Evaluation of DNA-based typing procedures for strain categorization of Candida spp.

DNA-based procedures have replaced earlier epidemiologic methodologies that relied on nonreproducible and insensitive measurements of phenotypic characteristics to identify a specific strain as the source of infection. The reliability (interlaboratory percent agreement for strain delineation) and sensitivity (recognition of subtle strain-to-strain variation) of similar DNA-based typing systems by different laboratories were evaluated. Ten isolates (five epidemiologic-related and five unrelated strains each) of Candida albicans, C. lusitaniae, C. parapsilosis, C. tropicalis, and Candida (Torulopsis) glabrata were characterized in a blinded fashion by three laboratories. All 50 isolates were subtyped in each laboratory by electrophoretic karyotyping (EK) analysis using contour-clamped homogenous electric field (CHEF) electrophoresis protocols. In addition, two laboratories also performed restriction endonuclease analysis of genomic DNA (REAG) using the restriction endonucleases SfiI and BssHII followed by CHEF electrophoresis separation of resulting fragments. DNA strain identification of the 50 isolates by the three different laboratories using similar CHEF methodologies demonstrated the following species-dependent, interlaboratory reproducibility: C. tropicalis (82%), C. parapsilosis (83%), C. albicans (90%), C. lusitaniae (93%), and C. glabrata (100%). In addition, agreement a,ns higher by the CHEF method (83 to 100%), when compared with the strain types identified by the REAG (60 to 100%) method. Five to seven strains of each Candida species evaluated were detected by the different methodologies used for this study. This study indicates that these procedures are relatively discriminatory and reliable tools to study strain-to-strain variations in epidemiologic evaluations of these yeasts. (C) 1999 Elsevier Science Inc.