Methicillin-resistant Staphylococcus aureus (MRSA) have been reported to be an increasingly common cause of post-partum mastitis [Reference Moazzez1, Reference Reddy2]. While 2–33% of breastfeeding women may develop some form of post-partum mastitis, in 3–11% of these cases the mastitis may progress to breast abscesses [Reference Amir3, Reference Barbosa-Cesnik, Schwartz and Foxman4]. In general, community-associated MRSA (CA-MRSA) infections have different risk factors and have been caused by different bacterial clones than traditional healthcare-associated MRSA (HA-MRSA) infections, although migration of CA-MRSA clones into healthcare facilities may erode these distinctions [Reference Popovich, Weinstein and Hota5, Reference David and Daum6]. While CA-MRSA have been reported to cause complicated post-partum mastitis [Reference Saiman7, Reference Stafford8], much about the epidemiology of these infections remains unknown.

Recently, D'Souza et al. [Reference D'Souza, Rodrigues and Mehta9] genetically characterized a mixed population of CA-MRSA and HA-MRSA collected from various tissue sources between October 2006 and June 2009 at Hinduja Hospital in north Mumbai, India, and elsewhere in the city. Their data highlighted the prevalence of the CA-MRSA clones, ST22-MRSA-IV and ST772-MRSA-V, and further indicated that these two clones may be replacing the HA-MRSA clone, ST239-MRSA-III, in this region [Reference D'Souza, Rodrigues and Mehta9]. A cluster of MRSA breast abscesses in women who had given birth at Breach Candy Hospital Trust, a tertiary care medical centre in south Mumbai, following the sampling period of D'Souza et al. [Reference D'Souza, Rodrigues and Mehta9], prompted this study of the infecting clone(s).

The sample included 50 non-duplicate MRSA isolates (four duplicate isolates were excluded) that were collected between June 2009 and January 2010 from Breach Candy Hospital Trust. In this sample, 20 isolates were from breast abscesses of women who had delivered at the obstetrics and gynaecology (OBGYN) ward, 20 isolates were from various infections of patients of other wards (type and number of specimens: unspecified pus n=5, endotracheal secretion n=3, blood n=2, ear n=2, lip abscess n=2, conjunctiva n=1, perinephric abscess n=1, nasal abscess n=1, synovial fluid n=1, toe n=1, vagina n=1), and 10 isolates were from nasal swabs of healthcare workers of the OBGYN ward, where the outbreak occurred, and the adjacent neonatal intensive care unit (NICU). The 40 infection isolates represented all of the non-duplicate MRSA isolates collected in this hospital during this time period.

The 20 breast abscess cases had previously come to the hospital for delivery. Length of initial stay was available for 14/20 of these cases and was 3–4 days for vaginal delivery and 5–6 days for caesarean section. These cases were healthy at discharge but subsequently returned as outpatients with MRSA breast abscesses (number of cases presenting by month: June n=1, July n=2, October n=2, November n=7, December n=4, January n=4). Thus, a majority of these cases (17/20) presented in the second half of the sampling period, with a peak in November 2009. Incubation period from date of delivery to date of culture was available for 17/20 of these cases (range 17–165 days, median 36 days). These cases meet an epidemiological definition of healthcare-associated, community-onset MRSA (HACO-MRSA) infections, as they were cultured ⩽48 h after admission to the outpatient department and occurred in patients with healthcare-associated risk factors (recent delivery in a healthcare setting) [Reference Klevens10].

To identify the genetic backgrounds of the MRSA isolates, multilocus sequence typing (MLST) and staphylococcal protein A (spa) typing was performed for all isolates [Reference Robinson and Enright11]. Both eGenomics nomenclature (www.egenomics.com) and Ridom nomenclature (www.spaserver.ridom.de) was used to name spa types. Staphylococcal chromosomal cassette mec (SCCmec), which is a mobile genetic element that encodes the methicillin resistance determinant, was typed for all isolates using PCR assays that score the mec gene complex (classes A, B, C) and the ccr gene complex (ccrAB1, ccrAB2, ccrAB3, ccrAB4, ccrC) [Reference Robinson and Enright11, Reference Kondo12]. In addition, sequence-based subtyping of SCCmec was done for all isolates using a fragment of the ccrB gene and the mec-associated direct repeat units (dru) [Reference Goering13, Reference Smyth14]. lukS-PV and lukF-PV genes, which encode the Panton–Valentine leukocidin (PVL), were detected by PCR [Reference Lina15]. Example sequences of each unique multilocus sequence type were submitted to the MLST database (www.mlst.net), sequences of new spa and dru types were submitted to appropriate databases (www.egenomics.com and www.dru-typing.org), and example sequences of each unique ccrB allele were submitted to the GenBank database (accession numbers JQ746656–JQ746658).

Four spa types (eGenomics 648, 113, 1373, 1375; Ridom t852, t005, t9310, t6717) belonged to multilocus sequence type (ST) ST22 and were represented by 39 isolates (Table 1). Two spa types (eGenomics 692, 1080; Ridom t345, t657) belonged to ST772 and were represented by eight isolates. Single spa types were found among the two isolates of ST239 (eGenomics 351, Ridom t030) and the single isolate of ST30 (eGenomics 43, Ridom t021). All of the ST22 isolates had SCCmec IV and pvl (Table 1). Moreover, all of the ST772 isolates had SCCmec V and pvl, with the exception of a single isolate that had SCCmec IV and pvl. The two isolates of ST239 had SCCmec III but lacked pvl, and the single isolate of ST30 had SCCmec IV and pvl (Table 1). These results confirmed the presence of these clones in this region as previously reported by D'Souza et al. [Reference D'Souza, Rodrigues and Mehta9].

Table 1. Genetic characteristics of methicillin-resistant Staphylococcus aureus from a hospital in Mumbai, India

* Nomenclature: eGenomics/Ridom.

† The ccrB gene is absent and thus non-typable (nt) for SCCmec type V.

While spa typing did not reveal substantial diversity differences within the ST22-MRSA-IV and ST772-MRSA-V backgrounds, subtyping of their SCCmec elements with ccrB:dru sequences revealed contrasting patterns of diversity. SCCmec IV from ST22 was exclusively of the ccrB:dru subtype 3:0, whereas SCCmec V from ST772 presented six ccrB:dru subtypes in the seven isolates and the single ST772 isolate with SCCmec IV also had a different ccrB:dru subtype (Table 1). These results suggested that ST22-MRSA-IV was a more genetically homogeneous background than ST772-MRSA-V in this hospital.

Nearly all breast abscess cases (19/20) were caused by a single clone: pvl-positive, spa type 648 (Ridom t852), ccrB:dru subtype 3:0, ST22-MRSA-IV (Table 1). The exception was one isolate with spa type 113 (Ridom t005), which differed by a single base pair from spa type 648. The MRSA isolates from breast abscesses were resistant to ciprofloxacin, half were resistant to gentamicin, and all but two isolates were susceptible to co-trimoxazole, erythromycin, and clindamycin. The outbreak clone was also the most frequent cause of infections in other wards (6/20) and was the most frequently carried clone by healthcare workers in the OBGYN and NICU wards (8/10) (Table 1). Thus, these results provided genetic confirmation of the outbreak and showed that the outbreak clone was common in the hospital environment.

A previous study from northern Italy reported a cluster of S. aureus post-partum mastitis, including six breast abscesses, and neonatal skin infections that preceded additional infections in the community [Reference Tinelli16]. The genetically characterized isolates from subsequent infections were pvl-positive, spa type 113 (Ridom t005), ST22 methicillin-susceptible S. aureus (MSSA) [Reference Tinelli16]. The Indian cluster investigated here differed from the Italian cluster by the lack of concurrent neonatal infections and all isolates were MRSA. The ST22-MRSA-IV background is common in parts of Asia and Europe, but it exhibits variation in spa types and in carriage of pvl and it has been described from healthcare and community sources [Reference D'Souza, Rodrigues and Mehta9, Reference Amorim17–Reference Rossney20]. Our results clearly indicated that the pvl and SCCmec type IV markers alone would not be accurate markers of a CA-MRSA infection source in India. The insufficiency of these two markers for classifying the infection source of MRSA has also been observed previously in Ireland [Reference Rossney20]. Finally, isolates with the same genetic characteristics as the outbreak clone described here, pvl-positive, spa type 648 (Ridom t852), ST22-MRSA-IV, were recently reported to cause serious soft-tissue infections, including rare brain abscesses, in Bengaluru, southern India [Reference Nadig, Ramachandra Raju and Arakere19]. These observations indicate that this particular clone is highly virulent and has the capacity for wider geographical transmission in India. Given this clone's link to the hospital environment, infection control practices may have an important role in limiting its spread.