The study protocol was reviewed and approved by the National Defense Medical Center Institutional Review Board. Two collections of S. aureus isolates were used. A list of all children <14 years of age hospitalized with various staphylococcal infections during the period from December 2003 to November 2005 was compiled from records at the clinical microbiology laboratory at the Tri-Service General Hospital in Taipei. The first collection of 144 infecting strains was further categorized into 11 types of staphylococcal infection on the basis of the clinical details provided. A case was considered community acquired if MRSA was isolated from cultures of specimens obtained within 72 hours after hospitalization. Risk factors for MRSA infection included hospitalization <12 months before the date of MRSA isolation; history of any surgical procedure; history of endotracheal intubation; underlying chronic disorder; antimicrobial drug therapy <12 months before the date of MRSA isolation; presence of an indwelling venous or urinary catheter; or household contact with a person with an identified risk factor or a worker in a healthcare environment (6).

A second collection of 300 colonizing strains was obtained during the same period by culturing samples from the anterior nares of 1,195 healthy children in the community. Eligible participants were <14 years of age with no acute medical problem who either visited a healthcare facility for a well-child checkup or attended 1 of 7 kindergartens in Taipei.

MRSA identification and antimicrobial drug susceptibility were determined according to the Clinical Laboratory Standards Institute (formerly known as the National Committee for Clinical Laboratory Standards) guidelines (7,8). Staphylococcal cassette chromosome mec (SCCmec) elements were typed and PVL genes were detected as described (2,9,10). Sequences specific for sea to see, seg to sei, eta, etb, and tst, which encode staphylococcal enterotoxins (SEA to SEE, and SEG to SEI), exfoliative toxins (ETA and ETB), and toxic shock syndrome toxin-1, respectively, were detected by using methods described by Jarraud et al. (11). Pulsed-field gel electrophoresis (PFGE) was performed by using a CHEF Mapper XA system (Bio-Rad Laboratories, Hercules, CA, USA) according to a published protocol (12).

Data were analyzed by the Mantel-Haenszel test and χ² test with SPSS version 10.0 software (SPSS, Chicago, IL, USA). A p value <0.05 was considered significant.

Of the 444 isolates examined, PVL-positive isolates constituted 23% of all S. aureus isolates analyzed (Table 1). Among 144 isolates (67 methicillin-susceptible S. aureus isolates and 77 MRSA isolates) from different staphylococcal infections, 82 (56.9%) were PVL positive, and most were associated with skin and soft tissue infections (especially furuncles). In contrast, only 18 (6%) of 300 colonizing isolates had the PVL locus. PVL genes were also found in isolates associated with other deep-space infections including pyomyositis, osteomyelitis, and septic arthritis. Of 1,195 healthy children who were screened, 89 (7.4%) had cultures with MRSA. Only 15 (16.9%) of 89 community MRSA-colonizing strains were PVL positive.

Among the 144 S. aureus isolates obtained from various types of staphylococcal disease, 32 (22.2%) isolates were further confirmed as CA-MRSA according to the inclusion criteria (abscess [n = 5], carbuncles [n = 6], cellulitis [n = 9], furuncles [n = 3], pyomyositis [n = 1], osteomyelitis [n = 1], pneumonia [n = 1], and staphylococcal scarlet fever [n = 6]); all contained the genes encoding PVL. Antibiograms of 15 PVL-positive MRSA-colonizing strains did not differ significantly from those of 32 CA-MRSA–infecting strains with respect to clindamycin, erythromycin, gentamicin, and chloramphenicol. All 47 PVL-positive MRSA isolates were resistant to penicillin G, and none had reduced susceptibility to vancomycin, teicoplanin, trimethoprim-sulfamethoxazole, fusidic acid, or ciprofloxacin.

To gain insight into the association between nasal strains of PVL-positive MRSA and community-acquired disease, we designed a comparative study in which the colonization and clinical samples were collected during the same period. Results of SCCmec typing and exotoxin profiling for both community MRSA-colonizing strains and CA-MRSA–infecting strains are shown in Table 2. Of 74 colonizing strains that did not have the PVL locus, most (94.6%) had SCCmec IV, and only 1 (1.35%) had SCCmec V_T. Conversely, irrespective of origin, PVL-positive MRSA strains were more likely to have SCCmec V_T than were the PVL-negative MRSA-colonizing strains (p<0.001). Regarding the exotoxin profiles, the most frequently encoded toxin gene among PVL-positive MRSA isolates was seb (97.9%). PVL-positive MRSA-colonizing strains and CA-MRSA–infecting strains were more likely to have genes that encoded SEB than were PVL-negative CA-MRSA–colonizing strains (p = 0.006). Genes for SEG/SEI were found only in PVL-negative MRSA-colonizing strains (p<0.001).

Figure

Figure. Pulsed-field gel electrophoresis patterns and phylogenetic tree of 47 methicillin-resistant Staphylococcus aureus (MRSA) isolates with Panton-Valentine leukocidin (PVL) genes. Banding patterns were digitalized and analyzed with Molecular Analyst Fingerprinting, Fingerprinting Plus,...

Diverse pulsotypes were found among the 47 PVL-positive MRSA strains subjected to PFGE typing (Figure). Two clusters that included 33 (70.2%) isolates were distinguished at the 70% similarity level. None of the isolates were linked on review of epidemiologic data derived from medical records. Except for 1 colonizing isolate (C7) and 2 infecting isolates (I22 and I23) that carried SCCmec IV, the other 30 (90.9%) isolates from clusters I and II had SCCmec V_T.

This is the first epidemiologic study of PVL-positive S. aureus in Taiwan. The prevalence of PVL-positive S. aureus among isolates collected from various types of staphylococcal infections was 56.9%; previous surveys reported rates of <5% to 12.4% (13–15). This higher prevalence is probably related to the greater proportion of pediatric patients with cutaneous infections. The frequencies of infections associated with these organisms were similar to those in previous studies (2,13,14). Overall, the outcomes of infections with PVL-positive strains were excellent and comparable to that of PVL-negative strains. Only 1 death occurred.

Four factors that support the relationship between nasal strains of PVL-positive MRSA and community-acquired disease. First, both PVL-positive MRSA-colonizing strains and CA-MRSA–infecting strains had consistent antibiograms. Second, most PVL-positive MRSA-colonizing strains and CA-MRSA–infecting strains had SCCmec V_T, whereas a high prevalence of SCCmec IV was found among PVL-negative MRSA-colonizing strains. Third, PVL-positive MRSA-colonizing strains and CA-MRSA–infecting strains had consistent exotoxin profiles, which differed from those of PVL-negative MRSA-colonizing strains. Fourth, PFGE findings indicate that some PVL-positive MRSA isolates that colonized children who remained asymptomatic were clonally related to clinically isolated CA-MRSA–infecting strains, especially those with SCCmec V_T, compared with SCCmec IV (6/8, 75% and 1/7, 14.3%, respectively).

Several study limitations merit consideration. First, our study is a snapshot in time because the molecular epidemiology of CA-MRSA is constantly changing. Second, we were unable to determine risk for infection because children colonized with community PVL-positive MRSA were not followed-up longitudinally. Finally, our results are geographically distinct and may not be generalized to the global population.

Our study showed that PVL genes are carried by a large number of S. aureus isolates, especially among those causing disease. We provide evidence that links community PVL-positive MRSA-colonizing strains to CA-MRSA–infecting strains from various types of staphylococcal infection.