|Home | About | Journals | Submit | Contact Us | Français|
Invasive disease following methicillin-resistant Staphylococcus aureus (MRSA) detection is common, regardless of whether initial detection involves colonization or infection. We assessed the genetic relatedness of isolates obtained ≥2 weeks apart representing either repeated infections or colonization-infection sets to determine if infections are likely to be caused by previously harbored strains. We found that MRSA infection following initial colonization or infection is caused by the same strain in most cases, suggesting that a single successful attempt at decolonization may prevent the majority of later infection.
Methicillin-resistant Staphylococcus aureus (MRSA) causes >278,000 hospitalizations and 56,000 septic episodes annually . Hospitalized patients with newly identified MRSA colonization or infection have a 35% risk of subsequent invasive disease within 1 year [2, 3]. Strategies to prevent infection have included decolonization [–11], administration of anti-staphy-lococcal antibodies , and vaccination [13–14].
Decolonization has had variable success . Its benefit may depend on whether subsequent MRSA infections are caused by the original infecting or colonizing strain, indicating persistent carriage or repeated acquisition from the local environment. Under these scenarios, a single successful decolonization of patients and potentially their immediate surroundings could prevent future infection. However, if subsequent infections are caused by different strains, decolonization would be of limited value unless sustained over long periods.
Colonizing strains are responsible for infections occurring soon after detected colonization . Less is known about the association between colonizing strains and subsequent infections occurring in the intermediate or distant future, but 40%– 60% of carriers demonstrate prolonged carriage with the same strain [14–17]. We assessed the genetic relatedness of MRSA isolates obtained >2 weeks apart. We evaluated isolates representing both repeated infections and asymptomatic carriage followed by infection.
We conducted a multicenter cohort study in 4 academic hospitals identifying patients with MRSA invasive disease following prior MRSA infection or colonization (table 1). Hospitals selected a 1–2-year study period during which all patients with sterile site MRSA cultures >2 weeks apart were identified (the infection-infection group). Associated MRSA isolates were retrieved. Two institutions additionally collected isolates from patients with MRSA-positive nares surveillance cultures followed by invasive clinical cultures (the colonization-infection group). Patients could contribute to both groups. Decolonization was not routinely employed. This study was approved by each hospital’s institutional review board.
Medical records were reviewed to confirm infection by Centers for Disease Control and Prevention (CDC) criteria . Sequential infections were required to be discrete and unrelated, as determined through review by an infectious diseases physician. Sequential device-related bacteremia required device removal between events and complete resolution of the initial infection.
Data collected included age, sex, hospitalization dates, length of hospital stay(s), hospital day of cultures, pre-admission location(s), and postdischarge disposition(s). Risk factors for MRSA infection were collected, such as recent surgery, wounds, and comorbidities (hemodialysis, diabetes).
Isolates were sent to the Molecular Epidemiology and Fungus Testing Laboratory at the University of Iowa (Iowa City) for confirmation of methicillin resistance (mecA gene probe), antibiotic susceptibility testing, and SCCmec typing . PCR assays for the Panton-Valentine leukocidin (PVL) gene were performed. Antibiotic susceptibility testing involved broth microdilution [21, 22] for oxacillin, erythromycin, clindamycin, tetracycline, rifampin, moxifloxacin, gentamicin, vancomycin, linezolid, and quinupristin-dalfopristin; mupirocin, tetracycline, and trimethoprim-sulfamethoxazole susceptibility were determined by Etest (AB Biodisk) . D-tests for inducible clindamycin resistance were performed when isolates were resistant to erythromycin but susceptible to clindamycin.
Strain types based on PFGE  were classified as indistinguishable, similar, or different . Identical banding patterns were considered to be indistinguishable, those differing by ≤3 bands were considered to be similar (includes indistinguishable strains), and those differing by >3 bands were considered to be different. PFGE patterns were compared to defined USA types  to distinguish common hospital and community-associated MRSA strains (BioNumerics). Total and hospital-specific dendrograms were created to display strain relatedness.
Patients’ characteristics were reported as proportions. Repeated infections were summarized by culture site, infection type by CDC criteria , duration between events, and location between events (inpatient, home, or skilled nursing facility or rehabilitation center [SNF/rehab]). Location between events was characterized by the highest acuity center.
Within the infection-infection group, strain relatedness was assessed by the proportion of the first 2 infection events per person that had indistinguishable, similar, or different strain types. Strain relatedness among the colonization-infection group was similarly assessed. Associations between strain relatedness and patient location or duration between events were assessed using Fisher’s exact test.
MRSA isolates were additionally described by SCCmec type and presence of PVL. We also assessed whether subsequent infecting strains exhibited increased antibiotic resistance, compared with original strains.
Thirty-seven patients had ≥2 discrete MRSA infections, for a total of 83 infections. By study design, nearly all infections (79 [95%] of 83) involved bacteremia. Twenty-nine patients had sets of isolates representing colonization followed by infection, 27 (93%) of which involved bacteremia. Eight of the 29 had another subsequent MRSA infection and contributed to both groups. Patient characteristics and infection types are found in table 2 and table 3.
We found considerable strain diversity, with 97%–100% of patients having PFGE subtypes that were unique within each institution (example institution shown in figure 1). Nevertheless, despite population diversity, strains from a single patient representing either sequential infection or colonization-infection were highly related (table 4). Among the infection-infection group, only 4 (11%) of 37 patients had subsequent MRSA infections caused by a different strain. Patients with intervening SNF/rehab stays were more likely to have infections caused by different strains than were those who went home (P=.02) or remained in the hospital (P=.02). Longer durations between MRSA infections were not predictive of differing strain types (table 5).
Six patients had >2 MRSA infections. These infections involved 20 bacteremic events spanning 30–302 days (median, 182.5 days). Five of 6 patients’ infections were repeatedly caused by indistinguishable strain types.
USA100 was the most common strain type in either group (54 of 66 sets), followed by USA500 (3 sets) and USA200 (2 sets). One patient each had repeated infection with USA300 and USA400, accounting for the 2 PVL positive, SCCmec IV isolates.
The percentage of initial invasive MRSA isolates that were susceptible to the following antibiotics is as follows: moxifloxacin, 6% of isolates; erythromycin, 6%; clindamycin, 33%; trimethoprim- sulfamethoxazole, 92%; tetracycline, 95%; gentamicin, 92%; rifampin, 100%; vancomycin, 100%; linezolid, 100%; quinupristin-dalfopristin, 100%; and mupirocin, 79%. Nine strains had inducible clindamycin resistance. Among the 26 repeated infections caused by identical strains, 4 strains (15%) showed increased antibiotic resistance (moxifloxacin, intermediate to resistant [1 strain]; gentamicin, susceptible to resistant [1 strain]; rifampin, susceptible to intermediate [1 strain]; and clindamycin, susceptible to resistant [1 strain]).
With a 35% risk of infection in the year following detection [2, 3], there is a need to prevent disease among MRSA carriers. We show that most MRSA infections following initial colonization or infection are caused by identical strains. This likely represents ongoing risks of infection from persistently carried strains, because diverse strain types at participating institutions made repeated acquisition of an identical strain unlikely from any source other than a patient’s immediate surroundings. It also supports lack of strain-specific immunity. This finding suggests that single successful decolonization of patients and possibly their immediate surroundings may obviate the majority of later MRSA infections.
Patients who remained hospitalized and those discharged to home were significantly more likely to experience repeated infections due to identical strains, compared with patients who had intervening SNF/rehab stays. Given the low frequency (<1%) with which healthy people carry MRSA , this was not unexpected for MRSA carriers who returned home, where there was unlikely to be a preexisting reservoir of diverse strains among family members. This is in contrast to those with an intervening stay at an SNF/rehab, where MRSA prevalence is often high and where social networking is common and encouraged. Among hospitalized patients, contact precautions and the lack of patient interaction may explain the tendency for a single strain to cause repeated infections despite evidence that hospital strains are diverse. Thus, decolonization of patients in their home or hospital room may be more successful at preventing infection than similar attempts at an SNF/rehab, where repeated screening and decolonization may be needed to ensure lack of carriage.
In 24% of subsequent infections, strains were not identical, although most differed by <3 bands. These strains may represent new environmental acquisition. Another possibility is that the duration between infections was sufficiently long for minor genetic changes to produce highly related, but non-identical, strains.
We found that a patient’s location between infections was more predictive of strain relatedness than was time between infections. Nearly all isolates recovered from persons with repeated infection were of similar strain types, even when infections were separated by >6 months.
MRSA strains causing repeated infections generally retained the same susceptibility profile. Of the 15% of strains with a more resistant antibiotic profile at the time of the second infection, only 2 had a substantial change in their MIC, and 1 was confirmed to be related to inducible macrolide-lincosamide-streptogramin resistance.
There are several limitations to this study. First, the sample size is small, despite multiple participating centers. Second, all institutions were academic centers, and results may not be generalizable to other settings. In addition, these results are largely based on health care-associated MRSA strains. Further data are needed to assess the relatedness of strain types in recurrent community-associated MRSA infections. Third, because we did not perform environmental cultures, we do not know the contribution of a patient’s immediate surroundings in maintaining carriage with the same strain. Nevertheless, we recognize that several studies have shown that MRSA easily contaminates the environment [28–30].
Repeated MRSA infection and MRSA infection occurring after colonization are not uncommon. We show that the majority of these infections are due to strains identical to the initial strain, even when the events occur many months apart. This evidence of persistent carriage suggests that time-limited decolonization regimens for patients, possibly coupled with disinfection of their surroundings, may substantially reduce the risk of subsequent infection.
Financial support. The CDC Prevention Epicenters Program and the National Institutes of Health (K23AI64161-01).
Potential conflicts of interest. D.K.W. receives research support from Sage Products, and has served as a consultant to Enturia and 3M Healthcare. R.P. has received research grants from Sanofi-Aventis, GlaxoSmithKline, Pfizer, and TAP Pharmaceuticals in the past 2 years. All other authors: no conflicts.