Evaluating the usefulness of spa typing, in comparison with pulsed-field gel electrophoresis, for epidemiological typing of methicillin-resistant Staphylococcus aureus in a low-prevalence region in Sweden 2000–2004

Authors


Corresponding author and reprint requests: A. C. Petersson, Clinical Microbiology and Immunology, Lund University Hospital, SE-221 85 Lund, Sweden
E-mail: ann-cathrine.petersson@skane.se

Abstract

Clin Microbiol Infect 2010; 16: 456–462

Abstract

The usefulness of spa typing was evaluated in relation to pulsed-field gel electrophoresis (PFGE), as a tool for epidemiological typing of methicillin-resistant Staphylococcus aureus (MRSA) in a low-prevalence region in southern Sweden. Bacterial isolates from 216 MRSA cases, newly identified in 2000–2004, were studied. The isolates were obtained from infected patients (31%), and from colonized individuals found by screening (69%). In total, 49 spa types and 73 PFGE patterns were identified. The discriminatory power of spa typing was lower (94.9 ± 1.8%) than that of PFGE (97.3 ± 1.2%). For two spa types (t002 and t008) the Panton–Valentine leukocidin results added useful discriminatory information. The most common spa types were t044 (n = 31; four PFGE patterns), t002 (n = 24; 10 PFGE patterns), t067 (n = 12; four PFGE patterns), t050 (n = 12; one PFGE pattern), and t324 (n = 11; one PFGE pattern). Epidemiological investigations identified 91 single cases and 39 transmission chains, each involving two to 13 cases. All the transmission chains were held together both by spa and PFGE typing. Among the 91 single-case isolates, 33 spa types and 50 PFGE patterns were unique (matchless) at the time of identification. The low prevalence of MRSA, the low number of outbreaks, and the wide spectrum of strains due to frequent acquisitions abroad (49% of the cases), makes spa typing a useful complement to epidemiological investigations in our setting. However, we still recommend the continued use of PFGE for further discrimination of isolates with identical spa types when epidemiological data can not exclude possible transmission.

Introduction

The increasing prevalence of methicillin-resistant Staphylococcus aureus (MRSA) is a cause for concern in many countries. In 2004, MRSA accounted for 25–50% of S. aureus septicaemia cases in southern Europe (EARSS; http://www.rivm.nl/earss). Although the MRSA rate is low in Sweden, the number of cases reported has been increasing each year (http://www.smittskyddsinstitutet.se/) [1]. Since 2000 MRSA has been notifiable in Sweden, both from infection and colonization. The Swedish Institute for Infectious Disease Control (SMI) continually surveys the prevalence and geographical spread of MRSA, and isolates from every newly reported case are referred to SMI for epidemiological typing.

Pulsed-field gel electrophoresis (PFGE) of macro restriction fragments is a commonly used technique for epidemiological typing of many bacterial species. Analysis of SmaI-digested DNA is still considered to be the reference standard for typing MRSA, and the method has proven superior to most other typing methods [2,3]. It was the first molecular typing method for which guidelines for interpretation of data were suggested [4]. Even though the effects of such standardization have been far-reaching, low interlaboratory comparability is still a problem [5].

Sequence-based typing methods such as multilocus sequence typing (MLST) [6], and spa typing, sequencing of the polymorphic X-region of the S. aureus protein A gene [7–9], have become frequently used alternatives to PFGE. This is due to the portability of sequence data and ease of exchanging results via databases available on the internet (http://www.mlst.net and http://www.spaserver.ridom.de). However, MLST, which relies on sequence analysis of fragments of seven housekeeping genes, is best suited for studying the evolutionary history of S. aureus [10,11]. The single locus analysis performed in spa typing gives information that has proven adequate in hospital settings [9,12,13].

The usefulness of spa typing, in comparison with PFGE, for early detection of transmission has, to our knowledge, so far not been evaluated in a low-prevalence region. In this study, we included all MRSA consecutively referred to the department of Clinical Microbiology and Immunology (CMI), Lund University Hospital, Sweden, during 2000–2004. Detailed epidemiological information, collected by the local infection control units, formed the basis of our evaluation. All isolates were further characterized by the presence or absence of genes coding for Panton–Valentine leukocidin (PVL). Representative isolates were subjected to MLST.

Materials and Methods

Bacterial isolates

A total of 216 MRSA isolates were analysed, each representing a case detected through clinical or screening samples referred to CMI from 2000 to 2004. The isolates were identified as MRSA by an in-house conventional PCR using the primers nuc N1-f (5′-GCGATTGATGGTGATACGGTT-3′), and nuc N2-r (5′-CAAGCCTTGACGAACTAAAGC-3′), adapted from Brakstad et al. [14], and mecA P3-f (5′-GGTACTGCTATCCACCCTCAA-3′) and mecA P4-r (5′-CTTACTGCCTAATTCGAGTGCTA-3′). Staphylococcus aureus CCUG 35601 served as positive control for the nuc/mecA PCR and S. aureus ATCC 49775 for the lukS-PV/lukF-PV PCR.

Epidemiological investigations

CMI serves a region of c. 700 000 inhabitants. MRSA cases were categorized as either ‘Infected’ (cases with clinical infection) or ‘Colonized’ (cases found through screening). Cases that had recently (defined here as within the preceding 6 months) been employed or treated in a hospital or nursing home were defined as ‘Healthcare related’. Cases with no such contacts were defined either as ‘Community related’ or ‘Unknown’ when a hospital connection could not be excluded. Recent immigration, or travel to a foreign country, was defined as ‘Acquired abroad’.

Screening samples were routinely taken from (i) patients and medical staff who had recently been employed or treated in a hospital or nursing home outside Scandinavia, or in a hospital or nursing home in Scandinavia known to have MRSA patients, (ii) fellow patients and staff at the same ward(s) as healthcare-related MRSA cases, and (iii) household members of community-related cases. Samples were taken from the anterior nares, throat, perineum, skin lesions, intravenous and stoma sites and urine from catheterized patients. Newly discovered MRSA cases were diagnosed at or referred to the Departments of Infectious Diseases at Lund University Hospital, Kristianstad Central Hospital, or Helsingborg Hospital. Epidemiological investigations were performed in collaboration with the Hospital Infection Control Units.

spa Typing

spa Typing was performed at CMI, as described elsewhere [9], using the primers SPA1-f 5′-AAGACGATCCTTCGGTGA-3′ (adapted from [8]), and SPA2-r 5′-CACCAGGTTTAACGACAT-3′ [7]. In 2004, SPA2-r was substituted by SPA3r 5′-AGCAGTAGTGCCGTTTGC-3′ (in-house). SPA-F9-f 5′-AACGTAACGGCTTCATCC-3′ was introduced when insufficient PCR product was obtained from spa type t355 isolates. These primers correspond to nucleotides 1094–1111, 1492–1475, 1533–1516 and 1067–1084, respectively, of S. aureus NCTC 8325-4 (GenBank J01786). All sequencing reactions were carried out using the ABI PRISM BigDye terminator cycle sequencing ready reaction kit (Applied Biosystems, Foster City, CA, USA). The ridom staphtype® software (Ridom GmbH, Würzburg, Germany) was used for sequence analysis and assignment of spa types [9].

PFGE

PFGE analysis was performed at SMI according to standard procedures [15]. Briefly, SmaI-digested DNA was electrophoresed in 1% agarose in 0.5× TBE at 14°C for 23 h, using the CHEF Mapper XA system (Bio-Rad Laboratories, Hercules, CA, USA) set at 6 V/cm, with pulse times linearly increased from 5 s initial switch time to 60 s final switch time. SmaI-digested DNA from S. aureus NCTC 8325 was included as normalization standard on every gel. Ethidium bromide-stained gels were photographed over UV light with a charge-coupled device camera. The DNA banding patterns were included in a national MRSA–PFGE database, using the bionumerics software, version 4.61 (Applied Maths NV, Sint-Martens-Latem, Belgium). Pair-wise similarities were calculated using the Dice coefficient, and the algorithm UPGMA (Unweighted Pair Group Method using Arithmetic averages) was used for constructing dendrograms. Position tolerance and optimization were both set at 1%. Each distinguishable banding pattern, within the size range 48–679 kb, was assigned a PFGE pattern name as in the following examples: (i) Bel EC-3a, UK E15, UK E16 (patterns indistinguishable from DNA banding patterns of MRSA in the HARMONY collection [5]), (ii) SE03-5, SE03-5b (patterns not seen in the HARMONY collection, but in two or more MRSA isolates in the national collection; SE = Sweden, 03 = 2003, the year when MRSA with this PFGE pattern was first isolated in Sweden, 5 = a serial number for each unrelated SE03 pattern, b = an alphabetical suffix added to denote close relationship (>80% pair-wise Dice similarity) to SE03-5 or SE03-5a), (iii) Unique 001, Unique 002, etc. (patterns unrelated to any other pattern within the database).

MLST

MLST was performed, as previously described [6], on MRSA isolates representative of each SE prototype PFGE pattern. Sequence types (STs) were determined using the MLST database (http://www.mlst.net).

PVL genes

The PVL genes lukS-PV and lukF-PV were detected by PCR as described [16].

Statistical analysis

The discriminatory capacities of spa typing and PFGE were measured with Simpson’s index of diversity [17] with 95% CI as previously described [18]. Concordance between the two methods was calculated as described elsewhere [19]. All calculations were performed using all single-case isolates (n = 91) and the first isolate in each epidemiologically defined transmission chain (n = 39). Chi-square test was used for comparison of proportions.

Results and discussion

The number of new cases, identified at CMI, increased each year, from 11 in 2000 to 81 in 2004, corresponding to an increase in incidence of c. 2–12 new cases per 100 000 inhabitants per year. This low incidence allowed for the implementation of an extensive infection control programme including early case-finding through extensive screening of patients with risk factors, and contact tracing combined with infection control measures such as isolation of MRSA cases. Generous inclusion of people in the community seeking medical care made culturing from this group feasible, and allowed early identification of new cases [1]. This may explain why a minority (66/216; 31%) of the isolates in this study were from infected cases. Skin and soft tissue infections dominated (53/66; 80%), followed by urinary tract infections and respiratory tract infections (6/66; 9% each). Only one isolate was obtained from a blood culture. The epidemiological investigations identified 105 (49%) cases as community related and 96 (44%) as healthcare related. The origin of acquisition could not be established for 15 (7%) cases, 12 of whom were immigrants who had recently visited their countries of origin. Infected cases were more frequently community related (44/66; 67%) than healthcare related (15/66; 23%), whereas colonized cases were more often healthcare related (81/150; 54%) than community related (61/150; 41%). The proportion of infected cases was significantly higher among community-related than healthcare-related cases, 42% vs. 16% (p <0.001). Half of the cases (107/216; 49%) had acquired their MRSA abroad, from a wide range of countries (Fig. 1).

Figure 1.

 Characteristics of 216 MRSA isolated from infected or colonized cases in a low-prevalence region in Sweden 2000–2004. An Unweighted Pair Group Method using Arithmetic averages (UPGMA) dendrogram shows the relatedness between the 72 pulsed-field gel electrophoresis patterns seen among the isolates. The scale shows percent similarity. a: (S), single case; (#n), transmission chain number. b: Country of acquisition when not Sweden: AF, Afghanistan; AU, Australia; BA, Bosnia-Herzegovina; CL, Chile; CO, Colombia; DE, Germany, EG, Egypt; ES, Spain; ET, Ethiopia; FR, France; FRY, Federal Republic of Yugoslavia until February 2003, thereafter Serbia-Montenegro; GM, Gambia; GR, Greece; HR, Croatia; HU, Hungary; IN, India; IQ, Iraq; IR, Iran; IT, Italy; KOR, Korea; LK, Sri Lanka; LV, Latvia; MT, Malta; PA, Panama; PH, the Philippines; PL, Poland; PRC, People’s Republic of China; RU, Russia; SA, Saudi-Arabia; SO, Somalia; SY, Syria; TN, Tunisia; TT, Trinidad and Tobago; UAE, United Arab Emirates; UK, United Kingdom; USA, United States of America. c: 2 (#4); 2 (S), 5 (#5), 2 (#6), 2 (#7). d: 6 (S), 2 (#13), 2 (#14), 2 (#15), 2 (#16). e: DE, ES, FR, GR, HU, TT. f: 2 (S), 2 (#17), 2 (#18); 1 (S); 12 (#19); 2 (#20); 1 (S). g: 1 (S); 7 (S), 3 (#22), 3 (#23), 3 (#24), 1 (#25), 2 (#26), 3 (#27), 3 (#28), 2 (#29). h: FRY, IQ, MT, SA, SY, TN, UAE.

Among the 216 MRSA isolates, we identified 49 spa types and 72 PFGE patterns (Fig. 1). One of the PFGE patterns, SE97-4, was represented by two DNA banding patterns differing in the high molecular weight part of the gel but indistinguishable within the normalized part of the gel. The concordance between spa and PFGE typing was 96% (both when using 72 and 73 PFGE patterns in the calculation), which is in agreement with other reports [20,21].

In total, 88 spa type/PFGE pattern combinations were recorded. t044/SE94-1 (ST80) was the most common combination (n = 27). t044 was also seen in combination with SE94-1b (n = 2), SE94-1d (n = 1) and SE94-1m (n = 1). A similar situation, with different PFGE patterns among t044 isolates, has been described in a Danish report [22]. SE94-1 was also described in combination with t042 (n = 1) and t131 (n = 4), which are both closely related to t044. All t044, t042 and t131 isolates were PVL-positive. ST80/t044 PVL-positive MRSA is widely spread, and often community acquired in Europe [22–26]. The majority of our t044/SE94-1 isolates were found in families with connection to the Middle East, and nearly all of the imported cases were from countries around the Persian Gulf.

t002/SE99-1 (ST5) was the second most common combination (n = 14). t002 was seen in another ten isolates showing nine different PFGE patterns that clustered with SE99-1 at 67% similarity in the UPGMA dendrogram. In total, 11 spa types and 23 PFGE patterns were recorded among 51 isolates belonging to MLST clonal complex 5 (CC5; here STs 5, 125, 149 and 228). This great variation was not surprising considering the evolution and worldwide spread of MRSA belonging to CC5 [10,11]. All but two of the isolates belonging to CC5 were PVL-negative. The two PVL-positive isolates were from cases with a history of recent travel to Asia (t002/Bel EC-3a) and Africa (t311/SE01-23).

The third most common combination (n = 12) was t050/SE97-3f (ST45). Twenty isolates had spa types (n = 8) and PFGE patterns (n = 5) closely related to t050 and SE97-3f, respectively. A majority of cases with ST45 isolates (24/32) had acquired their MRSA in Sweden, and none of them had connections with countries outside Europe. ST45 MRSA isolates have frequently been reported from many European countries [20–22,27]. All ST45 isolates were PVL-negative.

Also common during the study period were PVL-negative ST72 MRSA with PFGE pattern SE98-3, in combination with either spa type t324 (n = 11) or t148 (n = 2). These isolates were from children adopted from Korea, and from members of their Swedish families. In total, four transmission chains, no. 4 (2 cases), no. 5 (5 cases), no. 6 (2 cases) and no. 7 (2 cases), involved ST72 MRSA.

PVL genes were detected in 13/49 spa types and 17/72 PFGE patterns. In total, 31% of the isolates (25% of the single-case isolates and 38% of the transmission-chain isolates) were PVL-positive. Isolates with the same PFGE pattern were either all PVL-positive or all PVL-negative. The same homogeneity was seen for all spa types except t002 and t008, for which the PVL results added useful discriminating information. The PVL-positive MRSA variants found in our material have been reported from countries around the world [23,26–28]. The PVL-positive t008/SE03-5 (ST8; n = 7) MRSA was identified as a variant of USA300 [Hæggman, et al. 47th Intersci Conf Antimicrob Agents Chemother, abstract C2-143], the predominant community-associated MRSA in the USA [28].

Epidemiological investigations identified 39 transmission chains, involving two to 13 cases, and 91 single cases (Table 1 and Fig. 1). Twenty-three PFGE patterns and 25 spa types were identified among the 39 transmission chains. Twenty-eight of the 36 small chains (two to five persons) involved transmission within families only. Most of these family members (85/97) had had no known recent contact with the healthcare system. The remaining eight small chains (no. 11: t010/SE02-1, no. 12: t312/SE02-1, no. 14: t002/SE99-1, no. 15: t002/SE99-1, no. 20: t295/SE97-3f, no. 21: t032/UK E15, no. 25: t044/SE94-1, -1b, and no. 28: t044/SE94-1) involved one or more healthcare-related cases, and in five of them (nos 12, 20, 21, 25 and 28) the index case had a clinical infection. The three larger chains all involved healthcare-related cases. Chain no. 3 (t174/SE04-2, ST1) involved two patients and four personnel in a nursing home. Chain no. 10 (t067/SE00-9c, ST125) involved seven patients and two personnel in a hospital ward. The index patient in this chain had recently been hospitalized in Spain, where ST125 MRSA was already prevalent at that time [29]. The largest chain (no. 19) involved five patients and five members of the personnel in a nursing home, and three relatives of the personnel. Twelve of the isolates had PFGE pattern SE97-3f (ST45) and spa type t050 (repeat pattern 08-16-02-16-34-34-17-34-16-34), and one isolate, obtained from one of the relatives, differed slightly both by PFGE (SE97-3aq) and spa typing (t026, repeat pattern 08-16-34). This was the only transmission chain that included more than one spa type. More than one PFGE pattern was seen in another three chains, nos 2 (t030), 16 (t002) and 25 (t044). In total, 35 of the 39 chains were held together by PFGE and 38 by spa typing (Table 1). However, allowing closely and possibly related PFGE patterns within a transmission chain [4] and related spa repeat patterns, yielded full concordance between epidemiological data and typing results (Fig. 1).

Table 1.   Concordance between epidemiological data and typing results for 125 MRSA isolates involved in 39 epidemiologically defined transmission chains
No. of cases per transmission chainNo. of transmission chainsNo. of spa typesNo. of PFGE patternsNo. of transmission chains comprising only isolates indistinguishable by
spa TypingPFGE
  1. aThe three deviant transmission chains were chain nos 2, 16 and 25, all three comprising closely/possibly related PFGE patterns (Fig. 1).

  2. bOne isolate, given the same classification with both typing methods, differed slightly from the other twelve isolates in this transmission chain (no. 19).

 21914121918a
 31047108a
 466666
 511111
 611111
 911111
1312b2b0b0b

When considering the 91 single-case isolates, the concordance between epidemiological data and typing results is not very high, although it is much higher for PFGE than for spa typing. The number of spa types and PFGE patterns that were unique (matchless) at the time of identification were 33 and 50, respectively.

In our study, the discriminatory power of PFGE was higher than that of spa typing (97.3 ± 1.2% vs. 94.9 ± 1.8%). Numerous studies, based on various collections of MRSA isolates, have been designed to compare different methods and their discriminatory power. Grundmann et al. [3] stated that PFGE was as discriminatory as MLST (97.6% vs. 95.7%). Two studies found spa typing to be as discriminatory as PFGE (97% vs. 96%) [21,30], whereas Stommenger et al. [20] found spa typing (96.9%) less discriminatory than PFGE (99.3%) but still better than MLST (93.1%). Even though the discriminatory power of spa typing was lower than PFGE in our study, we found the method useful as a complement to the detailed epidemiological investigations that were feasible in our low-prevalence setting, characterized by few outbreak situations and high numbers of acquisitions abroad. However, we still recommend continued use of PFGE, for example in reference laboratories, for further discrimination of isolates with identical spa types whenever a potential chain of transmission, which can not be disregarded based on available epidemiological data, needs to be investigated.

Acknowledgements

This study was presented in part at the 15th ECCMID, Copenhagen, Denmark, 2005. We thank the Hygiene department and P. J. Hugo Johansson at Lund University Hospital, and H. Elmrud and B. Carlsson, Kristianstad Hospital, for providing epidemiological data, and A. Lindmark, SMI, for skilful technical assistance performing PFGE.

Transparency Declaration

The authors have no conflict of interest to declare.

Ancillary