Empiric Therapy in the Persistently Febrile Patient With Neutropenia: Caspofungin vs Liposomal Amphotericin B

Abstract & Commentary

Synopsis: Overall, caspofungin was at least as effective as liposomal amphotericin B for empiric therapy in persistently febrile neutropenic patients. Caspofungin therapy was associated with greater survival 7 days after the end of therapy, greater efficacy in treatment of baseline fungal infections, and it was better tolerated.

Source: Walsh TJ, et al. Caspofungin vs Liposomal Amphotericin B For Empirical Antifungal Therapy in Patients With Persistent Fever and Neutropenia. N Engl J Med. 2004;351: 1391-1402.

Walsh and colleagues at 116 sites in 20 countries randomized 1095 patients > 16 years of age with chemotherapy or HSCT associated neutropenia (< 500/mm3) and persistence of fever, despite at least 96 hours of parenterally administered empiric antibacterial therapy. All but 6% had an underlying hematologic malignancy, the majority with acute myelogenous leukemia. Patients received, in a double-blind fashion, either caspofungin (70 mg on day 1, then 50 mg IV daily) or liposomal amphotericin B (LAB) (3 mg/kg/d)—each with an accompanying placebo resembling the other medication. No premedication was allowed on day 1, but was thereafter. The dose of caspofungin could be increased to 70 mg daily if fever persisted for 5 or more days, while that of LAB could be increased to 5 mg/kg/d. In the absence of baseline or breakthrough fungal infection, the drugs were administered until the ANC was > 500/mm3.

Patients received caspofungin for a mean of 13 days, and LAB for a mean of 12.5 days. The overall success rate (see Table 1 for definition of success), which was the primary efficacy end point, was 34.2% among caspofungin recipients and 33.6% of those given LAB, after adjustment for strata (risk and prior receipt of antifungal prophylaxis) (see Table 2). Prior antifungal prophylaxis did not significantly affect the response rates. Caspofungin recipients had significantly better outcomes for 3 of the 5 components of the primary efficacy end point; survival for at least 7 days after treatment (92.6% vs 89.2%; P = 0.05), successful treatment of baseline fungal infection (51.9% vs 25.9%; P =0.04), and absence of premature study discontinuation (89.7% vs 85.5%; P = 0.03). With regard to the last, caspofungin and LAB were discontinued because of lack of efficacy in 5.4% and 6.3%, respectively, and because of toxicity in 4.9% and 8.2%, respectively. There was no significant difference between the treatment arms, with regard to the remaining end point components, resolution of fever, or absence of breakthrough fungal infection. The latter occurred in 29 caspofungin and 24 LAB recipients, with approximately equal numbers of aspergillosis (10 and 9) and invasive candidiasis (16 and 15).

Successful resolution of fungal infections present at baseline was observed in 14 of 27 casofungin recipients and in 7 of 27 LAB recipients. Baseline infection due to Aspergillus spp. was successfully treated with caspofungin in 5 of 12 (41.7%) cases, and by LAB in only 1 of 12 (8.3%). Caspofungin was successful in 8 of 12 (66.7%) patients with baseline invasive candidiasis, while LAB was successful in 5 of 12 (41.7%).

Caspofungin was better tolerated than LAB, with less nephrotoxicity (2.6% vs 11.5%; P < 0.001), fewer infusion related events (35.1% vs 51.6%; P < 0.001), and lesser frequency of discontinuation because of a drug related adverse event (5% vs 8%; P = 0.04).

Comment by Stan Deresinski, MD, FACP

We have come a long way from the initial report by Pizzo and colleagues that appeared to demonstrate a benefit to the use of empirically administered amphotericin B deoxycholate in febrile neutropenia, in a study that involved a total of only 50 patients.1

In an editorial accompanying the paper by Walsh and colleagues, Klastersky summarizes data from more recent large randomized trials in persisting febrile neutropenia using the same composite end point as in this study (see Figure). One study demonstrated superiority of LAB over amphotericin B deoxycholate with, however, no difference in survival.2 A comparison of voriconazole with LAB failed to meet criteria for non-inferiority of voriconazole, although this triazole was better tolerated.3 In the current study, caspofungin easily met criteria for non-inferiority vis-a-vis LAB, with regard to the composite end point. Caspofungin was, however, superior to LAB, with regard to 3 of 5 of the individual elements of the end point. Perhaps of most significance, was the superiority of caspofungin over LAB in the treatment of baseline fungal infections, mostly due to Aspergillus spp. or Candida spp.

In the studies involving empiric therapy with LAB, the dose utilized has been 3 mg/kg/d, a lesser dose than that commonly used for therapeutic purposes. It could be argued that a higher dose would be more efficacious (in the current study Walsh and colleagues could increase the dose to 5 mg/kg/d for fever persisting for 5 or more days). An increase in dose would inevitably, however, be associated with greater toxicity.

At this time, the only thing going for amphotericin B deoxycholate is its insignificant acquisition cost. The latter may, however, be offset by its apparent lesser efficacy and costs related to toxicity, especially nephrotoxicity. Voriconazole, although apparently of lesser efficacy than LAB, has the advantages over it of lesser cost and toxicity, as well as the potential for oral administration. Its inhibition of cytochrome CYP450 enzymes must always, however, be taken into account in patients receiving multiple pharmaceutical agents. Finally, although the overall assessment is of equivalent efficacy, the study reviewed here demonstrates superiority of caspofungin over LAB in persistent febrile neutropenia, with regard to 7 day mortality, tolerability, and treatment of baseline fungal infections. Caspofungin would appear to be the overall winner to date, but one cannot argue with Klastersky’s call for a head-to-head comparison of this echinocandin with voriconazole.

Comment by J. Peter Donnelly, PhD

Walsh et al are to be complemented for being able to complete such a complex study and produce evidence of the highest level that caspofungin and liposomal amphotericin B are equally effective for the empirical treatment of persistent fever in neutropenic patients. However, in Europe at least, there has been an intellectual move away from the very idea of empirical antifungal therapy, at least in the sense of treating every patient with persistent fever, irrespective of other findings that might be consistent with a systemic fungal infective disease such as lung involvement. For instance, in the first of these studies, pulmonary infiltrates were identified in 10% of patients.1 Although not reported in the 2 later studies, there is no reason to think the incidence would be much different. Hence, about 9 of every 10 patients given empirical therapy may not actually need it, certainly not for an average of almost 2 weeks.

Interestingly, when these studies are lined up, there is little difference between them in terms of survival a week after therapy stopped (see Figure), successful treatment of fungal infections present at the start, the prevention of breakthrough fungal infections, or in the premature stopping of therapy. However, the proportion of patients becoming afebrile was lower in the current study than in either of the 2 previous ones. It is also obvious that this criteria is unduly important in defining the composite end point, since the rate of success is virtually identical to the rate of defervescence in each case, irrespective of the antifungal.

What are we to make of this? In the first place, we need to refine the criteria for starting empirical antifungal therapy. Clearly the increased use of high-resolution CT scans will enable us to detect pulmonary infiltrates, and the screening for galactomannan and fungal DNA in blood may help select those patients with likely fungal disease who can be preemptively treated. Conversely, empirical therapy might be reserved only for those with persistent fever who are also in a critical clinical condition and are unable to undergo diagnostic tests. In any event, if treatment is started empirically, it should not release the clinicians from attempting to make a diagnosis, but should be used rather as a holding operation until the picture becomes clearer and fungal disease is either excluded, in which case it should be stopped, or confirmed, in which case treatment should be continued. Importantly, clinicians should not expect such patients to become afebrile. I, for one, hope that the need for empirical therapy will diminish as diagnosis becomes better, and confidence in the strategies involved becomes greater.


1. Pizzo PA, et al. Empiric Antibiotic and Antifungal Therapy For Cancer Patients With Prolonged Fever and Granulocytopenia. Am J Med. 1982;72:101-111.

2. Walsh TJ, et al. Liposomal Amphotericin B For Empirical Therapy in Patients With Persistent Fever and Nuetropenia. N Engl J Med. 1999;340:764-771.

3. Walsh TJ, et al. Voriconazole Compared With Liposomal Amphotericin B For Empirical Antifungal Therapy in Patients With Neutropenia and Persistent Fever. N Engl J Med. 2002;346:225-234.

4. Walsh TJ, et al. Liposomal Amphotericin B For Empirical Ttherapy in Patients With Persistent Fever and Neutropenia. National Institute of Allergy and Infectious Diseases Mycoses Study Group. N Engl J Med. 1999;340(10):764-771.

Stan Deresinski, MD, FACP, Clinical Professor of Medicine, Stanford; Associate Chief of Infectious Diseases, Santa Clara Valley Medical Center, is Editor for Infectious Disease Alert.

J. Peter Donnelly, PhD, Clinical Microbiologist, University Hospital, Nijmegen, The Netherlands, Section Editor, Microbiology, is Associate Editor for Infectious Disease Alert.