Epidemiological cutoff values for azoles and Aspergillus fumigatus based on a novel mathematical approach incorporating cyp51A sequence analysis

Epidemiological cutoff values (ECV) are commonly used to separate wild-type isolates from isolates with reduced susceptibility to antifungal drugs, thus setting the foundation for establishing clinical breakpoints for Aspergillus fumigatus. However, ECVs are usually determined by eye, a method which lacks objectivity, sensitivity, and statistical robustness and may be difficult, in particular, for extended and complex MIC distributions. We therefore describe and evaluate a statistical method of MIC distribution analysis for posaconazole, itraconazole, and voriconazole for 296 A. fumigatus isolates utilizing nonlinear regression analysis, the normal plot technique, and recursive partitioning analysis incorporating cyp51A sequence data. MICs were determined by using the CLSI M38-A2 protocol (CLSI, CLSI document M38-A2, 2008) after incubation of the isolates for 48 h and were transformed into log(2) MICs. We found a wide distribution of MICs of all azoles, some ranging from 0.02 to 128 mg/liter, with median MICs of 32 mg/liter for itraconazole, 4 mg/liter for voriconazole, and 0.5 mg/liter for posaconazole. Of the isolates, 65% (192 of 296) had mutations in the cyp51A gene, and the majority of the mutants (90%) harbored tandem repeats in the promoter region combined with mutations in the cyp51A coding region. MIC distributions deviated significantly from normal distribution (D'Agostino-Pearson omnibus normality test P value, <0.001), and they were better described with a model of the sum of two Gaussian distributions (R(2), 0.91 to 0.96). The normal plot technique revealed a mixture of two populations of MICs separated by MICs of 1 mg/liter for itraconazole, 1 mg/liter for voriconazole, and 0.125 mg/liter for posaconazole. Recursive partitioning analysis confirmed these ECVs, since the proportions of isolates harboring cyp51A mutations associated with azole resistance were less than 20%, 20 to 30%, and >70% when the MICs were lower than, equal to, and higher than the above-mentioned ECVs, respectively.