|Home | About | Journals | Submit | Contact Us | Français|
The role of Panton-Valentine leukocidin (PVL) in determining the severity and outcome of complicated skin and skin structure infections (cSSSI) caused by methicillin (meticillin)-resistant Staphylococcus aureus (MRSA) is controversial. We evaluated potential associations between clinical outcome and PVL status by using MRSA isolates from patients enrolled in two large, multinational phase three clinical trials assessing telavancin for the treatment of cSSSI (the ATLAS program). MRSA isolates from microbiologically evaluable patients were genotyped by pulsed-field gel electrophoresis (PFGE) and PCR for pvl and 31 other putative virulence determinants. A single baseline pathogen of MRSA was isolated from 522 microbiologically evaluable patients (25.1%) among 2,079 randomized patients. Of these MRSA isolates, 83.2% (432/519) exhibited the USA300 PFGE genotype and 89.1% (465/522) were pvl positive. Patients with pvl-positive MRSA were more likely than those with pvl-negative MRSA to be young, to be North American, and to present with major abscesses (P < 0.001 for each). Patients were significantly more likely to be cured if they were infected with pvl-positive MRSA than if they were infected with pvl-negative MRSA (91.6% versus 80.7%; P = 0.015). This observation remained statistically significant after adjustment for presence of abscess, fever, or leukocytosis; infection size; diabetes; patient age; and study medication received. The fnbA, cna, sdrC, map-eap, sed, seg, sei, sej, SCCmec type IV, and agr group II genes were also associated with clinical response (P < 0.05). This contemporary, international study demonstrates that pvl was not the primary determinant of outcome in patients with MRSA cSSSI.
Methicillin (meticillin)-resistant Staphylococcus aureus (MRSA) is a leading cause of complicated skin and skin structure infections (cSSSI) in the United States (4, 18, 23). In many regions of the United States, the genetically distinct community-associated MRSA (CA-MRSA) clone USA300 is now the predominant cause of cSSSI (13, 18, 25). Most CA-MRSA isolates isolated from cSSSI carry pvl, the gene encoding Panton-Valentine leukocidin (PVL) (6, 19, 26). PVL is a pore-forming, bicomponent exotoxin known to induce cell death by necrosis or apoptosis (8). Infections caused by S. aureus strains carrying pvl are commonly thought to be associated with worse clinical outcome (6, 16). For example, the presence of the PVL toxin has been shown to cause necrotizing pneumonia in animal models (14) and is associated with necrotizing S. aureus pneumonia in humans (9). However, other laboratories have found that PVL is not a virulence determinant in murine models of CA-MRSA infection (1, 27). In addition, recent studies by our group have suggested that the presence of pvl was associated with a better clinical outcome in patients with cSSSI (2) or bacteremia (15) due to S. aureus. Thus, the role of PVL in determining the clinical severity and outcome of MRSA cSSSI remains unresolved.
As part of two phase 3 clinical trials, we have established a large collection of contemporary, geographically diverse S. aureus isolates from a clinically well-characterized population of cSSSI patients. Using this resource, the current study sought to accomplish two objectives, (i) to validate our previous observation that the presence of pvl is not the primary determinant of clinical outcome in patients with cSSSI due to MRSA and (ii) to identify potential associations between other putative bacterial virulence genes and clinical outcome in patients with cSSSI due to MRSA.
(These data were presented, in part, at the 48th Annual ICCAC/IDSA 46th Annual Meeting, 25 to 28 October 2008, Washington, DC.)
The ATLAS (assessment of telavancin in cSSSI) clinical trials were two methodologically identical, double-blind, randomized, active-controlled, parallel group, multinational phase 3 studies investigating the efficacy and safety of telavancin versus those of vancomycin for treatment of cSSSI caused by gram-positive organisms. The study designs of the ATLAS trials have been published in detail elsewhere (24).
A total of 2,079 patients, enrolled from 129 participating centers in 21 countries, were studied from September 2004 through June 2006. Men and nonpregnant women aged ≥18 years (one patient was aged 17 years) and diagnosed with cellulitis, a major abscess requiring surgical drainage, an infected wound or ulcer, or an infected burn caused by a suspected or confirmed gram-positive organism that warranted ≥7 days of parenteral antibiotic therapy were eligible for the study. Purulent drainage and/or collection or the presence of at least three of the following signs or symptoms was also required for participation: erythema, heat and/or localized warmth, fluctuance, swelling and/or induration, pain and/or tenderness to palpation, fever (temperature of >38°C), or white blood cell (WBC) count of >10,000 cells/mm3.
Patients were randomized to receive either telavancin (10 mg/kg of body weight every 24 h) or vancomycin (1 g every 12 h). Bacterial isolates were obtained from all patients at baseline by needle aspiration or a surgical procedure. Test-of-cure evaluation was performed 7 to 14 days after administration of the last dose of the study drug. The clinical response at the test of cure was classified as cure, failure, or indeterminate (24). The microbiologically evaluable population was defined as clinically evaluable patients who had a baseline gram-positive pathogen recovered from pretreatment cultures. In the present investigation, the analysis population was defined as microbiologically evaluable patients with cSSSI due to monomicrobial S. aureus. The investigation was approved by the Duke University Medical Center Institutional Review Board.
Pulsed-field gel electrophoresis (PFGE) with SmaI was performed as previously described (17). The PFGE profiles were analyzed using BioNumerics software (Applied Maths, Kortrijk, Belgium). A similarity coefficient of 80% was used to define pulsed-field type clusters. Dice coefficients (pairwise similarity) were calculated for each pair of isolates, and a dendrogram was constructed using an optimization value of 0.50% and a position tolerance ranging from 1.25% to 1.35% (the end of the fingerprint). PFGE profiles were interpreted according to a published typing schema (17).
S. aureus genomic DNA was extracted using an ultraclean microbial DNA kit (MO BIO Laboratories, Inc., Carlsbad, CA) in accordance with the manufacturer's instructions. PCR assays were used to screen the S. aureus genome for 33 putative bacterial virulence determinants, including toxins (sea, seb, sec, sed, see, seg, seh, sei, sej, tst, eta, etb, hlg, and pvl), adhesins (bbp, clfA, clfB, cna, ebpS, fnbA, fnbB, map-eap, sdrC, sdrD, sdrE, and spa), agr groups I to IV, SCCmec types I to IV, and other virulence genes (chp, efb, icaA, V8, and arcA). The primers and conditions used to amplify the genes of interest have previously been described (2).
Of the 33 genes screened using PCR, 30 were evaluated using multiplex PCR. To minimize the possibility of false-negative results, genes that were not detected in the multiplex PCR assay were subsequently reanalyzed by uniplex PCR. Uniplex PCR alone was used to detect spa, arcA, and agr groups I to IV due to the presence of multiple repeats in the spa gene that interfered with the detection of other genes. agr groups I to IV were detected serially using the primers and conditions described by Peacock et al. (22).
To verify the designation of the USA300 genotype, the arcA locus of the arginine catabolic mobile element was evaluated using PCR (5). The positive control for the arginine catabolic mobile element was NRS384. SCCmec typing was performed using multiplex PCR as described by Oliveira and de Lencastre (21). SCCmec types I and IV were further validated using uniplex PCR as previously described (20, 29).
Cure rate by pvl status was stratified by markers of severity, including abscess versus nonabscess infection type, presence of fever, leukocytosis (>10 × 109 cells/liter), infection size (baseline infection area of >100 cm2), patient age, and surgical intervention. Infection size was calculated as the product of the length and width of the primary infection site at its longest and widest axes, respectively. “Surgical intervention” was defined as any surgical procedure conducted between 7 days before and 2 days after the initiation of the study drug, inclusive, that the sponsor's blinded review assessed as a significant procedure undertaken to treat the infection.
Continuous variables were compared between groups by using the two-sample t test. Categorical variables were analyzed using Fisher's exact test, its 2-by-K extension, or the Cochran-Mantel-Haenszel test for stratified analyses. Our a priori hypothesis, that the presence of pvl in MRSA cSSSI is associated with a better clinical outcome, was tested using Fisher's exact test. All reported P values are two sided. P values of ≤0.05 were deemed statistically significant. Where indicated, P values were adjusted for multiple comparisons to control the false-discovery rate at 20%. Since subjects were not randomized on the basis of pvl status, all P values calculated should be considered descriptive and not inferential. Results were obtained using SAS software, version 9.1.3.
A total of 2,079 patients were recruited to the ATLAS studies. Bacterial genetic profiles were available for 769 microbiologically evaluable patients with a single baseline pathogen of S. aureus from 92 centers in 15 countries. Of these, 522 patients (25.1% of the randomized patients) with MRSA were enrolled in this study. A total of 243 of these 522 patients (46.6%) received telavancin, and 279 (53.4%) received vancomycin. These patients were predominantly white, male, and North American, with an average age of 43 years (Table (Table1).1). A total of 63.8% of patients had at least one chronic illness, and 18.4% had a prior MRSA infection.
The pvl gene was detected in 465/522 (89.1%) of the MRSA isolates. MRSA isolates positive for the pvl gene were more frequently isolated from younger patients (P < 0.001) and patients from North America than pvl-negative strains (P < 0.001) (Table (Table1).1). Major abscesses were significantly more common among patients infected with pvl-positive MRSA than among those infected with pvl-negative MRSA (72.0% versus 17.5%, respectively; P < 0.001). Patients with pvl-positive MRSA isolates were less likely to have been hospitalized within the previous 6 months (P < 0.001) or to have used antimicrobial agents within the previous 3 months (P < 0.001) and were more likely to reside in an area of CA-MRSA endemicity (P < 0.001).
Overall, 91.6% (426/465) of patients with cSSSI due to pvl-positive MRSA were cured of their infection, compared with 80.7% (46/57) of patients with pvl-negative MRSA isolates (Fig. (Fig.1)1) (P = 0.015). The higher cure rate among patients with pvl-positive MRSA cSSSI persisted when the results were stratified by infection type (abscess versus nonabscess; P = 0.029), presence of fever (P = 0.016) or leukocytosis (P = 0.002), infection size (baseline infection area of >100 cm2; P = 0.021), surgical intervention (P = 0.118), patient age (P = 0.010), diabetes (P = 0.037), and randomized study medication (P = 0.008) (Table (Table22).
Next, we evaluated potential associations between the presence of other putative virulence genes and clinical outcome. Some putative virulence genes, including clfA, spa, sdrD, bbp, ebpS, hlg, efb, icaA, and V8, were highly conserved in the majority of bacterial strains. In contrast, most toxin genes were present in a minority of MRSA isolates. Ten virulence genes were significantly associated with clinical outcome after adjustment for multiple comparisons, including four adhesin genes (fnbA, cna, sdrC, and map-eap), four toxin genes (sed, seg, sei, and sej), SCCmec type IV, and agr group II (Table (Table3).3). The presence of three adhesins (fnbA, sdrC, and map-eap) and SCCmec type IV was associated with higher cure rates than those observed when these genes were not present (false-discovery-rate-adjusted P < 0.05).
A total of 519 isolates underwent PFGE genotyping, of which 465 (89.6%) were typeable by the USA typing schema. Of these 519 isolates, 432 (83.2%) were USA300 and 21 (4.0%) were USA100. Patients infected with the USA300 MRSA clone were significantly more likely to be cured of their cSSSI than patients infected with non-USA300 isolates (398/432 [92.1%] versus 71/87 [81.6%]; P = 0.005). Conversely, patients with cSSSI due to USA100 MRSA were significantly less likely to be cured than patients infected with non-USA100 MRSA (13/21 [61.9%] versus 456/498 [91.6%]; P < 0.001).
Although pvl-positive MRSA is now a leading cause of cSSSI in many parts of the world, the impact of pvl in determining the severity of these infections is controversial. Using a large international collection of well-characterized MRSA cSSSI isolates from two phase 3 clinical trials, the current investigation has provided additional insights into the contribution of pvl to the outcome of patients with cSSSI.
The present study demonstrates that cSSSI due to pvl-positive MRSA is significantly more likely to be cured than cSSSI due to MRSA not containing this gene. This finding persisted when the analysis was adjusted for patient comorbidity as well as for infection type, severity, and surgical intervention. The results of this study validate previous observations for patients with cSSSI (2) and bacteremia (15) and are also consistent with a recent study showing that the presence of pvl did not adversely influence the outcome of uncomplicated SSSI (10). Taken together, these results clearly indicate that factors other than the simple presence of pvl determine the clinical outcome of cSSSI caused by MRSA.
The presence of pvl was associated with important clinical characteristics in patients with MRSA cSSSI. Patients with pvl-positive MRSA cSSSI were younger, more likely to have major abscess, and less likely to have health care contact. In contrast, pvl-negative strains were more likely to have originated from older patients with nonabscess infections and health care-associated strains. These findings are consistent with previous reports (19). It is likely that these patient characteristics may contribute, in part, to the better outcome seen in patients infected with pvl-positive strains.
The role of pvl in the pathogenesis of S. aureus infection remains incompletely understood. A large body of epidemiologic evidence links the presence of pvl with the presence of skin and soft tissue infections (19, 26). Several clinical observations also support the potential role of pvl in the initiation of skin and soft tissue infections, although this role is still undefined. For example, Ellis et al. (7) showed that persons asymptomatically colonized with pvl-positive S. aureus were significantly more likely to develop soft tissue infections with these same strains than persons who were colonized with S. aureus isolates that did not contain pvl. In addition, outbreak transmission (11), including interfamilial transmission (3) of skin infections, is a prominent feature of pvl-positive MRSA. While the role of pvl in the initiation of skin infections is an area of active research, the current investigation demonstrates that its presence is not the primary determinant of clinical outcome in patients with MRSA cSSSI. Thus, factors other than pvl should be considered when potential therapeutic or diagnostic laboratory-based prognostic strategies for cSSSI are targeted.
Our investigation showed that the presence of some toxin genes (sed, seg, sei, and sej) was associated with poor clinical outcome and that the presence of some adhesin genes (fnbA, sdrC, and map-eap) was associated with good clinical outcome. Our previous study showed that the toxin gene seh was associated with persistent S. aureus bacteremia (15). Peacock et al. (22) reported that several virulence genes were significantly more common in S. aureus isolates from invasive disease than in those from nasal carriage. These findings may indicate that several toxin genes are independently or jointly associated with the clinical outcome of cSSSI due to MRSA. On the other hand, the genes found in this study to be associated with clinical outcome may simply be markers of particular bacterial clones and are not themselves directly involved in the pathogenesis of more-severe infection. This explanation is supported by our finding of significant differences in outcome among different PFGE types. Because these genotyping strategies were unadjusted for confounding effects, it is likely that these associations between PFGE and outcome are also due in part to epidemiologic and infection-related characteristics.
The strengths of this study include the size, scope, and detail of the clinical material. The MRSA isolate collection used in this study is large, contemporary, international, cSSSI specific, and characterized using clinical data generated to the level of rigor of a U.S. Food and Drug Administration (FDA) registrational trial. Limitations include the fact that we did not evaluate quantitative expression or the presence of polymorphisms within pvl, as these properties may influence the function of gene products (12, 28). This evaluation focused, by design, only upon infections in which the causative pathogen was available for culture and was a subgroup analysis; therefore, it provided data that can lead only to a limited conclusion. We were also unable to consider potential relationships between vancomycin trough levels and outcome. Patients with pvl-negative MRSA isolates were relatively infrequent and differed significantly from patients with pvl-positive MRSA. Because of the small number of patients in the pvl-negative group, we were unable to employ multivariable analysis to simultaneously adjust for these differences. However, when we examined the relationship between pvl status and clinical outcome by adjusting for a single covariate at a time, our findings of a worse cure rate in patients with pvl-negative MRSA CSSI remained consistent and robust.
In summary, this study has demonstrated that pvl-positive MRSA strains appear to be associated with better outcome in patients with cSSSI. These findings indicate that factors other than the presence of pvl are the primary determinants of clinical outcome in cSSSI. Future efforts are required for a better understanding of the role of pvl in the pathogenesis of serious infections due to MRSA.
The ATLAS clinical program was funded by Theravance, Inc., South San Francisco, CA.
Editorial support was provided by Ivo Stoilov and Emily Hutchinson, medical writers at Envision Pharma, funded by Astellas Pharma, Inc. Charlotte L. Nelson served in the lead data management role for this project.
M.E.S. is a consultant for Theravance, Inc., and has previously consulted for Astellas Pharma, Inc. He has received a research grant from Theravance, Inc. S.L.B. and F.C.G. are employees of Theravance, Inc. G.R.C. is a consultant and scientific advisor (review panel or advisory committee) for Theravance, Inc., Cubist Pharmaceuticals, AstraZeneca, Astellas Pharma US, Inc., Cerexa, Inc., Merck & Co., Inc., Cempra Pharmaceuticals, Inc., Pfizer, Arpida, GSK, Inimex, Targanta, and Trius. He has received research grants from Theravance, Inc., Cerexa, Inc., Merck & Co., Inc., Cempra Pharmaceuticals, Inc., and Innocoll, Inc. V.G.F. has received grants or research support from Astellas Pharma, Inc., Cubist Pharmaceuticals, Merck & Co., Inc., Theravance, Inc., Inhibitex, Cerexa, Inc., and NIH. He is a consultant for Astellas Pharma, Inc., Cubist Pharmaceuticals, Inhibitex, Merck & Co., Inc., Johnson & Johnson, and Leo Pharma and is on the speaker's bureau and advisory committee for Cubist Pharmaceuticals. He has received honoraria from Arpida, Astellas Pharma, Inc., Cubist Pharmaceuticals, Inhibitex, Leo Pharma, Merck & Co., Inc., Pfizer, Targanta, Theravance, Inc., and Ortho-McNeil. I.-G.B., G.T.T., L.F.R., and T.H.R. have no conflicts of interest to declare.
Published ahead of print on 21 October 2009.