We conducted prospective, active population-based surveillance for candidemia (defined as any Candida species isolated from blood) in Atlanta and San Francisco (total population, 5.34 million) during 1992-1993. The average annual incidence of candidemia at both sites was 8 per 100,000 population. The highest incidence (75 per 100,000) occurred among infants

Candida dubliniensis is a newly described species that is closely related phylogenetically to Candida albicans and that is commonly associated with oral candidiasis in human immunodeficiency virus-positive patients. Several recent studies have attempted to elucidate phenotypic and genotypic characteristics of use in separating the two species. However, results obtained with simple phenotypic tests were too variable and tests that provided more definitive data were too complex for routine use in the clinical laboratory setting. The objective of this study was to determine if reproducible identification of C. dubliniensis could be obtained with commercial identification kits. The substrate reactivity profiles of 80 C. dubliniensis isolates were obtained by using the API 20C AUX, ID 32 C, RapID Yeast Plus, VITEK YBC, and VITEK 2 ID-YST systems. The percentages of C. dubliniensisisolates capable of assimilating or hydrolyzing each substrate were compared with the percentages from the C. albicans profiles in each kit's database, and the results were expressed as percentC. dubliniensis and percent C. albicans. Any substrate that showed >50% difference in reactivity was considered useful in differentiating the species. In addition, assimilation of methyl-α-d-glucoside (MDG), d-trehalose (TRE), and d-xylose (XYL) by the same isolates was investigated by the traditional procedure of Wickerham and Burton (L. J. Wickerham and K. A. Burton, J. Bacteriol. 56:363–371, 1948). At 48 h (the time recommended by the manufacturer for its new database), we found that the assimilation of four carbohydrates in the API 20C AUX system could be used to distinguish the species, i.e., glycerol (GLY; 88 and 14%), XYL (0 and 88%), MDG (0 and 85%), and TRE (15 and 97%). Similarly, results with the ID 32 C system at 48 h showed that XYL (0 and 98%), MDG (0 and 98%), lactate (LAT; 0 and 96%), and TRE (30 and 96%) could be used to separate the two species. Phosphatase (PHS; 9 and 76%) and α-d-glucosidase (23 and 94%) proved to be the most useful for separation of the species in the RapID Yeast Plus system. While at 24 h the profiles obtained with the VITEK YBC system showed that MDG (10 and 95%), XYL (0 and 95%), and GLY (26 and 80%) could be used to separate the two species, at 48 h only XYL (6 and 95%) could be used to separate the two species. The most useful substrates in the VITEK 2 ID-YST system were TRE (1 and 89%), MDG (1 and 99%), LAT (4 and 98%), and PHS (83 and 1%). While the latter kit was not yet commercially available at the time of the study, it would appear to be the most valuable for the identification of C. dubliniensis. Although assimilation of MDG, TRE, and XYL proved to be the most useful for species differentiation by the majority of commercial systems, the results with these carbohydrates by the Wickerham and Burton procedure were essentially the same for both species, albeit following protracted incubation. Thus, it is the rapidity of the assimilation achieved with the commercial systems that allows the differentiation of C. dubliniensis from C. albicans.

Thirty-two Malassezia spp. isolates from human clinical specimens represented M. furfur and M. pachydermatis. Both species reportedly were obtained from patients with similar febrile systemic syndromes, including infections of the lungs or other tissues.

The conventional identification (assimilation profiles in the API 20C [bioMérieux Vitek, Inc., Hazelwood, Mo.] identification system, colony color on CHROMagar Candida [CHROMagar, Paris, France], and morphology on cornmeal-Tween 80 agar) of 32 atypical isolates of Candida albicans recovered during a study of oropharyngeal candidiasis in human immunodeficiency virus-infected individuals could not be confirmed by genetic fingerprinting (1, 3). Comparative studies of the nucleotide divergence in the 5Ј end of the 26S ribosomal DNA gene conducted at the National Center for Agricultural Utilization Research, Peoria, Ill., revealed that 10 of these 32 isolates were Candida dubliniensis and the remaining 22 were atypical C. albicans isolates (2). This constitutes, to the authors' knowledge, the first recovery of C. dubliniensis in association with human infections in the United States (4). When the 10 C. dubliniensis isolates were reexamined with the API 20C identification system, none were found to assimilate ␣-methyl-D-glucoside (MDG). Similar results were reported by Sullivan et al. in their initial description of the organism's phenotypic characteristics (4, 5). However, MDG cannot be utilized as the sole carbon source by many isolates of C. albicans (1). It became clear upon closer examination of the API 20C carbohydrate assimilation profiles that none of the 10 C. dubliniensis isolates assimilated xylose (XYL) and MDG (API 20C biocodes of 2172170 or 6172170). In contrast, all 22 C. albicans isolates utilized one or both of these carbon sources (API 20C biocodes of 2542170, 2552170, 2556170, 2572170, or 2576170). In summary, we have recovered the first clinical isolates of C. dubliniensis in the United States and found that all could be differentiated from C. albicans by their inability to assimilate XYL and MDG in the API 20C identification system. The absence of simple, reproducible tests for phenotypic characteristics of use in diagnostic microbiology laboratories to differentiate C. albicans from C. dubliniensis has greatly curtailed our clinical and epidemiological understanding of C. dubliniensis. Furthermore, the finding of distinct API 20C assimilation profiles associated with C. dubliniensis may stimulate clinical laboratories to evaluate our results and, if our results are substantiated, to initiate the use of assimilation profiles in their definitive identification of presumptive C. albicans isolates recovered from clinical specimens. The authors are currently investigating this and other phenotypic characteristics with larger experimental populations in an effort to provide additional simple and effective methods for the differentiation of C. dubliniensis from C. albicans.

Isoenzyme and protein profiles of clinical isolates initially identified as Candida haemulonii demonstrated the presence of two distinct groups. DNA relatedness studies with representative cultures confirmed the presence of two species. Physiological features that can be used to separate two groups within C. haemulonii are reported.