|
|
| |||||||||||||||||||||||||||||||||||||||
Background Amphotericin B has long been the standard treatment for candidemia, but its use is complicated by its toxicity. More recently, fluconazole, a water-soluble triazole with activity against candida species and little toxicity, has become available. We conducted a multicenter randomized trial that compared amphotericin B with fluconazole as treatment for candidemia.
Methods To be eligible, patients had to have a positive blood culture for candida species, a neutrophil count
Results Of the 237 patients enrolled, 206 met all entry criteria. The most common diagnoses were renal failure, nonhematologic cancer, and gastrointestinal disease. There was no statistically significant difference in outcome: of the 103 patients treated with amphotericin B, 81 (79 percent) were judged to have been treated successfully, as were 72 of the 103 patients treated with fluconazole (70 percent; P = 0.22; 95 percent confidence interval for the difference, -5 to 23 percent). The bloodstream infection failed to clear in 12 patients in the amphotericin group and 15 in the fluconazole group; the species most commonly associated with failure was Candida albicans. There were 41 deaths in the amphotericin group and 34 deaths in the fluconazole group (P = 0.20). Intravascular catheters appeared to be the most frequent source of candidemia. There was less toxicity with fluconazole than with amphotericin B.
Conclusions In patients without neutropenia and without major immunodeficiency, fluconazole and amphotericin B are not significantly different in their effectiveness in treating candidemia.
500 per cubic millimeter, and no major immunodeficiency. Patients were randomly assigned to receive either amphotericin B (0.5 to 0.6 mg per kilogram of body weight per day) or fluconazole (400 mg per day), each continued for at least 14 days after the last positive blood culture. Outcomes were assessed by a group of investigators blinded to treatment assignment.
Methods
Patients
To be enrolled in the trial, patients had to be at least 13 years old and to have had at least one positive blood culture for candida species during the previous four days. In addition, at some time between the first positive blood culture and enrollment the patient had to have had a temperature of at least 38.6 °C on one occasion or a temperature of at least 37.8 °C on two occasions at least four hours apart, a systolic blood pressure of less than 100 mm Hg or a decrease in systolic blood pressure of at least 30 mm Hg from base line, or some sign of inflammation (swelling, heat, erythema, or purulence) at a site infected with candida. Patients were excluded if they had neutrophil counts of less than 500 per cubic millimeter, had a hematologic cancer, had a congenital immunodeficiency syndrome, were known to be seropositive for human immunodeficiency virus type 1, had received organ transplants, had burns on more than 30 percent of the body, had histories of allergy or intolerance to imidazoles or azoles, or had moderate-to-severe liver disease (defined as a level of alkaline phosphatase, aspartate aminotransferase, or alanine aminotransferase greater than five times the upper limit of normal, a total serum bilirubin level higher than 8 mg per deciliter [137 µmol per liter], or a prothrombin time exceeding the upper limit of normal by at least five seconds and not correctable with vitamin K). Patients were also excluded if they were deemed unlikely to survive for at least 24 hours, had received more than 0.6 mg of amphotericin B per kilogram of body weight or 400 mg of fluconazole within the previous 14 days, had already undergone unsuccessful systemic antifungal therapy for the same infection, had previously participated in this trial, or were receiving investigational drugs, approved drugs for investigational indications, immunostimulants, lymphocyte-replacement therapy, or cancer chemotherapy that was likely to cause neutropenia during the study period. Women of childbearing age could not be pregnant or lactating, were required to have negative results on a pregnancy test, and were required to use adequate contraception during the course of the trial. Informed consent was obtained from all patients or their families, and the study was reviewed and approved by the institutional review board of each center.
Study Protocol
After enrollment, the patients were randomly assigned in equal numbers to receive fluconazole or amphotericin B on the basis of a sequential list of block-randomized assignments maintained by the pharmacy at each participating institution. The treatment assignments were not known to investigators or patients before enrollment and randomization, but because of the disparate properties of the two drugs it was not possible to maintain blinding after randomization. Patients were to continue to be given the assigned study drug for at least 14 days after the resolution of clinical findings related to the episode of candidemia or at least 14 days after the last positive deep-site (normally sterile) culture was obtained, whichever was longer. Patients assigned to receive fluconazole were given 400 mg per day in a single dose. Fluconazole was administered intravenously for the first seven days of therapy and then by mouth unless the patient's condition precluded reliable oral therapy. If the patient weighed more than 90 kg or less than 50 kg, the dose of fluconazole was the multiple of 50 that was closest to, but did not exceed, 6 mg per kilogram. The dose of fluconazole was decreased by 50 percent in patients whose estimated creatinine clearance was less than 50 ml per minute and by 75 percent in patients whose estimated clearance was less than 20 ml per minute. Patients assigned to receive amphotericin B were given 0.5 to 0.6 mg per kilogram per day intravenously for the first seven days of therapy. After seven days, the drug could be administered approximately three times per week, as long as the average daily dose was 0.5 to 0.6 mg per kilogram. We recommended that therapy begin with a 1-mg test dose, followed by prompt escalation to the full daily dose over no more than three days. Concomitant use of any combination of acetaminophen, ibuprofen, diphenhydramine, hydrocortisone (up to 50 mg per day), and meperidine was allowed.
Patients were evaluated at the time of entry into the study, daily during therapy, and 2, 6, and 12 weeks after therapy was completed. Two blood cultures were obtained at base line, on days 2, 4, and 7 of therapy, twice weekly during the remainder of the therapy, and again at the two- and six-week follow-up examinations. After enrollment, blood cultures were performed according to the Isolator system (Wampole Laboratories, Cranbury, N.J.). Hemoglobin, hematocrit, leukocyte count, platelet count, prothrombin time, serum creatinine, electrolytes, alkaline phosphatase, total bilirubin, alanine aminotransferase, and aspartate aminotransferase were measured twice weekly during the course of therapy. Either drug could be discontinued if the patient's infection was unresponsive or progressive after five or more days of therapy. Fluconazole was to be stopped if there were marked adverse effects on the liver (defined as a level of alkaline phosphatase 1.5 times the greater of the upper limit of normal or the base-line value, a level of aspartate or alanine aminotransferase 3 times the greater of the upper limit of normal or the base-line value, or a total serum bilirubin level 3 times the greater of the upper limit of normal or the base-line value). If the serum creatinine level rose above 3.5 mg per deciliter (309 µmol per liter) in a patient treated with amphotericin, the investigator was to alter the timing and amount of amphotericin B therapy. Modification of the dosage could include either temporary or permanent cessation of amphotericin B therapy. In the event either drug was discontinued, treatment was scored as a failure only if additional systemic antifungal therapy was subsequently administered.
After the conclusion of therapy and follow-up, case-report forms were purged of information that might identify the study drugs by two of the investigators and then scored in a blinded fashion by four other investigators. Treatment was considered successful if all signs and symptoms of the candida infection resolved, blood cultures obtained at the end of therapy and at the follow-up visits were negative, and cultures from normally sterile sites previously positive for candida became negative and any associated signs of local inflammation resolved. Treatment was considered to have failed if there was unresponsive or progressive infection after five or more days of therapy or if unacceptable side effects necessitated a change in antifungal therapy. Treatment was also considered to have failed if a patient withdrew from the study before improvement or failure had been clearly documented. Cases considered to be successes at the end of study therapy were reclassified as relapses if there was a positive blood culture for candida, a visceral candida infection was discovered, or more than four days of antifungal therapy was given for an asymptomatic candida urinary tract infection or any amount of antifungal therapy was given for any other cause. Finally, the local investigators identified three cases in which there was strong evidence that the patients had induced their episodes of candidemia by manipulating intravenous catheters. The blinded reviewers scored these patients on the basis of their responses to the initial course of therapy and ignored later events. The blinded reviewers also excluded data on these patients from the analysis of therapy lasting five or more days but retained them in the primary and intention-to-treat analyses.
Statistical Analysis
The primary analysis in this trial was specified in advance as a comparison of the response rates (successes vs. failures and relapses) for the patients meeting all entry criteria. The null hypothesis was that the difference between the proportions of patients with favorable responses in the two groups would be less than 20 percent. To test this hypothesis with a power of 80 percent and a significance level of 5 percent, 90 patients were needed in each group for the primary analysis. Unless otherwise specified, all analyses pertain to this sample. Two secondary analyses were specified in advance: an intention-to-treat analysis of all patients who enrolled in the trial and had candidemia, whether or not they met all entry criteria, and an analysis of the outcome in patients who received at least five days of therapy. Outcome analyses were performed as two-sided Cochran-Mantel-Haenszel tests. Other post hoc tests were performed as indicated. Data are reported as means ±SE and sometimes include the 80 percent range that spans the 10th to 90th percentiles.
Results
Enrollment and Base-Line Characteristics of the Patients
Between December 1989 and April 1993, 237 patients were enrolled at 24 centers (Table 1). The patients in the two study groups in the primary analysis had similar characteristics at base line (Table 2). Acute Physiology and Chronic Health Evaluation (APACHE II) scores12 were computed at study entry.
|
|
Candida albicans was the most common infecting species, and 59 percent of the patients were infected with this species alone (Table 3). The first positive blood culture was obtained 2.6 ±0.1 days before study entry for patients in both treatment groups.
|
Vascular catheters were judged to be the most probable source of candidemia in 149 patients (72 percent) (Table 4).
|
Patients in the amphotericin B group were treated for 17 ±1 days (80 percent range, 6 to 30) and received 0.5 ±0.01 mg of drug per kilogram per day (80 percent range, 0.3 to 0.6), for a total dose of 570 ±40 mg (80 percent range, 140 to 1080). Patients in the fluconazole group were treated for 18 ±1 days (80 percent range, 6 to 34), and received an average of 5.3 ±0.2 mg of drug per kilogram per day (80 percent range, 2.7 to 7.8) during this time. The number of days of therapy did not differ significantly in the two study groups (P = 0.56 by Student's t-test).
Toxicity
Elevations in blood urea nitrogen or serum creatinine levels were judged by the blinded review committee to have occurred in 37 percent of the patients treated with amphotericin B and 2 percent of those treated with fluconazole (P<0.001 by Fisher's exact test). Hypokalemia developed in 10 percent of the patients in the amphotericin B group and 2 percent of those in the fluconazole group (P = 0.006). Liver enzymes became elevated in 10 percent of the amphotericin B group and 14 percent of the fluconazole group (P = 0.43).
Outcome
There was no statistically significant difference in outcome between the two study groups in the primary analysis: 81 of the patients treated with amphotericin B (79 percent) had successful outcomes, as did 72 of the patients treated with fluconazole (70 percent) (P = 0.22; 95 percent confidence interval for the difference, -5 to 23 percent). Similarly, in the intention-to-treat analysis 89 of 111 patients treated with amphotericin B (80 percent) and 81 of 113 patients treated with fluconazole (72 percent) had successful outcomes (P = 0.17; 95 percent confidence interval for the difference, -3 to 20 percent). Finally, in the secondary analysis of patients treated for at least five days, 81 of 94 patients given amphotericin B (86 percent) and 70 of 93 patients given fluconazole (75 percent) had successful outcomes (P = 0.05; 95 percent confidence interval for the difference, 0 to 18 percent).
Analysis of Failures and Relapses
The bloodstream infection failed to clear in 12 patients treated with amphotericin B and 15 treated with fluconazole. The organism most commonly associated with failure in both groups was C. albicans (Table 5). The patients treated with amphotericin B received 0.5 ±0.04 mg of drug per kilogram per day (80 percent range, 0.25 to 0.65), for a total of 4.5 ±0.9 mg per kilogram (80 percent range, 0.5 to 9.6) given over a period of 8.7 ±1.5 days (80 percent range, 2 to 17) before the patient died or treatment failure was declared. Eight of these patients died, seven as a direct result of the episode of candidemia (see below). The 15 patients treated with fluconazole received 5.5 ±0.5 mg per kilogram per day (80 percent range, 3.1 to 8.4) for 8.3 ±1.4 days (80 percent range, 4 to 17). Nine of these patients died, five as a direct result of the episode of candidemia. The mean duration of therapy before the declaration of failure did not differ between treatment groups (P = 0.9). There was also no statistically significant difference in the rate of failure to clear any species, although the power to detect a difference was limited for species other than C. albicans.
|
The bloodstream infection cleared in four patients receiving fluconazole, but their therapy was altered because fever either persisted or recurred late in the course of treatment. In three of these patients, no site was positive for candida species; C. albicans was found in a sputum culture from the fourth. In one patient treated with fluconazole a blood culture drawn on day 4 yielded Saccharomyces cerevisiae on day 11. Since the patient was also still febrile, therapy was changed to amphotericin B. The fever persisted for an additional two weeks. Two patients treated with fluconazole had persistent peritoneal candida infections. Finally, four patients given amphotericin B and three given fluconazole withdrew from the study after four days or less of therapy. The reviewers judged them impossible to evaluate, and they were scored as having had treatment failures.
In one patient given amphotericin B, a perinephric abscess due to candida developed after a 47-day course of therapy, and one patient given fluconazole presented with a candida abscess at the site of a recent glioma resection after a 16-day course of therapy. Two patients in the amphotericin B group had unexplained hypotension early in the follow-up period, and systemic antifungal therapy was reinstituted. Neither patient had positive cultures for candida during the episode. Finally, one patient in the amphotericin B group and three in the fluconazole group had candiduria during follow-up without fever or leukocytosis and were treated with more than four days of systemic antifungal agents. Although the initial courses of therapy were successful in these patients, the use of systemic antifungal agents during follow-up required that their treatments be scored as failures. In a post hoc analysis, these four technical failures were rescored as successes. Under these scoring rules, there was no difference in outcome in the primary analysis: 82 patients treated with amphotericin B (80 percent) had successful outcomes, as did 75 patients treated with fluconazole (73 percent) (P = 0.40; 95 percent confidence interval for the difference, -9 to 23 percent). Similarly, there were no statistically significant differences in the intention-to-treat analysis (success rates of 81 percent and 74 percent, respectively; P = 0.31; 95 percent confidence interval for the difference, -6 to 20 percent) or the five-or-more-days-of-therapy analysis (success rates of 87 percent and 78 percent; P = 0.14; 95 percent confidence interval for the difference, -3 to 21 percent).
Follow-up and Deaths
The patients given amphotericin B were followed for 61 ±5 days after the end of therapy, and those given fluconazole were followed for 65 ±5 days (P = 0.6). During this time, 41 of 103 patients in the amphotericin B group (40 percent) and 34 of 103 in the fluconazole group (33 percent) died (P = 0.20), with no statistically significant difference in the length of time until death (P = 0.20) or in the number of deaths in the first seven days of therapy (10 in the amphotericin B group and 5 in the fluconazole group, P = 0.28). If a patient had a positive blood culture within 48 hours of death or had evidence of visceral candidiasis at autopsy, the episode of candidemia was judged to be directly contributory to death. There were eight such deaths in the amphotericin B group and seven in the fluconazole group. The results of the intention-to-treat and five-or-more-days-of-therapy analyses were also not statistically significant (data not shown).
The predetermined scoring rules made it possible for treatment to be scored as a success although the patient died on or shortly after the last day of therapy. A post hoc analysis was performed in which all the patients who died within 14 days after therapy ended were scored as having had treatment failures. Under these rules, 68 of 103 patients treated with amphotericin B (66 percent) had successful outcomes, as did 64 of 103 patients treated with fluconazole (62 percent) (P = 0.66; 95 percent confidence interval for the difference, -14 to 22 percent). Results were similar for the intention-to-treat analysis, the five-or-more-days-of-therapy analysis, and an analysis in which death within seven days after the end of therapy was considered to indicate treatment failure (data not shown).
Management of Intravascular Catheters
Complete catheter records were available for 91 percent of both study groups. Sixty-two patients in each group underwent complete catheter exchanges (a complete catheter exchange was defined as the removal and replacement of all catheters on a single day) on or before the first day of treatment with the study drug. The exchange was performed over a guidewire placed through a preexisting catheter in 10 patients in the amphotericin B group and 7 in the fluconazole group.
Endophthalmitis
Ninety-four percent of the patients underwent ocular examinations before the end of therapy. Thirteen in the amphotericin B group and 16 in the fluconazole group had retinal findings consistent with candida endophthalmitis13. In one patient in the fluconazole group with persistent fungemia, a lesion persisted and enlarged despite subsequent amphotericin B therapy. Vitrectomy was ultimately required for cure. In no other patient did a lesion cause problems, and no patient had candida endophthalmitis after the end of therapy; of the 127 patients who reached the 2-week or 6-week follow-up point, 108 (85 percent) had either normal ophthalmologic results or no ocular symptoms at 12 weeks. No such follow-up information was available for the remaining 19 patients.
Discussion
We studied a large series of patients with candidemia. Beyond a slight disparity in sex, the two study groups were comparable with respect to age and risk factors for candidemia. It is noteworthy that the mean APACHE II scores were identical (16 ±1).
There were no statistically significant differences between amphotericin B and fluconazole as treatments for candidemia in the primary and intention-to-treat analyses. Amphotericin B had a marginally significant higher success rate than fluconazole in a secondary analysis of patients treated for at least five days, but this analysis is potentially biased because of early withdrawal from this unblinded trial. Analyzed in detail (Table 5), the differences between the two groups were not due to persistent candidemia (12 percent in the amphotericin B group and 15 percent in the fluconazole group), toxicity (3 percent and 2 percent), or deaths (40 percent and 33 percent), but to therapeutic modifications late in the course of therapy or after the end of therapy. Particularly troubling were patients who received systemic antifungal therapy for asymptomatic funguria during follow-up. These patients represented treatment failures according to our predefined scoring system, but this determination obscures the patients' initial responses to therapy and otherwise benign follow-up courses. Another aspect of our scoring system was that it did not require survival beyond the end of therapy. When, in a post hoc analysis, success was redefined to require survival for 7 or 14 days after therapy ended, the success rates for the two drugs became essentially identical. The most frequent site of secondary spread appears to have been the eye, with retinal lesions in 29 patients. Other than one late abscess in each study group, no other hematogenous sequelae of candidemia were noted.
Seventy-five of the 206 patients in our primary analysis died, for an overall mortality rate of 36 percent. Although the lack of comparable demographic data in other published studies of patients with candidemia makes comparisons difficult, this rate is below the 46 to 75 percent reported for patients with primarily nononcologic illnesses2,3,4,5 and the 63 to 72 percent reported for studies that included primarily patients with cancer2,6,7.
This study had several limitations. First, the properties of the study drugs made blinding impossible. Although records were reviewed in a blinded fashion by an oversight committee, local study investigators were not blinded as regards treatment drugs, and decisions to change antifungal regimens would cause the patients' treatments to be scored as failures under all circumstances. Second, we were unable to control the management of intravascular catheters as part of the trial. Fortunately, however, analysis of catheter management demonstrated that a similar proportion of catheters was exchanged in each study group. Finally, it should be emphasized that our data apply only to patients without neutropenia and with relatively little immunosuppression.
In conclusion, fluconazole and amphotericin B do not differ significantly as therapy for candidemia in patients without neutropenia who are not severely immunocompromised. The most common source of candidemia in our study appeared to be intravascular catheters, and the investigators at all our centers attempted to change all intravascular catheters that were in place at the time of candidemia. Our study was not large enough to determine whether patients infected with species other than C. albicans were less likely to respond to either drug, but no striking discrepancies were apparent. Too few late complications were seen to permit meaningful comparison. We do not know whether timely catheter exchanges or systemic antifungal therapy accounts for the low incidence of late complications, but we believe both to be essential components of management.
Supported by grants from Roerig-Pfizer and in part by a contract (N01-AI-15082) to the Mycoses Study Group from the National Institute of Allergy and Infectious Diseases.
Source Information
From the University of Texas Medical School at Houston, Houston (J.H.R.); Laboratory of Clinical Investigation, National Institute of Allergy and Infectious Diseases, Bethesda, Md. (J.E.B.); Boston University Medical Center Hospital, Boston (A.M.S.); University of Alabama, Birmingham (P.G.P.); University of North Carolina, Chapel Hill (C.M.V.); Harbor-UCLA Medical Center, Torrance, Calif. (J.E.E.); Bowman Gray School of Medicine, Winston-Salem, N.C. (R.G.W.); University of Virginia, Charlottesville (W.M.S.); New England Deaconess Hospital, Boston (A.W.K.); and Roerig-Pfizer, New York (A.P.D., M.J.L., C.D.W.). Additional contributing authors include Harold C. Neu, M.D., John J. Stern, M.D., Carmelita U. Tuazon, M.D., Robert H. Rubin, M.D., Bryan P. Simmons, M.D., Jack D. Sobel, M.D., Michael F. Parry, M.D., James M. Horton, M.D., Issa E. Ephthimios, M.D., Allan R. Tunkel, M.D., Ph.D., Ronald A. Greenfield, M.D., Winkler G. Weinberg, M.D., David S. McKinsey, M.D., and Corstiaan Brass, M.D.Presented in part at the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, October 18-20, 1993 (abstract no. 805) and at the 12th Congress of the International Society for Human and Animal Mycology, Adelaide, Australia, March 13-18, 1994 (abstract no. S11.1).
Address reprint requests to Dr. Rex at the University of Texas Medical School, 6431 Fannin 1728 JFB, Houston, TX 77030.
References
The following investigators also participated in this study: Boston University Medical Center Hospital, Boston: C. Saunders; University of Alabama, Birmingham: L. Riser and M.E. Bradley; University of North Carolina, Chapel Hill: J. Serody and B. Longmire; Harbor-UCLA Medical Center, Torrance, Calif.: S. Filler; Bowman Gray School of Medicine, Winston-Salem, N.C.: J. Sinclair; University of Virginia, Charlottesville: C. Sable and J. White; New England Deaconess Hospital, Boston: M. Golden and E. Levin; Columbia University, New York: P. Doerney; Pennsylvania Hospital, Philadelphia: M. Buckley, M.N. Braffman, and N. Pietroski; George Washington University Medical Center, Washington, D.C.: S. McMullen; Massachusetts General Hospital, Boston: A. Delvecchio; Methodist Hospital of Memphis, Memphis, Tenn.: M. Gelfand and B. Craft; Wayne State University, Detroit: L. Flatt; Stamford Hospital, Stamford, Conn.: D. Adler-Klein; Carolinas Medical Center, Charlotte, N.C.: J. Connell; University of Iowa College of Medicine, Iowa City: B.N. Doebbeling, R.P. Wenzel, and C. Leick; Medical College of Pennsylvania, Philadelphia: C.C. Johnson and P. Walsh; University of Oklahoma Health Sciences Center, Oklahoma City: D. Filippo; Infectious Diseases Research Consortium of Georgia, Atlanta: D. Blum; Research Medical Center, Kansas City, Mo.: B. Lee; Buffalo Medical Group, Buffalo, N.Y.: J. Fenner; Middleton Memorial Veterans Affairs Medical Center, Madison, Wis.: J.M. Jones; National Institute of Allergy and Infectious Diseases Mycoses Study Group, University of Alabama, Birmingham: W.E. Dismukes, J.Y. Lee, and G.A. Cloud; and Roerig-Pfizer, New York: S. Elkin, P. Eisman, A. Feldman, J. Foulkes, S. Morrison, and C. Petrucelli.
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
The Treatment of Candidemia
DiNubile M. J., Girmenia C., Martino P., Cofsky R. D., Tang C.M., Howe P., Crook D.W., Rex J. H., Bennett J. E., Sugar A. M., Meunier F.
Extract |
Full Text
N Engl J Med 1995;
332:1100-1102, Apr 20, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
