Prospective evaluation of the new chromogenic medium CandiSelect 4 for differentiation and presumptive identification of the major pathogenic Candida species

The incidence of invasive fungal infections has increased steadily over the past two decades (Eggimann et al., 2003; Marr, 2004; Richardson, 2005). With recent advances in medical and surgical intervention and the increasing population of immunocompromised patients, the diversity and list of human fungal pathogens continue to grow (Walsh et al., 2004). It is widely accepted that Candida species are a common cause of nosocomial infection with significant associated morbidity, mortality and increased health-care costs (Eggimann et al., 2003; Nucci & Marr, 2005; Pasqualotto et al., 2005; Richardson, 2005). A recent SCOPE study indicated that Candida is the fourth most-frequent cause of nosocomial bloodstream infection accounting for 9 % of bloodstream infections overall (Wisplinghoff et al., 2004). Candida albicans is the most common Candida species implicated in these infections, but non-albicans species such as Candida glabrata and Candida krusei are increasing in frequency. Furthermore, infections caused by these latter species often vary in their antifungal susceptibility (Sanglard & Odds, 2002). According to most published guidelines for the treatment of candidiasis, rapid and accurate diagnosis of infection is required before appropriate treatment can be instigated (Buchner et al., 2002; Pappas et al., 2004; Rex et al., 2000). Several chromogenic media have been developed to help in the rapid identification of yeasts (Cardenes et al., 2002; Cooke et al., 2002; Fricker-Hidalgo et al., 2001; Letscher-Bru et al., 2002; Odds & Bernaerts, 1994; Rousselle et al., 1994). These media contain chromogenic substrates that react with enzymes secreted by yeast cells, resulting in colonies with various pigmentations. These enzymes are species-specific, allowing organisms to be identified to species level by their colour and colony characteristics. The first generation of these media was designed for the isolation and identification of C. albicans in a single step according to the specific pigmentation of yeast colonies. The use of such media also makes it easier to detect mixed yeast cultures (Cardenes et al., 2002; Letscher-Bru et al., 2002; Odds & Bernaerts, 1994). CandiSelect 4 (CS4) chromogenic medium (Bio-Rad) has recently been developed for the identification of C. albicans and the presumptive identification of the major pathogenic yeast species (C. glabrata, Candida tropicalis and C. krusei). This prospective study evaluated the performance of CS4 in comparison with that of CHROMagar Candida (CA), another chromogenic medium used widely in the clinical microbiology laboratory. This study was designed to evaluate CS4 as a primary isolation medium to identify yeasts directly from clinical specimens. Three different studies were conducted to evaluate the performance of CS4 medium. (i) One thousand five hundred and forty-nine clinical specimens [474 urine, 319 tracheal, 173 oral, 129 anal, 122 sputa, 114 bronchial aspirates, 71 vaginal, 43 stools, 30 ear, nose and throat, 27 bronchoalveolar lavage and 47 miscellaneous (e.g. catheter, drains, ascites)] were plated onto CS4 and CA in a prospective study conducted between February and July 2004. Each non-fluid specimen was suspended in sterile 0.85 % physiological saline and 0.02 ml of this suspension was then streaked onto plates of medium. Both media were used for primary isolation in parallel with Sabouraud's dextrose agar (SDA). (ii) Eighty-eight strains, originally isolated from clinical specimens in the bacteriology laboratory and transferred to the mycology laboratory for species identification, were subcultured onto CS4 and CA media. These strains were isolated from 85 samples [24 urine, 14 tracheal, 13 vaginal swabs, 11 stools, eight bronchial aspirates or bronchoalveolar lavage, five sputa, four oral and six miscellaneous (e.g. blood cultures, peritoneal)]. Eighteen strains from our reference collection were also tested (stock isolates). (iii) The growth of less common yeasts and filamentous fungi was also tested using strains from our stock collection.

Plates were incubated at 37 °C for 2448 h; if no growth was observed, incubation was extended up to 72 h. If an infection with Cryptococcus neoformans was suspected, a second plate was inoculated in parallel and incubated at 30 °C. Each plate was read with emphasis placed on recording the colony colour, the number, size and texture of colonies, and the presence of colour diffusion into the surrounding agar.

The reference procedures for identifying yeast species comprised germ tube production, micro-fermentation, microscopic morphology, chlamydosporulation on rice-agar-Tween medium, rapid identification tests (Bichrolatex albicans, GLABRATA RTT and Krusei Color; Fumouze Laboratory) and API 32C assimilation tests, if required (bioMérieux).

Statistical analysis was performed using SPSS 11.0 software, and McNemar's test (for paired observations) was used for analysis of fungal c.f.u. (CS4 vs CA) at 24, 48 and 72 h. Statistical significance was defined as P<0.05.

With these two chromogenic tests, the most specific morphology is seen with 48 h colonies. Typically on CS4, colonies produced by C. albicans are pink to purple, with a purple pigmentation which diffuses out around the colonies. Intense turquoise pigmented colonies, with a mat, uniformly coloured, convex, smooth morphotype, are indicative of C. tropicalis. Light-turquoise colonies (white periphery and turquoise centre), with a flat, shiny, smooth morphotype, are suggestive of C. glabrata. Turquoise-blue colonies, with a characteristically rough morphotype, a dry appearance and an irregular outline, are typical of C. krusei (Fig. 1). Following incubation for 48 h, colonies that remained white or pale-blue in colour were identified by conventional methods. A total of 1549 clinical samples (primary cultures) and 106 stock strains (subcultures) were analysed.

(133K): Fig. 1. Appearance of Candida species isolated directly from monomicrobial or polymicrobial clinical specimens after 48 h incubation at 37 °C. Colonies produced by C. albicans (a, c) are pink to purple, with a purple pigmentation which diffuses out around the colonies. Intense turquoise pigmented colonies, with a mat, uniformly coloured, convex, smooth morphotype, are indicative of C. tropicalis (b, c). Light-turquoise colonies (white periphery and turquoise centre), with a flat, shiny, smooth morphotype, are suggestive of C. glabrata (c). Turquoise-blue colonies, with a characteristically rough morphotype, a dry appearance and an irregular outline, are typical of C. krusei (d).

For pure subcultures, among the 106 isolates, 21 were identified as C. albicans, 33 as C. glabrata, 21 as C. tropicalis and five as C. krusei. The other isolates were Candida parapsilosis (n=8), Candida kefyr (n=7), Candida lusitaniae (n=3), Candida sp. (n=3) and Saccharomyces cerevisiae (n=5). From 24 h, all strains of C. albicans were directly identifiable on the two chromogenic media CA and CS4. At 48 h, the proportion of typical strains observed on the two chromogenic media was identical for C. glabrata (85 %) and C. krusei (100 %). A slight difference in favour of CS4 was observed for C. tropicalis (100 vs 95 %).

For primary cultures, among the 1549 specimens, 502 (32.4 %) were culture-positive for one or more yeast species. Thirty-seven samples (7.4 %) contained two (n=34) or three (n=3) yeast species in combination and the other 465 specimens were monomicrobial. A total of 542 yeast isolates were recovered on at least one of the three media and identified by conventional methods. The distribution of the isolated yeasts or yeast-like species is summarized in Table 1. The sensitivity of direct identification of C. albicans in mono- and polymicrobial samples using CS4 and CA media is shown in Table 2. After 72 h incubation at 37 °C, the sensitivity of the two media was 92.1 and 91.1 %, respectively, with no false-positive results. The presumptive identification of other species is shown in Table 3. CS4 was superior to CA for C. glabrata (80 vs 75 %) and C. krusei (92 vs 83 %), but less effective for C. tropicalis (68 vs 76 %). Statistical analysis of these data did not demonstrate a significant difference between CS4 and CA. Concerning the extent of fungal growth (for all species), there was no significant difference between CS4 and CA at 48 h, but CS4 was significantly better after 72 h (P=0.0083).

Table 1. Yeast species isolated from 502 mono- and polymicrobial clinical specimens after 72 h incubation at 37 °C on chromogenic media

Table 2. Primary cultures: direct identification of C. albicans isolated from monomicrobial and polymicrobial specimens using CS4 and CA plates

Table 3. Primary cultures: presumptive identification of C. glabrata, C. tropicalis and C. krusei in monomicrobial/polymicrobial specimens using CS4 and CA plates after 48 h incubation at 37 °C

When the data obtained for C. albicans are considered alone, the difference in growth density was in favour of CS4 after 48 h and 72 h incubation (P=0.04 and P=0.02, respectively). This was particularly obvious for specimens with an initial low fungal load. Thus after 72 h incubation, 346 strains of C. albicans were isolated on CS4 compared with only 331 on CA; the proportion of samples that were negative was therefore 5.5 and 9.6 %, respectively. Other Candida species producing creamy colonies that are white or pale-blue on CS4 were not distinguishable to the species level with this chromogenic medium. The isolates tested (either as subcultures or primary cultures) that fell into this group included C. parapsilosis (n=32), C. kefyr (n=23), C. lusitaniae (n=5), Candida norvegensis (n=2) and Candida guilliermondii (n=1). The selectivity of CS4 was determined with several polymicrobial specimens; only one bacterial isolate, Klebsiella pneumoniae, was recovered from bronchoalveolar lavage.

CS4 also allowed the growth of Cryptococcus neoformans, Geotrichum spp., Trichosporon spp., Scedosporium apiospermum, Fusarium oxysporum, Aspergillus fumigatus, Aspergillus flavus, Aspergillus niger, Mucor spp. and Rhizopus spp.

It should be noted that the majority of isolates that did not develop the expected colour on CS4 (Table 3) were recovered from tracheal-bronchial specimens, mucous expectorations or stool samples. In addition, some of these specimens stained the agar blue at the point at which they were deposited; it is therefore recommended that these types of specimen should be streaked at closely spaced intervals to obtain well-isolated colonies away from the deposit. Concerning the specificity of colony colour, no false-positive reaction was observed for C. albicans. Phosphatase activity is particularly intense in C. tropicalis; however, some strains of C. parapsilosis, C. kefyr, C. guilliermondii and C. lusitaniae which have more intense phosphatase activity can give pale-turquoise colonies after 48 h incubation. Some strains of C. kefyr and C. parapsilosis can mimic the morphotype of C. glabrata (pale-blue colonies). The other yeast species produce colonies which have a similar colour and texture to those obtained on SDA. Conventional identification tests such as Bichrolatex albicans, Krusei Color and GLABRATA RTT, API 32C, Auxacolor 2 and antifungal susceptibility tests (E-test and Fungitest) can be performed directly using colonies isolated on CS4. CS4 also allowed the growth of other yeast or yeast-like species or filamentous fungi; however, subculture on Czapek agar or SDA is necessary for the identification of these moulds.

In conclusion, this prospective study demonstrated that CS4 is an effective, selective and useful new chromogenic medium for the presumptive identification of the most frequently isolated Candida species.