Tolerance to amphotericin B in clinical isolates of Candida tropicalis.

A broth microdilution method was used for testing amphotericin B against 33 clinical isolates of Candida tropicalis. All isolates were in vitro susceptible to the polyene (MIC [minimal inhibitory concentration] < or = 1.0 microg/mL). However, when the isolates were cultured in a medium containing amphotericin B at a concentration of 1.5 microg/mL, a wide interstrain variation of growth rate was observed. Five isolates (15%) proved to be highly tolerant to the drug and grew at a frequency ranging from 1 x 10(-1) to 2 x 10(-2). Twenty-three isolates (70%) grew at a frequency ranging from 1 x 10(-5) to 1 x 10(-8). The remaining five isolates (15%) failed to grow in drug-containing medium. In general, this growth variation was not associated with amphotericin B MICs displayed by the single isolates. In addition, the strains grown in drug-containing medium did not represent amphotericin B-resistant mutants, as shown by the maintenance of MICs similar to those of their respective parent isolates. Killing experiments conducted in selected isolates confirmed a variation of fungicidal activity of amphotericin B. To see whether this phenomenon was associated with a variation of amphotericin B response in vivo, we established an experimental model of systemic murine candidiasis in CD1 mice by intravenous injection of cells belonging to Candida tropicalis 3147 (growth rate at a frequency of 1 x 10(-1) in amphotericin B medium) and Candida tropicalis 4055 (no growth). Low (0.3 mg/kg/day) and high (1 mg/kg/day) doses of amphotericin B were both effective at reducing the fungal burdens in the kidneys of mice infected with either strain (p, 0.01 to 0.02). However, whereas the burden of mice infected with isolate 3147 and treated with the polyene at 0.3 mg/kg/day was reduced by 1.2 +/- 0.25 (mean +/- standard deviation) log10 cfu/g compared to untreated mice, the same dosing regimen yielded a burden reduction of 2.6 +/- 0.07 log10 cfu/g in mice infected with isolate 4055 (p < 0.001). Similarly, amphotericin B at 1 mg/kg/day yielded a burden reduction of 1.8 +/- 0.20 vs. 2.5 +/- 0.30 log10 cfu/g in mice infected with isolates 3147 and 4055, respectively (p < 0.001). Our data revealed a variable pattern of tolerance to amphotericin B among isolates of Candida tropicalis and showed that this phenomenon might influence the rate of organ clearance during therapy.