the third most prevalent nosocomial pathogens (2). Vancomycin-resistant enterococci (VRE) have spread rapidly since their discovery in the late 1980s (3). The European Antimicrobial Resistance Surveillance System (EARSS) reports VRE frequencies ranging from <2% to >20% (4). In the United States, according to the National Healthcare Safety Network (NHSN), in 2009 and 2010 roughly 80% of all E. faecium isolates associated with health care-associated infections (HAIs) were resistant to vancomycin (5).

VRE are an important cause of HAIs that cause significant mortality and that result in high economic costs in hospitals. The average increase in the rate of VRE-associated mortality has been published to be 2-fold (4). VRE infections cause diseases more serious than those caused by vancomycin-susceptible enterococci (4). Patients that are specifically vulnerable include those with hematological diseases (6), neutropenia (7), or immunosuppression (8) and patients with liver transplants (9, 10). Thus, increased mortality has been described when at-risk patients are colonized with VRE (8–10). Recent reports mainly emphasize implementation of correct infection control measures (11, 12). Scientific evidence on how to handle VRE is still low, and it is suggested that management should be done according to “local needs and resources” (13). On the other hand, active surveillance for at-risk populations in hospital settings has been suggested (4, 14), and surveillance programs are thought to be an important cornerstone in the prevention of infections with VRE. Active screening may help in preventing VRE infections (recently reviewed in reference 14). Screening is recommended for patients on critical care units and transplant and hematology wards as well as patients undergoing chronic dialysis (14). Active screening reduces VRE colonization rates and may save costs in the hospital.

The standard for active surveillance is detection of VRE by culture. However, this procedure is time-consuming, and rapid tests employing real-time PCR have been developed. Commonly used commercial tests include, e.g., the Cepheid GeneXpert System or the Becton Dickinson (BD) GeneOhm VanR assay. Molecular assays allow rapid reporting, yet a common limitation is the decreased specificity for vanB detection due to the presence of vanB-like genes in bacteria other than enterococci (15). For the BD GeneOhm VanR assay, a sensitivity of 98% but a specificity of only 87% has been observed when it is used with rectal swab samples, and it has been suggested that the assay is a good screening test in a population with mainly vanA-colonized patients, yet vanB-positive results require confirmation (16). In other studies, the sensitivity and specificity of this assay were in similar ranges (92 to 100% and 82 to 93%, respectively) (17–20), and most often it had decreased specificity for detection of vanB. Depending, of course, on the prevalence of VRE, it can generally be concluded that by now molecular vanA and vanB detection has a good negative predictive value but positive test results require confirmation by an independent method. Those properties allow molecular assays to be used for screening purposes.

Although various assays for the molecular detection of VRE are available, only one fully automated system currently allows easy and rapid detection without considerable molecular expertise (Cepheid GeneXpert). Other available assays need specific infrastructure, equipment for molecular assays, and trained staff, thus reducing the accessibility of rapid screening assays. A recent new device for fully automated molecular detection is the BD Max system, which combines nucleic acid extraction, PCR setup, and 5-color real-time PCR detection. The flexible programming allows user-developed assays to be run in an automated manner (21–25). However, those assays still require liquid handling and thus need a laboratory capable of performing molecular assays. In contrast, commercial in vitro diagnostic assays available for the BD Max platform make use of completely lyophilized reagents and do not require further pipetting.

Currently, no commercial assay for detection of VRE on the BD Max platform is available, and laboratory-developed protocols for primary samples have not yet been evaluated (26). Moreover, laboratory-developed PCR assays performed on the BD Max system have the limitations in handling outlined above. We therefore thought to establish a PCR suitable for detection of VRE on the fully automated BD Max platform. This assay should use commercially available, custom-made dry-down reagents to omit the need for liquid handling and allow optimal usage of the automated BD Max platform.

MATERIALS AND METHODS

Samples.

Three real-time PCRs for the detection of VRE (the BD GeneOhm VanR and type 1 and type 3 BD Max laboratory-developed PCRs) were evaluated. Therefore, rectal swab samples that had been sent to the medical microbiology department of a 2,000-bed tertiary care university hospital with a request for VRE detection were used in a retrospective analysis. Rectal swab samples were collected using liquid Amies elution swab (eSwab) collection systems (Mast Diagnostica, Reinfeld, Germany). For all samples, culture results obtained by 48 h of growth on the chromogenic medium VRE Select (Bio-Rad, Munich, Germany) followed by identification with matrix-assisted laser desorption ionization–time of flight mass spectrometry (Bruker Daltonik, Bremen, Germany) and confirmation of the presence of vanA or vanB by PCR (17) were available. The correctness of rectal swab sampling was assessed by positive growth on Columbia–5% sheep agar plates (BD, Heidelberg, Germany). Samples were included in this study on the basis of an available culture result. The first set contained 86 rectal eSwabs from 82 individual patients (4 samples were from repeat patients during different hospital stays) which were obtained between December 2014 and March 2015. The second set included 61 rectal eSwabs from individual patients collected between September 2015 and January 2016. Neither of the collections included samples from a proven or suspected outbreak. In the case of discrepancies between the results of culture and molecular detection, samples were replated once and enriched in BBL enriched thioglycolate medium (BD) prior to plating on chromogenic agar. For the final analysis, the culture data were taken as the “gold standard.” For spiking experiments, negative rectal eSwab samples were chosen and DNA isolated from a vanB-positive strain (ATCC 51229) using a QIAamp DNA blood minikit (Qiagen, Hilden, Germany) according to the manufacturer's recommendations was added in defined amounts (10 μl DNA plus 190 μl eSwab matrix per sample buffer tube [SBT]).

Primers and probes.

Primers and dually labeled probes were ordered from Eurogentec (Cologne, Germany). Primer stock solutions were prepared at 50 pmol/μl, except for the sample process control (SPC) primers, which were prepared at 10 pmol/μl. All dually labeled probes were diluted to 10 pmol/μl and stored at −20°C. The sequences of the primers used to detect vanA, vanB, and SPC are depicted in Table 1. The sample process control is included in the BD Max ExK DNA-2 kit (BD Diagnostics, Sparks, MD, USA) and encodes the Drosophila melanogaster scaffold protein gene (AC246497.1) cloned in a pUC119 vector sequence (U07650).

TABLE 1.

Recipes of type 1 and 3 BD Max assays for detection of vanA and vanB

Primer, probe, or component	Sequence (5′–3′)	Type 3 assaya			Primer concn (pmol)/tubeb in type 1 assay
		Stock solution concn (μM)	Vol (μl)/reaction mixture	Final concnc (nM)
Primer vanA-fw	ATG AAT AGA ATA AAA GTT GCA ATA CT	50	0.25	500	12.5
Primer vanA-rv	GGC GAG AGT ACA GCT GAA TA	50	0.25	500	12.5
vanA probe	HEX-CTC AGA GGA GCA TGA CGT ATC GGT-BHQId	10	1	400	10
Primer vanB-fw	GGA CAA ATC ACT GGC CTA CAT TC	50	0.25	500	12.5
Primer vanB-rv	CGC CGA CAA TCA AAT CAT C	50	0.25	500	12.5
vanB probe	FAM-ACC TAC CCT GTC TTT GTG AAG CCG-BHQI	10	1	400	10
Primer SPC fw	GGA TCT AGC CGT GTG CCC GCT	10	0.25	100
Primer SPC rv	GGC ATG GAG GTT GTC CCA TTT GTG	10	0.25	100
SPC probe	Cy5-TTG ATG CCT CTT ACA TTG CTC CAC CTT TCC T-BHQ2	10	0.5	200
5× MM			4.8
Distilled water			3.7
Total			12.5

^aLiquid MM with primers and probes.

^bThe assay mixture is aliquoted, lyophilized, and sealed in ready-to-use BD Max snap-in tubes. BD MMK (SPC) and dried primers and probes were used. BD MMK (SPC) contains the polymerase enzyme and primers and probe for amplification of SPC in channel 5 of the BD Max system. BD MMK (SPC) is separately added to the BD Max extraction strip.

^cAfter addition of 12.5 μl DNA extract by the BD Max system.

^dBHQ1, black hole quencher 1.

For maximal ease of use, we also obtained vanA and vanB primers and probes that were commercially (Eurogentec) prealiquoted and dried down in tubes that fit directly into the extraction strip of the BD Max system (for the composition of these ready-to-use snap-in tubes, see Table 1). In this format, SPC primers were not included, as they are contained in the BD MMK (SPC) enzyme mix (Becton Dickinson).

Preparation of BD Max PCR reagents.

Two different real-time PCR mixtures were established on the BD Max system. According to the system's manual, these are called (i) type 3, which includes liquid master mix (MM) with primers and probes and (ii) type 1, which includes MMK (SPC) and dried primers and probes (see Fig. S1 in the supplemental material).

The type 3 assay uses a liquid 2-fold-concentrated master mix in 12.5 μl that contains all necessary reagents. This was prepared as depicted in Table 1. For the polymerase, we used the real-time ready DNA probes master 5× (catalog number 05502381001; Roche Applied Science, Mannheim, Germany) that was combined with the primers and probes indicated in Table 1. Of note, this specific polymerase mix did not require a specific neutralization buffer to neutralize the alkaline DNA eluate (as otherwise recommended by Becton Dickinson for type 3 assay mixtures). The master mix was prepared as a stock solution, aliquoted, and stored at −20°C for up to 4 weeks. Interchangeably, the master mix was dispensed at 12.5 μl directly into snap-in tubes fitting the BD Max extraction strip and sealed using cover foil (Becton Dickinson) and a PlateMax sealer (2 times for 8 s each time at 180°C; Axygen).

In the second approach, we ran the real-time PCR as a type 1 assay. Commercially obtained dry-down primers and probes (Table 1) were used together with the BD MMK (SPC) polymerase (BD, Heidelberg, Germany), which includes a primer for detection of the sample process control (SPC).

BD Max PCR.

A total of 150 μl of eSwab medium was added into the SBT of the BD Max ExK DNA-2 extraction kit. SBTs were covered with a septum cap, vortexed, and placed into the sample rack. Reaction strips of the BD Max ExK DNA-2 kit were supplemented either (i) with the 2-fold-concentrated PCR master mix in position 3 (type 3) or (ii) with the BD MMK (SPC) tube in position 2 and the snap-in tubes containing dry-down primers and probes in position 3 (type 1) (see Fig. S1 in the supplemental material). For type 3 assays, the BD Max system combined 12.5 μl of the nucleic acid eluate with the 2-fold-concentrated PCR master mix, and the combination was then loaded into the PCR cartridge. For type 1 assays, the contents of the dry-down primer/probe tube were rehydrated in 12.5 μl neutralization buffer, the nucleic acid eluate (12.5 μl) was added, and the mixture was transferred to hydrate the BD MMK (SPC) tube (containing a polymerase and SPC primers and probes) prior to loading of the PCR cartridge. Approximately 4 μl of the PCR mixture is detected in the amplification chamber. The assay used default settings for the extraction. The PCR protocol for type 3 assays was 95°C for 60 s and 45 cycles of 98°C for 15 s and 60°C for 21.9 s. For type 1 assays, the initial heat activation was 95°C for 600 s and the other parameters were the same as those for type 3 assays. Fluorescence gains for channels 475/520 nm (6-carboxyfluorescein [FAM]), 530/565 nm (hexachlorofluorescein [HEX]), 630/665 nm (Cy5), and 680/715 nm (SPC control for the type 1 assay) were 40, 80, 40, and 80, respectively. The fluorescence thresholds used for threshold cycle (C_T) determination for vanA and vanB are indicated in footnote a of Table 2, and the SPC threshold was 75 (channel 630/665 nm, type 3 assays) or 200 (channel 680/715 nm, type 1 assays). A color compensation of 5% was programmed for FAM to HEX. The Roche PCR mix has a fluorescent dye as a pipetting control signaling in channel 680/715 nm, which was the reason to use a SPC probe detectable in channel 630/665 nm. For calculation of threshold cycles, we used the C_T at threshold crossing algorithm.

TABLE 2.

Combined sensitivities and specificities for BD GeneOhm VanR and type 1 and 3 BD Max assays

Gene and evaluation no.a	Assay	No. of samples with the following resultb:				Sensitivityc (%)	Specificityc (%)
		TP	FP	FN	TN
vanA
No. 1 (n = 86)	Type 3 BD Max	20	2	3	61	87.0	96.8
	BD GeneOhm VanR	20	1	3	62	87.0	98.4
No. 2 (n = 61)	Type 3 BD Max	21	0	4	36	84.0	100.0
	Type 1 BD Max	25	0	0	36	100.0	100.0
vanB
No. 1 (n = 86)	Type 3 BD Max	36	4	7	39	83.7	90.7
	BD GeneOhm VanR	36	5	7	38	83.7	88.4
No. 2 (n = 61)	Type 3 BD Max	27	1	1	32	96.4	97.0
	Type 1 BD Max	28	6	0	27	100.0	81.8

^aEvaluation no. 1 was done with fluorescence thresholds of 100 and 200 for vanA and vanB, respectively, for the type 3 BD Max assay; evaluation no. 2 used thresholds of 50 for both vanA and vanB for the type 3 BD Max assay and 100 and 50 for vanA and vanB, respectively, for the type 1 BD Max assay.

^bResults obtained in comparison to those obtained by culture. TP, true positive; FP, false positive; FN, false negative; TN, true negative.

^cInhibited samples were retested once at a 1:10 dilution. Resolved data were included in the calculation of sensitivity and specificity.

BD GeneOhm VanR PCR.

eSwab medium (150 μl) was used to inoculate the sample buffer tube of the BD GeneOhm VanR kit (BD, Heidelberg, Germany). The assay was done by exactly following the protocol of the manufacturer. Quantitative C_T data were taken from the licensed software with the help of BD; otherwise, the results were used as obtained from the results report.

Statistical analysis.

Calculations and graphs, including receiver operating characteristic (ROC) analysis, were done with Prism (version 5) software (GraphPad Software). McNemar's or Fisher's exact test was used for the analysis of 2-by-2 contingency tables.

RESULTS

The BD Max type 3 laboratory-developed assay shows sensitivity and specificity for detection of VRE equal to those of the BD GeneOhm VanR kit.

We thought to establish a PCR protocol able to detect vanA- and vanB-positive enterococci that could be used on the fully automated BD Max system. The sequences of the primers and hydrolysis probes for the real-time PCR were published previously (17). Recipes for the reagents are shown in Table 1. Initially, we evaluated a PCR that used a fully supplemented liquid master mix that was run as a type 3 assay. For simplified handling, this master mix could be stored in tubes fitting the BD Max extraction strip, as reported for other BD Max laboratory-developed PCR assays (21).

The PCR was first tested on 36 vanA- and vanB-positive isolates, all of which were detected correctly (data no shown). The intra-assay coefficients of variation (CV) were 1.0% for vanA and 1.7% for vanB, and the interassay CV were 2.1% for vanA and 5.8% for vanB. Next, we compared the type 3 BD Max PCR to the BD GeneOhm VanR Assay for direct detection of VRE from 86 rectal swab samples. Both molecular assays were run from 150 μl eSwab medium. Culture data were used as the gold standard for comparison and calculation of sensitivity and specificity. Twenty-three swab samples were positive for vanA, 43 were positive for vanB, and 20 were negative for VRE.

The type 3 BD Max assay used with the ExK DNA-2 kit for extraction showed higher inhibition rates than the BD GeneOhm VanR kit: 14.0% versus 0%. Most of the inhibited samples contained some amount of stool, and the PCR result could be solved upon repetition with a diluted sample. The final inhibition rate was 1.2% for the type 3 BD Max assay. The sensitivities of the BD Max assay and the BD GeneOhm VanR assay were equal: 87% for vanA and 83.7% for vanB. The specificities of the BD Max assay and the BD GeneOhm VanR assay were 96.8% and 98.4%, respectively, for vanA and 90.7% and 88.4%, respectively, for vanB (Table 2). Differences in sensitivity and specificity were not significant (McNemar's test, P > 0.05).

Comparison of the C_T values of both assays showed that the BD Max assay had slightly lower C_T values, but the sensitivity for detection of vanA and vanB was equal (Fig. 1A). Both assays showed a good correlation for detection of vanA (r² = 0.66) and a slightly lower correlation (r² = 0.57) for detection of vanB (Fig. 1B and C). For detection of vanA, each assay missed 3 positive specimens. Samples that were positive by only one PCR showed high C_T values (Fig. 1B). For detection of vanB, each assay missed 7 positive specimens, resulting in sensitivities below 85%. Samples that were positive only by the BD GeneOhm PCR and negative by the BD Max PCR also included samples with C_T values of <30, indicating at least moderate DNA input (Fig. 1C). Analyzing the samples that were missed by the corresponding type 3 BD Max PCR in more detail, we observed that those samples showed positive amplification yet had a low endpoint fluorescence below the threshold of 200. Both PCRs also gave false-positive results (4 and 5 for the type 3 BD Max PCR and the BD GeneOhm PCR, respectively), yet of those, 3 were positive by both PCRs.

FIG 1.

The type 3 BD Max VRE assay shows performance equal to that of the BD GeneOhm VanR assay. (A) Distribution of C_T values for 86 rectal swab samples analyzed by the type 3 BD Max VRE assay and the BD GeneOhm VanR assay. SPC, sample process control; IC, internal control. (B, C) Side-by-side comparison of C_T values for detection of vanA (B) and vanB (C) for both assays. neg, VRE-negative sample.

Side-by-side comparison of type 1 and type 3 BD Max PCRs.

To further simplify handling of the BD Max laboratory-developed PCR and in order to make full use of the automated setup, we next ran the BD Max PCR as the type 1 assay. Therefore, primers and probes of the same sequence and concentration were ordered commercially and directly aliquoted and dried in tubes fitting the BD Max extraction strip. For amplification we used the commercial BD MMK (SPC) polymerase (Table 1). Of note, in this mode no manual handling of liquids is required anymore. Based on the previous experiences, we lowered the thresholds for detection of vanA and vanB, as indicated in Table 2. We then evaluated custom-made, commercially ordered dry-down primers (type 1) against the type 3 BD Max PCR with lowered thresholds for detection of VRE from 61 rectal eSwab samples. The culture result was used as the gold standard, as before. Nineteen samples were positive for vanA, 22 were positive for vanB, and 6 showed both vanA- and vanB-positive VRE. With the type 3 assay, testing of 9 samples (14.8%) had to be repeated, confirming the relatively high inhibition rate observed before. In contrast, running of the assay as the type 1 assay with dry-down primers/probes and the BD MMK (SPC) enzyme showed only 2 (3.3%) inhibited samples. After one reanalysis, 2 and 0 samples remained inhibited by the type 3 and type 1 assays, respectively, indicating the increased robustness of the type 1 assay. The distribution of the C_T values showed that detection of the SPC was much more reliable and equal with the type 1 assay, confirming its increased robustness (Fig. 2A).

FIG 2.

Comparison of the in-house BD Max VRE assay run in two formats, type 1 and type 3. (A) Distribution of C_T values for 61 rectal swab samples analyzed by the BD Max VRE assay run either as the type 1 assay (dry primer mix combined with BD MMK [SPC] polymerase) or the type 3 assay (liquid master mix). (B, C) Side-by-side comparison of C_T values for detection of vanA (B) and vanB (C) for both types of assays. neg, VRE-negative sample.

Sensitivities for the BD Max type 1 assay were 100% for vanA and vanB in the second evaluation (Table 2). With the type 3 assay chemistry, we partly observed weakly developing curves when the amount of input DNA was low, specifically for vanB detection. We therefore lowered the thresholds for the type 3 assay for C_T value determination. This resulted in an increased sensitivity for vanB detection in the second evaluation compared to that in the first evaluation (96.4% versus 83.7%; P = 0.09, Fisher's exact test), whereas the sensitivity of vanA detection was equal in both evaluations (84% versus 87.0%; P = not significant). Nevertheless, the type 1 assay was still superior in sensitivity for detection of vanA (P = 0.045, McNemar's test), which might be related to the different enzyme (BD MMK [SPC]) used. For the type 1 and type 3 assays, specificities were 100% and 100%, respectively, for vanA detection and 81.8% and 97.0%, respectively, for vanB detection (Table 2). The difference in specificity for vanB was significant (P = 0.025, McNemar's test). C_T values showed a good correlation in a linear regression analysis, with r² values being 0.79 and 0.78 for vanA and vanB, respectively (Fig. 2B and C).

Technical adaptations and optimization for detection of VRE by the type 1 BD Max PCR.

A closer analysis of the type 1 format of the BD Max laboratory-developed assay revealed that, especially with the BD MMK (SPC) polymerase, we consistently observed a drop in endpoint fluorescence when small amounts of DNA were added in the assay (exemplified for detection of vanB in Fig. 3A). The resulting differences in the slopes of the curves might affect the C_T calculation. Therefore, we decided to do the C_T determination using the C_T by threshold calculation method instead of the default C_T by inflection point method, with the latter being a second derivative maximum algorithm. For the type 1 assay, we had used a low threshold of 50 for determination of positive vanB results. This assay showed a relatively low specificity of only 81.8% for detection of vanB. We therefore analyzed the C_T and endpoint fluorescence distribution of true- and false-positive results (Fig. 3B and C). We observed that samples with false-positive results (negative by culture and by the type 3 BD Max PCR) had higher C_T values (Fig. 3B) and a typically low endpoint fluorescence (Fig. 3C), although no perfect discrimination of true and false positives was possible with either of the parameters. We next performed a receiver operator characteristic analysis (Fig. 3D). From these data it can be concluded that use of a C_T value of 35 would result in improved positive/negative discrimination with a sensitivity of 100% (95% confidence interval [CI], 87.6% to 100%) and a specificity of 87.9% (95% CI, 71.8% to 96.6%) for detection of vanB.

FIG 3.

Detailed analysis of detection of vanB by type 1 BD Max VRE assay. (A) Amplification curves for vanB for a serial dilution of DNA extracted from a vanB-positive E. faecium isolate. Indicated is the amount of DNA added to the sample buffer tube (SBT). ΔF_n, normalized fluorescence of the FAM dye detected in channel 475/520 nm of the BD Max system. (B, C) Distribution of C_T values (B) or endpoint fluorescence (C) for samples positive by PCR and confirmed to be positive by culture (samples with true-positive [tp] results) or negative by culture (samples with false-positive [fp] results). Indicated are single data points and medians. (D) ROC curve for detection of vanB. Sensitivity and 1 − specificity for different C_T thresholds are shown, as is the area under the curve (AUC). (E) C_T values for detection of vanB for DNA extracted from a vanB-positive E. faecium isolate spiked into vanA- and vanB-negative rectal swab specimens (clinical matrix). The results are means + SDs for duplicates from two independent experiments.

Finally, the performance of the two different types of assays for vanB was also assessed by use of a clinical matrix spiked with pure DNA. Both PCRs showed nearly equal performance over a 4-log fold dilution, with efficiencies of 0.97 and 1.05 for the type 1 and type 3 assays, respectively (Fig. 3E).

DISCUSSION

In this report we describe the development and evaluation of an automated laboratory-developed PCR for the detection of vanA- or vanB-positive enterococci with minimal requirements for manual handling. We made use of the flexible programming option of the fully automated BD Max system to run the PCR assay. So far, this option is unique to the BD Max device. Whereas multiple PCR protocols for detection of VRE are available, only one system, the Cepheid GeneXpert system, allows the fully automated detection of VRE. The latter system, however, does not allow user-developed protocols to be run. Here we publish a protocol to be used for VRE detection on the BD Max system. Option 1 is to use a liquid master mix that contains all necessary reagents for the detection of vanA and vanB. This can be used by the BD Max as the so-called type 3 assay. For ease of use, the fully complemented master mix can be aliquoted in tubes that are directly compatible with the DNA extraction strip of the BD Max system. In this way, at the time of assay request, no liquid handling is necessary. However, the reagents have to be prepared in advance in an environment suitable for molecular diagnostics. We have shown before that the PCR enzyme used here is stable when being aliquoted and frozen at −20°C for up to 12 weeks (21).

We compared this assay to the well-established BD GeneOhm VanR assay using culture results as the standard. Both PCR assays showed comparable performance (Table 2), yet the BD Max type 3 assay had significantly higher inhibition rates when rectal eSwab specimens were used. We did not use the available stool extraction kit from BD, and we could resolve the results for most of the inhibited samples by use of a 1:10 predilution. In our study, data obtained with the BD GeneOhm VanR assay were worse than those published initially (overall sensitivity, 93.8%; specificity, 87.5%) (16); we also consistently observed a limited specificity for vanB detection (19). This is a limitation of many VRE assays due to the presence of vanB in bacteria other than enterococci (27). The sensitivity of vanB detection could be improved when the fluorescence threshold for a positive result was lowered, as was done in the second evaluation. Use of a threshold of 50 instead of 200 fluorescence units resulted in an increase in sensitivity for vanB to 96%. Thus, the type 3 BD Max assay was comparable to the BD GeneOhm VanR assay. The latter assay had been shown to be less sensitive but more specific for detection of vanA in comparison to the fully automated GeneXpert assay, whereas for vanB the BD GeneOhm assay was more sensitive (15). In the latter study, both assays had a very low specificity for detection of vanB (20.6% versus 14.7%). It was concluded that positive vanB results in particular have to be confirmed by culture. On the basis of the prevalence of either vanA or vanB enterococci, this will affect the way in which the assay can be used. We conclude that the type 3 BD Max assay shows sensitivities and specificities for VRE in the range of those of other commercially available PCRs and allows automated detection of VRE with no liquid handling at the time of assay request but has considerable inhibition rates when used with rectal swab specimens.

In a second step, we further improved the handling of the BD Max VRE assay by using custom-made, commercially available primers/probes dried down in tubes fitting the BD Max DNA extraction strip. In this format, the assay is run with a lyophilized BD PCR enzyme and the machine itself does all necessary pipetting steps, including hydration of primers and enzymes. This type of assay is run as the type 1 assay. We compared the two laboratory-developed versions (type 1 versus type 3) side by side. Of note, with the type 1 format we observed much reduced inhibition rates of only 3.3%, which are in the range of those of the BD GeneOhm VanR assay or the GeneXpert assay (15, 19). An important finding was that the BD enzyme especially showed a considerable drop in endpoint fluorescence with low-positive samples (Fig. 3). This resulted in miscalculation of the C_T values with the original C_T by inflection point algorithm. When we switched to the C_T at threshold crossing algorithm, more reliable threshold cycles were calculated. The type 1 assay showed sensitivity for the detection of vanA superior to that of the type 3 format, yet its specificity for vanB was lower. A closer analysis (Fig. 3B) revealed that false-positive vanB detection correlated with higher C_T values. An ROC analysis revealed that use of a C_T of 35 for discrimination of positive and negative samples results in an increase of the specificity to 87.9% (95% CI, 71.8% to 96.6%). Increased discrimination by adjusting the fluorescence threshold (Fig. 3C) might be an alternative. Alternatively, PCR-positive, culture-negative samples might also indicate a low bacterial burden, as has been suggested in a discrepant analysis using the GeneXpert system (28). Similar to our study, samples with C_T values of >34 often showed negative culture results. Whereas for Staphylococcus aureus skin carriage rates and, potentially, transmissions are associated with a higher nasal load (29, 30), so far no relationship between the VRE load and transmission has been reported. However, given that transmission of VRE does not occur easily (31), it might be reasonable to speculate that low-positive results are of reduced clinical importance and that maximum sensitivity in molecular VRE detection is not of upmost priority.

The limited sensitivity of the BD Max VRE assay might also have to do with the extraction efficacy for the DNA. Rectal swab samples can contain stool, a known difficult matrix for extraction. Moreover, we envisage that the DNA extraction in NaOH used by the BD Max system might interfere with downstream PCR amplification, despite the intrinsic neutralization step. Specifically when using the BD MMK (SPC) polymerase enzyme (type 1 assay), we observed a decreasing amplification efficacy for vanB at a low DNA input. Amplification curves with very low endpoint fluorescence might be missed when the threshold is set too high, and that seems to be the reason why we observed increased sensitivity when lowering the threshold (evaluation no. 1 versus evaluation no. 2).

Running the assay in a type 1 format omitted the need of any liquid handling within the laboratory and might provide an interesting alternative for user-developed assays. Meanwhile, this type of assay has already been commercialized for some BD Max assays. The comparison of the BD Max VRE type 1 and type 3 assays might also serve as blueprint for the facilitated usage of other, published laboratory-developed PCRs on the BD Max system.

The presented PCR can be used for the direct detection of VRE from rectal swab specimens, as evaluated in this report. Additionally, the same PCR can also be used for the confirmation of resistance conferred by vanA and vanB from culture isolates (as was done here on a limited number of isolates) or from positive blood culture bottles with Gram-positive cocci (data not shown). Similar approaches of combining culture methods or mass spectrometry with PCRs for the rapid detection of VRE have been reported (32, 33). The BD Max VRE PCR facilitates such diagnostic procedures by using automated PCR with minimal handling times.

Besides reduced manual handling, the BD Max assay also proved to be cheaper. Whereas the GeneOhm VanR assay costs $35/sample, the cost of the type 3 BD Max assay was about $23/sample. Running the assay in the type 1 format requires commercial primers/probes as well as the BD MMK (SPC) polymerase and would result in a cost of $31/sample. In addition to the savings for the assay reagents, additional benefit would come from reduced handling times, which should be about 15 min for 8 samples, as published for the respective assays of methicillin-resistant S. aureus (BD GeneOhm MRSA versus BD Max MRSA) (34). A limitation of the study is that it was done at only one center. Typing data were not available, but from occasional pulsed-field gel electrophoresis analyses, we assume a sufficient level of diversity exists in the analyzed cohorts.

In summary, the BD Max vanA and vanB PCR was evaluated for direct detection of VRE from rectal eSwabs. Its sensitivity equaled that of the BD GeneOhm VanR assay, yet specificity for vanB remains limited. The assay requires the use of threshold adaptations and the C_T at threshold crossing method for increased performance. Use of dry-down primers further simplifies the assay and omits any need to handle liquid PCR reagents. In this way, the assay combines both flexibility and automation at best for laboratory-developed tests on the BD Max system.