Articles. Vol 366 October 22,

Transcription

1 versus a regimen of amphotericin B followed by fluconazole for candidaemia in non-neutropenic patients: a randomised non-inferiority trial B J Kullberg, J D Sobel, M Ruhnke, P G Pappas, C Viscoli, J H Rex, J D Cleary, E Rubinstein, L W P Church, J M Brown, H T Schlamm, I T Oborska, F Hilton, M R Hodges Summary Background has proven efficacy against invasive aspergillosis and oesophageal candidiasis. This multicentre, randomised, non-inferiority study compared voriconazole with a regimen of amphotericin B followed by fluconazole for the treatment of candidaemia in non-neutropenic patients. Methods Non-neutropenic patients with a positive blood culture for a species of candida and clinical evidence of infection were enrolled. Patients were randomly assigned, in a 2:1 ratio, either voriconazole (n=283) or amphotericin B followed by fluconazole (n=139). The primary efficacy analysis was based on clinical and mycological response 12 weeks after the end of treatment, assessed by an independent data-review committee unaware of treatment assignment. Findings Of 422 patients randomised, 37 were included in the modified intention-to-treat population. was non-inferior to amphotericin B/fluconazole in the primary efficacy analysis, with successful outcomes in 41% of patients in both treatment groups (95% CI for difference 1 6% to 1 6%). At the last evaluable assessment, outcome was successful in 162 (65%) patients assigned voriconazole and 87 (71%) assigned amphotericin B/ fluconazole (p= 25). cleared blood cultures as quickly as amphotericin B/fluconazole (median time to negative blood culture, 2 days). Treatment discontinuations due to all-cause adverse events were more frequent in the voriconazole group, although most discontinuations were due to non-drug-related events and there were significantly fewer serious adverse events and cases of renal toxicity than in the amphotericin B/fluconazole group. Interpretation was as effective as the regimen of amphotericin B followed by fluconazole in the treatment of candidaemia in non-neutropenic patients, and with fewer toxic effects. Relevance to practice There are several options for treatment of candidaemia in non-neutropenic patients, including amphotericin B, fluconazole, voriconazole, and echinocandins. can be given both as initial intravenous treatment and as an oral stepdown agent. Introduction The choice of the best first-line treatment for candidaemia remains controversial. Clinical studies have shown that amphotericin B, fluconazole, and caspofungin have similar efficacy in the treatment of candida bloodstream infections. 1 6 Whereas fluconazole may be selected on the basis of its safety, the increasing proportions of patients infected with candida strains resistant to fluconazole (eg, Candida krusei or fluconazole-resistant strains of other species including C glabrata) draw attention to the need for initial treatment with a broader-spectrum agent. 6,7 In accordance with recent guidelines, many experts favour initial treatment with amphotericin B in severely ill or clinically unstable patients. 6 In light of the concerns about azole resistance, initial treatment with a short course of amphotericin B, followed by fluconazole once the isolate is known to be likely susceptible, has become a common clinical approach. Thus, the current choice of treatment for candidaemia is guided by weighing the risk of fluconazole resistance and the greater activity of amphotericin B for these species against the substantial toxic effects and risk of renal failure associated with amphotericin B treatment 8 or the lack of an oral dosing option for caspofungin. 6, an azole antifungal available as intravenous and oral formulations, has proven efficacy in the treatment of invasive aspergillosis and other mould infections The drug has potent activity against candida species in vitro, including many isolates with reduced fluconazole susceptibility, and is as effective as fluconazole in the treatment of oesophageal candidiasis in immunocompromised patients. 16 has also been studied in patients with candidaemia refractory to previous antifungal treatment. 17,18 This study was undertaken to assess the efficacy and safety of primary treatment with intravenous or oral voriconazole compared with amphotericin B followed by fluconazole in nonneutropenic patients with candidaemia. Lancet 25; 366: Published online October 12, 25 DOI:1.116/S (5) See Comment page 1413 Nijmegen University Centre for Infectious Diseases, Department of Internal Medicine (484), Radboud University Nijmegen Medical Centre (541), PO Box 911, 65 HB Nijmegen, Netherlands (Prof B J Kullberg MD); Wayne State University School of Medicine, Detroit, MI, USA (J D Sobel MD); Charité, Humboldt University, Berlin, Germany ( M Ruhnke MD); University of Alabama, Birmingham, AL, USA (P G Pappas MD); University of Genova and National Institute for Cancer Research, Genova, Italy (Prof C Viscoli MD); University of Texas Medical School, Houston, TX, USA (J H Rex MD); University of Mississippi School of Pharmacy, Jackson, MS, USA (J D Cleary PharmD); Tel Aviv University, Tel-Hashomer, Israel (Prof E Rubinstein MD); Medical University of South Carolina, Charleston, SC, USA (L W P Church MD); Chris Hani Baragwanath Hospital, Johannesburg, South Africa (J M Brown MD); Pfizer Global Research and Development, New York, NY, USA (H T Schlamm MD); and Pfizer Global Research and Development, Sandwich, UK (I T Oborska PhD, F Hilton MSc, M R Hodges MD) Correspondence to: Prof B J Kullberg b.kullberg@aig.umcn.nl Vol 366 October 22,

2 Methods Patients Patients were aged 12 years or older, had one or more positive blood culture for candida species obtained up to 96 h before study entry, and had clinical findings consistent with infection in the 48 h before study entry defined as: temperature of more than 37 8 C on two occasions at least 4 h apart, or of more than 38 6 C once; systolic blood pressure less than 9 mm Hg, or a decrease in systolic blood pressure of at least 3 mm Hg from the baseline value; or signs of inflammation at a site infected with candida species. 19 Patients were excluded if they had received more than 2 days of systemic antifungal therapy during the 96 h before the study, had failed previous antifungal treatment for the candida infection under study, were unlikely to survive longer than 24 h, were receiving any drug known to interact appreciably with azole antifungals, were pregnant, or had an allergy to azoles or amphotericin B. Patients were also excluded if they had a neutrophil count of less than /L, AIDS, aplastic anaemia, chronic granulomatous disease, moderate or severe liver disease (alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, or total bilirubin more than five times the upper limit of normal), or severe renal impairment (serum creatinine 22 mol/l). This study was approved by the ethics review boards of the institutions in which it was undertaken, and written informed consent was obtained from each patient or legal representative before study entry. Procedures Patients were randomly assigned by a central, computerised randomisation system in a 2:1 ratio to receive treatment with either voriconazole or amphotericin B followed by fluconazole (amphotericin B/fluconazole). The randomisation procedure was stratified by administrative region (region 1, Americas and India; region 2, Europe and southeast Asia). The treatment assignments were not known to investigators or patients before enrolment and randomisation, but because of the disparate properties of the two drugs blinding could not be maintained after randomisation. was given intravenously at 6 mg/kg every 12 h for 24 h, and then at 3 mg/kg every 12 h. After 3 days, patients could be switched to oral voriconazole 2 mg twice daily. Amphotericin B was given at 7 1 mg/kg per day infused over 2 6 h. The protocol stated that amphotericin B should be replaced by oral or intravenous fluconazole at 4 mg/day after a minimum of 3 days and a maximum of 7 days with the following exceptions: patients infected with an isolate shown (or predicted on the basis of species) to have a fluconazole minimum inhibitory concentration of 16 mg/l or more were to remain on amphotericin B; patients infected with C lusitaniae were allowed to switch from amphotericin B to fluconazole earlier than day 3; and patients unable to tolerate 3 days of amphotericin B despite a reduction in the infusion rate and use of any combination of paracetamol, ibuprofen, diphenhydramine, hydrocortisone, and pethidine could be switched to fluconazole, even before day 4. Patients received treatment for at least 2 weeks after the last positive blood culture, with a maximum duration of 8 weeks. Patients were followed up until 12 weeks after the end of treatment. To allow for an average of 2 weeks of treatment, all-cause mortality was assessed at 14 weeks after enrolment. Clinical assessments were undertaken daily during treatment, at the end of treatment, and at 2 weeks, 6 weeks, and 12 weeks after the end of treatment. Mycological investigations and clinical laboratory tests were done twice weekly during treatment, at the end of treatment, and at 2 weeks, 6 weeks, and 12 weeks after the end of treatment. Blood culture samples were taken at baseline, on days 1, 2, 3, 4, and 7, twice weekly during antifungal treatment thereafter, at 2 weeks after the end of treatment, and, if clinically indicated, at additional time points. At each time point, two sets of blood cultures (total 2 ml) were taken for culture by BacTec high-blood volume fungal media (BD Diagnostics, Sparks, MD, USA), BacT-Alert (Bio Merieux, Hazelwood, MO, USA), the lysis centrifugation method or equivalent methods. Samples were incubated for at least 7 days, or until positive. Response to treatment was assessed by an independent data-review committee unaware of the study medication and of adverse events that could reveal treatment assignment. Efficacy analyses were prespecified and were based on the modified intention-to-treat population of all patients with a blood culture positive for Candida species within 96 h of enrolment who received at least one dose of study drug. The primary efficacy endpoint was response to treatment, defined as mycological eradication and clinical cure or improvement, as assessed by the datareview committee at the 12-week follow-up visit. In this analysis, only patients who reached follow-up at 12 weeks after the end of treatment could be counted as successes. An additional efficacy analysis comprised patients who were assessed by the data-review committee as successfully treated at their final study visit. In this analysis, patients who had a successful outcome at the end of treatment but who did not reach the assessment 12 weeks after the end of treatment could be counted as a success. This analysis parallels the primary analysis undertaken in previous randomised studies of candidaemia. 1,2,5,19 Additional endpoints were time to first negative blood culture and time from randomisation to death. Statistical analysis The primary objective was to show that voriconazole was non-inferior to amphotericin B/fluconazole in the Vol 366 October 22, 25

3 11 not treated 24 no positive blood culture for candida within 96 h of randomisation Figure 1: Trial profile 4 withdrawn before randomisation 1 given study treatment but not randomised 283 assigned voriconazole 272 received study drug (safety population) 248 in modified intentionto-treat population 427 patients included 422 randomised 8 not treated 9 no positive blood culture for candida within 96 h of randomisation 139 assigned amphotericin B/ fluconazole 131 received study drug (safety population) 122 in modified intentionto-treat population modified intention-to-treat population. On the basis of an estimated success rate of 65% at the last evaluable follow-up assessment for both treatment groups, a power of 8, and the requirement to show that voriconazole is non-inferior to the comparator group with a of 15%, a total sample size of 36 patients (24 assigned voriconazole and 12 assigned amphotericin B/ fluconazole) satisfying the criteria for the modified intention-to-treat population were calculated to be needed. Taking into account an estimated ineligibility rate before the start of the study of 15%, a total of 426 patients were needed. would be judged non-inferior to amphotericin B/fluconazole if the lower limit of the two-sided 95% CI for the difference in response rates (stratified by administrative recruitment region) between voriconazole and amphotericin B/ fluconazole was above 15%. Time to first negative blood culture and time from randomisation to death were analysed by Kaplan-Meier methods. Role of the funding source The study was sponsored by Pfizer Global Research and Development, and the sponsor was involved in the study design, data collection, and statistical analysis, in accordance with the guidelines of the US Food and Drug Administration and good clinical practice. The independent data-review committee (BJK, JHR, JS, MR, PGP, CV) was responsible for the data analysis, data interpretation, and writing of the manuscript. Neither Amphotericin B/ (n=248) fluconazole (n=122) Mean age, years (range) 53 6 (13 9) 53 3 (13 87) Male 145 (59%) 71 (58%) Mean APACHE II score (SD; range) 13 8 (6 5; 41) 14 7 (6 6; 2 36) Predisposing factors Abdominal surgery 95 (38%) 46 (38%) Non-abdominal surgery 32 (13%) 15 (12%) Non-surgical 121 (49%) 61 (5%) In intensive care unit 119 (48%) 61 (5%) Mechanically ventilated 89 (36%) 47 (39%) Candidaemia only 239 (96%) 117 (96%) Candidaemia and other sites of 9 (4%) 5 (4%) invasive candidiasis Pathogen* C albicans 17 (43%) 63 (51%) Non-albicans candida species 15 (61%) 61 (5%) C tropicalis 53 (21%) 16 (13%) C parapsilosis 45 (18%) 19 (16%) C glabrata 36 (15%) 21 (17%) C krusei 4 (2%) 1 (1%) Other candida species 15 (6%) 5 (4%) 2 candida species 12 (5%) 3 (3%) Data are number (%) unless otherwise stated. *Some patients were infected with more than one candida species. Table 1: Baseline characteristics of patients in modified intention-totreat analysis the sponsor nor the corporate authors had a role in the data-review committee procedures or decisions, the final interpretation of the results, or the wording of the report. All authors had full access to all the data in the study, and the corresponding author held final responsibility for the decision to submit the paper for publication. Results Between September, 1998, and January, 23, more than 4 patients from 27 countries were included in the study (figure 1). The characteristics of the patients in the two groups were similar except that patients assigned amphotericin B/fluconazole had a slightly higher mean Amphotericin B/ p (n=248) fluconazole (n=122) Primary success rate* 11 (41%) 5 (41%) 96 Success by pathogen C albicans 46/17 (43%) 3/63 (48%) C glabrata 12/36 (33%) 7/21 (33%) C parapsilosis 24/45 (53%) 1/19 (53%) C tropicalis 17/53 (32%) 1/16 (6%) 32 C krusei 1/4 (25%) /1 Secondary success rate 162 (65%) 87 (71%) 25 Success rate at end of treatment 173 (7%) 9 (74%) 42 Success rate 2 weeks after end of treatment 13 (52%) 64 (53%) 99 Success rate 6 weeks after end of treatment 11 (44%) 56 (46%) 78 All-cause 14-week mortality 88 (36%) 51 (42%) 23 *Sustained successes as assessed by data-review committee at 12-week follow-up visit only. Successes assessed by data-review committee at latest available study visit (including end of therapy, 2 weeks, or 6 weeks after end of treatment if 12-week assessment after end of treatment was not available). Successes assessed by data-review committee. Table 2: Favourable responses to treatment and all-cause mortality Vol 366 October 22,

4 1 8 Probability of positive culture Numbers at risk Positive Negative Censored Time (days) Figure 2: Kaplan-Meier plot of time to first negative blood culture for modified intention-to-treat population with positive blood cultures on day APACHE II score at study entry (table 1). Sites of deep tissue candidiasis complicating candidaemia were documented in only 4% (14/37) of patients. At baseline, around 6% of patients in the voriconazole group and about a half of those assigned amphotericin B/fluconazole were infected with a 1 8 Probability of survival Numbers at risk Alive Time (weeks) Dead Censored Figure 3: Kaplan-Meier plot of time to death from randomisation for modified intention-to-treat population Vol 366 October 22, 25

5 non-albicans candida species. There was a similar distribution of the various non-albicans candida species in the two treatment groups (table 1). The median duration of treatment was 15 days in both groups (range 1 57 days for voriconazole; 1 66 days for amphotericin B/fluconazole). The median duration of amphotericin B treatment was 4 days. All intravenous catheters present at the time of study enrolment were removed by day 4 in 86% of patients in the voriconazole group and 88% of those in the amphotericin B/ fluconazole group. In the primary efficacy analysis, 11 of 248 (41%) patients assigned voriconazole reached the 12-week follow-up after the end of treatment and were judged as a success compared with 5 of 122 (41%) patients assigned amphotericin B/fluconazole (95% CI for difference when stratified by geographic region: 1 6 to 1 6; table 2). Because the lower limit of the two-sided 95% CI was above 15%, voriconazole was deemed non-inferior to amphotericin B/fluconazole. Similar efficacy was noted for both treatment groups for infections caused by all candida species apart from C tropicalis, for which a higher success rate was reported with voriconazole than with amphotericin B/fluconazole (p= 32). Efficacy assessed by the data-review committee at the last evaluable follow-up was 65 5% in the voriconazole group and 71 3% in the amphotericin B/fluconazole treatment group (stratified difference 5 6%; 95% CI 15 8 to 4 2). Efficacy at earlier study visits was also similar in both groups (table 2). Kaplan-Meier plots showed that voriconazole cleared blood cultures as quickly as amphotericin B/fluconazole (median time to negative blood culture was 2 days in both groups; figure 2). Likewise, mortality was similar in the two groups: 36% of patients in the voriconazole group and 42% of patients in the amphotericin B/ fluconazole group died within the 14 weeks after enrolment (figure 3). The hazard ratio for dying in the voriconazole group was 82 (95% CI 58 to 1 16). All 43 patients who took one or more doses of study medication were monitored for adverse events (table 3). The overall rates of serious adverse events were higher in the amphotericin B/fluconazole group than in the voriconazole group, as were the numbers of adverse events judged to be treatment related (51% vs 36%, p= 3). The frequency of hepatic events and rash was similar between the two groups. Renal adverse events were more frequent in the amphotericin B/fluconazole group than in the voriconazole group, even though amphotericin B was given for only a short period (median 4 days). Serum creatinine concentrations from baseline doubled in 17% of patients on amphotericin B/ fluconazole compared with 12% in the voriconazole group, and serum creatinine concentrations remained persistently raised in the amphotericin B/fluconazole group, despite the short exposure to amphotericin B (figure 4). Of patients receiving amphotericin B for Amphotericin B/ p (n=272) fluconazole (n=131) All-cause adverse events Patients with serious adverse events 125 (46%) 74 (57%) 48 Hepatic events* 63 (23%) 32 (24%) 78 Sepsis 57 (21%) 33 (25%) 34 Renal events 22 (8%) 28 (21%) 2 Fever 41 (15%) 24 (18%) 41 Chills 8 (3%) 1 (8%) 3 Vomiting 24 (9%) 17 (13%) 2 Rash 16 (6%) 7 (5%) 83 Visual events 11 (4%) 1 (1%) 7 Reasons for discontinuation from study drug Patient died 56 (21%) 35 (27%) 168 Insufficient clinical response 8 (3%) 3 (2%) 77 Adverse events (all-causes) 4 (15%) 9 (7%) 24 Related to study drug 17 (6%) 6 (5%) 5 Not related to study drug 23 (9%) 3 (2%) 18 Laboratory abnormality (all-causes) 13 (5%) 5 (4%) 661 Related to study drug 8 (3%) 2 (2%) 393 Not related to study drug 5 (2%) 3 (2%) 761 Lost to follow-up 7 (3%) 3 (2%) 864 Other reasons 26 (1%) 17 (13%) 298 Total discontinued 15 (55%) 72 (55%) 972 Patients with dose reduced or temporarily 11 (4%) 18 (14%) 4 discontinued owing to adverse events *Increase in concentration of alkaline phosphatase, aspartate aminotransferase, alanine aminotransferase, or bilirubin. Increase in creatinine concentration, acute kidney failure. As judged by local investigator. Switches from amphotericin B to fluconazole or dose reductions because of adverse events or laboratory abnormalities were not reported as discontinuations from study or study failures. Including protocol or entrance criteria violation and withdrawn consent. Table 3: All-cause adverse events and reasons for discontinuation from study drug in patients who received at least one dose of study drug longer than 7 days, a renal adverse effect was reported in 76% (13/17). For the 21 patients infected with C glabrata, the median duration of amphotericin B treatment was 7 days (vs 4 days in other patients), and an adverse renal effect was reported in more than half these patients. Visual adverse events were more frequent in the voriconazole group than in the amphotericin B/ Median change in creatinine ( mol/l) Number of patients Number of samples Time since randomisation (weeks) Figure 4: Median change in serum creatinine concentration from randomisation Vol 366 October 22,

6 fluconazole group (table 3). Although the number of treatment discontinuations due to all-cause adverse events was higher in patients receiving voriconazole (table 3), discontinuations in those assigned amphotericin B are under-reported since patients could be switched to fluconazole without being classified as treatment failures. Specifically, an additional 13 patients (1%) were switched prematurely from amphotericin B to fluconazole owing to toxic effects, as was allowed per protocol. Discussion In this comparative, randomised, multicentre study, voriconazole was shown to be non-inferior to a regimen of amphotericin B followed by fluconazole in the treatment of candidaemia in non-neutropenic patients. Additionally, clearance of candida from the bloodstream with voriconazole was as rapid as with amphotericin B. There are several strategies for treatment of candidaemia, although all have some limitations. Strategies based on initial treatment with amphotericin B are favoured by many experts, especially in severely ill patients, because of the risk that the infecting candida strain will have reduced susceptibility to fluconazole. 4,6 The shift in epidemiology toward non-albicans candida species observed in many centres has made the possibility of antifungal resistance a major concern when selecting initial treatment. 7,2 However, amphotericin B is associated with an increased risk of infusion-related reactions and renal toxic effects. 8 The present study compared the efficacy and safety of voriconazole with a strategy commonly used in clinical practice, consisting of initial treatment with amphotericin B followed by fluconazole to complete a therapeutic course. Although the regimen of amphotericin B followed by fluconazole has not been specifically investigated in previous trials, randomised studies have confirmed that either drug alone has equal activity against candidaemia 1,2 and that combined treatment with amphotericin B and fluconazole does not lead to antagonism. 19 In accordance with treatment guidelines, 6 amphotericin B is favoured by many in severely ill or clinically unstable patients, and it has become common practice to switch to fluconazole if the candida isolate is found to be susceptible. Anticipating an incidence of non-albicans candida species of at least 5%, 7,2 this comparator arm was selected to maximise the spectrum of activity until the candida species and susceptibility were known, while specifically attempting to keep to a minimum the risk of amphotericin B toxic effects reported in previous studies 1,2,5 by limiting the duration of exposure to this drug. We chose as a primary endpoint the success rate that was sustained at the 12-week follow-up visit after the end of treatment. This endpoint differs from those used in earlier randomised studies in its requirement that a patient reach the 12-week follow-up visit all patients not reaching this endpoint for any reason were deemed treatment failures. Although this endpoint leads to reporting of low success rates, it was chosen to capture any late occurrences of persistence, relapse, adverse effects, or excess mortality of either treatment group. Thus, the response rates for this study s primary endpoint (41%) were lower than those in earlier randomised studies of candidaemia, in which success All-cause p Success rate at p Success rate at end p Success rate at p Success rate at p mortalitylast available of studydrug 6 8 weeks after 12 weeks after follow-up visit administration end of treatment end of treatment Candidaemia Amphotericin B ( 5 6 mg/kg per day) 4%* 79%** 83% Fluconazole (4 mg per day) n=26 33% 7% 74% Canadian Candidaemia Study Amphotericin B ( 6 mg/kg per day) 34% 62%** Fluconazole (4 mg per day) n=16 38% 57% Candidaemia Amphotericin B ( 7 mg/kg per day) 4%* 69%** fluconazole (8 mg per day) Fluconazole (8 mg per day) n=219 39% 56% Candidaemia 3 (present study) Amphotericin B ( 7 1 mg/kg per day) 42%* 71% 74% 46% 41%** Fluconazole (4 mg per day) (intravenously 3 mg/kg 36% 65% 7% 44% 41% twice daily, oral 2 mg twice daily) n=427 Caspofungin Study Amphotericin B ( 6 1 mg/kg per day) 3% 62%** 41% Caspofungin (5 mg per day ) n=224 34% 73% 43% Itraconazole Study Fluconazole (4 mg per day) 4%* 69% 41%** Itraconazole (2 mg per day ) n=1934% 67% 34% *9 98-day mortality day mortality. Loading dose, 6 mg/kg bid. Loading dose, 7 mg. Intravenous only; loading dose 2 mg twice daily on days 1 2. Follow-up was up to 12 weeks after end of study drug administration. **Reported as primary study endpoint. Sustained successes as assessed at 12-week follow-up visit only. Table 4: Randomised trials for primary treatment of candidaemia in non-neutropenic patients and effect of efficacy endpoint definitions, by antifungal agents studied Vol 366 October 22, 25

7 rates varied from 56% to 79% (table 4). 1,5,19 The only other study to use a similarly strict primary endpoint of sustained success at the 12-week follow-up after end of treatment had a similar success rate of 35 41%. 21 The overall mortality in the current study (36 42%) was similar to that reported previously (3 4%), 1,5,19 suggesting that the effect of antifungal treatment in all these studies was similar, despite the apparent differences in the proportions of patients with protocoldefined successful responses to treatment. The secondary analysis in the present study, in which the response was assessed by the blinded data-review committee at the latest available study visit, provides an analysis similar to that of previous randomised studies with fluconazole and amphotericin B (table 4). 1,19 This analysis resulted in successful response rates of 65 71% in the two study groups, which are similar to those reported from those earlier clinical trials (56 79%). 1,19 Furthermore, the response for the voriconazole group at the end of treatment (7%) was similar to that in a recent trial of amphotericin B versus caspofungin (62% and 73%, respectively), which used investigator-assessed success at the end of treatment as the primary endpoint. 5 This study has several limitations. The properties of the study drugs made blinding impossible because the infusion-related effects of amphotericin B, its effects on laboratory measurements, and need for co-medication would have revealed the treatment allocation. However, response to treatment, eligibility, protocol violations, and reason for death were assessed by an independent data-review committee who were unaware of the study medication and of laboratory results or adverse events that could have revealed the treatment allocation. There was an inherent bias in this open-label study, in that patients prematurely discontinuing voriconazole owing to toxic effects (irrespective of causal relation to study drug) were regarded as treatment failures, whereas those discontinuing amphotericin B because of toxic effects could still be classified as successes, since a switch to fluconazole was allowed in the study protocol. Thus, 13 patients (1%) needed an early switch from amphotericin B to fluconazole owing to toxic effects, including ten patients who were ultimately judged to be successes. This is an important difference between this study and earlier candidaemia trials that used amphotericin B alone as a comparator group, 1,2,5 for in these earlier studies patients discontinuing amphotericin B because of toxic effects were classified as treatment failures. As a result, more patients on voriconazole failed treatment before the end because of adverse events, especially those that were not drug-related, than did patients receiving amphotericin B followed by fluconazole. The proportion of patients with renal toxic effects was higher in the amphotericin B/fluconazole group, indicating that patients on amphotericin B were still at substantial risk of nephrotoxicity, even though this drug was given for a median duration of only 4 days. In patients on amphotericin B/fluconazole, the mean serum creatinine concentrations remained raised for at least 5 weeks, despite early discontinuation of amphotericin B in most patients. More voriconazole-group patients reported abnormal visual events, but the proportion affected in this study was substantially lower than reported previously (45%). 11 Because the patients were seriously ill, they might not have experienced or reported visual events; almost 4% were sedated and receiving mechanical ventilation. There were many infections due to non-albicans candida species (55%). Consequently, the results from this study are a useful indicator of the optimum treatment strategy for these species, some of which show reduced susceptibility to fluconazole. For C albicans, C glabrata, and C parapsilosis, successful response rates were similar between the voriconazole and amphotericin B/ fluconazole groups, but for C tropicalis, the proportion of patients responding successfully was substantially higher in the voriconazole group. There is no apparent explanation for the poor outcome of C tropicalis infections treated with amphotericin B/fluconazole in this study, because the strains were generally susceptible to the study drugs given (voriconazole MIC9 (minimum inhibitory concentration for 9% of organisms], 25 mg/l; amphotericin B MIC9, 1mg/L), and the risk factors, including APACHE II score, were well balanced. In light of the small sample size in the amphotericin B/ fluconazole group, this result does not seem robust and thus is of uncertain relevance. rendered blood cultures negative as quickly as amphotericin B followed by fluconazole. In both groups, the median time to negative blood culture was 2 days, which fell well within the initial amphotericin B phase in the amphotericin B/fluconazole group. This finding suggests that voriconazole, which is thought to be fungistatic against candida, is as effective as amphotericin B, which has fungicidal activity in vitro. Table 4 summarises the randomised trials of primary treatment of candidaemia in non-neutropenic patients. Although the primary endpoints reported differ, the outcomes at comparable treatment endpoints are very similar, confirming that there are now seven regimens supported by randomised trials: amphotericin B, fluconazole, amphotericin B plus fluconazole, amphotericin B followed by fluconazole, voriconazole, caspofungin, and itraconazole. Non-inferiority could not be statistically confirmed only for itraconazole. 21 Whereas the use of fluconazole as primary treatment could have become undesirable in view of the increasing incidence of potentially resistant species, amphotericin B has a poorer safety profile. Thus, caspofungin or voriconazole are currently the most useful options for primary treatment of candidaemia in non-neutropenic patients, whereas fluconazole remains the choice in selected patients in settings with a low rate of resistant species. Vol 366 October 22,

8 In summary, the results of this large, multicentre study show that voriconazole is as effective as the commonly used strategy of amphotericin B followed by fluconazole in non-neutropenic patients for the treatment of candidaemia, including C albicans and non-albicans candida species. The broader spectrum of voriconazole, its better safety than amphotericin B, and the availability of intravenous and oral formulations, provide an important new treatment option for candidaemia. Contributors J H Rex, H T Schlamm, I T Oborska, and B J Kullberg designed the study; J D Cleary, E Rubinstein, L W P Church, J M Brown, H T Schlamm, I T Oborska, and M R Hodges collected the data; F Hilton did the statistical analysis; B J Kullberg, J D Sobel, M Ruhnke, P G Pappas, C Viscoli, and J H Rex analysed the data and were the writing committee; B J Kullberg chaired the data-review committee and wrote the draft report. Local investigators Argentina A O Gallesio, R Lamberghini. Australia S C A Chen, G J Dobb. Austria E Presterl. Belgium M Aoun, F Jacobs, D M H Spapen, P Damas, P Biston. Brazil M Nucci, F de Queiroz Telles, A S S Levin, L Z Goldani, L C Severo. Canada S Shafran, R Pelletier, P Phillips, E J Bow. Czech Republic P Sevcik. France Y Cohen, J-P Dupeyron, P Balvay, R Grelier. Germany M Ruhnke, P Kujath, G Just-Nübling. Hong Kong C Gomersall, K K Young. India C B Sridhar, S C Varma, H Kaul, J Philip, G C Rajadhyaksha, G L Avasthi, V N Kamat, V K Kapoor, M K Daga, V Shantaram. Israel E Rubinstein, R Lang, M Shapiro. Malaysia L H Tan, A B Kamarulzaman. Mexico S Ponce deleon Rosales, A L Halpert. Morocco A Sbihi, O Kerkeb. Netherlands B J Kullberg. Panama M Suarez. Philippines J Chua, M Mendoza, T Tupasi. Poland D Maciejewski, A Dyaczynska-Herman. Slovak Republic L Drgona. South Africa J M Brown, R I Raine, G E Naude, J G Kilian, J P Pretorius. South Korea J-H Song, J-M Kim, W-S Shin, J-H Woo. Spain K Aguirrebengoa, J de la Torre Cisneros, F Rodriguez Villanova, G Seller Peres, F Alvarez Lerma, M Palomar Martinez. Switzerland T F Calandra, J Garbino, W Zimmerli. Taiwan Y-C Chen. Thailand P Mootsikapun, K Malathum. UK N R Webster. USA J Kruse, M G Seneff, R P Gonzalez, H V Dedhia, J Rosman, W G Gardner, D Stuart, L M Sloan, M G Schuster, L W P Church, S D O Marro, P Kamitsuka, C L V Anderson, L M Brumble, S W Chapman, A E Bacon III, S G Filler, J Vazquez, M J Zervos, A W Karchmer, D H Kett, D S McKinsey, J E Lutz, T C Samo, K O Cleveland, P P Cook, P G Pappas. Conflict of interest statement B J Kullberg, J D Sobel, M Ruhnke, P G Pappas, C Viscoli, and J H Rex have served as consultants for Pfizer. L W P Church has received speakers fees from Pfizer. H T Schlamm, I T Oborska, F Hilton, and M R Hodges are employees of Pfizer. The other authors declare that they have no conflict of interest. Acknowledgments This study was presented in part at the 14th European Congress of Clinical Microbiology and Infectious Diseases, Prague, Czech Republic, May 1 4, 24. The study was sponsored by Pfizer Global Research and Development. References 1Rex JH, Bennett JE, Sugar AM, et al. A randomized trial comparing fluconazole with amphotericin B for the treatment of candidemia in patients without neutropenia. N Engl J Med 1994; 331: Phillips P, Shafran S, Garber G,et al. Multicenter randomized trial of fluconazole versus amphotericin B for treatment of candidemia in non-neutropenic patients. Eur J Clin Microbiol Infect Dis 1997; 16: Viscoli C, Girmenia C, Marinus A, et al. Candidemia in cancer patients: a prospective, multicenter surveillance study in Europe by the Invasive Fungal Infection Group (IFIG) of the European Organization for Research and Treatment of Cancer (EORTC). Clin Infect Dis 1999; 28: Edwards JE Jr, Bodey GP, Bowden RA, et al. International conference for the development of a consensus on the management and prevention of severe candidal infections. Clin Infect Dis 1997; 25: Mora-Duarte J, Betts R, Rotstein C, et al. Comparison of caspofungin and amphotericin B for invasive candidiasis. N Engl J Med 22; 347: Pappas PG, Rex JH, Sobel JD, et al. Guidelines for treatment of candidiasis. Clin Infect Dis 24; 38: Pfaller MA, Diekema DJ, Jones RN, et al. International surveillance of bloodstream infections due to Candida species: frequency of occurrence and in vitro susceptibilities to fluconazole, ravuconazole, and voriconazole of isolates collected from 1997 through 1999 in the SENTRY antimicrobial surveillance program. J Clin Microbiol 21; 39: Bates DW, Su L, Yu DT, et al. Mortality and costs of acute renal failure associated with amphotericin B therapy. Clin Infect Dis 21; 32: Denning DW, Ribaud P, Milpied N, et al. Efficacy and safety of voriconazole in the treatment of acute invasive aspergillosis. Clin Infect Dis 22; 34: Walsh TJ, Lutsar I, Driscoll T, et al. in the treatment of aspergillosis, scedosporiosis and other invasive fungal infections in children. Pediatr Infect Dis J 22; 21: Herbrecht R, Denning DW, Patterson TF, et al. versus amphotericin B for primary therapy of invasive aspergillosis. N Engl J Med 22; 347: Ruhnke M, Schmidt-Westhausen A, Trautmann M. In vitro activities of voriconazole (UK-19,496) against fluconazolesusceptible and -resistant Candida albicans isolates from oral cavities of patients with human immunodeficiency virus infection. Antimicrob Agents Chemother 1997; 41: Cuenca-Estrella M, Diaz-Guerra TM, Mellado E, Monzon A, Rodriguez-Tudela JL. Comparative in vitro activity of voriconazole and itraconazole against fluconazole-susceptible and fluconazoleresistant clinical isolates of Candida species from Spain. Eur J Clin Microbiol Infect Dis 1999; 18: Ostrosky-Zeichner L, Rex JH, Pappas PG, et al. Antifungal susceptibility survey of 2 bloodstream Candida isolates in the United States. Antimicrob Agents Chemother 23; 47: Pfaller MA, Diekema DJ, Messer SA, Boyken L, Hollis RJ, Jones RN. In vitro activities of voriconazole, posaconazole, and four licensed systemic antifungal agents against Candida species infrequently isolated from blood. J Clin Microbiol 23; 41: Ally R, Schurmann D, Kreisel W, et al. A randomized, double-blind, double-dummy, multicenter trial of voriconazole and fluconazole in the treatment of esophageal candidiasis in immunocompromised patients. Clin Infect Dis 21; 33: Ostrosky-Zeichner L, Oude Lashof AML, Kullberg BJ, Rex JH. salvage treatment of invasive candidiasis. Eur J Clin Microbiol Infect Dis 23; 22: Perfect JR, Marr KA, Walsh TJ, et al. treatment for less-common, emerging, or refractory fungal infections. Clin Infect Dis 23; 36: Rex JH, Pappas PG, Karchmer AW, et al. A randomized and blinded multicenter trial of high-dose fluconazole plus placebo versus fluconazole plus amphotericin B as therapy for candidemia and its consequences in nonneutropenic subjects. Clin Infect Dis 23; 36: Pfaller MA, Jones RN, Doern GV, et al. International surveillance of blood stream infections due to Candida species in the European SENTRY Program: species distribution and antifungal susceptibility including the investigational triazole and echinocandin agents. Diagn Microbiol Infect Dis 1999; 35: Tuil O, Cohen Y. An open comparative multicenter study of intravenous (IV) itraconazole versus IV fluconazole in the treatment of candidemia in non-neutropenic patients. Crit Care 23; 7 (suppl 2): S Vol 366 October 22, 25