Prevention of fungal infections in cancer patients with amphotericin B or fluconazole.

Seventeen trials (3798 patients, 381 deaths) were included. In two large three-armed trials, results for amphotericin B were combined with results for nystatin in a "polyene" group. Because nystatin is an ineffective drug in these circumstances, this approach creates a bias in favour of fluconazole. Furthermore, most patients were randomised to oral amphotericin B, which is poorly absorbed and poorly documented. There was overlap among the "polyene" trials but were unable to obtain any information to clarify these issues further from the trial authors or from Pfizer, the manufacturer of fluconazole. There were no significant differences in effect between fluconazole and amphotericin B, but the confidence intervals were wide. More patients dropped out of the study when they received amphotericin B, but as none of the trials were blinded, decisions on premature interruption of therapy could have been biased. Furthermore, amphotericin B was not given under optimal circumstances, with premedication to reduce infusion-related toxicity, slow infusion, and with fluid, potassium and magnesium supplements to prevent nephrotoxicity. The major harms were hepatic impairment and gastrointestinal adverse effects with fluconazole and infusion-related toxicity, renal impairment and gastrointestinal adverse effects with amphotericin B. For the 2011 update no additional trials were identified for inclusion.