Effectiveness of Teicoplanin Versus β-lactam for Bacteremia Due to Methicillin-Susceptible Staphylococcus Aureus: A Propensity Score-Based Study

Background: Many studies have shown that vancomycin is inferior to β-lactam antibiotics in terms of effectiveness in the treatment of methicillin-susceptible Staphylococcus aureus (MSSA) bacteremia. However, limited data are available regarding the comparison of clinical outcomes between patients receiving initial teicoplanin and those receiving β-lactam antibiotics for MSSA bacteremia. Methods: Eighty-four adults with MSSA bacteremia were included: initial teicoplanin treatment group (n=28) and β-lactam treatment group (n=56). The two groups were further stratied based on propensity score matching according to the outcome analysis using a logistic regression model. We investigated the clinical outcomes between the groups before and after propensity score matching after treatment completion. Results: Pittsburgh bacteremia score ≥ 4 (odds ratio, 60.6; 95%CI, 7.4–496.8) was an independent risk factor for unfavorable outcome. After propensity score matching, the initial teicoplanin treatment group and the β-lactam treatment group consisted of 28 patients each. No statistically signicant differences were observed in the proportions of patients with favorable outcomes and 30-day overall mortality rates between the groups before and after propensity score matching after the completion of teicoplanin or β-lactam treatment. The Kaplan-Meier 30-day survival curve also showed no signicant difference between the patients receiving initial teicoplanin treatment and those receiving β-lactam treatment before and after matching (hazard ratio, 1.84, 95%CI, 0.60–5.64; and 3.12, 95%CI, 0.98–9.99, respectively). Conclusions: There were no signicant difference in clinical outcomes between initial teicoplanin treatment and β-lactam treatment among patients with MSSA bacteremia. Pittsburgh bacteremia score ≥ 4 was a signicant risk factor for mortality. the treatment of MSSA bacteremia. This retrospective study to examine whether the outcomes between patients with MSSA bacteremia who initially received teicoplanin and those who received initial β-lactam treatment. +Surgical intervention, drainage, central venous catheter removal, and urinary catheter removal were performed for source control. Patients with pneumonia or primary bacteremia were excluded. n: adequate and timely removal or debridement of the source of bacteremia, N: the source of bacteremia needed to be removed or debrided. The results of the present study indicated that there was no signicant difference in the clinical outcomes between patients with MSSA bacteremia receiving initial teicoplanin treatment (maintenance dose of 6 mg/kg every 12 hours) and those receiving β-lactam treatment before the availability of susceptibility report. Pittsburgh bacteremia score ≥ 4 was an independent risk factor for mortality.


Background
Staphylococcus aureus is one of the leading pathogens causing community-acquired and hospitalacquired bacteremia. Absence of appropriate antimicrobial treatment in patients with S. aureus bacteremia has a major impact on the outcomes [1]. The timing of antibiotic administration was recognized to be a major determinant of the outcome of bacteremia [2][3][4]. Initial empirical therapy for S. aureus infection may include a β-lactam or a glycopeptide such as vancomycin or teicoplanin, which is adjusted after the susceptibility test results are available [5,6]. Lodise et al reported an increase in mortality when treatment of nosocomial S. aureus bacteremia was delayed [6], and delayed initiation of appropriate therapy was common for patients with methicillin-resistant S. aureus (MRSA) infection [6].
Currently, it is clear that glycopeptides are not equivalent to β-lactams for the initial treatment of methicillin-susceptible S. aureus (MSSA) bacteremia. Many studies have indicated that vancomycin is an inferior treatment choice for MSSA strains [7,8]. Initial vancomycin treatment is associated with a higher incidence of delayed clearance (≥ 3 days) of MSSA bacteremia [8]. In a multicenter prospective observational study, Chang et al reported that nafcillin was superior to vancomycin in preventing bacteriological failure (persistent bacteremia or relapse) in patients with MSSA bacteremia [9]. Another prospective study focusing on hemodialysis-dependent patients with MSSA bacteremia. Patients who were treated with vancomycin were at a higher risk of treatment failure than those who received cefazolin [10]. Treatment with vancomycin was further shown to be associated with higher mortality in patients with bacteremic pneumonia caused by MSSA [11]. Initial vancomycin therapy with subsequent deescalation to a β-lactam antibiotic may lead to worse outcomes when compared with initial β-lactam therapy for MSSA-related infectious endocarditis in intravenous drug users [12,13].
While vancomycin and teicoplanin are used to treat gram-positive infections, there are differences in their structure, half-life, and e cacy. Teicoplanin has been widely reported to be comparable to vancomycin in e cacy but has fewer adverse effects than vancomycin [14]. Despite being structurally related to vancomycin, teicoplanin has a prolonged elimination half-life of approximately 60 hours [15]. In the Cochrane database system review, both teicoplanin and vancomycin were similarly effective in treating patients with proven or suspected infections, but the incidence of adverse effects including nephrotoxicity was lower with teicoplanin [16]. However, another meta-analysis of several trials that used adequate allocation concealment, the clinical outcome favored teicoplanin (relative risk, 0.82; 95% con dence interval [CI], 0.63-1.06) [14]. In one of the comparative studies involving febrile neutropenic patients undergoing hematopoietic stem-cell transplantation, time to attain an effective trough concentration was shorter and the rate of clinical failure was lower with teicoplanin than vancomycin [17]. Given the scarce information on the comparison between teicoplanin and β-lactams in the treatment of MSSA infections, it remains unclear whether teicoplanin is less effectiveness than β-lactam antibiotics, as observed for vancomycin in the treatment of MSSA bacteremia. This retrospective study aimed to examine whether the outcomes between patients with MSSA bacteremia who initially received teicoplanin and those who received initial β-lactam treatment.

Study design
This retrospective study included adult patients (aged ≥ 18 years) with MSSA bacteremia who were hospitalized at a 2,700-bed tertiary care hospital in southern Taiwan between 2012 and 2014. The study period and sample analysis were based on patient groups in our previous study [18,19]. Patients were included if they received either teicoplanin or a β-lactam antibiotic within 48 hours after the onset of infection and the duration of treatment with teicoplanin or β-lactam antibiotics was ≥ 72 hours. If the patients had more than one episode of MSSA bacteremia, only the rst episode was included. We collected clinical data using a standardized case report form. Adequate dosage was de ned as dosage based on the manufacturer's instructions (Sano -Aventis, Taiwan). Teicoplanin was prescribed at a loading dose of 6 mg/kg (three loading doses 12 hours apart) followed by a maintenance dose of 6 mg/kg every 12 hours or adjusted equivalent doses for patients with impaired renal function [20,21].
Adequate β-lactam therapy was de ned as treatment with adequate dosage of oxacillin, cephalosporins, or carbapenems according to patients' renal function [21]. The antimicrobial regimen was selected at the discretion of the treating clinicians. All teicoplanin and β-lactam prescriptions were approved by infectious disease specialists for their indications and dosages through the antimicrobial stewardship system [22,23]. Patients included in the study were classi ed into two groups (initial teicoplanin treatment group and β-lactam treatment group) based on initial empirical antibiotic selection for MSSA bacteremia. The physicians' decision regarding de-escalating the empirical antibiotic to de nite antibiotic was based on the results of drug susceptibility testing. We further strati ed the groups based on propensity score matching according to the outcome analysis using a logistic regression model. The study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (No. 201601482B0).

Microbiological methods
Each blood culture set (BACTEC Plus Aerobic/F and BACTEC Plus Anaerobic/F) was processed according to the BACTEC system (Becton Dickinson, Franklin Lakes, NJ, USA). MSSA was de ned as an isolate of S. aureus that was susceptible to oxacillin (using a cefoxitin disk), as determined by the disk diffusion method according to the Clinical and Laboratory Standards Institute recommendations [24]. The minimum inhibitory concentrations (MICs) of teicoplanin were determined using Etest® teicoplanin strips (AB Biodisk, Solna, Sweden).

De nitions
The clinical severity at the time of blood sampling for cultures was strati ed using the modi ed Pittsburgh bacteremia score. Patients with a Pittsburgh bacteremia score ≥ 4 points were considered to be in critical illness condition [25]. Sources of bacteremia were de ned according to microbiology, imaging ndings, and physicians' judgment. Catheter-related infection was identi ed if the inserted catheter was in place for ≥ 72 hours and the culture of the clipped distal 5-cm tip of the removed catheter yielded ≥ 15 colonies of MSSA after rolling it on the culture medium or if the culture of purulent discharge from the catheter exit site yielded MSSA [26]. Bone and joint infection was de ned based on clinical manifestations with consistent histopathological and/or radiographic ndings [27]. Infective endocarditis was identi ed if consistent histopathological ndings were observed in the valvular specimens obtained during surgery or if valvular vegetations were observed on echocardiography in patients with MSSA bacteremia [28]. Soft tissue infection was identi ed if the specimen sampled from the infected site yielded MSSA after culture [27]. Intra-abdominal infection was identi ed when cultures of the peritoneal uid, bile, or intra-abdominal abscess grew MSSA and the infection extended into the peritoneal space with abscess formation [29]. Urinary tract infection was identi ed when MSSA was the only identi ed pathogen with ≥ 10 5 colony-forming units per milliliter in urine culture [27]. Pneumonia was identi ed when clinical symptoms or signs of lower respiratory tract infection were accompanied by consistent radiographic ndings [27]. Primary bacteremia was de ned as growth of MSSA on blood culture in patients with no apparent focus of infection other than the blood culture. Adequate source control was de ned as timely percutaneous or surgical intervention to drain the infected uid, to debride the infected tissues, to drain the intra-abdominal sources of infection, or to remove the central venous catheter due to Patients were assessed on day 7 after β-lactam or teicoplanin treatment to de ne early clinical response. The nal clinical response was evaluated upon completion of the entire therapy. All patients were evaluated for the presence of septic shock, persistent bacteremia, fever, or leukocytosis. Favorable and unfavorable early clinical responses were based on the absence or presence of these ndings, respectively [30]. The primary endpoint in this study was favorable outcome, which was de ned as resolution of clinical signs and symptoms and a negative culture report at the end of therapy. Unfavorable outcomes were de ned as clinical progression of sepsis, mortality, and/or blood cultures positive for MSSA at the end of teicoplanin or β-lactam treatment. The 30-day overall mortality was de ned as all-cause mortality occurring within 30 days of hospitalization after the onset of MSSA bacteremia.

Statistical analysis
Continuous variables were compared using Student's t-test or Mann-Whitney U test. Dichotomous variables were compared using chi-squared test or Fisher's exact test. A p-value less than 0.1 was incorporated into a logistic regression model to determine independent variables associated with favorable outcome at the end of teicoplanin or β-lactam treatment. Hosmer-Lemeshow goodness-of-t test was used to evaluate the predictive performance of the logistic regression model. The propensity score matching based on 1:1 ratio was calculated using independent predictors of favorable outcome at the end of initial teicoplanin or β-lactam treatment, which were assessed using a multivariable logistic regression model. The 30-day survival was evaluated using Kaplan-Meier curves and log-rank test. All statistical analyses were performed using IBM SPSS Statistics version 21 (IBM Corp., Armonk, NY, USA). All tests were two-tailed and p-values < 0.05 were considered statistically signi cant.

Results
Among the 91 patients with MSSA bacteremia identi ed over the 3-year study period, two patients who did not receive antibiotic treatment, two patients who received initial vancomycin treatment, and three patients who were aged < 18 years were excluded. Among the 84 adult patients included in the study, 28 (33.3%) received initial teicoplanin treatment and 56 (66.7%) initial β-lactam treatment. Subsequently, we strati ed these two groups based on propensity score matching according to the outcome analysis using the logistic regression model (Fig. 1).
Comparisons of the demographics, comorbidities, severity of blood stream infection at onset, source of bacteremia, rate of adequate infection source control, infection by strains with teicoplanin MICs ≥ 1.5 mg/L, and clinical outcomes between the two groups before and after propensity score matching are presented in Table 1. Before propensity score matching, no statistically signi cant differences were observed in terms of sex, age, length of hospital stay, and the proportion of S, aureus strains with teicoplanin MICs ≥ 1.5 mg/L between the groups. A signi cantly higher prevalence of coronary artery disease (42.9% versus 8.9%, p < 0.01) and congestive heart failure (39.3% versus 7.1%, p < 0.01) was observed in the initial teicoplanin treatment group. There was no statistically signi cant difference in disease severity (Pittsburgh bacteremia score ≥ 4) between the groups or infection sources of bacteremia. The adequate infection source control rate was higher in the initial teicoplanin treatment group than in the β-lactam treatment group (78.9% versus 41.7%, p = 0.01). We did not nd statistically signi cant differences in short-term favorable outcome, favorable outcome at the time of completion of teicoplanin or β-lactam therapy, and the 30-day overall mortality rate between the two groups (Table 1). Assessment on day 7 after starting the initial teicoplanin or ß-lactam antibiotic therapy. d Evaluation at the time of completion of the initial teicoplanin or ß-lactam antibiotic therapy.
The risk factors for unfavorable clinical response at the completion of teicoplanin or β-lactam therapy in the unadjusted univariate analysis included higher disease severity (Pittsburgh bacteremia score ≥ 4 (94.1% versus 20.9%, p < 0.01) and infective endocarditis as the source of bacteremia (23.5% versus 3.0%, p = 0.01) ( Table 2). After adjustments were made in the multivariate analysis, we observed that the Pittsburgh bacteremia score ≥ 4 (odds ratio, 60.6; 95% CI, 7.4-496.8) was independently associated with unfavorable outcome at the time of completion of teicoplanin or β-lactam therapy ( Table 2). There was no signi cant evidence of lack of t in any of the nal models, as the p-values were > 0.05 in the Hosmer-Lemeshow goodness-of-t tests. *Multivariate analysis of the risk factors for an unfavorable clinical outcome in patients with MSSA bacteremia treated initially with teicoplanin or ß-lactam antibiotics showed that a Pittsburgh bacteremia score ≥ 4 (odd ratio, 60.6 [95% con dence interval, 7.4-496.8], p < 0.01) was an independent risk factor for an unfavorable outcome.
All patients included in the study were divided into the initial teicoplanin treatment (n = 28) and β-lactam treatment (n = 28) groups after 1:1 propensity score matching with Pittsburgh bacteremia score ≥ 4 (the independent risk factor for unfavorable outcome) and age. After matching, there were no statistically signi cant differences in terms of sex, length of hospital stay, and the proportion of S. aureus strains with teicoplanin MICs ≥ 1.5 mg/L between the groups. The incidence of coronary artery disease and congestive heart failure was higher (42.9% versus 10.7%, p = 0.03 and 39.3% versus 10.7%, p = 0.04, respectively) among patients receiving initial teicoplanin treatment than in those receiving β-lactam treatment even after propensity score matching. No statistically signi cant differences were observed in the source of bacteremia and the rate of adequate infection source control between the two groups after matching. We did not nd statistically signi cant differences in short-term favorable outcome, favorable outcome at the time of completion of teicoplanin or β-lactam therapy, and 30-day overall mortality rate between the groups after propensity score matching.
Among the patients in the initial teicoplanin treatment group, 21 (75.0%) switched to β-lactam treatment and three (14.3%) died within 30 days of hospitalization after the onset of MSSA bacteremia. Altogether, 71.5% (15/21) of the patients switched to β-lactam treatment within 4 days after the onset of bacteremia, which was consistent with the time of availability of nal susceptibility test results (Supplementary Figure). On the other hand, the 30-day mortality rate was 14.3% in patients with initial teicoplanin treatment without a switch to β-lactam treatment for MSSA bacteremia (  Fig. 3. The cases were grouped according to the initial treatment (teicoplanin or β-lactam) before propensity score matching (Fig. 3(i)) and were further strati ed as the initial teicoplanin treatment group or the βlactam treatment group after propensity score matching (Fig. 3(ii)). The 30-day survival was not signi cantly different between the two groups before propensity score matching (hazard ratio, 1.84; 95% CI, 0.60-5.64, p = 0.29) as well as after propensity score matching (hazard ratio, 3.12; 95% CI, 0.98-9.99, p = 0.06).

Discussion
Several studies regarding initial vancomycin treatment for MSSA bacteremia have revealed unfavorable outcomes [7][8][9][10][11][12][13]. Lodise et al reported that 72 patients with MSSA-related infective endocarditis who were initially treated with vancomycin and subsequently switched to a β-lactam antibiotic in a median duration of 3 days had a four-fold increase in the mortality compared to those who were initially treated with a β-lactam antibiotic (9/22 [40.9%] versus 5/44 [11.4%]) [6]. This nding suggests that the timing initiation of appropriate empirical therapy is critical for the outcomes, since even the patients who switched to β-lactam therapy after the availability of susceptibility results showed outcomes inferior to those treated with a β-lactam antibiotic initially. To the best of our knowledge, our study, which consisted of a cohort of 84 patients with MSSA bacteremia, is the rst study to examine the clinical outcomes in patients who received teicoplanin or β-lactam treatment. In our study, patients in the initial teicoplanin treatment group did not seem to exhibit poorer outcome than those in the β-lactam treatment group. Even after propensity score matching for disease severity, patients in the initial teicoplanin treatment group exhibited similar clinical outcomes when compared with patients in the β-lactam treatment group. No empirical regimen involving a switch from teicoplanin to β-lactam was found to be superior or inferior to the other even when patients switched to β-lactam treatment after the availability of culture results. This relationship persisted in the logistic regression analysis controlling for clinical characteristics. This nding may be related to the timing of the switch to β-lactam treatment in our cohort, which occurred less than 4 days after the onset of infection in 53.6% (21/28) (Supplementary Figure). Another issue regarding the relationship of treatment outcomes with teicoplanin MICs has been discussed by Chang et al [22].
They suggested that MRSA bacteremia with a teicoplanin MIC ≥ 1.5 mg/L was a predictive factor for unfavorable outcomes in patients receiving teicoplanin treatment [22]. In the present study, 78.6% (22/28) of the patients in the initial teicoplanin treatment group with MSSA bacteremia had teicoplanin MICs < 1.5 mg/L (Fig. 2). This result may partially explain the absence of statistically signi cant unfavorable clinical outcomes in the initial teicoplanin treatment group. Moreover, our recent study indicated that patients with MRSA bacteremia who received teicoplanin treatment with a maintenance dose of 6 mg/kg every 12 hours exhibited favorable outcomes irrespective of the teicoplanin MIC [31]. Further studies with larger sample sizes may be needed to con rm our ndings.
A previous study indicated that in vitro bactericidal activity of vancomycin is slower than that of nafcillin, with more frequent failures in animal models [32]. In our present study, patients with MSSA bacteremia in the initial β-lactam treatment group exhibited a higher rate of favorable short-term outcomes than those in the initial teicoplanin treatment group (60.7% versus 39.3%). However, the difference was not statistically signi cant, which might be due to our small sample size. Some studies conducted in special patient groups such as hemodialysis-dependent and bloodstream-related pneumonia patients with MSSA bacteremia demonstrated high rates of treatment failure and mortality among patients receiving vancomycin treatment [10,11]. Among patients with MSSA bacteremia, the 30-day mortality or unfavorable outcomes were not signi cantly higher in patients with infection due to pneumonia and in hemodialysis-dependent patients when compared with those having other sources of infection. This result seems to be inconsistent with the ndings of Stryjewski et al [10]. Although the lung penetration of teicoplanin may be lower than that of vancomycin [33,34], a higher dose of teicoplanin would be more effective [31], particularly in patients with bacteremic pneumonia. Other reasons may be related to the small sample size of this patient population in our study and the fact that only one patient was infected by a bacterial strain with teicoplanin MIC ≥ 1.5 mg/L. The present study still has some limitations. The retrospective study design has inherent limitations due to potential confounding and selection bias. Prospective randomized controlled trials are needed to validate our ndings. A small proportion of patients were diagnosed with endocarditis in the present study. Variations in the source of MSSA bacteremia and the small number of patients with endocarditis precludes us from making comparisons the outcomes of endocarditis with those of other more common sources of MSSA bacteremia. We did not analyze serum teicoplanin levels in the present study. However, all patients received teicoplanin treatment with a maintenance dose of 6 mg/kg every 12 hours. Our previous experience has shown that this dosage will produce favorable outcomes in patients with MRSA bacteremia irrespective of the teicoplanin MIC [31].