ABSTRACT
Vancomycin is the first-line therapy for methicillin-resistant Staphylococcus aureus (MRSA) bacteremia, but its efficacy in adult patients has been questioned. Less is known about the outcomes of MRSA bacteremia treated with vancomycin in pediatric patients. This study reviews the outcomes and clinical characteristics of MRSA bacteremia in children treated with vancomycin and characterizes the microbiologic and molecular features of the bloodstream isolates. A retrospective cohort study was conducted among pediatric patients with MRSA bacteremia treated with vancomycin for >5 days from 1 August 2005 to 31 May 2007 in a large tertiary care center. MRSA bloodstream isolates were characterized by antimicrobial susceptibility testing, PCR analysis of virulence genes, and Diversilab typing. Clinical records were reviewed for outcomes and comorbidities. A total of 22 pediatric patients with MRSA bacteremia were identified. Eleven cases (50.0%) were considered vancomycin treatment failures. Features significantly associated with vancomycin treatment failure were prematurity (P = 0.02) and isolates positive for Panton-Valentine leukocidin (PVL) (P = 0.008). Features typically associated with community-associated MRSA strains were identified in hospital-associated isolates. A dominant clone was not responsible for the high number of treatment failures. Further studies are needed to determine if vancomycin should be the first-line treatment for MRSA bacteremia in premature infants and for PVL-positive isolates.
Staphylococcus aureus is a major cause of invasive infections in both children and adults. Methicillin-resistant S. aureus (MRSA) infections have increased substantially since the first case was reported in 1961, with a prevalence greater than 50% in certain regions of the United States (39). With this overall increase in infections, the proportion of invasive S. aureus infections has increased in the pediatric population as well, especially those with methicillin-resistant strains (12). MRSA infections have emerged in the community among people without prior hospitalization or other traditional risk factors (2, 15, 27). Community-associated (CA) MRSA strains have traditionally differed from health care-associated (HA) strains in that they are more susceptible to antibiotics, contain the staphylococcal chromosome cassette mec (SCCmec) type IV, and are characterized by the production of specific toxins such as the Panton-Valentine leukocidin (PVL) (27, 37). Recent reports have described the occurrence of characteristics typically found in CA strains in isolates causing infections that are considered HA (6, 11, 13, 25). The clinical impact of MRSA infections is significant regardless of the origin of the infection. MRSA infections in adults are responsible for increased mortality rates, longer lengths of hospital stay, and higher rates of therapeutic failure compared to methicillin-susceptible S. aureus infections (1, 18, 38).
Vancomycin is generally considered the treatment of choice for most invasive MRSA infections. The outcomes of MRSA bacteremia treated with vancomycin have been well described for adult populations (23, 28). There are, however, little data on the outcomes of MRSA bacteremia in children treated with vancomycin. This report describes the outcome, factors associated with treatment failure, and microbiologic characterization of bloodstream MRSA isolates in a pediatric population treated adequately with vancomycin, defined as receiving vancomycin for at least 5 days.
MATERIALS AND METHODS
Patient population.The study was conducted at Memorial Hermann Hospital (MHH), a 700-bed, university-affiliated teaching hospital located in Houston, TX. Patients less than 18 years of age with MRSA bacteremia between 1 August 2005 and 31 May 2007 in which the organism was available for testing were identified. Patients were excluded if they had received vancomycin for ≤5 days. Data were gathered from each patient's medical record using a structured data instrument that included outcomes, demographic information, comorbidities, the source of the MRSA bacteremia (e.g., abscess, catheter), antibiotic treatment, the duration of time between the first positive blood culture collection and the administration of vancomycin, vancomycin trough levels, and microbiologic data. This study was approved by the Committee for the Protection of Human Subjects, the Institutional Review Board for the University of Texas Health Science Center at Houston (HSC-MS-09-0076).
Outcome measures were adapted from vancomycin outcome studies conducted with adults (23). Treatment failure was considered if any of the following end points occurred: (i) 30-day mortality, (ii) persistent bacteremia defined as bacteremia for >7 days, or (iii) recurrence of bacteremia within 30 days of the end of vancomycin treatment. These definitions for treatment failure were used to allow for an objective assessment of failure rather than classifying the end points based on a subjective interpretation of retrospective clinical data. Metastatic infection was defined as the presence of a focus of invasive infection, such as an abscess, during treatment with vancomycin. HA MRSA bacteremia was defined as a positive blood culture collected greater than 48 h after admission; MRSA bacteremia was considered community associated if the definition of HA MRSA was not met. Investigated comorbidities included prematurity, antibiotic use in the 30 days preceding the first positive blood culture, treatment with a systemic immunosuppressant (including corticosteroids), liver disease, receipt of total parenteral nutrition (TPN), mechanical ventilation, cardiovascular disease, and admission to the intensive care unit (ICU). The source of the MRSA bacteremia was determined by the presence of other positive MRSA cultures at the time of the index-positive culture and the physician's opinion in the clinical record.
Microbiologic and molecular data.The first organism obtained from the patient's bloodstream was used for all microbiologic and molecular assessments. The organisms were stored frozen in drug-free skim milk at −70°C. Prior to any microbiological evaluation the isolates were passaged twice on drug-free blood agar plates.
Agr typing.Accessory gene regulator (Agr) typing was performed as previously published (21). Briefly, control strains for the Agr typing included NRS 147 (agr I), NRS 149 (agr II), NRS 143 (agr III), and NRS 153 (agr IV). Genomic DNA was extracted from S. aureus strains by boiling or by using the UltraClean microbial DNA isolation kit (MoBio Laboratories, Inc., Carlsbad, CA) and used as an amplification template. PCR was performed using Qiagen HotStarTaq master mix (Invitrogen, Carlsbad, CA) in accordance with the manufacturer's instructions. PCR amplification was carried out under the following conditions: 5-min denaturation step at 95°C followed by 25 stringent cycles (1 min of denaturation at 94°C, 1 min of annealing at 55°C, and 1 min of extension at 72°C) and a final extension step at 72°C for 10 min. PCR products were resolved by electrophoresis through 2.0% agarose gels, followed by ethidium bromide staining and visual analysis.
PVL typing.PVL typing was performed as previously described (22). Briefly, luk-PV-1 and luk-PV-2 were used as the forward and reverse primers, respectively. The NRS 158 strain was used as the control stain. Genomic DNA was extracted from S. aureus strains as stated above. The PCR included deoxynucleoside triphosphates (dNTPs) (200 μM), luk-PV-1 (1 μM), luk-PV-2 (1 μM), 1× buffer, and Finnzymes DyNAzyme EXT (0.2 μM). Amplification was carried out under the following conditions: 5-min denaturation step at 94°C followed by 25 stringent cycles (30 s of denaturation at 94°C, 30 s of annealing at 55°C, and 1 min of extension at 72°C) and a final extension step at 72°C for 10 min. PCR products were resolved by electrophoresis through 2.0% agarose gels, followed by ethidium bromide staining and visual analysis.
SCCmec typing.SCCmec typing was performed as previously published (40). SCCmec typing standard MRSA control strains included type I (NCTC10442), type II (N315), type III (85/2082), type IVa (4744), type IVb (2172), type IVc (4788), type IVd (JCSC4469), type V (WIS/WB68318), and mecA (NRS 70) strains. Genomic DNA was extracted from S. aureus strains as described above. PCR was performed using Qiagen HotStarTaq master mix. PCR amplification was carried out under the following conditions: 5-min denaturation step at 94°C followed by 10 cycles of 94°C for 45 s, 65°C for 45 s, and 72°C for 1.5 min and another 25 cycles of 94°C for 45 s, 55°C for 45 s, and 72°C for 1.5 min, ending with a final extension step at 72°C for 10 min. PCR products were resolved by electrophoresis through 2.0% agarose gels, followed by ethidium bromide staining and visual analysis.
DL typing of isolates.DNA was extracted as described above. Isolates were typed using the Diversilab (DL) system according to the manufacturer's instructions. Strain relatedness was determined according to previously described criteria (31). Isolates were considered identical where there was >95% similarity and no differences in bands.
Vancomycin MICs.The vancomycin MIC for each isolate was determined by broth microdilution. The American Type Culture Collection (ATCC) S. aureus strain ATCC 29213 was used as a control. Broth microdilution trays were prepared by serial dilutions of vancomycin in Mueller-Hinton broth, with final concentrations ranging from 128 μg/ml to 0.125 μg/ml. A well with broth only was used as a growth control. Fifty microliters of a 1:100 dilution of a 0.5 McFarland standard suspension of each isolate in 0.9% NaCl was used to inoculate each well. The trays were incubated for 24 h at 35°C. The MIC was considered the lowest concentration of the drug in a well without growth.
The vancomycin MICs were also determined by the Etest method because this method has been found to provide a better prediction of treatment outcome for certain MRSA strains (16). Etest analysis was performed according to the manufacturer's instructions (bioMerieux, Durham, NC).
The MICs for trimethoprim-sulfamethoxazole, clindamycin, erythromycin, rifampin, and vancomycin were determined using the BD Phoenix automated microbiology system (Becton Dickinson, Franklin Lakes, NJ). Interpretation of the resulting MICs was according to the Clinical and Laboratory Standards Institute criteria (5). All pediatric isolates of MRSA that were erythromycin resistant and clindamycin susceptible underwent clindamycin inducibility testing (17).
hVISA macromethod.The heterogeneous vancomycin-intermediate S. aureus (hVISA) phenotype was determined by Etest as previously described (36). Two hundred microliters of a 2.0 McFarland standard suspension of each isolate was streaked on a brain heart infusion agar plate. Vancomycin and teicoplanin Etest strips were placed onto the agar plates and incubated for 48 h at 35°C. The hVISA phenotype was considered present in strains with vancomycin and teicoplanin MICs of ≥8 μg/ml or a teicoplanin MIC of ≥12 μg/ml. ATCC 700698 and ATCC 700699 were used as controls.
Statistical analysis.The data were managed and analyzed using SAS version 9.1.3 (Cary, NC). Categorical variables were assessed using Fisher's exact test, and continuous variables were analyzed with the Kruskal-Wallis test. A P value of <0.05 was considered significant.
RESULTS
During the study interval, 23 patients under the age of 18 with MRSA bacteremia were identified. Isolates were available from all 23 patients. Of these, 22 patients received vancomycin for greater than 5 days. The one patient that was excluded was treated with clindamycin. A total of 11 (50%) patients experienced treatment failure with vancomycin (Table 1). Three patients (13.6%) died during their hospitalization, 9 patients (40.9%) had documented persistence of MRSA bacteremia for > 7 days while receiving vancomycin, and 2 patients (9.1%) experienced a recurrence of bacteremia within 30 days after completion of vancomycin therapy. All three of the patients that died were premature infants (autopsy data were not available). Three patients experienced multiple end points; 2 patients had persistence of bacteremia for >7 days and died, and 1 patient experienced all three end points. Metastatic infections occurred in 2 patients. Both were due to abscesses, and both children were considered treatment failures due to either persistent bacteremia or a recurrence.
Outcomes of 22 pediatric patients with MRSA bacteremia treated with vancomycin for >5 days
A bivariate comparison of clinical characteristics between vancomycin treatment failures and successes is shown in Table 2. The mean age of the cohort was 31.4 ± 58 months, and ages ranged from 1 week to 16 years. The ages of the patients and proportions of males and females did not differ between the two groups. The majority of the isolates were considered HA MRSA (72.7%). Patients with CA and HA MRSA bacteremia were equally distributed between successes and failures. Premature infants were significantly more likely to fail vancomycin therapy (P = 0.02) than other pediatric patients. The numbers of comorbidities in these premature infants were high; seven of the eight infants had received prior antibiotic and immunosuppressant therapy, and all eight were on mechanical ventilation and were receiving TPN. Other comorbidities and the probable sources of the MRSA bacteremia did not differ between overall successes and failures. There were no cases of endocarditis; however, echocardiography was not routinely performed. Catheters were removed from all patients with a suspected catheter-associated infection within 48 h of collection of the first positive culture. All patients with an abscess underwent surgical drainage within 24 h of the abscess becoming clinically apparent. Vancomycin was generally administered within 24 h of the index culture collection. The times until a vancomycin trough of >15 μg/ml did not differ between treatment successes and failures.
Bivariate comparison of demographic, clinical, and microbiologic features between cases of failure and successa
Nine isolates were PVL positive and were significantly more likely to be associated with treatment failure than PVL-negative strains (P = 0.008). Of the cases of treatment failure due to PVL-positive isolates, 6 were premature infants. PVL was found in 3/6 (50.0%) cases of CA bacteremia and 6/16 (37.5%) isolates considered HA. Six of the 9 (66.7%) isolates that were positive for PVL were SCCmec type IV. agr I was present in 11 isolates (50.0%), agr II was found in 10 (45.5%), and agr III was present in 1 isolate (4.5%). SCCmec type II was found in 7 (31.8%) isolates, type IVa in 11 (50.0%), and type IVb in 1 (4.6%). Three organisms (13.6%) had an SCCmec that was nontypeable by the methods used in this study. SCCmec type IV, compared to all others, was not associated with vancomycin treatment failure (P = 0.67). Five of the 6 (83.3%) isolates that were considered CA were SCCmec type IV. The vancomycin MICs were either 0.5 μg/ml or 1 μg/ml by broth microdilution and 1 μg/ml or 1.5 μg/ml by Etest. The vancomycin MIC by either method was not associated with treatment failure. One isolate met the criteria for hVISA, and the patient with this isolate failed vancomycin therapy due to persistent bacteremia.
Diversilab typing of the isolates revealed five different clones, designated MHH 1 to 5 (Fig. 1). The characteristics of the clones are described in Table 3. The two most common clones were MHH 2 (n = 7) and MHH 4 (n = 9). Compared to all other clones, MHH 4 was not associated with vancomycin treatment failure (P = 0.39). When comparisons were made to the DL MRSA library, it was found that MHH 2 was closely related to the USA100 cluster, while MHH 4 resembled the cluster that contains the USA300 clone.
Dendrogram showing the MHH clones.
Characteristics of the MHH clones
Rifampin was added to vancomycin for four patients who were failing therapy due to persistent bacteremia. The blood cultures of these patients became negative within 1 to 3 days after the addition of rifampin. Gentamicin was used in combination with vancomycin in two treatment failures. The remaining surviving patients who were considered failures were treated with vancomycin monotherapy, with eventual clearance of the bacteremia at 10 to 14 days.
Antimicrobial susceptibility patterns of the MRSA isolates in this study were determined by the BD Phoenix automated microbiology system. All of the isolates were susceptible to trimethoprim-sulfamethoxazole, rifampin, and vancomycin. Eleven isolates (50.0%) were resistant to clindamycin, all of which were considered HA. Two isolates were resistant to clindamycin by an inducible mechanism. All study isolates were resistant to erythromycin.
DISCUSSION
Our study raises a number of important issues. First, few studies of MRSA infections in children have focused specifically on bacteremia treated with vancomycin. Second, 50% of the patients failed treatment with vancomycin in our pediatric population. The high number of failures is not likely due to differences in care between the two groups. There was a trend for earlier administration of vancomycin in the treatment failure patients, the times until a vancomycin trough of >15 μg/ml were not different between the two groups, and equal numbers of patients from each group received treatment in an ICU setting. Only one isolate had the hVISA phenotype; the remainder of the isolates were considered susceptible to vancomycin. Addition of rifampin to patients who were failing vancomycin appeared to be beneficial, as has been reported previously for pediatric patients (33).
Premature infants were significantly more likely to fail vancomycin therapy in our study. Such infants are likely at an increased risk of invasive MRSA infections because of the frequent use of glucocorticoids and invasive procedures and devices, prolonged hospital stays, and the immaturity of the immune system. Morbidity due to invasive staphylococcal infections in infants is significant, often leading to the development of septic shock and deep-seated infections, frequently despite the use of appropriate antimicrobial chemotherapy (13, 14). Though mortality was significant in previous reports (13), the overall mortality in our cohort was relatively low (13.6%) and all deaths occurred in premature infants.
A strong association between treatment failure and PVL-positive isolates was found (P = 0.008). Isolates harboring PVL have been associated with skin infections such as cutaneous abscesses, furunculosis, and necrotic skin infections (22), although more-serious infections such as necrotizing pneumonia have also been documented (9). Interestingly, one study of adult bacteremic patients found that PVL-positive MRSA isolates were less likely to cause persistent bacteremia than PVL-negative isolates (20). Another study of an adult population showed that infections with PVL-positive isolates were more likely to be cured than infections with PVL-negative isolates in patients with skin and soft-tissue infections (4). While the outcome of PVL-positive MRSA infections in adults appears to be favorable, our study results suggest that PVL-positive isolates have the potential to cause significant morbidity in the pediatric patient population, especially in premature infants. The precise mechanism by which PVL may serve as a virulence factor remains unclear. PVL is a toxin that consists of two components, LukS-PV and LukF-PV, that are secreted and assemble on polymorphonuclear leukocyte membranes, where they can cause cell lysis or apoptosis (2). It has also been demonstrated that PVL expression may result in upregulated expression of protein A, which provides evasion of the immune system by binding to the Fc portion of IgG, and that this plays a proinflammatory role in the immune response by binding to tumor necrosis factor receptor 1 (7, 19). Th1 responses, the production of certain cytokines, and immunoglobulin synthesis are all considerably reduced in premature infants (8, 29). This makes the innate immune system especially important for control of infection in infants while the adaptive immune system is developing, and, consequently, pediatric patients may be more dependent on neutrophils for clearance of MRSA bacteremia than adults. Alternatively, other investigations have suggested that α-hemolysin, but not PVL, is essential for the virulence of certain S. aureus strains (3). Whether PVL is a direct cause of pathogenesis or is a marker for other virulence factors remains to be determined.
PVL is generally epidemiologically linked to CA MRSA (35). This does not appear to be the case in the pediatric population at our institution. In our cohort, 37.5% of HA isolates were PVL positive and 43.8% harbored SCCmec IV. Five isolates that caused HA infections resembled the USA300 clone by DL typing. Recent reports have indicated that MRSA isolates with CA genotypes including PVL, SCCmec IV, and sequence type (ST) 8 have been found in infections that are considered HA (6, 11, 13, 25). These reports along with our results suggest that CA MRSA strains are becoming endemic in an increasing number of hospitals. A dominant clone was not responsible for the high number of vancomycin treatment failures in this population.
Numerous studies of adult populations have found that isolates with higher vancomycin MICs, even those in the “susceptible” range, are associated with a worse clinical outcome (23, 24, 26, 30, 32). We did not observe this in our study. This may reflect geographic differences in our isolates or differences between the pediatric and adult patient populations.
Limitations of this study include its small sample size and the retrospective design. It is entirely possible that additional factors, not identifiable because of the small number of patients in our study, may predict treatment failure. It must be noted that MRSA bacteremia in children is a rare disease. Only 14 cases of pediatric MRSA bacteremia were reported in a 4-year period in Buffalo, NY (12), and over a 3-year period in North Carolina (34). Similar case numbers are also reported elsewhere (10, 14). In contrast, our institution had 145 cases of MRSA bacteremia in the adult patient population over this same time period. Multi-institutional studies are likely needed to identify all of the characteristics predictive of vancomycin treatment failure in pediatric patients.
Further limiting our study is the lack of a standard definition of treatment failure of MRSA bacteremia in children. The definitions of treatment failure used in this study have been applied to adult patients (23) and were used to avoid bias by classifying outcomes based on a retrospective chart review. We cannot exclude the possibility that different definitions of failure may be needed in children. Only the first bloodstream isolate from each patient is stored by our clinical microbiology laboratory; therefore, isolates from patients with a recurrence of bacteremia could not be definitively determined to be caused by the same strain, and thus the definition of recurrence used by Lodise et al. (23) was applied. Nonetheless, we found a high rate of failure of vancomycin in children with MRSA bacteremia, especially in premature infants, that were treated at a large pediatric hospital in the fourth largest city in the United States. This may reflect a more generalized problem with the use of vancomycin for treatment of MRSA bacteremia in this population. Larger studies are needed to clarify the role of PVL in MRSA bacteremia. If PVL is confirmed to be a marker of vancomycin treatment failure, switching to an alternate agent or addition of a second antimicrobial such as rifampin once the presence of PVL is determined should be considered in pediatric patients, especially in premature infants. Additionally, further studies can be performed in areas where PVL is highly prevalent to evaluate the use of alternate agents as empirical therapy for presumed MRSA bacteremia.
ACKNOWLEDGMENTS
We thank T. Ito and K. Hiramatsu for the kind gift of the SCCmec control strains: type I, NCTC10442; type II, N315; type III, 85/2082; type IVa, 4744; type IVb, 2172; type IVc, 4788; type IVd, JCSC4469; type V, WIS/WB68318. We also thank James Murphy for his help in reviewing the manuscript.
FOOTNOTES
- Received 2 October 2009.
- Returned for modification 25 November 2009.
- Accepted 8 January 2010.
- Copyright © 2010 American Society for Microbiology
