Discriminative power of fatty acid methyl ester (FAME) analysis using the microbial identification system (MIS) for...

Candida (Torulopsis) glabrata is frequently isolated in cases of fungal infection and commonly shows acquired or innate fluconazole resistance. Saccharomyces cerevisiae, an emerging opportunistic yeast pathogen, causes serious systemic infections in immunocompromised, and vaginitis and superficial infections in immunocompetent patients. For both species reliable identification in the routine laboratory is mandatory, but species identification of strains, e.g. trehalose-negative C. glabrata, may be difficult. Therefore, gas-liquid chromatography (GLC) of whole cell fatty acid methyl ester (FAME) profiles, that is independent of assimilation profiles of strains and suitable for reliable and rapid identification of clinically important yeasts, was applied. However, frequent misidentification of C. glabrata as S. cerevisiae has been reported when using the Yeast Clinical Database of MIS. Accuracy of MIS identification may be strongly influenced by the amounts of cell mass analyzed. Therefore, the present study compared the MIS results of these two yeasts achieved with different cell masses. Primarily we optimized, especially with respect to cost-effectiveness, the recommended streaking technique yielding a maximal recovery of 90-130 mg of cell mass from one plate, enabling testing of poor growing strains of C. glabrata. For all C. glabrata strains tested (n = 10) the highest identification scores (SI [Similarity Index] range 0.525-0.963, median 0.832) were achieved with 30 to 45 mg of cell mass. Only 5 of 10 S. cerevisiae strains revealed good library comparisons (SI > or = 0.5) when using 30 mg of cell mass, whereas with 45 mg all strains but two revealed this SI-level. For S. cerevisiae a higher amount of cell mass processed (up to 90 mg) was correlated with better identification scores (SI range using 90 mg: 0.464-0.870, median, 0.737). Several passages prior to FAME analysis of C. glabrata strains on recommended media revealed narrowing of SI ranges, but differences in SI values were not statistically significant.