Background: Sensitivity analyses were incorporated in a Phase III study of caspofungin vs. liposomal amphotericin B as empirical antifungal therapy for febrile neutropenic patients to determine the impact of varying definitions of fever resolution on response rates.
Methods: The primary analysis used a 5-part composite endpoint: resolution of any baseline invasive fungal infection, no breakthrough invasive fungal infection, survival, no premature discontinuation of study drug, and fever resolution for 48 h during the period of neutropenia. Pre-specified analyses used 3 other definitions for fever resolution: afebrile for 24 h during the period of neutropenia, afebrile at 7 days post therapy, and eliminating fever resolution altogether from the composite endpoint. Patients were stratified on entry by use of antifungal prophylaxis and risk of infection. Allogeneic hematopoietic stem cell transplants or relapsed acute leukemia defined high-risk patients.
Results: In the primary analysis, 41% of patients in each treatment group met the fever-resolution criteria. Low-risk patients had shorter durations of neutropenia but failed fever-resolution criteria more often than high-risk patients. In each exploratory analysis, response rates increased in both treatment groups compared to the primary analysis, particularly in low-risk patients.
Conclusions: Response rates for the primary composite endpoint for both treatment groups in this study were driven by low rates of fever resolution. Requiring fever resolution during neutropenia in a composite endpoint can mask more clinically relevant outcomes.