Comparison of Caspofungin and Amphotericin B for Invasive Candidiasis

Selection of Patients

Patients were eligible for enrollment in the study if they were over the age of 18 years and had had one or more positive candida cultures from blood or another, sterile site within the previous four days. Patients with positive cultures of urine specimens, sputum specimens, bronchoalveolar-lavage specimens, oropharyngeal or esophageal specimens, or samples from indwelling drains were excluded. An additional criterion for enrollment was at least one of the following clinical signs of infection during the previous two days: fever (a temperature that exceeded 101°F [38.3°C] or two readings that exceeded 100°F [37.8°C]), clinically significant hypothermia (a temperature of less than 96.8°F [36.0°C]), hypotension (systolic blood pressure, <90 mm Hg or a decrease ≥30 mm Hg), or signs of inflammation at a candida-infected site. Patients with suspected endocarditis, osteomyelitis, or meningitis were excluded. Patients who had received antifungal therapy for more than two days (cumulative dose of amphotericin B, >2 mg per kilogram of body weight; lipid amphotericin B, >10 mg per kilogram; or fluconazole, >1600 mg) were also excluded. Patients receiving rifampin, ritonavir, or cyclosporine were also not enrolled.

The study protocol was approved by the institutional review board of each participating institution, and written informed consent was obtained from all patients before enrollment.

Study Design

Patients were stratified according to the presence or absence of neutropenia and the score on the Acute Physiology and Chronic Health Evaluation (APACHE II) (≤20 or >20). They were randomly assigned to receive either intravenous caspofungin or intravenous amphotericin B according to a schedule maintained by each participating institution's pharmacist. The schedules were generated by computer to ensure equivalent randomization at each site. Patients and investigators were unaware of the treatment assignments. Patients who were assigned to receive caspofungin were given a 70-mg loading dose, followed by 50 mg per day. Patients who were assigned to receive amphotericin B and who did not have neutropenia were given 0.6 to 0.7 mg per kilogram per day; those with neutropenia received 0.7 to 1.0 mg per kilogram per day. A double-dummy technique was used to maintain the blinding. The daily treatment regimen consisted of infusion of caspofungin or matching placebo (saline) for one hour, immediately followed by infusion of amphotericin B or matching placebo (saline with a multivitamin complex) for two or more hours. An increase (or reduction) in the dose was not permitted.

Patients were to receive antifungal therapy for 14 days after the most recent positive candida culture. A minimum of 10 days of intravenous therapy was required. After 10 days, intravenous therapy was continued or oral fluconazole was substituted (400 mg per day). Fluconazole was given only to patients who did not have neutropenia, whose clinical condition had improved, whose follow-up cultures had been negative for 48 hours, and whose candida isolates were susceptible to fluconazole. Patients with Candida krusei or C. glabrata infection continued to receive intravenous therapy.

Symptoms or signs of candida infection (including the most abnormal temperature) were documented daily during the treatment regimen and two weeks and six to eight weeks after treatment. Physical examination and laboratory tests were performed twice a week during treatment and at both follow-up visits. For patients with candidemia, two samples for blood cultures were obtained daily until the results had been negative for at least 48 hours. For nonblood infections, follow-up cultures were also routinely obtained; however, in certain patients, the infection was assumed to have been eradicated and follow-up cultures were not required if there was no longer any clinical or radiographic evidence of infection. Retinal examinations for candida endophthalmitis were performed by ophthalmologists before enrollment, at the end of intravenous therapy, and at the final follow-up visit.

Evaluation of Efficacy

Efficacy was assessed in terms of the overall response to treatment. A favorable overall response was defined as the resolution of all symptoms and signs of candida infection and culture-confirmed eradication (or presumptive eradication for certain nonblood infections). The outcome was considered to be unfavorable if the infection was clinically or microbiologically unresponsive, if the study drug was withdrawn before there was documented improvement, or if toxic effects necessitated a change in antifungal therapy. Evaluations were performed on day 10 of intravenous therapy, at the end of intravenous therapy, at the end of all antifungal therapy (intravenous therapy and oral fluconazole), and at both follow-up visits. The primary time point for the determination of efficacy was the end of intravenous therapy. During the six-to-eight-week period after treatment, a patient was considered to have a relapse if an invasive candida infection had recurred or if antifungal therapy for a proven or suspected candida infection was again administered.

Statistical Analysis

The study was primarily designed to determine whether caspofungin was as effective as amphotericin B for the treatment of invasive candidiasis, with efficacy measured in terms of the overall response at the end of intravenous therapy. The observed outcome was adjusted for the two stratification variables (the presence or absence of neutropenia and the APACHE II score) according to the Cochran–Mantel–Haenszel method. The noninferiority of caspofungin would be demonstrated if the two-sided 95.6 percent confidence interval for the difference in efficacy between the two treatment groups (the response to treatment in the caspofungin group minus the response in the amphotericin B group), adjusted for neutropenic status and the APACHE II score, included 0 and the lower boundary was not lower than –20.0 percent. The superiority of caspofungin would be demonstrated if the confidence interval was entirely above 0. Assuming a response rate of 70 percent in the amphotericin B group (α=0.05, β=0.1, and δ=0.20), we calculated that 110 patients per treatment group were needed to test for noninferiority. A noninferiority analysis was also performed for patients with candidemia; approximately 85 such patients per treatment group were required (β=0.2). Adjustments for multiple comparisons were not performed.

The two prespecified study populations for the analysis of efficacy were the patients included in the modified intention-to-treat analysis and the population of patients who met prespecified criteria for evaluation. The modified intention-to-treat analysis (the primary analysis) included patients who had a documented diagnosis of invasive candidiasis and who received the study treatment for at least one day. The prespecified criteria for evaluation were inclusion in the modified intention-to-treat analysis and no concomitant antifungal therapy, no protocol violations that might interfere with the assessment of efficacy, an appropriate evaluation at the end of treatment, and receipt of the study treatment for at least five days.

The study was also designed to compare certain end points in an analysis of safety, including nephrotoxicity. A nephrotoxic effect was defined as at least a doubling of the serum creatinine level, or an increase of at least 1.0 mg per deciliter (88.4 mmol per liter) if the base-line level was elevated. Other predefined end points in the analysis of safety included drug-related adverse events, discontinuation of treatment due to drug-related adverse events, infusion-related toxic effects, and hypokalemia requiring potassium supplementation. All patients who received the assigned study treatment were included in the safety analysis. An independent data safety monitoring board monitored both safety and efficacy during the study. The authors who are not affiliated with Merck had access to all the study data, take responsibility for the accuracy of the analysis, and had authority over the preparation of the manuscript and the decisions about publication.

Base-Line Characteristics of the Patients

Table 1. Table 1. Numbers of Patients Enrolled, Included in the Modified Intention-to-Treat Analysis, and Included in the Analysis of Patients Who Met Prespecified Criteria for Evaluation.

Table 2. Table 2. Base-Line Characteristics of the 224 Patients Included in the Modified Intention-to-Treat Analysis.

A total of 239 patients were enrolled in the study over a period of 44 months, 224 of whom were included in the modified intention-to-treat analysis (Table 1). The base-line characteristics of the patients were similar in the two treatment groups (Table 2). Because of the stratification method, the groups did not differ significantly with respect to neutropenic status (P=0.32) or the mean APACHE II score (P=0.46). The majority of the patients had candidemia, but peritonitis and intraabdominal abscesses were not uncommon. Approximately 60 percent of the patients had received prior antifungal therapy, but only for a day or less in most cases.

Candida Isolates

Table 3. Table 3. Base-Line Candida Isolates.

The most common candida isolate was C. albicans (Table 3), which accounted for 35.6 percent of infections in the caspofungin group and 54.1 percent of those in the amphotericin B group (P=0.009). C. parapsilosis was isolated in 19.8 percent of the patients in the caspofungin group and 18.3 percent of those in the amphotericin B group, C. tropicalis in 19.8 percent and 12.8 percent, C. glabrata in 12.8 percent and 9.2 percent, and C. krusei in 4.0 percent and 0.9 percent, respectively. Five patients had infections with both C. albicans and other candida species.

Duration of Treatment

The duration of treatment was similar in the two groups (P=0.60). Patients in the caspofungin group were treated for a mean of 12.1 days (median, 11.0; range, 1 to 28), and those in the amphotericin B group were treated for a mean of 11.7 days (median, 10.0; range, 1 to 28). A switch to oral fluconazole (after day 10) occurred in the cases of 27 caspofungin-treated patients (24.8 percent) and 40 amphotericin B–treated patients (34.8 percent).

Efficacy

Table 4. Table 4. Favorable Responses to Treatment.

In the modified intention-to-treat analysis, the proportion of patients with a favorable response at the end of intravenous therapy was 73.4 percent in the caspofungin group and 61.7 percent in the amphotericin B group (Table 4); after adjustment for neutropenic status and the APACHE II score, the difference in the proportion of patients with a favorable response was 12.7 percentage points (95.6 percent confidence interval, –0.7 to 26.0; P=0.09). In the analysis of patients who met the prespecified criteria for evaluation, 80.7 percent of the caspofungin-treated patients and 64.9 percent of the amphotericin B–treated patients had successful outcomes at the end of intravenous therapy. The difference between the treatment groups for this analysis was 15.4 percentage points (95.6 percent confidence interval, 1.1 to 29.7; P=0.03). The outcomes were consistent among the stratified subgroups. In both treatment groups, the response rate was lower among patients with indicators of a poor prognosis (neutropenia or an APACHE II score higher than 20) than among patients without these indicators, but there was still a trend in favor of caspofungin.

The outcomes stratified according to the candida pathogen were generally similar in the two groups. For C. albicans, the rate of a favorable response was 63.9 percent in the caspofungin group and 57.6 percent in the amphotericin B group. The response rate was higher among patients with non-albicans infections in both the caspofungin group (80.0 percent) and the amphotericin B group (68.0 percent). The responses were similar for the most common non-albicans species — namely, C. parapsilosis (70.0 percent and 65.0 percent), C. tropicalis (85.0 percent and 71.4 percent), and C. glabrata (76.9 percent and 80.0 percent).

At each of the four other time points (day 10, the end of antifungal therapy, two weeks after treatment, and six to eight weeks after treatment), the percentage of patients with successful outcomes was higher in the caspofungin group than in the amphotericin B group (Table 4).

Persistent or Recurrent Infection

Table 5. Table 5. Treatment Failures and Relapses (Modified Intention-to-Treat Analysis).

Similar proportions of patients in the two treatment groups had persistently positive cultures, persistent signs or symptoms, or new metastatic lesions or withdrew from the study after four or fewer days (Table 5). A larger proportion of patients in the amphotericin B group had toxic effects requiring a change in therapy (P=0.03).

The proportion of patients with a relapse was similar in the two treatment groups (Table 5). Only five patients had a relapse of candidemia (three in the caspofungin group and two in the amphotericin B group). In all five patients, the organism cultured at relapse and the base-line isolate were identical with respect to the species and the minimal inhibitory concentration.

Site of Infection

The study was specifically designed to compare the efficacy of caspofungin and that of amphotericin B for the treatment of candidemia. In the modified intention-to-treat analysis, the proportion of patients with candidemia who had a favorable outcome at the end of intravenous therapy was 71.7 percent in the caspofungin group and 62.8 percent in the amphotericin B group. The difference, adjusted for neutropenic status and the APACHE II score, was 10.0 percentage points (95.0 percent confidence interval, –4.5 to 24.5; P=0.22). In the analysis of patients with candidemia at base line who met the prespecified criteria for evaluation, 80.3 percent of the caspofungin-treated patients and 64.6 percent of the amphotericin B–treated patients had a successful outcome at the end of intravenous therapy. In this analysis, the difference was 15.2 percentage points (95.0 percent confidence interval, –0.6 to 31.0; P=0.06).

A larger proportion of patients in the caspofungin group than in the amphotericin B group had multiple positive blood cultures at base line (58.7 percent vs. 48.9 percent). However, the percentages of patients with blood cultures that were still positive on days 4 and 7 of intravenous therapy did not differ significantly between the two groups (day 4: 19.6 percent in the caspofungin group and 19.1 percent in the amphotericin B group; day 7: 12.0 percent and 8.5 percent, respectively).

The outcomes were also examined in patients without candidemia. Among those with peritonitis, the response rate was 100 percent (eight of eight patients) in the caspofungin group and 87.5 percent (seven of eight) in the amphotericin B group; among those with intraabdominal abscesses, the response rate was 75.0 percent (three of four) and 33.3 percent (three of nine), respectively. Of the nine patients with multiple sites of infection, four of the five patients treated with caspofungin (80.0 percent) and all four treated with amphotericin B (100 percent) had a favorable response.

Central Venous Catheters

Management of central venous catheters did not differ significantly between the two groups. Overall, 111 patients with candidemia (54 in the caspofungin group and 57 in the amphotericin B group) had an indwelling central venous catheter at the time of the first positive blood culture. By day 3, the central venous catheter had been removed in 41 of the caspofungin-treated patients (75.9 percent) and 42 of the amphotericin B–treated patients (73.7 percent); guide-wire changes were made in 7 patients (13.0 percent) and 10 patients (17.5 percent), respectively. The response rate among the 11 patients whose central venous catheters were not removed or changed (6 in the caspofungin group and 5 in the amphotericin B group) was similar to the rate among the patients whose central venous catheters were removed or changed.

Mortality

The mortality rate among all patients was similar in the two treatment groups. There were 39 deaths in the caspofungin group (34.2 percent) and 38 in the amphotericin B group (30.4 percent, P=0.53). The proportion of patients who died during intravenous therapy was also similar in the two groups. The death of one patient (a patient in the amphotericin B group who had a cardiac arrest) was judged to be drug-related.

A post hoc analysis was performed to determine mortality attributable to candida. Patients were considered to have died from candida infection if any of the following criteria were met: the investigator identified the candida infection as the cause of death, the patient had a positive candida culture within 48 hours of death, or there was histopathological evidence of candida or a positive culture at autopsy. Overall, five of the caspofungin-treated patients (4.4 percent) and nine of the amphotericin B–treated patients (7.2 percent) died from candida infection (P=0.57).

Candida Endophthalmitis

An eye examination was performed in 217 of the patients (96.9 percent); 187 patients (83.5 percent) underwent a base-line examination, and 155 (69.2 percent) underwent one or more follow-up examinations. Only seven patients (3.7 percent) had candida endophthalmitis at base line; resolution was noted in all patients with follow-up examination. One patient in the amphotericin B group who had normal findings at base line reported ocular disturbances on day 3; a follow-up eye examination confirmed the presence of endophthalmitis.

Susceptibility of Candida Isolates

The base-line candida isolates from 94 percent of the patients were evaluated for in vitro susceptibility to antifungal agents at a central microbiology laboratory according to the methods of the National Committee for Clinical Laboratory Standards (document 27A). The minimal inhibitory concentration of caspofungin for the various candida species ranged from 0.125 to more than 8 μg per milliliter, as follows: C. albicans (median, 0.5 μg per milliliter; range, 0.125 to >8), C. glabrata (1.0 μg per milliliter; range, 0.5 to 2), C. tropicalis (1.0 μg per milliliter; range, 0.5 to >8), C. parapsilosis (2 μg per milliliter; range, 2 to >8), and C. krusei (2 μg per milliliter; all values were 2). All the isolates for which the minimal inhibitory concentration of caspofungin exceeded 8 μg per milliliter were associated with trailing end points; this trailing effect was not noted when a different medium, antibiotic medium no. 3, was used for susceptibility testing. The minimal inhibitory concentration of amphotericin B ranged from 0.25 to 2.0 μg per milliliter. The outcome of treatment was not predicted by the base-line minimal inhibitory concentration; all isolates for which the minimal inhibitory concentration of caspofungin exceeded 2 μg per milliliter responded favorably to caspofungin.

Adverse Events

Table 6. Table 6. Drug-Related Adverse Events and Other Safety End Points.

The proportion of patients with drug-related adverse events was significantly higher in the amphotericin B group than in the caspofungin group (Table 6). Fever, chills, and infusion-related events were much more frequent in the amphotericin B group. One patient in the caspofungin group (0.9 percent), as compared with 40 patients in the amphotericin B group (32.0 percent), had infusion-related adverse events of moderate or severe intensity. Significantly more patients in the amphotericin B group had nephrotoxic effects (24.8 percent, vs. 8.4 percent in the caspofungin group) or hypokalemia (23.4 percent vs. 9.9 percent).

Our trial was designed to compare the efficacy of an echinocandin with that of amphotericin B in patients with invasive candidiasis at various sites, as well as in those with candidemia. Inclusion in the study required evidence of proven invasive candidiasis, according to established guidelines.³⁰ The study design and outcome evaluations mirrored those used in prior studies, except that our study was double-blinded and included patients with neutropenia.^4,5,11-13

Except for a difference in the proportion of patients with C. albicans infections, the two treatment groups were well matched. In the modified intention-to-treat analysis, caspofungin was as effective as amphotericin B at the end of intravenous therapy. In the analysis of patients who met the prespecified criteria for evaluation, caspofungin was superior to amphotericin B. Differences in efficacy between the two groups were mainly a reflection of failures due to toxic effects (in 3 percent of the caspofungin group and 17 percent of the amphotericin B group). There were no significant differences in relapse or mortality rates in the two groups at the final follow-up visit.

Over 80 percent of the patients in our study had candidemia, and the outcomes in this subgroup mirrored the overall results, as did the outcomes in patients with infections at sites other than the blood. The rate of a favorable response to caspofungin among patients with postsurgical infections, including candida peritonitis and intraabdominal abscesses, was high.

A significantly larger percentage of patients treated with amphotericin B had infusion-related adverse events, nephrotoxic effects, or hypokalemia. The incidence of adverse events in the two groups was similar to that in two previous, dose-ranging studies that compared caspofungin with amphotericin B.^27,28

Although patients with neutropenia were not excluded from the study, only 11 percent of the enrolled patients had neutropenia at base line. Thus, the outcome data for these patients must be interpreted with caution. The poor prognosis for patients with neutropenia, the fact that the study was blinded, and unfamiliarity with echinocandins among clinicians may have accounted for the low proportion of patients with neutropenia among those enrolled. As expected, the response rate was lower among patients with neutropenia than among those without neutropenia in both treatment groups. The most common cause of treatment failure in the cohort of patients with neutropenia was the development of metastatic (in most cases hepatosplenic) candida lesions after granulocytic recovery.

In our study, the incidence of non-albicans infections (55 percent) was considerably higher than that in the initial comparative studies of amphotericin B and fluconazole (25 to 35 percent).^11,12 Recent epidemiologic changes in the microbiologic characteristics of candida infections may account in large part for this difference.^15-18 The international composition of the study may also have contributed to the difference. For instance, the enrollment of patients in Latin American countries accounted for most C. parapsilosis infections.^15,31,32 Finally, there may have been a bias against the inclusion of patients with C. albicans infections, with clinicians opting to use fluconazole to treat immunocompetent patients infected with this organism.

Among patients with non-albicans candida infections, the outcomes were similar in the two groups. However, five of the nine caspofungin-treated patients with persistently positive cultures had C. parapsilosis candidemia. This finding was not associated with neutropenia, a high APACHE II score, or a delayed change of a central venous catheter. Four of these five patients were enrolled at the same institution within a four-month period. The minimal inhibitory concentrations of caspofungin (2 μg per milliliter), amphotericin B, and fluconazole were the same for the C. parapsilosis isolates from all four of these patients, suggesting that they were infected with the same strain or with closely related strains. In contrast, 12 of the 14 patients with C. parapsilosis infections at other institutions had favorable responses to caspofungin.

In this study, caspofungin was effective for the treatment of invasive candidiasis and, more specifically, candidemia. At each evaluation, the percentage of patients with a favorable outcome was higher in the caspofungin group than in the amphotericin B group. The outcomes at the end of intravenous therapy in the patients who could be evaluated suggest that caspofungin was superior to amphotericin B. Caspofungin is an effective, yet less toxic, alternative to amphotericin B for the treatment of invasive candidiasis.

In addition to the authors, the following investigators participated in the study: Argentina — P. Cahn (Fundacion Huesped, Buenos Aires), R.O. Ruvinski (Hospital Durand, Buenos Aires), A. Torres (Hospital Zenem Santillan, Tucuman); Australia — G. Dobb (Royal Perth Hospital, Perth), C. Eisen and J. Lipman (Royal Brisbane Hospital, Herston); Austria — W. Graninger (General Hospital of the City of Vienna, Vienna), W. Linkesh (University Clinic for Internal Medicine, Graz), N. Vetter (Pulmonologisches Zentrum der Stadt Wein, Vienna); Belgium — F. Jacobs (Hôpital Université Erasme, Brussels), B. Vandercam (Cliniques Universitaires Saint-Luc-Médecine, Brussels); Canada — G. Harding (St. Boniface General Hospital, Winnipeg, Man.), M. Laverdiere (Hospital Maisonneuve-Rosemont, Montreal), P. Phillips (St. Paul's Hospital, Vancouver, B.C.); Colombia — A. Villaneuva (Instituto de Seguros Sociales, Barranquilla); France — M. Leporrier (Centre Hospitalier Universitaire Clemenceau, Caen), Y. Ozier (Cochin Hospital, Paris), J.F. Rossi (Hôpital LaPeyronie, Montpellier); Germany — G. Just-Nuebling and W. Stille (University Hospital of Johann-Wolfgang-Goethe-University, Frankfurt); Hong Kong — R. Liang (Queen Mary Hospital, Hong Kong); Israel — M. Shapiro (Hadassah Medical Center, Jerusalem), A.M. Yinnon (Shaare-Zede Medical Center, Jerusalem); Mexico — J. Andrade-Perez (Santiago Nuevo Hospital Civil de Guadalajara, Guadalajara), J. Castenon-Gonzalez (Centro Medico Nacional Siglo XXII, Mexico Federal District); the Netherlands — I.M. Hoepelman (Academisch Ziekenhuis Utrecht, Utrecht), A. Voss (Academisch Ziekenhuis Nijmegen, Nijmegen); Russia — N. Klimko (Institute of Invasive Mycosis, St. Petersburg), V. Savchenko (Hematological Center of the Academy of Medical Sciences, Moscow); Spain — E. Bouza (Hospital Gregorio Maranon, Madrid), J.A. Carton (University of Oviedo, Oviedo), M. Gobernado-Serrano (Hospital La Fe, Valencia), A. Guerrero-Espejo, C.Q. Rodriquez-Navarro (Hospital Ramon y Cajal, Madrid), J.A. Martinez (Hospital Clinic I Provincial de Barcelona, Barcelona); United Kingdom — P. Hawkey (University of Leeds, Leeds); United States — R.D. Adam (University of Arizona Medical Center, Tucson), L. Cone (Eisenhower Medical Center, Rancho Mirage, Calif.), J.R. Graybill (Audie L. Murphy Memorial Veterans Hospital, San Antonio, Tex.), R. Greenberg (University of Kentucky Medical Center, Lexington), S. Gordon and K. Adal (Cleveland Clinic Foundation, Cleveland), E. Johnson (St. Michael's Medical Center, Newark, N.J.), S. Hadley (Beth Israel Deaconess Hospital, Boston), D. Hospenthal (Tripler Army Medical Center, Honolulu), D.G. Maki (University of Wisconsin Medical Center, Madison), J.E. Mangino (Ohio State University Hospital, Columbus), B. Myers (Mount Sinai Medical Center, New York), J. Pfundstein (Fairfax Hospital, Fairfax, Va.), I. Raad (M.D. Anderson Cancer Center, Houston), A.C. Reboli (Cooper Hospital–University Medical Center, Camden, N.J.), B. Segal (Buffalo General Hospital and Roswell Park Cancer Institute, Buffalo, N.Y.), J. Vazquez (Wayne State University, Detroit); Venezuela — J.R. Torres-Rojas (Clinica Santa Sofia, Caracas). Data Safety and Monitoring Board — H. Masur (chair, National Institutes of Health, Bethesda, Md.), M.P. Glauser (back-up chair, Centre Hospitalier Universitaire, Lausanne, Switzerland), S. Waldman (Thomas Jefferson University Hospital, Philadelphia), S. Petersdorf (University of Washington Medical Center, Seattle), C.E. Davis (University of North Carolina, Chapel Hill).