UBC Faculty Research and Publications

Antifungal treatment for invasive Candida infections: a mixed treatment comparison meta-analysis Mills, Edward J; Perri, Dan; Cooper, Curtis; Nachega, Jean B; Wu, Ping; Tleyjeh, Imad; Phillips, Peter Jun 26, 2009

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12941_2009_Article_159.pdf [ 738.74kB ]
JSON: 52383-1.0223049.json
JSON-LD: 52383-1.0223049-ld.json
RDF/XML (Pretty): 52383-1.0223049-rdf.xml
RDF/JSON: 52383-1.0223049-rdf.json
Turtle: 52383-1.0223049-turtle.txt
N-Triples: 52383-1.0223049-rdf-ntriples.txt
Original Record: 52383-1.0223049-source.json
Full Text

Full Text

ralAnnals of Clinical Microbiology and ssBioMed CentAntimicrobialsOpen AcceResearchAntifungal treatment for invasive Candida infections: a mixed treatment comparison meta-analysisEdward J Mills*1,2, Dan Perri3, Curtis Cooper4, Jean B Nachega5,6, Ping Wu2, Imad Tleyjeh7,8 and Peter Phillips9Address: 1Faculty of Health Sciences, Simon Fraser University, Burnaby, Canada, 2Department of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Canada, 3Department of Medicine, McMaster University, Hamilton, Canada, 4Division of Infectious Diseases, Ottawa Hospital, University of Ottawa, Ottawa, Canada, 5Departments of Epidemiology and International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA, 6Department of Medicine and Centre for Infectious Diseases, Faculty of Health Sciences, Stellenbosch University, Cape Town, South Africa, 7Division of Infectious Diseases, Department of Medicine, Research Center, King Fahd Medical City, Riyadh, Saudi Arabia, 8Division of Infectious Diseases, Department of Medicine, Mayo Clinic, Rochester, MN, USA and 9Division of Infectious Diseases, University of British Columbia, Vancouver, CanadaEmail: Edward J Mills* - emills@cfenet.ubc.ca; Dan Perri - dan.perri@utoronto.ca; Curtis Cooper - ccooper@ottawahospital.on.ca; Jean B Nachega - jnachega@jhsph.edu; Ping Wu - pwu@ccnm.edu; Imad Tleyjeh - Tleyjeh.Imad@mayo.edu; Peter Phillips - pphillips@cfenet.ubc.ca* Corresponding author    Objectives: Invasive fungal infections are a major cause of mortality among patients at risk. Treatmentguidelines vary on optimal treatment strategies. We aimed to determine the effects of different antifungaltherapies on global response rates, mortality and safety.Methods: We searched independently and in duplicate 10 electronic databases from inception to May2009. We selected any randomized trial assessing established antifungal therapies for confirmed cases ofinvasive candidiasis among predominantly adult populations. We performed a meta-analysis and thenconducted a Bayesian mixed treatment comparison to differentiate treatment effectiveness. Sensitivityanalyses included dosage forms of amphotericin B and fluconazole compared to other azoles.Results: Our analysis included 11 studies enrolling a total of 965 patients. For our primary analysis ofglobal response rates, we pooled 7 trials comparing azoles to amphotericin B, Relative Risk [RR] 0.87 (95%Confidence Interval [CI], 0.78–0.96, P = 0.007, I2 = 43%, P = 0.09. We also pooled 2 trials of echinocandinsversus amphotericin B and found a pooled RR of 1.10 (95% CI, 0.99–1.23, P = 0.08). One study comparedanidulafungin to fluconazole and yielded a RR of 1.26 (95% CI, 1.06–1.51) in favor of anidulafungin. Wepooled 7 trials assessing azoles versus amphotericin B for all-cause mortality, resulting in a pooled RR of0.88 (95% CI, 0.74–1.05, P = 0.17, I2 = 0%, P = 0.96). Echinocandins versus amphotericin B (2 trials) for all-cause mortality resulted in a pooled RR of 1.01 (95% CI, 0.84–1.20, P = 0.93). Anidulafungin versusfluconazole resulted in a RR of 0.73 (95% CI, 0.48–1.10, P = 0.34). Our mixed treatment comparisonanalysis found similar within-class effects across all interventions. Adverse event profiles differed, withamphotericin B exhibiting larger adverse event effects.Conclusion: Treatment options appear to offer preferential effects on response rates and mortality.Published: 26 June 2009Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 doi:10.1186/1476-0711-8-23Received: 21 January 2009Accepted: 26 June 2009This article is available from: http://www.ann-clinmicrob.com/content/8/1/23© 2009 Mills et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 11(page number not for citation purposes)When mycologic data are available, therapy should be tailored.Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23IntroductionInvasive fungal infections contribute importantly to mor-bidity and mortality in immunocompromised patientsincluding those with hematologic cancers, recent trans-plants, autoimmune disorders, and critical illness. Themost common fungal pathogen is Candida.For a number of reasons, early and accurate diagnosis ofinvasive fungal infections is often difficult and patientsfrom these high risk groups may have evidence of dissem-inated fungal infection at autopsy that was not identifiedprior to death[1]. Clinical manifestations of invasive fun-gal infections often occur at a late stage of infection con-tributing to diagnostic delays and higher case-fatalityrates. Immunocompromised patients may not generatedetectable antibody responses to specific fungal patho-gens[2]. Furthermore, non-invasive antigen detectionmethods such as the Fungitec G assay for beta-1–3 glucanare not widely available. Finally, culture techniques arenot highly sensitive, and invasive diagnostic techniquesmay be contraindicated or not applicable to the clinicalpresentation. Consequently, antifungal treatment for con-firmed invasive fungal infections is challenging and eval-uations of therapeutic interventions are limited[3].Several choices of antifungal agents exist that differ greatlywith respect to both toxicity and cost[4].There is an ever-growing literature on the use of antifungalagents in patients with candidemia. Previous systematicreviews have not looked at the relative effectiveness ofinterventions of confirmed infections [4-6]. Using a sys-tematic review of the literature and meta-analytic tech-niques, we aimed to quantify the effects of antifungaltherapy on confirmed systemic fungal infection responserates, associated mortality and safety when reserved forconfirmed cases only. Furthermore, we determined differ-ences in treatment effects across interventions using amixed treatment comparison meta-analysis.MethodsEligibility criteriaWe included any randomized trial of antifungal therapiesfor confirmed cases of invasive candidiasis among pre-dominantly adult (³ 18 years of age) populations. Weincluded randomized trials of any duration. Given thathead to head evaluations have existed for decades, studieshad to compare antifungal therapy to another antifungaltherapy (head-to-head evaluations). Studies had to reporton any of the following clinically-important outcomes:clinical response, all-cause mortality; fungal-attributabledeath, and adverse events. We excluded studies onlyreporting on dose-comparison or dosage form evalua-tions. As we were interested in disseminated disease, trialschest) were excluded, as were aspergillosis trials, crypto-coccosis and endemic mycoses trials. We excluded drugsno longer recommended by Infectious Disease Society ofAmerica Guidelines (IDSA) including ketoconazole[7].Search strategyIn consultation with a medical librarian, we (DP, EM)established a search strategy (available from correspond-ing author on request). We searched independently, induplicate, the following 10 databases (from inception toMay 2009): MEDLINE, EMBASE, Cochrane CENTRAL,AMED, CINAHL, TOXNET, Development and Reproduc-tive Toxicology, Hazardous Substances Databank, Psych-info and Web of Science, databases that included the fulltext of journals (OVID, ScienceDirect, and Ingenta, includ-ing articles in full text from approximately 1700 journalssince 1993). Key search words included words addressingthe infections: Fungus, fungal, fungemia, mycosis, candidia-sis, candidemia, candida; and words addressing the interven-tions: antifungal, amphotericin, azoles, triazoles, fluconazole,itraconazole, miconazole, voriconazole, posaconazole, ravuco-nazole, flucytosine, echinocandins, caspofungin, micafungin,anidulafungin, confirmed; and finally a word indicating arandomized trial: random* (wildcard). In addition, wesearched the bibliographies of published systematicreviews and collected papers. We contacted the authors oftrials for study clarifications, where required. Searcheswere not limited by language, sex or age.Study selectionTwo investigators (DP, EM) working independently, induplicate, scanned all abstracts and obtained the full-textreports of records, that indicated or suggested that thestudy was a randomized trial evaluating antifungal ther-apy on the outcomes of interest. After obtaining fullreports of the candidate trials, in full peer-reviewed publi-cation, the same reviewers independently assessed eligi-bility from full text papers.Data collectionThe same 2 reviewers conducted data extraction inde-pendently using a standardized pre-piloted form. Review-ers collected information about the antifungal therapyand type of interventions tested, the population studied(age, setting, underlying conditions), the treatment effecton specified outcomes, adverse events, and specificadverse events addressing renal toxicity and liver impair-ment. Study evaluation included general methodologicalquality features, including allocation concealment,sequence generation, a description of who was blinded,use of intention-to-treat analysis and proportion of studypopulation lost-to-follow-up. We entered the data into anelectronic database such that duplicate entries existed forPage 2 of 11(page number not for citation purposes)focused on single site fungal infections (mucocutaneous,esophageal, dermatologic, meningeal, bladder, or focaleach study; when the two entries did not match, weresolved differences through discussion and consensus.Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23Data analysisIn order to assess inter-rater reliability on inclusion of arti-cles, we calculated the Phi statistic, that provides a meas-ure of inter-observer agreement independent ofchance[8]. Our primary outcome of interest is responserates. Secondary outcomes include all-cause mortality,fungal-attributable mortality, and adverse events.We performed two specific analyses. First, we performed afrequentist fixed-effects meta-analysis of study outcomesacross classes of drugs, applying a Relative Risk [RR] andappropriate 95% Confidence Intervals [CIs] of outcomesaccording to the number of events reported in the originalstudies. In the event of zero outcome events in one arm ofa trial, we used the Haldane method and added 0.5 toeach arm[9]. Given the varied size of studies, we pooledstudies using a fixed effects approach that recognizes theprecision of studies and provides greater to weight tolarger studies[10]. We calculated the I2 statistic for eachanalysis as a measure of the proportion of the overall var-iation that is attributable to between-study heterogene-ity[11]. Given the varied interventions, and theconsideration that most trials were not no-treatment orplacebo controlled, we pooled studies assessing within-class interventions as head-to-head trials. Our first analy-sis examined drugs within-class and then examined indi-vidual drugs using the mixed treatment comparisons. Weinitially pooled all azole interventions versus all ampho-tericin B trials for response and conducted a multivariablemeta-regression using the unrestricted maximum likeli-hood method assessing the impact of individual azoles onthe overall estimate of effect and the individual deliverymethods of amphotericin B on overall estimate[12]. Ourregression covariates were chosen a priori and included:amphotericin delivery and allocation concealment. Anal-yses were conducted using StatsDirect and STATA.For our second analysis, we examined the relative effec-tiveness of each individual drug using the Lu-Ades fixedeffects method for combining direct and indirect evidencein mixed treatment comparisons, a Bayesianapproach[13]. We estimated the posterior densities for allunknown parameters using MCMC (Markov chain MonteCarlo) for each model in WinBUGS version 1.4 (MedicalResearch Council Biostatistics Unit, Cambridge). Eachchain used 100,000 iterations with a burn-in of 500, thinof 5, and updates varying between 80 and 110. We usedthe same seed number (SEED = 314159) for all chains.The choice of burn-in was chosen according to Gelman-Rubin approach[14]. We assessed convergence based ontrace plots and time series plots (available upon request).The accuracy of the posterior estimates was done by calcu-lating the Monte Carlo error for each parameter. As a ruleof thumb, the Monte Carlo error for each parameter ofinterest is less than about 5% of the sample standard devi-ation[15]. All results are reported as posterior means withcorresponding 95% credibility intervals (CrIs). Credibilityintervals are the Bayesian equivalent of classical confi-dence intervals.ResultsOur literature search identified 1284 potentially relevantabstracts of full text articles. Of these, 42 full text RCTswere obtained. We excluded 31, leaving 11 that met ourFlow-diagram of search and included studiesigure 1Page 3 of 11(page number not for citation purposes)Flow-diagram of search and included studies.Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23Page 4 of 11(page number not for citation purposes)Global response rate fixed-effects meta-analysis: Triazoles versus amphotericin BFigure 2Global response rate fixed-effects meta-analysis: Triazoles versus amphotericin B.0.2 0.5 1 2Van't Wout 0.90 (0.49, 1.63)Kullberg 0.99 (0.77, 1.30)Rex 0.69 (0.58, 0.81)Philips 0.85 (0.59, 1.22)Kujath 0.86 (0.52, 1.37)Anaissie 0.97 (0.76, 1.25)Abele-Horn 1.04 (0.75, 1.45)combined [fixed] 0.87 (0.78, 0.96)relative risk (95% confidence interval)All-cause mortality fixed-effects meta-analysis: Triazoles versus amphotericin BFigure 3All-cause mortality fixed-effects meta-analysis: Triazoles versus amphotericin B.0.2 0.5 1 2 5Van't Wout 0.67 (0.23, 1.82)Kullberg 0.85 (0.65, 1.12)Rex 0.82 (0.56, 1.19)Philips 1.06 (0.69, 1.63)Kujath 1.20 (0.45, 3.23)Anaissie 0.89 (0.39, 2.07)Abele-Horn 0.93 (0.51, 1.68)combined [fixed] 0.88 (0.74, 1.05)relative risk (95% confidence interval)Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23inclusion criteria[2,3,16-24], Phi = 0.91, n = 2,554, SeeFigure 1 and Additional File 1. The majority (n = 7) ofstudies assessed the head-to-head non-inferiority of azole-class drugs compared to amphotericin B. The azolesinclude fluconazole[2,3,17,18,23], itraconazole[16], andvoriconazole[19]. The other studies assessed anidulafun-gin to fluconazole[22], micafungin to amphotericinB[20]; caspofungin to amphotericin B[21];and micafun-gin to caspofungin[24]. Trials were predominantly con-ducted in populations dominated by patients withhematologic cancers experiencing infection with Candidaspecies. The median participant age was 57 years (IQR 56–59).In keeping with the time periods that the studies werepublished (1991–2007), reporting of study methodologi-cal features was moderate. Seven of eleven trials reportedintent-to-treat principles; and 6 of 11 provided details onparticipants lost-to-follow-up. An average of 10.8% ofstudy participants were lost-to-follow-up.Meta-analysisGlobal response ratesFor our primary outcome, we pooled 7 trials (n = 965, SeeFigure 2) assessing azoles to amphotericin B. Our pooledestimate is 0.87 (95% CI, 0.78-0.96, P = 0.007, I2 = 43%,P = 0.09). When we compared only fluconazole trials (5trials) to amphotericin B, we found similar effects (RR0.82, 95% CI, 0.74–0.92, P = 0.0009, I2 = 52%, P = 0.07).The itraconazole versus amphotericin B trial (RR 0.90,95% CI, 0.49–1.63, P = 0.61) and voriconazole versusamphotericin B trial (RR 0.99, 95% CI, 0.77–1.30, P =0.94) provided similar estimates.Table 1: Odds ratios and 95% CrIs for mixed treatment comparisons of confirmed infection studies, Response ratesTreatment Comparison Odds Ratio 95% Credible IntervalCaspofungin vs. Fluconazole 2.03 (0.98, 3.76)Amphotericin B Deoxycholate vs. Fluconazole 1.13 (0.78, 1.58)Amphotericin B Liposomal vs. Fluconazole 1.85 (0.65, 4.19)Voriconazole vs. Fluconazole 1.14 (0.62, 1.94)Micafungin vs. Fluconazole 2.13 (0.83, 4.55)Anidulafungin vs. Fluconazole 2.14 (1.19, 3.58)Itraconazole vs. Fluconazole 1.97 (0.32, 6.69)Amphotericin B Deoxycholate vs. Caspofungin 0.60 (0.32, 1.02)Amphotericin B Liposomal vs. Caspofungin 0.91 (0.45, 1.63)Voriconazole vs. Caspofungin 0.61 (0.27, 1.18)Micafungin vs. Caspofungin 1.04 (0.59, 1.70)Anidulafungin vs. Caspofungin 1.18 (0.45, 2.56)Itraconazole vs. Caspofungin 1.05 (0.16, 3.72)Amphotericin B Liposomal vs. Amphotericin B Deoxycholate 1.64 (0.63, 3.52)Voriconazole vs. Amphotericin B Deoxycholate 1.01 (0.63, 1.54)Micafungin vs. Amphotericin B Deoxycholate 1.88 (0.80, 3.79)Anidulafungin vs. Amphotericin B Deoxycholate 1.96 (0.96, 3.58)Itraconazole vs. Amphotericin B Deoxycholate 1.74 (0.30, 5.77)Voriconazole vs. Amphotericin B Liposomal 0.75 (0.25, 1.74)Micafungin vs. Amphotericin B Liposomal 1.18 (0.81, 1.68)Anidulafungin vs. Amphotericin B Liposomal 1.44 (0.42, 3.66)Itraconazole vs. Amphotericin B Liposomal 1.28 (0.16, 4.86)Micafungin vs. Voriconazole 1.95 (0.72, 4.31)Anidulafungin vs. Voriconazole 2.03 (0.85, 4.13)Itraconazole vs. Voriconazole 1.81 (0.28, 6.22)Anidulafungin vs. Micafungin 1.21 (0.38, 2.94)Itraconazole vs. Micafungin 1.08 (0.14, 3.99)Itraconazole vs. Anidulafungin 0.99 (0.14, 3.56)Page 5 of 11(page number not for citation purposes)on sequence generation; 6 of 11 on allocation conceal-ment; 3 of 11 on who was blinded; 9 of 11 reported usingWe also pooled 2 trials of echinocandins (micafungin[20]and caspofungin[21]) versus amphotericin B and found aAnnals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23pooled RR of 1.10 (95% CI, 0.99–1.23, P = 0.08). The ani-dulafungin to fluconazole trial yielded a RR of 1.26 (95%CI, 1.06–1.51, P = 0.001) in favor of anidulafungin[22];and micafungin to caspofungin (RR 1.00, 95% CI, 0.94–1.08, P = 0.21)[24].All-cause mortalityOur secondary outcomes included all-cause mortality. Wepooled 7 trials (n = 965, figure 3) assessing azoles versusamphotericin B for all-cause mortality, resulting in apooled RR of 0.88 (95% CI, 0.74–1.05, P = 0.17, I2 = 0%,P = 0.96). This was also found when individual azoleswere analyzed: fluconazole (5 trials) RR 0.92 (95% CI,0.73–1.17, P = 0.51, I2 = 0%, P = 0.90; itraconazole (1 tri-als) RR 0.67, 95% CI, 0.74–1.05, P = 0.20; voriconazole (1trials) RR 0.85, 95% CI, 0.65–1.12, P = 0.67).When we assessed echinocandins versus amphotericin Bpreparations (2 trials) for all-cause mortality, we found apooled RR of 1.01 (95% CI, 0.84–1.20, P = 0.93).Micafungin versus caspofungin resulted in a RR of 0.85(95% CI, 0.96–1.11) in the direction of favour of caspo-fungin. Anidulafungin versus fluconazole resulted in a RRof 0.73 (95% CI, 0.48–1.10, P = 0.34) in the direction ofanidulafungin.Fungal-attributable mortalityWe also assessed deaths attributable to the fungal infec-tions. When we pooled 5 azole trials versus amphotericinB, we found a pooled RR of 0.84 (95% CI, 0.49–1.42, P =0.51, I2 = 0%, P = 0.74). When we pooled the 3 echinoc-andin trials versus amphotericin B, we found a pooled RRof 1.16 (95% CI, 0.75–1.79, P = 0.50). Anidulafungin ver-sus fluconazole yielded a RR of 0.84 (95% CI, 0.48–1.47,P = 0.88).Adverse eventsTo assess serious adverse events, we pooled 2 trials ofazoles versus amphotericin B assessing serious adverseevents and found a pooled RR of 0.67 (95% CI, 0.55–0.81, P = <0.0001) in favour of azoles. We also pooled 2trials of echinocandins versus amphotericin B and foundNetwork of evidence formed by the eight antifungal treatments compared on the basis of mortality data from 11 studiesFigure 4Network of evidence formed by the eight antifungal treatments compared on the basis of mortality data from 11 studies. Each treatment is a node in the network. The links between nodes are trials or pairs of trial arms. The numbers AGCDBEHFA = FluconazoleB = Caspofungin C = Amphotericin B Deoxycholate D = Amphotericin B LiposomalE = VoriconazoleF = MicafunginG = AnidulafunginH = Itraconazole151111 1Table 2: Absolute treatment efficacy and the probability that each treatment is best in the mixed treatment comparisons analysis using the response data from the confirmed infection studies.Response RatesResponse % Probability bestFluconazole 63.00 0.000Caspofungin 76.10 0.139Amphotericin B Deoxycholate 65.40 0.000Amphotericin B Liposomal 72.98 0.070Voriconazole 65.03 0.004Micafungin 75.98 0.200Anidulafungin 77.49 0.345Itraconazole 69.33 0.241Page 6 of 11(page number not for citation purposes)along the link lines indicate the number of trials or pairs of trial arms for that link in the network.Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23a pooled RR of 0.49 (95% CI, 0.37–0.66, P = <0.0001) infavour of the echinocandins. Micafungin and caspofunginwere not dissimilar in their safety profiles (RR 0.94, 95%CI, 0.70–1.29). We found no significant differencebetween anidulafungin versus fluconazole (RR 0.90, 95%CI, 0.60–1.36, P = 0.66).When we assessed nephrotoxicity defined variably accord-ing to the different studies, we pooled 6 trials of azolescompared to amphotericin B. We found a pooled RR of0.22 (95% CI, 0.15–0.32, P = <0.0001, I2 = 74%, P =0.001) in favour of azoles. We also pooled 3 echinocan-din compared to amphotericin trials and found a pooledRR 0.31 (95% CI, 0.17–0.57).Finally, we assessed hepatic enzyme elevations beyondnormal. We pooled 3 trials assessing azoles compared toamphotericin B and found a pooled RR of 1.08 (95% CI,0.79–1.47, P = 0.64, I2 = 0%, P = 0.63). The 2 echinocan-din versus amphotericin B trials yielded a pooled RR of1.03 (95% CI, 0.17–6.26). The single anidulafungin ver-Mixed treatment comparisonsFigure 4 displays the geometric distribution of the mixedtreatment comparison. Figures 5 and 6 display the cater-pillar plots. Table 1 reports the odds ratios of responserates for all the pairwise comparisons of the antifungaltreatment regimens and table 2 presents estimates of theabsolute efficacy for each treatment, along with the esti-mated probability that each treatment is best (responserate).Table 3 reports the odds ratios of all-cause mortality for allthe pairwise comparisons of the antifungal treatment reg-imens and table 4 presents estimates of the absolute effi-cacy for each treatment, along with the estimatedprobability that each treatment is best (mortality).DiscussionThe results of our systematic review and meta-analysisshould be of interest to clinicians, policy-makers andpatient groups. Our study found similar effects acrosswithin-class interventions. Safety profiles indicate that theCaterpillar plots of the odds ratios and 95% CrIs for mixed treatment comparisons, response ratesFigu e 5Caterpillar plots of the odds ratios and 95% CrIs for mixed treatment comparisons, response rates.       	 			  	 	!						!	  	 	!						!	 	!						!	!						!						!				!	!	!			  	 	!								"	"	Page 7 of 11(page number not for citation purposes)sus fluconazole trial found a RR of 0.21 (95% CI, 0.05–0.83, P = 0.001) in favour of anidulafungin.class of interventions, azoles and echinocandins, offerprotection over amphotericin B in terms of adverse events.Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23There are several important strengths to our meta-analysesthat should be considered when interpreting this study.We used extensive searching of electronic databases toidentify studies. Thus, we identified more studies than anyother systematic reviews[4,5,25]. In order to reduce bias,we conducted our searches independently, in duplicate.We extensively searched the bibliographies of publishedtrials and reviews in order to identify unpublished orobscure papers. Finally, we used methodologicallyadvanced approaches to pool and conducted sensitivityanalyses across a priori defined covariates.There are also several limitations to consider when inter-preting our analysis. Despite our extensive searching, it ispossible that we were unable to identify unpublished tri-als. Indeed, this issue affects every meta-analysis. Weattempted contact through email with 12 study authors toaddress inclusion and methodological questions butreceived responses from only 4; a common occurrencewith systematic reviews [26]. We examined 3 major clini-cal outcomes and their sub-categories: global response,class effects of drugs. These comparisons provide compel-ling comparisons, but only head-to-head trials will pro-vide stronger inferences[27,28].We considered response according to the original papers'definition of response. We considered response accordingto the original papers' definition of response. Notably,there is substantial variability in the timing and defini-tions of response across trials which may limit compara-bility. Timing of assessments ranged from 7 days after startof therapy[2] to up to 12 weeks after the end of ther-apy[19], with multiple variations in between. The criteriafor response also included a wide variety of clinical andmicrobiological response definition, as well as considera-tion of the ability to tolerate randomized ther-apy[2,21,23]. However, we feel our analysis is useful andas we saw no evidence of discrepancies based on our sen-sitivity analysis. More similarity in endpoint definitions infuture trials would be useful to facilitate across-trial com-parisons and evaluation of future therapies.Caterpillar plots of the odds ratios and 95% CrIs for mixed treatment comparisons, all-cause mortalityFigu e 6Caterpillar plots of the odds ratios and 95% CrIs for mixed treatment comparisons, all-cause mortality.      			  	 	!						!	  	 	!						!	 	!						!	!						!						!				!	!	!			  	 	!								"	"	Page 8 of 11(page number not for citation purposes)mortality and adverse events. It is possible that other out-comes would yield differing effects. We conducted mixedtreatment comparisons and demonstrated similar within-There has been an ongoing debate over the quality ofindustry-funded trials of antifungal agents, predomi-nantly in the empiric and prophylaxis trials [29-31]. ThisAnnals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23contention has predominantly concerned clarificationspost-publication, the use of oral amphotericin B com-pared to intravenous, and the reporting of all-cause mor-tality compared to cause-specific mortality. In our analysiswe have aimed to overcome this discussion throughextensive sensitivity analysis, through focusing on con-firmed fungal infections, through evaluating relative effec-tiveness and through presenting both all-cause and cause-specific mortality.A conundrum in evaluating antifungal therapy is assessingfungle attributable mortality. In most trials, the patientsenrolled were at a high risk of mortality due to illness.When we assess all-cause mortality, we recognize thatmany of the patients would have died from their originaldisease and not fungal infections. Clinical practice maydiffer from clinical trial procedures as clinicians maycourses of other therapies. However, we could not displaya difference between these drug classes when we exam-ined deaths attributable to fungal infections.We examined the impact of different dosage forms ofamphotericin B to determine if their results change andfound it did not. A prior systematic review reported atrend towards all-cause mortality benefit and reduction innephrotoxicity risk in lipid-based formulations of ampho-tericin B as compared with conventional amphotericinB,[6] however there was no statistically significant differ-ence in efficacy (clinical response) between the dosageformulations. The reasons for discrepancy between all-cause mortality and efficacy were likely a result of clinicaland methodologic problems. Dosing and duration oftherapy of lipid-based formulations vary widely fromstudy to study, making results difficult to interpret. TrialTable 3: Odds ratios and 95% CrIs for mixed treatment comparisons of confirmed infection studies, all-cause mortalityTreatment Comparison Odds Ratio 95% Credible IntervalCaspofungin vs. Fluconazole 1.34 (0.65, 2.48)Amphotericin B Deoxycholate vs. Fluconazole 1.14 (0.78, 1.62)Amphotericin B Liposomal vs. Fluconazole 1.81 (0.71, 3.90)Voriconazole vs. Fluconazole 0.90 (0.48, 1.53)Micafungin vs. Fluconazole 1.73 (0.74, 3.49)Anidulafungin vs. Fluconazole 0.67 (0.36, 1.14)Itraconazole vs. Fluconazole 0.83 (0.11, 2.94)Amphotericin B Deoxycholate vs. Caspofungin 0.92 (0.50, 1.54)Amphotericin B Liposomal vs. Caspofungin 1.35 (0.77, 2.20)Voriconazole vs. Caspofungin 0.72 (0.33, 1.38)Micafungin vs. Caspofungin 1.28 (0.86, 1.85)Anidulafungin vs. Caspofungin 0.56 (0.21, 1.22)Itraconazole vs. Caspofungin 0.67 (0.08, 2.44)Amphotericin B Liposomal vs. Amphotericin B Deoxycholate 1.58 (0.68, 3.17)Voriconazole vs. Amphotericin B Deoxycholate 0.78 (0.49, 1.19)Micafungin vs. Amphotericin B Deoxycholate 1.51 (0.72, 2.81)Anidulafungin vs. Amphotericin B Deoxycholate 0.60 (0.29, 1.12)Itraconazole vs. Amphotericin B Deoxycholate 0.72 (0.10, 2.50)Voriconazole vs. Amphotericin B Liposomal 0.57 (0.21, 1.26)Micafungin vs. Amphotericin B Liposomal 0.98 (0.67, 1.38)Anidulafungin vs. Amphotericin B Liposomal 0.44 (0.13, 1.08)Itraconazole vs. Amphotericin B Liposomal 0.53 (0.06, 2.02)Micafungin vs. Voriconazole 2.02 (0.82, 4.18)Anidulafungin vs. Voriconazole 0.81 (0.33, 1.68)Itraconazole vs. Voriconazole 0.97 (0.12, 3.46)Anidulafungin vs. Micafungin 0.45 (0.15, 1.05)Itraconazole vs. Micafungin 0.54 (0.06, 2.02)Itraconazole vs. Anidulafungin 1.34 (0.15, 5.05)Page 9 of 11(page number not for citation purposes)reserve more toxic agents (amphotericin B) for salvagetherapy of more severely ill patients who failed shortheterogeneity and small sample size also make it difficultto draw conclusions from comparative studies. An earlierAnnals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/23systematic review comparing amphotericin dosage formsfound no difference in mortality between lipid and con-ventional formulations[4]. To our knowledge, no defini-tive high quality randomized controlled trial comparingamphotericin B dosage formulations has been publishedsince the most recent systematic review.While we did not include pediatric trials in our analysis,results from a recent systematic review assessing antifun-gal therapy in children with invasive fungal infectionsreported congruent conclusions[25]. Although not ameta-analysis, the authors' review (including a supple-mentation with adult studies) led them to conclude thatthere was no difference in clinical response between dif-ferent classes of antifungal agents in the treatment of pro-longed febrile neutropenia (empiric therapy) or invasivecandidal infection. They did report significant differencesin toxicity (particularly nephrotoxicity) between classesthat favoured azoles and echinocandins over amphoter-icin B.In conclusion, our study suggests that azoles and echino-candins are equally effective interventions for treatinginvasive candidiasis and confirms the Infectious DiseaseSociety of America (IDSA) guidelines[7], that recom-mends azoles or echinocandins as the first line treatmentfor Candida infections. Our analysis found similar within-class effects. Amphotericin B offers an effective, but moretoxic alternative.to Pfizer Ltd., Glaxo-Smithkline, Korean HIRA, and Inter-national Society for Clinical Trials. Peter Phillips has beena consultant for Pfizer, Merck Frosst Canada,, and Hoff-mann-La Roche; and is a member of the speakers' bureaufor Pfizer, Merck Frosst Canada, and Schering-PloughPharmaceuticals.Authors' contributionsEM, DP, PW, CC, IT conceived of the study. EM, DP, PW,CC, IT developed the study protocol. EM, DP, PW, CC, ITconducted the searches and data abstraction. EM, DP, PW,CC, IT, JN, PP analyzed and/or interpreted the data. EM,DP, PW, CC, IT, JN, PP wrote the initial drafts of the man-uscript and approved submission. EM, DP, PW, CC, IT,JN, PP revised the manuscript. All authors read andapproved the final manuscript.Additional materialAcknowledgementsThe authors thank Dr. Indra Tumur for important critical reviews. We thank Mr. Chris O'Regan for early comments on the protocol and manu-script.Funding: This study received unrestricted funding from Pfizer Ltd to evalu-ate antifungal agents. They had no role in the conduct, interpretation or writing of this manuscript.References1. Empiric antifungal therapy in febrile granulocytopenicpatients. EORTC International Antimicrobial TherapyCooperative Group.  Am J Med 1989, 86:668-672.2. Phillips P, Shafran S, Garber G, Rotstein C, Smaill F, Fong I, Salit I,Miller M, Williams K, Conly JM, Singer J, Ioannou S: Multicenterrandomized trial of fluconazole versus amphotericin B fortreatment of candidemia in non-neutropenic patients. Cana-dian Candidemia Study Group.  Eur J Clin Microbiol Infect Dis 1997,16:337-345.3. Anaissie EJ, Darouiche RO, Abi-Said D, Uzun O, Mera J, Gentry LO,Williams T, Kontoyiannis DP, Karl CL, Bodey GP: Management ofinvasive candidal infections: results of a prospective, rand-omized, multicenter study of fluconazole versus amphoter-icin B and review of the literature.  Clin Infect Dis 1996,23:964-972.4. Johansen HK, Gotzsche PC: Amphotericin B lipid soluble formu-lations vs amphotericin B in cancer patients with neutrope-nia.  Cochrane Database Syst Rev 2000:CD000969.5. Gafter-Gvili A, Vidal L, Goldberg E, Leibovici L, Paul M: Treatmentof invasive candidal infections: systematic review and meta-analysis.  Mayo Clin Proc 2008, 83:1011-1021.6. Barrett JP, Vardulaki KA, Conlon C, Cooke J, Daza-Ramirez P, EvansAdditional file 1Characteristics of included studies. Table addressing study populations, interventions and fungal species measured.Click here for file[http://www.biomedcentral.com/content/supplementary/1476-0711-8-23-S1.doc]Table 4: Absolute treatment efficacy and the probability that each treatment is best for all-cause mortality in the mixed treatment comparisons in confirmed infection studies.MortalityMortality % Probability bestFluconazole 28.44 0.006Caspofungin 33.83 0.008Amphotericin B Deoxycholate 30.93 0.001Amphotericin B Liposomal 39.99 0.004Voriconazole 25.8 0.090Micafungin 39.16 0.001Anidulafungin 20.75 0.385Itraconazole 21.82 0.504Page 10 of 11(page number not for citation purposes)Competing interestsCurtis Cooper, Ping Wu, Dan Perri, and Imad Tleyjehdeclare no conflict of interest. Edward Mills has consultedEG, Hawkey PM, Herbrecht R, Marks DI, Moraleda JM, Park GR, SennSJ, Viscoli C: A systematic review of the antifungal effective-ness and tolerability of amphotericin B formulations.  ClinTher 2003, 25:1295-1320.Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central Annals of Clinical Microbiology and Antimicrobials 2009, 8:23 http://www.ann-clinmicrob.com/content/8/1/237. Pappas PG, Rex JH, Sobel JD, Filler SG, Dismukes WE, Walsh TJ,Edwards JE: Guidelines for treatment of candidiasis.  Clin InfectDis 2004, 38:161-89.8. Meade MO, Guyatt GH, Cook RJ, Groll R, Kachura JR, Wigg M, CookDJ, Slutsky AS, Stewart TE: Agreement between alternativeclassifications of acute respiratory distress syndrome.  Am JRespir Crit Care Med 2001, 163:490-493.9. Sheehe PR: Combination of log relative risk in retrospectivestudies of disease.  Am J Public Health Nations Health 1966,56:1745-1750.10. Pocock SJ: Safety of drug-eluting stents: demystifying networkmeta-analysis.  Lancet 2007, 370:2099-2100.11. Higgins JP, Thompson SG: Quantifying heterogeneity in a meta-analysis.  Stat Med 2002, 21:1539-1558.12. Thompson SG, Higgins JP: How should meta-regression analysesbe undertaken and interpreted?  Stat Med 2002, 21:1559-1573.13. Lu G, Ades A: A combination of direct and indirect evidencein mixed treatment comparisons.  Stat Med 2004, 23:3105-3124.14. Gelman A, Rubin DB: Inferences from iterative simulationusing multiple sequences.  Stat Sci 1992, 7:.15. Brooks SP, Gelman A: Alternative methods for monitoring con-vergence of iterative simulations.  J Computational and GraphicalStatistics 1998, 7:434-455.16. van't Wout JW, Novakova I, Verhagen CA, Fibbe WE, de Pauw BE,Meer JW van der: The efficacy of itraconazole against systemicfungal infections in neutropenic patients: a randomised com-parative study with amphotericin B.  J Infect 1991, 22:45-52.17. Abele-Horn M, Kopp A, Sternberg U, Ohly A, Dauber A, RusswurmW, Buchinger W, Nagengast O, Emmerling P: A randomized studycomparing fluconazole with amphotericin B/5-flucytosine forthe treatment of systemic Candida infections in intensivecare patients.  Infection 1996, 24:426-432.18. Kujath P, Lerch K, Kochendorfer P, Boos C: Comparative study ofthe efficacy of fluconazole versus amphotericin B/flucytosinein surgical patients with systemic mycoses.  Infection 1993,21:376-382.19. Kullberg BJ, Sobel JD, Ruhnke M, Pappas PG, Viscoli C, Rex JH, ClearyJD, Rubinstein E, Church LW, Brown JM, Schlamm HT, Oborska IT,Hilton F, Hodges MR: Voriconazole versus a regimen of ampho-tericin B followed by fluconazole for candidaemia in non-neutropenic patients: a randomised non-inferiority trial.  Lan-cet 2005, 366:1435-1442.20. Kuse ER, Chetchotisakd P, da Cunha CA, Ruhnke M, Barrios C, Rag-hunadharao D, Sekhon JS, Freire A, Ramasubramanian V, Demeyer I,Nucci M, Leelarasamee A, Jacobs F, Decruyenaere J, Pittet D, UllmannAJ, Ostrosky-Zeichner L, Lortholary O, Koblinger S, Diekmann-Berndt H, Cornely OA: Micafungin versus liposomal amphoter-icin B for candidaemia and invasive candidosis: a phase IIIrandomised double-blind trial.  Lancet 2007, 369:1519-1527.21. Mora-Duarte J, Betts R, Rotstein C, Colombo AL, Thompson-MoyaL, Smietana J, Lupinacci R, Sable C, Kartsonis N, Perfect J: Compar-ison of caspofungin and amphotericin B for invasive candidia-sis.  N Engl J Med 2002, 347:2020-2029.22. Reboli AC, Rotstein C, Pappas PG, Chapman SW, Kett DH, KumarD, Betts R, Wible M, Goldstein BP, Schranz J, Krause DS, Walsh TJ:Anidulafungin versus fluconazole for invasive candidiasis.  NEngl J Med 2007, 356:2472-2482.23. Rex JH, Bennett JE, Sugar AM, Pappas PG, Horst CM van der, EdwardsJE, Washburn RG, Scheld WM, Karchmer AW, Dine AP, et al.: A ran-domized trial comparing fluconazole with amphotericin Bfor the treatment of candidemia in patients without neutro-penia. Candidemia Study Group and the National Institute.N Engl J Med 1994, 331:1325-1330.24. Pappas PG, Rotstein CM, Betts RF, Nucci M, Talwar D, De Waele JJ,Vazquez JA, Dupont BF, Horn DL, Ostrosky-Zeichner L, Reboli AC,Suh B, Digumarti R, Wu C, Kovanda LL, Arnold LJ, Buell DN:Micafungin versus caspofungin for treatment of candidemiaand other forms of invasive candidiasis.  Clin Infect Dis 2007,45:883-893.25. Blyth CC, Palasanthiran P, O'Brien TA: Antifungal therapy in chil-dren with invasive fungal infections: a systematic review.Pediatrics 2007, 119:772-784.26. Mullan RJ, Flynn DN, Carlberg B, Tleyjeh IM, Kamath CC, LaBella ML,Erwin PJ, Guyatt GH, Montori VM: Systematic reviewers com-27. Glenny AM, Altman DG, Song F, Sakarovitch C, Deeks JJ, D'Amico R,Bradburn M, Eastwood AJ: Indirect comparisons of competinginterventions.  Health Technol Assess 2005, 9:1-134.28. Song F, Altman DG, Glenny AM, Deeks JJ: Validity of indirect com-parison for estimating efficacy of competing interventions:empirical evidence from published meta-analyses.  Bmj 2003,326:472.29. Johansen HK, Gøtzsche PC: Problems in the design and report-ing of trials of antifungal agents encountered during meta-analysis.  JAMA 1999, 282:1752-1759.30. Glasmacher A, Prentice A, Gorschlüter M, Engelhart S, Hahn C, Dju-lbegovic B, Schmidt-Wolf IGH: In Reply.  J Clin Oncol 2005,23:9429-9432.31. Gotzsche PC, Johansen HK: Misleading statements in industry-sponsored meta-analysis of itraconazole.  J Clin Oncol 2005,23:9428-9429.yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 11 of 11(page number not for citation purposes)monly contact study authors but do so with limited rigor.Journal of clinical epidemiology 2009, 62:138-142.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items