Cochrane meta-analysis : teicoplanin versus vancomycin for proven or suspected infection

Objective: To compare efficacy and safety of vancomycin versus teicoplanin in patients with proven or suspected infection. Methods: Data Sources: Cochrane Renal Group’s Specialized Register, CENTRAL, MEDLINE, EMBASE, nephrology textbooks and review articles. Inclusion criteria: Randomized controlled trials in any language comparing teicoplanin to vancomycin for patients with proven or suspected infection. Data extraction: Two authors independently evaluated methodological quality and extracted data. Study investigators were contacted for unpublished information. A random effect model was used to estimate the pooled risk ratio (RR) with 95% confidence interval (CI). Results: A total of 24 studies (2,610 patients) were included. The drugs had similar rates of clinical cure (RR: 1.03; 95%CI: 0.98-1.08), microbiological cure (RR: 0.98; 95%CI: 0.93-1.03) and mortality (RR: 1.02; 95%CI: 0.79-1.30). Teicoplanin had lower rates of skin rash (RR: 0.57; 95%CI: 0.35-0.92), red man syndrome (RR: 0.21; 95%CI: 0.08-0.59) and total adverse events (RR: 0.73; 95%CI: 0.53-1.00). Teicoplanin reduced the risk of nephrotoxicity (RR: 0.66; 95%CI: 0.48-0.90). This effect was consistent for patients receiving aminoglycosides (RR: 0.51; 95%CI: 0.30-0.88) or having vancomycin doses corrected by serum levels (RR: 0.22; 95%CI: 0.10-0.52). There were no cases of acute kidney injury needing dialysis. Limitations: Studies lacked a standardized definition for nephrotoxicity. Conclusions: Teicoplanin and vancomycin are equally effective; however the incidence of nephrotoxicity and other adverse events was lower with teicoplanin. It may be reasonable to consider teicoplanin for patients at higher risk for acute kidney injury.


INTRODUCTION
Methicillin-resistant Staphylococcus aureus (MRSA) is a leading cause of bloodstream and other invasive infections worldwide (1,2) .Between 48 and 57% of S. aureus isolates from inpatients are resistant to methicillin in United States (3,4) and around 30% in many European countries (5) .Vancomycin remains the drug of choice for the treatment of infections caused by MRSA; however one of the major limitations for its use is its potential nephrotoxicity (6) .Teicoplanin, another glycopeptide, has essentially the same efficacy of vancomycin, and with some advantages, such as once-daily bolus administration, intramuscular use, lack of requirement for routine serum monitoring and possibly less nephrotoxicity (7) .However teicoplanin is more expensive.
There is uncertainty as to whether vancomycin causes permanent or temporary kidney damage.Many studies have shown an increased risk of kidney failure after vancomycin treatment (8)(9)(10)(11)(12)(13) , although others have not found an association (14)(15)(16) .In fact, adverse kidney effects were common with earlier vancomycin preparations, but the significance of this problem is less well-established with current purified formulations (8) .Furthermore, other factors, such as association with nephrotoxic drugs, especially aminoglycosides, and different nephrotoxicity definitions may have blurred the real impact of vancomycin on kidney function in some previous studies (17) .
Vancomycin might lead to nephrotoxicity due to its effects on proximal tubular cells, where it accumulates inside lysosomes (18,19) .There, it inhibits the activity of many enzymes, such as sphingomyelinase, resulting in vacuolization and necrosis (20) .As aminoglycosides accumulate in the same cells and are also nephrotoxic, using both drugs simultaneously may lead to a faster and more severe loss of kidney function (21) .
To date, just one meta-analysis of randomized controlled trials (RCTs) has been published on this issue (7) .The authors found no difference between vancomycin and teicoplanin regarding clinical or bacteriological response.However, 10.7% of vancomycin treated patients developed nephrotoxicity compared to 4.8% of those treated with teicoplanin (p < 0.001).Nevertheless, methods used to conduct this meta-analysis were poorly reported, seriously hindering interpretation of its results.

OBJECTIVE
This systematic review of RCTs aimed to investigate the efficacy and safety of vancomycin compared to teicoplanin, in patients with proven or suspected infection.

Types of studies
We included all RCTs and quasi-RCTs (RCTs in which allocation to treatment was obtained by alternation, use of alternate medical records, date of birth or other predictable methods) comparing intravascular (IV) vancomycin to IV or intramusculra (IM) teicoplanin.Studies were considered for inclusion regardless of their publication status, language, blinding, size, duration of patient follow-up, or their primary objectives and reported outcomes.
RCTs in which there were no relevant or adverse events in both the treatment and control groups were excluded, because these studies provide no information on the magnitude of the treatment effect (22) .

Types of participants Inclusion criteria
-Patients of all ages with suspected or proven Grampositive infection.
Exclusion criteria -Use of teicoplanin or vancomycin for prophylaxis (rather than for suspected or proven infection).

Types of interventions
-At least one arm allocated to receive IV or IM teicoplanin, and another arm to receive IV vancomycin.

Types of outcome measures
Primary outcomes -Nephrotoxicity: an elevation of serum creatinine (SCr) greater than or equal to twice the basal level, or urine output less than 0.5 mL/kg/h over a 12-hour period.In case data were not available according to this definition and after contacting authors, a similar definition used in the original study was accepted.-Clinical cure: patients who showed resolution or significant improvement of signs and symptoms by the end of study drug treatment.
-Microbiological cure defined as a negative culture from a material in which it had been previously positive.-Mortality.-Infusion reactions.
-Other adverse events reported in the studies.

Search methods for identification of studies
The search strategy included all languages.The following sources were searched.

Electronic searches
1.The Cochrane Renal Group specialized register and the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library.CENTRAL and the Cochrane Renal Group's specialized register contain the hand-searched results of conference proceedings from general and specialty meetings.This is an ongoing activity across the Cochrane Collaboration and is both retrospective and prospective.Therefore we did not specifically search conference proceedings.
Please refer to The Cochrane Renal Group's Module in The Cochrane Library for the most up-to-date list of conference proceedings (23) .2. MEDLINE (from 1966) using the optimally sensitive strategy developed for the Cochrane Collaboration for the identification of RCTs (23) together with a specific search strategy, developed with input from the Cochrane Renal Group Trial Search Coordinator.3. EMBASE (from 1980) using a search strategy adapted from that developed for the Cochrane Collaboration for the identification of RCTs (23) together with a specific search strategy developed with input from the Cochrane Renal Group Trial Search Coordinator.
Check appendix 1 for search terms used.
Searching other resources 1. Reference lists of nephrology textbooks, review articles and relevant studies.2. Letters seeking information about unpublished or incomplete studies to investigators known to be involved in previous studies.

Assessment of study eligibility
The review was undertaken by five authors (AC, AG, DB, CA and ES).The search strategy described was used to obtain titles and abstracts of studies that might be relevant to the review.Two authors (DB and CA) independently selected the abstracts identified in our search.
If any of the authors considered a citation might possibly include a relevant RCT the full text article was assessed.After obtaining the full text articles, each potential was evaluated independently by two authors (groups of two formed by AC, AG, DB, CA or ES).In the case of a disagreement, the authors discussed the reasons for their decisions.If the disagreement was not resolved during this process, a third author would make the final decision (AC or ES or AG).In case of any doubts about the study design (e.g.observational study compared to RCTs), the author of the publication was contacted.

Data extraction
Data extraction was carried out independently by AC and ES using standard data extraction forms.Disagreements were resolved by consensus.Studies reported in non-English language were translated before assessment.Duplicate publications or sub-studies of included studies were listed under the primary reference, since they may have provided information on relevant outcomes not available in the original publication.Any further information required from the original author was requested by written correspondence.

Study quality
The quality of studies included was assessed independently by AC and ES without blinding to authorship or journal using the checklist developed for the Cochrane Renal Group.Discrepancies were resolved by discussions aimed at a consensus.

Statistical assessment
Dichotomous data (e.g.AKI needing dialysis, or nephrotoxicity as defined above) from all included RCTs was combined to estimate the pooled risk ratio (RR) with 95% confidence interval (CI) using a randomeffects model (24) .
The analyses were based on intention-to-treat data from the individual studies, whenever possible.Every effort was made to obtain complete information about patients' outcomes, including contacting authors.However, we did not include in the denominator patients with no follow-up.
The presence of heterogeneity across studies was evaluated using I² statistics (25) and standard χ 2 tests for homogeneity for each outcome analysis.An I² value Bugano DDG, Cavalcanti AB, Goncalves AR, Almeida CS, Silva E represents the percentage of total variation across studies due to heterogeneity rather than chance.We considered an I² value less than 25% as low and an I² value more than 75% as high.We looked for potential publication bias and other biases associated with small study effects by constructing funnel plots (26) .Funnel plots are simple scatter plots of the treatment effects obtained from individual studies on the vertical axis (for example, log OR) against some measure of study size on the horizontal axis (for example, standard error of log OR).
We had originally planned to carry out univariate and multivariate random-effects meta-regression models to analyze potential clinical and study quality factors that might influence treatment effects, that is, in an attempt to explain heterogeneity (27,28) .The following variables were to be considered: standard error of log odds ratio, publishing status (MEDLINE indexed or not), study quality (generation of allocation sequence, allocation sequence concealment, follow-up, intention-to-treat analysis), definition of nephrotoxicity, dose adjustment guided by vancomycin serum measurement, clinical sub-groups (critically ill patients, kidney failure patients, elderly patients or concomitant aminoglycoside use).However, as we have not found substantial heterogeneity for any of the primary outcomes, meta-regression was not performed.We conducted simple sub-group analyses instead (serum vancomycin-guided dose adjustment and concomitant aminoglycoside use).We had planned to look to other sub-groups (according to age or baseline kidney function), but that was not feasible because we were unable to obtain appropriate data.
Adverse effects were tabulated and assessed with descriptive techniques.Whenever possible, the pooled RR with 95%CI was calculated for each adverse effect.
All p values reported were two-tailed and values lower than 0.05 were considered significant, except for the χ 2 test for homogeneity.This method has low sensitivity for detecting heterogeneity using few studies, therefore we considered a p value lower than 0.10 as statistically significant.

Description of studies
We initially identified 909 potentially relevant studies (Figure 1).After evaluating their abstracts (or titles) we excluded 796 reports because they were not RCTs or did not compare teicoplanin to vancomycin.The full-text articles of the remaining 113 studies were evaluated, with a further 82 considered ineligible.This left 31 potentially relevant RCTs.Five reports were duplicate publications of included (29)(30)(31) and excluded studies (32) ; one report was a subset of a larger study (33) and one study used just one dose of vancomycin or teicoplanin and was excluded (34) .The 24 studies finally included enrolled 2,610 patients.Most were published between 1988 and 2000, with 3 studies published between 2001 and 2004 (Table 1).The median sample size was 72 patients, ranging from 20 to 635.Most evaluated adults, with only two studies including pediatric patients.Ten of 24 studies evaluated febrile neutropenic patients, the remaining included several other infections related or probably related to Gram-positive bacteria.Sixteen studies did not include patients with previously elevated SCr, although cut-off levels for exclusion varied.Definitions of nephrotoxicity were also not uniform across the studies.
Most studies administered 6 to 10 mg/kg of teicoplanin IM or IV, every 12 hours, for 3 doses, then once daily (Table 1).Several schemes of vancomycin were used, varying from 24 to 40 mg/kg/d, divided into 2 to four doses or a fixed dose of 2 g/d divided into 2 to four doses.Vancomycin was adjusted according to serum levels in seven studies, although only for selected patients in two of these.

Risk of bias in included studies
In general, the quality of included studies was poor (Appendix 3).Only 6 out of 24 studies reported allocation concealment.Blinding of participants, healthcare personnel and outcome assessors was adequately described in 5 out of 24 studies.Intention-totreat analysis was performed in only 7 out of 24 studies.Post-randomization exclusions or losses to follow-up were greater than 10% in 13 out of 24 studies.
In six studies the unit of randomization and analysis was an infection episode.That is, the same patient could be included twice or more in the study.This is inappropriate because statistical methods used assume independency of observations.

Effects of interventions
The main results are summarized in table 2 and in the appendix 4. Teicoplanin reduced the risk of nephrotoxicity (Table 3: RR: 0.66; 95%CI: 0.48-0.90;I² = 10%).Ordering the studies according to the year of publication data did not suggest a pattern of decreasing nephrotoxicity related to vancomycin in the more recent studies.Clinical cure was similar with teicoplanin or vancomycin (Appendix 5: RR: 1.03; 95%CI: 0.98-1.08;I² = 0%) as well as microbiological cure (RR: 0.98; 95%CI: 0.93-1.03;I 2 = 0%).Funnel plots for nephrotoxicity or clinical cure did not suggest either a small studies' effect or reporting bias (graphs not shown in this manuscript).
We did not carry out meta-regression analysis because there was no evidence of substantial heterogeneity between the study results for the main endpoints (nephrotoxicity and clinical cure).
Sub-group analyses according to clinical indication (febrile neutropenia, catheter-associated infection, Gram-positive bacteraemia, endocarditis, bone/joint infection or other Gram-positive infections) did not show any evidence of superiority of either vancomycin or teicoplanin for any indication (Appendix 6).With respect to nephrotoxicity, subgroup analysis suggested no difference in the treatment effect for the comparisons of studies with adequate allocation concealment versus unclear or no allocation concealment (test for subgroup differences, p = 0.56), studies with blinding of participants, healthcare personnel and outcome assessors and studies with unclear or no blinding (test for subgroup differences, p = 0.70) and studies with versus without intention-to-treat analysis (test for subgroup differences, p = 0.48).
Data on AKI with an indication for dialysis was available in only 6 studies (786 patients).No patient in either the vancomycin or teicoplanin group needed dialysis, therefore it was impossible to estimate the RR.There was no evidence of a higher nephrotoxic effect of vancomycin compared to teicoplanin in patients receiving concomitant aminoglycosides (Appendix 7).A post-hoc analysis of nephrotoxicity limited to studies in which all patients had vancomycin administered according to serum levels provided results similar to the overall estimate (RR: 0.22; 95%CI: 0.10-0.52;I² = 0%).However, this analysis was based on only 32 nephrotoxic events in 5 studies.Data on other subgroups was unavailable (critically ill patients, kidney failure patients and elderly patients).
The effect of teicoplanin on microbiological cure was similar to vancomycin.Mortality was similar with both antibiotics (RR: 1.02; 95%CI: 0.79-1.3;I 2 = 0%), but due to serious imprecision and poor quality of included studies, this is low quality evidence.Skin rash (RR: 0.57; 95%CI: 0.35-0.92;I 2 = 5%) and red man syndrome were observed much less often with teicoplanin than with vancomycin.The incidence of any adverse effect was 27% lower with teicoplanin, although heterogeneity was very high (RR: 0.73; 95%CI: 0.53-1.0;I 2 = 52%).Comment: only six studies reported this outcome.No event was observed, therefore no pooled effect could be estimated.GRADE Scoring system: high quality: further research is very unlikely to change our confidence in the estimate of effect; moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate; low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; very low quality: we are very uncertain about the estimate.

Summary of main results
In this systematic review and meta-analysis, we found a similar effect of teicoplanin compared to vancomycin on clinical and microbiological cure.However the RR of nephrotoxicity was reduced by 34% when using teicoplanin.This represents a number needed to harm of 25 (assuming a risk of nephrotoxicity with vancomycin of 9%).The reduced nephrotoxicity of teicoplanin compared to vancomycin was similarly observed in patients with or without aminoglycosides, and also in studies in which vancomycin administration was guided by serum levels.Skin rash, red man syndrome and total adverse events were also less common with teicoplanin than vancomycin.Mortality was similar with both drugs, but the total number of deaths was low.Thus, there is inadequate precision in the estimate of effect on mortality.

Overall completeness and applicability of evidence
The results of this systematic review are applicable to most patients for whom teicoplanin or vancomycin is being considered for treatment of a Gram-positive infection, in particular due to MRSA.
However, some groups of patients may not have been adequately represented in this review.Most studies excluded patients with kidney failure and none included only critically ill patients.Data specific for the subgroups of kidney failure, critically ill or elderly patients were not available from the publications of the original studies and could not be obtained from the authors.Data on AKI needing dialysis was available in only six studies, but no patient (0/786) developed this complication in either antibiotic group.Thus, it was not possible to evaluate whether the lower risk of nephrotoxicity with teicoplanin than with vancomycin translates into a lower risk of AKI requiring dialysis.The absence of cases needing dialysis is most likely explained by the selection of patients at lower risk for this event, for instance under-representation of previous kidney failure or critically ill patients.Also, vancomycininduced nephrotoxicity is mild.However, it is possible that progression to dialysis may be precipitated by vancomycin among higher risk patients.
Comparative evaluations of clinical cure according to clinical site showed a consistent effect for the sites of infection/indications evaluated.Some previous studies suggest that the failure rate in endocarditis may be unacceptable with teicoplanin at usual doses (6 mg/kg every 12 hours for 3 doses, then once a day) compared to vancomycin (33,35,36) .Teicoplanin, even at higher doses, does not penetrate the vegetations; thus, success may be achieved only for small vegetations or when aminoglycosides are associated (37) .The totality of evidence from RCTs regarding endocarditis suggests teicoplanin is similar to vancomycin; however, a small study (38) had discrepant results, which were unfavorable to teicoplanin.This resulted in large inconsistent (I² = 52%) between-study effects.Thus, it is not possible to conclude on the efficacy of teicoplanin for this condition.

Quality of the evidence
The RCTs included in this review are generally small and only a few are free of methodological problems, Table 3. Pooled data for rate of nephrotoxicity of vancomycin and teicoplanin thereby increasing the risk of biased results.There was low heterogeneity between estimates of effect from the included studies for all outcomes, except occurrence of any adverse event.This last result is probably a consequence of the very different definitions of "any adverse event" used in the primary studies.
The quality of the evidence regarding the effect of teicoplanin compared to vancomycin on nephrotoxicity is moderate according to the GRADE system (39) .Limitations in design of primary studies downgraded the quality of evidence.The GRADE quality of evidence is also moderate for the evaluation of clinical cure.The level of evidence was downgraded due to methodological limitations of primary studies.

Potential biases in the review process
In order to ensure a high degree of internal and external validity, we followed a systematic approach for study identification, selection, data abstraction and analysis.Bugano DDG, Cavalcanti AB, Goncalves AR, Almeida CS, Silva E We searched for all relevant studies using sensitive and validated search strategies in several bibliographic databases.Studies were included independent of publication status or language.Original investigators were contacted, and some, but not all, contributed additional information.Data on the main outcome nephrotoxicity was obtained from 23 out of 24 studies and on clinical cure from 20 out of 24 studies.We looked for and found no evidence of reporting or small studies' bias using funnel plots for these outcomes.
Limitations in this review include the lack of a uniform definition of nephrotoxicity in the original studies.In fact, until recently there was not a universally recognized definition of AKI and several definitions were used in the literature (40) .The current definition of AKI proposed by the Acute Kidney Injury Network (AKIN) includes an elevation of at least 0.3 mg/dL in baseline levels of creatinine or a 50% increase in two different measurements, or a urine output lower than 0.5 mL/kg/h for over 6 hours (41) .The AKIN definition had not been published when we prepared this review's protocol.Therefore, we defined nephrotoxicity in our review according to the "injury" component of the RIFLE criteria for AKI (42) .However, we were unable to obtain data on nephrotoxicity according to our definition from the study authors.Therefore, we abstracted nephrotoxicity data as defined in the original studies, with the most common definition being an increase in SCr > 0.5 mg/dL above baseline.In spite of no uniformity in the definition of this outcome, there was no evidence of substantial heterogeneity among studies regarding the effect of teicoplanin versus vancomycin on nephrotoxicity.

Agreements and disagreements with other studies or reviews
One meta-analysis evaluating teicoplanin versus vancomycin was previously published; however, the author did not report any structured method for study identification, selection and analysis (7) .In that study, both drugs achieved similar probabilities of clinical cure (72.7% for teicoplanin versus 77.2% for vancomycin); nonetheless, teicoplanin had significantly less adverse events (21.9% versus 13.9%, p = 0.0003), especially less nephrotoxicity (4.8% versus 10.7%, p = 0.0005).A formal approach was followed in the present review and ten additional studies were included.Despite these differences, we found similar results for clinical cure (74.3 versus 72.0%) and nephrotoxicity (4.7 versus 9.2%).
A recurrent issue in the literature on teicoplanin is the relation between dose and its clinical efficacy (36,43) .Currently the recommended dose is 6 mg/kg (or 400 mg) every 12 hours, for 3 doses, then 6 mg/kg (or 400 mg) once daily, doubling this dose for endocarditis (36) .Initial studies with teicoplanin used a much lower dose, generally half of that currently used (31,44,45) .Most studies in this review used the current larger dose (400 mg/kg every 12 hours for 3 doses, then once daily), or changed to the larger dose during the study.The results of these studies present a very similar and consistent effect of teicoplanin versus vancomycin on clinical or microbiological cure.Recently a loading dose of 6 mg/kg every 12 hours, for 4 doses, then once daily, has been recommended to speedily achieve optimal concentrations of serum teicoplanin (46) .

CONCLUSIONS Implications for practice
This review summarizes the best available evidence on the use of teicoplanin versus vancomycin for infected or suspected to be infected patients.The overall quality of evidence across all comparisons is low to moderate using the GRADE system (39) .Teicoplanin is as efficacious as vancomycin regarding clinical and microbiological cure, although it is associated with a lower risk of nephrotoxicity and skin rash.Since no patient on either antibiotic required dialysis, the effect of teicoplanin compared to vancomycin on this outcome could not be determined.Thus it remains unclear whether teicoplanin has a clinically relevant advantage over vancomycin, although it may be reasonable to consider teicoplanin a better choice for patients at higher risk for AKI needing dialysis.
There is no consistent evidence of efficacy of teicoplanin compared to vancomycin for treating endocarditis.Therefore, teicoplanin cannot be currently recommended for this condition.

Implications for research
Investigators should conduct studies to evaluate antibiotics for Gram-positive infections with a sound design and adequate power to evaluate outcomes relevant to patients.Studies with vancomycin should report the incidence of AKI needing dialysis.Future studies involving vancomycin should use serum levels to guide dose adjustments.This review showed that the risk of nephrotoxicity was also higher in patients receiving vancomycin guided by serum levels, but this analysis was based on only a few events from four studies.
No RCT evaluated vancomycin versus teicoplanin exclusively in critically ill patients.We were also unable to obtain data specific for this subgroup in our review.Nevertheless, antibiotics to treat MRSA and other Grampositive infections are widely used in the intensive care setting.The effects of vancomycin versus teicoplanin in patients with previous kidney injury are also unclear from the available evidence.Thus, studies involving critically ill and kidney injury patients are necessary.Finally, adequately powered RCTs are warranted to evaluate the efficacy of teicoplanin compared to vancomycin for the treatment of endocarditis.

Figure 1 .
Figure 1.Selection of studies for inclusion in the systematic review of teicoplanin versus vancomycin for proven or suspected infection

Appendix 5 .
Pooled analysis of rates of clinical cure or improvement for vancomycin and teicoplanin

Table 2 .
Summary of findings for the main comparison