|Home | About | Journals | Submit | Contact Us | Français|
Health care-associated infections with methicillin-resistant Staphylococcus aureus (MRSA) contribute to significant hospitalization costs. We report here a droplet digital PCR (ddPCR) assay, which is a next-generation emulsion-based endpoint PCR assay for high-precision MRSA analysis. Reference cultures of MRSA, methicillin-susceptible S. aureus (MSSA), and confounders were included as controls. Copan swabs were used to sample cultures and collect specimens for analysis from patients at a large teaching hospital. Swab extraction and cell lysis were accomplished using magnetic-driven agitation of silica beads. Quantitative PCR (qPCR) (Roche Light Cycler 480) and ddPCR (Bio-Rad QX100 droplet digital PCR system) assays were used to detect genes for the staphylococcal protein SA0140 (SA) and the methicillin resistance (mecA) gene employing standard TaqMan chemistries. Both qPCR and ddPCR assays correctly identified culture controls for MRSA (76), MSSA (12), and confounder organisms (36) with 100% sensitivity and specificity. Analysis of the clinical samples (211 negative and 186 positive) collected during a study of MRSA nasal carriage allowed direct comparison of the qPCR and ddPCR assays to the Cepheid MRSA GeneXpert assay. A total of 397 clinical samples were examined in this study. Cepheid MRSA GeneXpert values were used to define negative and positive samples. Both the qPCR and ddPCR assays were in good agreement with the reference assay. The sensitivities for the qPCR and ddPCR assays were 96.8% (95% confidence interval [CI], 93.1 to 98.5%) and 96.8% (95% CI, 93.1 to 98.5%), respectively. Both the qPCR and ddPCR assays had specificities of 91.9% (95% CI, 87.5 to 94.9%) for qPCR and 91.0% (95% CI, 86.4 to 94.2%) for ddPCR technology.
Health care-associated infections are a major burden on hospitals worldwide and contribute to significant morbidity, mortality, and health care costs. In the United States, methicillin-resistant Staphylococcus aureus (MRSA) has increased in both community- and health care-associated infections (1, 2), and the threat of community-associated MRSA entering the hospital environment has lent an urgency to efforts to limit the spread of this organism (3). MRSA has become the most frequent cause of skin and soft tissue infections in emergency rooms, and it is associated with a variety of other infections, including nosocomial pneumonia and surgical-site and bloodstream infections (1, 4). Estimates of the impact of MRSA on morbidity and mortality vary from 100,000 to 250,000 hospital infections and 18,650 deaths per year (1, 2). Many of these infections are associated with the colonization of mucous membranes, skin, wounds, or the gastrointestinal tract.
One way to bring down health care costs and improve medical care is to demonstrate advances in the technologies that are needed to improve the efficiency and cost effectiveness of detection. Molecular assays have gradually begun to replace traditional microbiological culture methods in the identification of medically important pathogens. Several factors must be considered in the selection of a rapid MRSA screening test: the prevalence of MRSA in the patient population, the testing resources available, and the presence of a strong infection control policy. Once colonized patients are identified, a health care institution can isolate colonized patients, begin decolonization, and notify staff of the need for increased attention to hand hygiene (5, 6, 7, 8). The major contributor to the success of infection control policies is the turnaround time (TAT) advantage of PCR-based methods, which is typically 2 to 4 h, compared to standard microbiological cultures using selective agars, which is usually 24 to 48 h. In one study in surgical wards, MRSA transmission was shown to be significantly reduced when culture-based testing in the laboratory was replaced by real-time PCR assays (9). Real-time PCR-based assays have been proven to lead to a reduction in the incidence of MRSA disease (10, 11, 12).
This report introduces droplet digital PCR (ddPCR) technology, a next-generation emulsion-based endpoint PCR assay for high-precision MRSA analysis. ddPCR technology was compared to culture methods and two real-time PCR platforms, the Roche light cycler and the Cepheid GeneXpert. The assays were compared for their ability to distinguish MRSA from methicillin-susceptible S. aureus (MSSA) organisms and other potentially confounding organisms, as well as their ability to identify MRSA isolates of various genetic backgrounds. The evaluation was performed with collections of clinical isolates containing clonal lineages that are prevalent in community-associated and nosocomial MRSA infections. In addition, we implemented a single-step magnetic swab extraction and lysis protocol to rapidly process samples from nasal swabs for ddPCR analysis. The results represent the first demonstration of the ddPCR approach, a new quantitative assay, for sensitive and specific detection of MRSA.
Reference strains were obtained from the American Type Culture Collection and included methicillin-susceptible and methicillin-resistant S. aureus strains, as well as coagulase-negative staphylococcal species (Table 1). Additional reference strains representing well-characterized U.S. strains of MRSA described by Tenover and Goering (13, 14) were obtained from Ashley Robinson. The MRSA strains employed in this study represented five staphylococcal cassette chromosome mec element (SCCmec) types and subtypes (I, II, III, IV, and IVa).
We selected 60 clinical isolates of methicillin-resistant S. aureus from an epidemiological surveillance collection at a large Veterans Affairs (VA) hospital. Isolates were catalogued by their genus and species, year of collection, and site of isolation. Isolates were initially identified by conventional microbiological methods as coagulase positive and DNase positive. Resistance to methicillin was determined by measuring the oxacillin MIC for the organism, with resistance defined as an MIC of >2 μg per ml (15). Colonies were picked from Trypticase soy agar plates supplemented with 5% sheep erythrocytes (blood agar plates [BAP]) and stored at −80°C in cryovials containing a preservative (Microbank; Pro-Lab Diagnostics, Austin, TX).
Remnant samples from a university hospital were also employed. For specimen collection, the Copan dual swabs were used to sample each nostril as is recommended for testing with the Cepheid Xpert MRSA kit. Following processing, extra swabs were stored at 4°C until all molecular testing was complete. All patient identifiers were removed, and samples were coded by SCCmec target cycle threshold (CT) value.
Liquid Stuart Copan swabs were used to collect specimens for analysis. Swab extraction and cell lysis were accomplished using magnetic-driven agitation of silica beads. To perform this, the tip of each swab specimen was placed in a 2.0-ml cryovial containing 200 μl Tris-EDTA (TE) buffer, 100-μm silica beads, and a magnetic tumble stir element. The tube was inserted into an eight-tube holder that was positioned above a horizontal magnetic cylinder unit. The magnetic cylinder was powered to 5,000 rpm for 3 min to achieve coupled swab extraction and cell lysis. The supernatant was collected and used immediately for amplification studies or was stored at −80°C for future analysis.
For some experiments, DNA was prepared from a large-volume culture of S. aureus following cell lysis and spin column purification performed according to the manufacturer's directions (Qiagen DNA extraction kit). DNA concentrations were determined spectrophotometrically using a NanoDrop spectrometer (Fisher Scientific). Three concentrations of MRSA DNA, in genomic equivalents (copy number) of 106, 104, and 102 per ml, were prepared from this stock and stored in aliquots at −80°C. The molecular weights of MRSA DNA were estimated based on published data for 11 completely sequenced S. aureus genomes (accession no. NC_002951.fna, NC_013450.fna, NC_002952.fna, NC_002953.fna, NC_003923.fna, NC_009782.fna, NC_002758.fna, NC_002745.fna, NC_007795.fna, NC_007622.fna, NC_009641). The calculation of genomic equivalents (copy number) was based on the assumption that the average mass of a base pair is 650 Da.
The primers and probes were designed to detect sequences within the staphylococcal SA0140 protein gene (SA) and the gene for methicillin resistance (mecA). S. aureus SA0140 protein VIC-labeled 6-carboxytetramethylrhodamine (TAMRA)-quenched primer-probe sets were obtained from Applied Biosystems, Inc. The primers and 6-carboxyfluorescein (FAM)-labeled-black hole quencher 1 TaqMan probes were synthesized by Integrated DNA Technologies (IDT), Inc. Of three mecA and five SA primer-probe sets tested, the following primer-probe sets were selected for use in all subsequent assays: for mecA, forward primer 5′AACCACCCAATTTGTCTGCC3′ and reverse primer 5′ TGATGGTATGCAACAAGTCGTAAA3′ (nucleotide positions 38690 to 39416, 134-bp region in mecA gene, MRSA sequence type 228 [ST228], GenBank accession no. HE579073.1) and internal probe 5′CCTTGTTTCATTTTGAGTTCTGCAGTACCGG3′; for SA, forward primer 5′TGGGATTGTTACTAGCGAATCATGT3′ and reverse primer 5′CGTACCAAAGTCGAATGCCA3′ (nucleotide positions 136247 to 136428, 181-bp region in MRSA ST228 gene for SA0140 protein, GenBank accession no. HE579073.1) and internal probe 5′TGCGCGATTAAGTA3′.
Two types of amplification reactions were performed. Standard qPCR assays were performed in a Roche Light Cycler 480, and ddPCR assays were performed using a Bio-Rad QX100 droplet digital PCR system as described in reference 16. Standard TaqMan primer-probe chemistry was used for both amplification reactions. The reaction mixture was prepared with a 2× ddPCR Master mix (ddPCR Supermix for probes; Bio-Rad), 20× primer-probe set (900/250 nM), and DNA template in a final volume of 20 μl, as described previously (16). For qPCR, an initial denaturation step (95°C, 10 min) was followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. For ddPCR, the 20-μl reaction mixture was loaded into the sample well of an eight-channel disposable droplet generator cartridge (Bio-Rad). A total volume of 60 μl of droplet generation oil was loaded into the oil well for each channel, and the cartridge was placed in a droplet generator as recommended by the manufacturer (Bio-Rad Laboratories, CA). Following the generation of water-in-oil droplets, droplets collected in the droplet well were transferred with a multichannel pipette to a 96-well PCR plate. The plate was heat sealed with foil and placed in an Eppendorf thermocycler, and an initial denaturation step (95°C, 10 min) was followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. After PCR, the 96-well plate was loaded on a droplet reader (Bio-Rad Laboratories, CA), which automatically aspirated droplets from each well of the plate.
Droplets containing duplexed TaqMan assays were streamed through a droplet reader with a detector capable of simultaneous two-color detection (Bio-Rad Laboratories, CA). All droplets were gated based on their detector peak width to exclude rare outliers (e.g., doublets, triplets). Specific cleavage of the TaqMan probes for either mecA sequences (FAM-labeled probes) or SA sequences (VIC-labeled probes) generated a strong fluorescence signal. A threshold was assigned to each droplet based on the fluorescence amplitude (16). The Quantasoft analysis software (Bio-Rad Laboratories) was used to analyze the ddPCR data, as described previously (16).
Limit of detection (LoD) assays were performed on two MRSA strains (BAA-39 and BAA-40). As a potential confounder, the MSSA quality control strain ATCC 27659 was used. Staphylococcus epidermidis strain ATCC 12228 (an mecA-negative quality control strain) was used for the Cepheid GeneXpert MRSA validation protocol (as found in the product insert) and was also employed as a potential confounder in LoD assays. Saline suspensions of isolates from overnight cultures grown on BAP were prepared to a density equivalent to a 0.5 McFarland turbidity standard (Densichek Plus; bioMérieux, Inc., Durham, NC). Following an initial 1:100 dilution, suspensions were made in a series of 10-fold dilutions. Colony counts (in CFU/ml) were determined by plating on BAP. The LoD of each MRSA isolate in pure culture, as well as in the presence of confounder MSSA or mecA-negative S. epidermidis strains, were determined. Confounders were grown overnight on BAP, suspended in saline to a density equivalent to a 0.5 McFarland standard, and diluted to approximately 106 CFU/ml in each MRSA test dilution. Dilutions were absorbed onto two liquid Stuarts Copan swabs (part no. 900-0370; Cepheid) and DNA was extracted for analysis by PCR. To express LoD as CFU/reaction mixture, the volume absorbed by the swabs (reported by the manufacturer as approximately 75 μl) was confirmed by measuring the difference in the weights of 100-μl aliquots of saline before and after absorbance for 10 s by each of 10 swabs. The average of 10 replicates was taken as the average absorbed sample. By definition, LoDs were the lowest concentrations that gave positive results in the qPCR and ddPCR assays.
Some remnant samples producing discordant results were further analyzed. DNA from the Copan swabs, obtained by magnetic lysis, was used. In these studies, an independent set of mecA gene primers (forward primer 5′ATTGGGATCATAGCGTCA and reverse primer 5′ATTTATTATATTCTTCGTTACTCA, nucleotide positions 39393 to 40163, 770-bp region in mecA gene of MRSA ST228, accession no. HE579073.1) was used to validate discordant results from qPCR and ddPCR assays. These primers were specific for the mecA gene and overlapped the original mecA primer-probe set by 26 bases. Amplicons were purified and subjected to fluorescent dye terminator sequencing on an ABI Prism 3730xl DNA analyzer (SeqWright, Houston, TX). Sequences from both strands were obtained and identified by an NCBI BLAST nucleotide search to verify amplicon origins.
Primers and probes were designed to detect mecA sequences in the qPCR and ddPCR assays. All primer-probe sets were initially screened, employing various annealing times and temperatures. Well-characterized ATCC isolates of MRSA and MSSA were employed for optimization of the assay conditions. Screening of reference staphylococcal species other than S. aureus showed that the SA1 primer-probe set was specific for MRSA and MSSA isolates. Of the five potential mecA primer-probe sets tested, all detected known MRSA isolates. One mecA primer-probe set was found to have compatible annealing and extension profiles with the SA1 primer-probe set and was used in subsequent assays.
A number of reference nonstaphylococcal species were tested for reactivity to the two selected primer-probe sets, including both Gram-positive and Gram-negative microorganisms (Table 2). The qPCR and ddPCR assays had 100% sensitivities for the five SCCmec types and subtypes (I, II, III, IV, and IVa). Ten isolates that are representative of various U.S. strains were also tested and detected by both the mecA and SA primer-probe sets. S. epidermidis ATCC 35984, a mecA-positive isolate, was detected by the mecA primer-probe set but not the SA primer-probe set as a control.
Reproducibility between assays was assessed using a panel of dilutions of highly purified MRSA DNA. Assays were conducted with genomic copies equivalent to 1 × 106 (3.1 ng DNA/reaction mixture), 1 × 104 (31 pg/reaction mixture), and 1 × 102/reaction mixture (0.31 pg/reaction mixture). These three samples and a negative template control (Fig. 1) were run in triplicate for each qPCR and ddPCR assay. For the mecA assay, the percentages (±standard deviations [SD]) of positive droplets were 95.6 (±1.8%), 18.0 (±5.2%), and 0.23% (±0.06%) at 1 × 106, 1 × 104, and 1 × 102 genomic equivalents, respectively. For the SA assay, the percentages (±SD) of positive droplets were 99.4 (±0.6%), 18.7 (±5.7%), and 0.23% (±0.08%) at 1 × 106, 1 × 104, and 1 × 102 genomic equivalents, respectively.
Known concentrations of MRSA organisms were absorbed on swabs and extracted using magnetic lysis. Both qPCR and ddPCR assays, with three replicas per sample, consistently detected MRSA alone at 6 copies per reaction mixture. Limit-of-detection studies indicate that the ddPCR assay achieves a positive result with 95% confidence for a swab containing 90 to 100 CFU. Both the qPCR and ddPCR assays were linear throughout the range of 100 CFU/swab to 1 × 106 CFU/swab (data not shown). When MSSA (105 CFU/reaction mixture) was added to a known concentration of MRSA, no change in the limit of detection was observed for the mecA gene in either the qPCR or ddPCR assays, although there was an increase in the SA1 baseline signal, as expected. Substitution of S. epidermidis ATCC 12228 as a potential confounder at 105 CFU/reaction mixture for MSSA resulted in no change in the mecA or SA1 signals.
MRSA, MSSA, and microbiologically identified organisms were selected from our collection of clinical isolates (Table 2). These isolates were obtained from purity cultures in a clinical laboratory as part of the standard of care for patients. All patient identifiers were removed prior to collection and long-term storage of the isolates at −80°C. A total of 60 culture-identified MRSA isolates and 12 culture-identified MSSA isolates were grown overnight on blood agar plates. Individual colonies were selected and applied to Copan swabs. These swabs were tested by both qPCR and ddPCR for the presence of mecA- and SA-specific sequences. Of the 60 MRSA culture isolates tested, 2 showed no evidence of the presence of mecA by PCR. These two discordant samples were retested by microbiological culture techniques, including growth on CHROMagar and catalase and coagulase tests; susceptibility to methicillin was determined by disc diffusion testing, as described above. Based on this retesting, both discordant culture specimens were reclassified as MSSA, consistent with results from both the qPCR and ddPCR assays. There were no differences between qPCR and ddPCR measurements. The analytical validation parameters of the qPCR and ddPCR assays compared to direct culture and biochemical testing revealed a sensitivity (agreement with positive results) and specificity with culture methods of 100%.
Analyses of remnant samples collected for the molecular detection of MRSA nasal carriage allowed the direct comparison of both qPCR and ddPCR assays (CT value and droplet counts, respectively) to the Cepheid MRSA GeneXpert assay (CT value). A total of 397 clinically relevant unique samples were collected in the study. Cepheid MRSA GeneXpert values were used to define reference negative and positive samples. As shown in Table 3, both the qPCR and ddPCR assays were in good agreement with the reference assay. The sensitivities for the qPCR and ddPCR assays were 96.8% (95% confidence interval [CI], 93.1 to 98.5%) and 96.8% (95% CI, 93.1 to 98.5%), respectively. The qPCR assay had a specificity of 91.9% (95% CI, 87.5 to 94.9%), while the ddPCR had a specificity of 91.0% (95% CI, 86.4 to 94.2%).
Both qPCR and ddPCR produced false-positive results (17 for qPCR and 19 for ddPCR) compared to the reference Cepheid MRSA-negative samples. Since it was not possible to culture these remnant samples, we sought to confirm the presence of the mecA gene by amplification of the discordant samples, employing an additional set of mecA primers and sequence analysis. Of the 17 samples that were positive in both qPCR and ddPCR analyses, seven produced amplicons of the appropriate size for the second set of mecA primers. The amplicons were purified and sequenced. All seven sample sequences showed >99% homology to known mecA sequences in the NCBI database.
We have developed a sensitive and specific assay for MRSA using third-generation PCR technology: the droplet digital PCR system. The superior sensitivity and specificity of PCR has made it the method of choice for the rapid detection of MRSA in a clinical setting (17, 18). A duplex PCR assay was designed that combined a set of primers for chromosomal mecA and an unlinked S. aureus-specific gene (SA0140) for the detection of MRSA. Our initial studies employed culture-proven organisms, including control strains from the ATCC, as well as from our collection of clinical MRSA isolates that were identified by culture techniques. Testing of 60 stored MRSA isolates resulted in the confirmation of 58 as MRSA, while two were reclassified as MSSA. While these two discordant isolates might represent misidentifications by microbiologic techniques, it is possible that they might have lost the mecA gene during prolonged storage, as has been reported previously (19). The presence of mecA has been reported in several coagulase-negative staphylococcal (CoNS) species, including S. epidermidis (20, 21), Staphylococcus haemolyticus, and Staphylococcus simulans (22). For this reason, specificity was further examined by testing of MSSA strains, as well as both methicillin-susceptible and methicillin-resistant CoNS strains. Our assay readily detected mecA in a known methicillin-resistant strain of S. epidermidis. The SA-specific gene probe did not cross-react with any CoNS species. No nonstaphylococcal Gram-positive or Gram-negative bacterial species were detected by either the mecA or SA primer-probe sets with this assay. Previous studies have employed methicillin-resistant CoNS (MRCoNS) and methicillin-sensitive CoNS (MSCoNS) strains to evaluate false-positive detection with in-house real-time PCR targeting the orfX-SCCmec junction (23), as well as the GeneXpert and GeneOhm assays (24, 25, 26). In our case, we lost no sensitivity or specificity for our assay when MRSA cultures were spiked with either MSSA or MSCoNS.
The analytical sensitivity of our assay is comparable to the LoD values reported for other PCR-based assays. The LoDs for the Cepheid GeneXpert have been reported for four strains of MRSA collected on Copan swabs and were compared to values for direct and enrichment cultures employing Transwabs with Amies transport medium plated on MRSASelect chromogenic agar and tryptic soy broth (27). The Cepheid GeneXpert detected 610 CFU/ml, corresponding to 58 CFU/swab, compared to 750 CFU/ml and 40 CFU/ml with the direct and enrichment cultures (27). Subsequent studies have reported the LoD for MRSA in the GeneXpert assay to be 109.4 CFU/assay (24) and 3,300 CFU/ml and 250 CFU/swab (26). In comparison, our assays, both qPCR and ddPCR, detected MRSA at 4 × 103 CFU/ml, equivalent to 3 × 102 CFU/swab and 6 CFU/reaction mixture in a background of 105 CFU/ml of S. epidermidis.
The presence of false-positive samples has been reported previously for the GeneXpert MRSA assay. Compared to direct agar culture methods, Wolk and colleagues (24) reported 15.7% PCR-positive culture-negative specimens. There are several explanations for this discrepancy, such as nonviable organisms that are present at cultivation detected by PCR, the presence of SCCmec variants, or low densities of organisms (24, 25). Both the GeneXpert MRSA and GeneOhm assays were found to have false positives due to mecA deletion strains (28, 29). The detection of methicillin-resistant coagulase-negative (MRCoN) organisms, such as S. haemolyticus, S. epidermidis, Staphylococcus hominis, or Staphylococcus sciuri, might also account for false positives. Although generally regarded as apathogenic, such organisms have been increasingly recognized in central nervous system shunt infections, native or prosthetic valve endocarditis, urinary tract infections, and endophthalmitis (30), as well as in bloodstream infections (31, 32, 33). In our studies of remnant samples, it was not possible to confirm the presence of MRSA by culture, since all viable organisms in the sample were consumed in the extraction process. However, it was possible to evaluate the remaining DNA samples for the presence of the mecA gene using an independent set of primers. Of 17 discordant samples tested, seven were ultimately shown to contain mecA by sequence analysis and were not empty-cassette variants, as has sometimes been noted (23, 34). If these additional seven samples do indeed represent true-positive samples, then the sensitivity and specificity of our assay should improve compared to the GeneXpert MRSA assay. The selection of an appropriate screening assay for MRSA should take into consideration the constantly changing epidemiology of MRSA, as well as MSSA and CoNS, given the presence of mecA elements in such organisms.
Recently, interventions to prevent the spread of MRSA in health care settings have resulted in a decline in the number of MRSA infections (35, 36, 37, 38). However, prevalence rates of MRSA remain unacceptably high, and there has been a push to implement expanded surveillance of this pathogen (39). In the Veteran's Administration system, this has taken the form of a “MRSA bundle,” which includes universal screening by PCR-based MRSA assay of nasal swabs, use of contact precautions, and heightened awareness of appropriate hand hygiene practices. The goal of this broad-based effort was to make infection control and prevention a priority for everyone involved in patient care. Following its implementation, the frequency of health care-associated MRSA declined sharply, with a 62% reduction of infections in intensive care unit (ICU) patients and a 45% reduction in non-ICU patients (40). Other hospitals and institutions have taken a more targeted approach and screen certain high-risk ICU patients, employing a slightly different protocol (41). Unlike in the VA study, the later NIH-based study did not show a major decrease in MRSA infections. As discussed by Platt (42), these two studies leave the issue largely unresolved. There are major differences in the design and execution of each, which might contribute to the opposing results.
D. Sullivan has a research contract with Bio-Rad Laboratories (Hercules, CA). P. Belgrader is currently employed by Bio-Rad Laboratories (Pleasanton, CA).
Funding for this work was obtained from the National Institutes of Health, grant 1R01EB010106-01A1.
Published ahead of print 17 April 2013