High rates of misidentification of uncommon Candida species causing bloodstream infections using conventional phenotypic methods.

Background Candidemia caused by uncommon Candida species is increasing and misidentification may compromise optimal antifungal therapy. This multicenter study aimed to evaluate the accuracy of species-level identification of uncommon Candida. Methods Uncommon causative species of candidemia identified in routine laboratories using CHROMagar, API-32C and VITEK-2 Yeast ID system were collected from July 2011 to June 2014. These isolates were further identified using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) system and sequencing of the internal transcribed spacer and 28S rRNA gene. Susceptibility of the isolates was determined. Results Of 85 isolates evaluated, Candida guilliermondii (n = 36) was the most common, followed by Candid sake (n = 7) and Candida famata (n = 4). Using DNA-sequencing analysis as standard, none of C. sake and C. famata was correct, while VITEK MS correctly identified 10 of the 11 isolates. With the exclusion of one unspecified Candida by DNA-sequencing methods, the accuracy of conventional methods and VITEK MS was 64.3% and 86.9%, respectively (p = 0.001). Eight isolates were confirmed to be yeasts other than Candida. Compared with other Candida species, C. guilliermondii showed elevated minimal inhibitory concentration of echinocandins. Conclusion Misidentification of uncommon Candida species was common using the conventional methods, especially for C. sake and C. famata. MALDI-TOF MS assisted by DNA-sequencing methods should be considered.