INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Nosocomial infections caused by methicillin-resistant strains of Staphylococcus aureus (MRSA) have become an important clinical problem worldwide. Monitoring and limiting the intra- and interhospital spread of MRSA strains require the use of efficient and accurate epidemiologic typing systems. A large number of DNA-based methods have been developed for typing MRSA strains. Pulsed-field gel electrophoresis (PFGE) analysis is an accurate, reliable, and discriminatory method used by many hospital and reference laboratories (1), but it is technically demanding and time-consuming. Compared to PFGE analysis, PCR-based typing methods offer the advantages of rapidity and simplicity. However, when evaluated according to recommended criteria (18, 20), several PCR-based methods used for typing MRSA strains have shown limitations, either in intercenter pattern reproducibility, as described for arbitrarily primed PCR (AP-PCR) analysis (22), or in discrimination, as found with PCR-restriction fragment length polymorphism (RFLP) analysis of the coagulase or protein A gene (7). Compared to the latter assays, repetitive-element PCR (rep-PCR) analysis based on multicopy elements of the staphylococcal genome has shown good reproducibility and discriminatory power in single-center studies (4-6, 15, 25). There is therefore a need to identify the most efficient of these recently developed PCR-based typing methods.

Ideally, laboratories conducting regional surveillance of MRSA infections should adopt a common genotyping system to compare typing results. This requires using a system with good interlaboratory reproducibility, implying standardization of protocols, reagents, PCR equipment, and DNA pattern analysis. Fragment pattern analysis using automated laser fluorescence analysis (ALFA) systems is promising, as it provides enhanced resolution and better normalization of patterns than agarose gel electrophoresis (3, 10, 11).

The aims of the present study were to (i) compare the performance of three rep-PCR typing methods targeting repetitive elements IS256 (6), 16S-23S rRNA (15), and rep-MP3 (5, 25), (ii) assess the reproducibility of these PCR assays when performed by different laboratories using a standard protocol and the same batch of reagents, and (iii) evaluate the advantages of PCR pattern analysis using an automated laser sequencer over analysis by agarose electrophoresis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. A collection of 50 MRSA isolates (Table 1), already characterized by PFGE analysis (SmaI), were selected from previously described collections (8, 19, 21, 23, 24; R. De Ryck, A. Deplano, C. Nonhoff, B. Jans, C. Suetens, and M. J. Struelens, Prog. Abstr. 9th Eur. Cong. Clin. Microbiol. Infect. Dis., 1999, abstr. P123, p. 117). These strains were subdivided into four groups: a group of isolates from a single hospital epidemic showing indistinguishable PFGE profiles (strains 1 to 5); two groups of epidemiologically related isolates (strains 6 to 11 and 12 to 15) displaying closely related PFGE patterns which differed by 3 DNA fragments (18, 20); and 30 epidemiologically unrelated isolates (strains 16 to 45) displaying PFGE patterns which differed by 7 DNA fragments, including the reference S. aureus strain NCTC 8325. This collection included five duplicate isolates (strains 46 to 50) used as a reproducibility panel.

PFGE analysis. SmaI macrorestriction analysis resolved by PFGE was performed by the coordinating center as previously described (8). Major PFGE types were designated by arabic numerals, and subtypes were designated by letter suffixes (Table 1).

Study design. This multicenter study was organized by the European Study Group on Epidemiological Markers. Twelve centers representing eight countries participated in the study. The study was coordinated at the National Reference Laboratory for Staphylococci, University of Brussels-Hôpital Erasme, Brussels. The MRSA strains were distributed by the coordinating center to the participating centers labeled with code numbers. Each center tested the coded isolates without knowledge of isolate origin or PFGE type using one or more rep-PCR typing methods. Each rep-PCR typing method was performed according to a standard protocol for DNA extraction, amplification, and electrophoresis and using the same batch of PCR primers with the local equipment and minor modifications of protocols (Table 2). For data analysis, each investigator first classified the coded isolates visually into PCR types based on previously defined criteria of pattern interpretation. Tables of type distribution and gel images were then reported to the coordinating center, where the codes were broken and centralized analysis was performed.

Inter-IS256 PCR typing. Inter-IS256 typing was performed on a bacterial lysate obtained in a three-step procedure with lysostaphin, proteinase K, and boiling, as described previously (6). Amplification was performed using Ready-To-Go RAPD beads (Amersham Pharmacia Biotech, Roosendaal, The Netherlands). The bacterial lysate was diluted 1:10, and 2 µl (40 to 80 ng) of this solution were used as the DNA template in the Ready-To-Go tube for the PCR as previously described (6).

Inter-16S-23S rRNA PCR typing. Typing of 16S-23S rRNA was performed on a bacterial lysate obtained in a three-step procedure with lysostaphin, proteinase K, and boiling, as described previously (15).

Rep-MP3 PCR typing. Rep-MP3 typing was performed on phenol-chloroform-purified DNA as described recently, and 200 ng of DNA was added to 25 µl of PCR mix as previously described (25).

Agarose analysis of PCR products. Electrophoresis of IS256 and Rep-MP3 PCR products was performed on a 1.5% ultrapure agarose gel (Life Technology, Merelbeke, Belgium), and 16S-23S rRNA PCR products were electrophoresed on 3% NuSieve gel (FMC Bioproducts, Rockland, Maine) in 0.5× TBE (Tris-borate-EDTA) buffer containing 0.5 µg of ethidium bromide per ml. A 100-bp DNA ladder (size range, 100 to 2,600 bp; catalog number 27-4001-01; Amersham Pharmacia Biotech, Roosendaal, The Netherlands) was included in every sixth lane as molecular size markers.

ALFA of PCR products. DNA extraction and PCR protocols were as above except that CY-5' fluorescently labeled primers were used and PCR products were resolved in the ALF-Express automated DNA sequencer (Amersham Pharmacia Biotech). Electrophoresis was performed in 5% Long Ranger gel containing 7 M urea (catalog number 50660; FMC Bioproducts) in 0.6× TBE buffer. One microliter of amplicon was mixed with 5 µl of gel loading buffer containing two internal CY-5'-labeled DNA size markers, 0.4 µl of a 100-bp fragment (Amersham Pharmacia Biotech) and 0.4 µl of a 1,064-bp fragment of the Escherichia coli small-subunit ribosomal DNA (rDNA) (kindly provided by H. Grundmann). The external size marker was a CY-5'-labeled 100-bp ladder (size range, 100 to 1,500 bp) (kindly provided by H. Grundmann). Electrophoresis conditions were 1,600 V, 38 mA, and 45 W at 45°C for 490 min.

Local analysis of agarose-generated PCR patterns. PCR patterns resolved by agarose electrophoresis were visually interpreted by each investigator. Determination of DNA banding pattern differences and classification of coded isolates into PCR types was done without knowledge of strain origin or PFGE typing results. Classification was performed twice, following two distinct interpretation rules: (i) patterns differing by one or more band were categorized into distinct types and (ii) patterns differing by two or more bands were categorized into distinct types, whereas patterns differing by a single band were considered as belonging to the same type (6).

Centralized visual analysis of data. Gel photographs and/or TIFF files of gel images were sent to the coordinating center together with a table showing the results of visual classification of the coded strains. Classification of patterns performed at the participating centers was checked at the coordinating center. The performance characteristics of each PCR method were evaluated after breaking the isolate codes by calculation of three performance indices (18) for each participating center. First, intracenter reproducibility (R) was defined as the proportion of duplicate strains assigned to the same PCR type (18). Second, discriminatory power was estimated by calculation of the discrimination index (D) (13) based on PCR type distribution among the 33 unrelated strains with distinct PFGE types. Third, the agreement of PCR type distribution of all MRSA isolates with PFGE type classification was assessed.

Centralized computer-assisted analysis of data. ALF files and photographs of agarose gels were converted into TIFF files using Image Master software (Amersham Pharmacia Biotech) and the ScanJet IIP system (Hewlet Packard, Brussels, Belgium), respectively. All TIFF files were imported as fingerprint type into a BioNumerics software database (Applied Maths, Kortrijk, Belgium). For each set of gel patterns produced by a participating laboratory, data were stored as a separate file of fingerprint experiment type, each containing 50 fingerprints. The database included a total of 19 fingerprint types combining different centers, PCR methods, and post-PCR analytic methods.

Computer-assisted analysis of PFGE patterns. TIFF files of SmaI PFGE patterns were imported into the BioNumerics database as fingerprint types and analyzed using the dice coefficient and the unweighted pair group method using the arithmetic averages clustering method. PFGE patterns with a dice coefficient of 80% (corresponding to 6 DNA fragment differences between patterns) were assigned to the same type (20).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Performance of rep-PCR typing methods resolved by agarose gel electrophoresis. Figure 1 shows rep-PCR patterns obtained at the best agarose center, defined as having obtained a PCR type classification closest to the PFGE classification. The performance characteristics of each rep-PCR method, based on visual type classification, varied widely among participating centers (Table 3). Using the one-band difference interpretation rule, inter-IS256 PCR and rRNA PCR typing achieved a reproducibility of 100% in two and three centers, respectively. Inter-IS256 PCR typing showed a D index ranging between 97 and 100%, whereas that of 16S-23S rRNA PCR ranged between 72 and 82%. The lower discrimination of the latter method seemed to be related to production of a limited diversity of DNA fragments which clustered in a narrow size range (Fig. 1). Rep-MP3 PCR typing showed low reproducibility and intermediate-level discrimination, compared with the two other methods. Using the interpretation criterion of a two-band difference, there was limited improvement in reproducibility but at the cost of a significant decrease in discrimination (Table 3).

View larger version (105K): [in this window] [in a new window]   FIG. 1.   Patterns, clustering, and type classification of the 45 MRSA strains by PFGE analysis and inter-IS256, 16S-23S rRNA, and MP3 PCR patterns resolved by agarose electrophoresis in the center obtaining the best performance with each method.

ALFA. ALFA of rep-PCR products revealed additional amplified DNA fragments in the range of 100 to 200 bp compared with the patterns obtained by agarose electrophoresis (Table 2). The level of intralaboratory reproducibility on duplicate strains analyzed with ALF was similar to that obtained in the best agarose center with each PCR typing method (Table 4). The reproducibility between these post-PCR analysis methods could not be formally compared because ALFA was performed by fewer centers than agarose analysis.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Dendrogram of similarity of MRSA strains based on ALFA of inter-IS256 PCR, 16S-23S rRNA PCR, and MP3 PCR patterns produced by laboratories with optimal performance score compared with classification in PFGE type. The cut-off value for type identity was based on mean duplicate similarity values (Table 4).

Interlaboratory reproducibility. A low interlaboratory reproducibility was observed with rep-PCR patterns resolved by agarose electrophoresis (Table 2). In general, MRSA strains exhibited different patterns by a given PCR method in different laboratories. In most cases, the same strain showed a common core of high-intensity amplified DNA fragments by a given PCR method in all laboratories, whereas less abundantly amplified DNA fragments were less reproducible. Among the rep-PCR typing methods tested, inter-IS256 PCR appeared to be less variable by center than the other methods (Table 2).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Standardization of molecular typing methods is important in order to support large-scale epidemiological surveillance of infectious diseases with epidemic potential. Collaborative studies have shown limited interlaboratory reproducibility of AP-PCR and PFGE analysis for typing S. aureus (2, 22, 24). This study aimed at comparing the intralaboratory performance and interlaboratory reproducibility of three rep-PCR methods proposed for typing MRSA strains. In principle, these rep-PCR techniques offer the advantages of rapidity, low cost, and ease of use compared with PFGE analysis.

In the present study, a collection of well-defined MRSA strains characterized by epidemiological data and PFGE types were tested in 12 laboratories using their own thermocyclers but standard rep-PCR protocols, with minor deviations in only five centers. These rep-PCR methods produced arrays of DNA fragments which differed markedly between laboratories for the same strain and method. However, the clustering of strains into related patterns was more reproducible, a finding similar to that of a previous intercenter comparison of AP-PCR fingerprinting of MRSA strains (22). As various models of thermocyclers were used in different centers, this parameter may be partly responsible for the lack of precise pattern reproducibility between laboratories, underlining the key role of accurate temperature cycling in this respect. Minor deviations from the standard protocol in some centers possibly also contributed to this low reproducibility among laboratories. In contrast, a recent multicenter study of AP-PCR typing of Acinetobacter spp. using very controlled and standardized reagents, markers, and protocols showed good reproducibility between laboratories, with an average duplicate strain reproducibility of patterns of above 85% (11). The degree of experience with PCR methods tested in different laboratories was also quite variable, and better results might have been achieved if the experiments had been carried out after a training period in each center.

In this study, the best-performing center with inter-IS256 PCR typing achieved intracenter reproducibility and discrimination similar to those of PFGE. The finding that MRSA strains from different parts of Europe were better discriminated by inter-IS256 PCR typing than by rep-MP3 PCR typing is in discordance with a previous study (25) on S. aureus strains from the United States, which had concluded that rep-MP3 PCR typing was more discriminatory than inter-IS256 PCR typing. We also previously observed a lower discriminatory power of the inter-IS256 PCR method on U.S. strains compared with European strains of MRSA (6). These conflicting results seem to indicate that the repeat elements of the S. aureus genome used as a target in this assay may exhibit geographic variation in their polymorphism. Additional mapping and sequencing studies would be required to test this hypothesis.

The use of automated sequencers for fragment analysis moderately improved the resolution and interlaboratory reproducibility of inter-IS256 PCR patterns compared with agarose gel analysis. It should be noted, however, that the two centers using ALFA of inter-IS256 patterns made no deviation from the standard protocol. Previous studies have demonstrated the advantages of pattern analysis using a fluorescently labeled oligonucleotide primer on an automated sequencer for rep-PCR typing of several gram-negative bacilli (12), Acinetobacter species (10, 11), and MRSA (3, 5, 12).

Results obtained in the present study confirm that inter-IS256 PCR typing, used in an experienced laboratory, can reach excellent performance for typing MRSA strains, comparable to that obtained with PFGE analysis (6). The inter-IS256 PCR method can be used as a rapid typing tool for the local epidemiology of MRSA strains. Its turn-around time is only 5 h, compared to 2 or 3 days for PFGE. Rep-PCR typing based on the 16S-23S rDNA spacer and rep-MP3 element provided less discrimination but could be used as alternative rapid MRSA screening methods. It does not appear useful, however, to combine these rep-PCR techniques, as this does not increase typing resolution significantly.

Our results suggest that it would be difficult to fully standardize these rep-PCR typing methods using current technology. More technically complex genotyping techniques, like amplified fragment length polymorphism and restriction-sequence analysis of multiple polymorphic genes, seem more promising in terms of reproducibility (14, 17). Full sequence determination of a polymorphic locus, such as the protein A gene (spa) (16), or multilocus sequence typing (10) appear very promising for S. aureus typing. However, these assays are more labor intensive and less appropriate for routine use than rep-PCR techniques. Their main advantages, on the other hand, are complete reproducibility, good discriminatory power, and concordance with epidemiological data. Furthermore, sequence data provide easy comparative interpretation combined with ease of exchange via electronic networks (9, 16). The performance of these novel techniques still needs to be compared to current DNA fingerprinting techniques, such as rep-PCR and PFGE. Multicenter studies should be designed to verify that these methods allow interlaboratory exchange of genotypic data for monitoring geographic spread of MRSA strains.