Abstract
We compared direct inoculation of the Auxacolor yeast identification system from positive blood culture vials to standard identification with the API 20C AUX (API 20C), using 44 prospectively collected clinical specimens and 25 seeded blood culture vials. Direct inoculation of the Auxacolor system was accurate and more rapid than standard identification with the API 20C.
Rates of fungemia in North America have risen markedly, with Candida species now reported as the fourth most common cause of nosocomial bloodstream infections (4). Candida spp. other than Candida albicans, once rare, now account for almost half of the yeasts isolated from sterile sites in most centers (3, 4). Because some of these species have intrinsic or potential resistance to antifungal agents, rapid presumptive species level identification is crucial in allowing for directed antifungal therapy (9).
While the germ tube test provides rapid identification of C. albicans, non-C. albicans isolates from blood cultures are most commonly identified by commercial micromethod systems such as the API 20C AUX (API20C) (Biomerieux-Vitek) (9). This requires 24 h to obtain sufficient growth upon subculturing before inoculation and then another 48 to 72 h to identify the majority of isolates (6, 9). Direct inoculation of an identification system from positive blood culture vials could reduce this delay by obviating the need for subculturing.
The Auxacolor (AUX) system (Sanofi Diagnostics Pasteur) is an identification kit that is more rapid and accurate than the API20C when used to identify common yeast species from solid media (1, 6). This system is chromogenic, not turbidometric, and is little affected by changes in inoculum. It therefore has the potential to function when inoculated directly from positive blood culture vials, in which the inoculum size is variable and turbidity is affected by the variable opacity of the vial contents.
We compared the identification of medically relevant yeasts from positive blood culture vials using direct inoculation of the AUX system to the standard identification method using the API20C. We performed a two-part study using (i) prospectively collected positive blood cultures and (ii) blood culture vials seeded with uncommon yeast isolates.
Prospective arm.
The prospective arm of the study was conducted from 1 June to 1 December 1998 at five tertiary-care health centers in Montreal, Canada: the Royal Victoria Hospital, the Montreal General Hospital, the Jewish General Hospital, the Montreal Children’s Hospital, and the Hôpital Maisonneuve-Rosemont (enrollment at the Hôpital Maisonneuve-Rosemont began on September 20). During this period, all blood cultures identified as positive by an automated blood culture system with yeast forms demonstrated upon Gram staining were enrolled. In the event of multiple positive vials during a single episode of infection, only the first positive bottle was studied. Within 24 h of identification of a positive blood culture, the AUX was inoculated by using a modification of the manufacturer’s instructions in which 3 drops (approximately 30 μl) of the blood culture vial contents was suspended in the supplied AUX medium and this mixture was used as the inoculation broth. Two drops of this suspension were inoculated per well. The identification kits were incubated at 30°C and read at 24, 48, and 72 h. Simultaneously, the original bottle was subcultured, and after 24 h of incubation, isolated colonies were identified by using the API20C according to the manufacturer’s instructions. Times to identification with both strategies were recorded. Discrepancies were resolved by morphology and macrotube carbohydrate assimilation testing.
Seeded blood culture bottles.
To evaluate a broader range of strains, we also evaluated seeded blood cultures using 25 isolates comprising 9 species of clinically relevant yet rarely encountered yeasts. Strains were clinical sterile-site isolates supplied by the Québec Public Health Laboratory. Isolates were identified by code number and then subcultured to Sabouraud dextrose agar. Isolated colonies were serially diluted in sterile saline to a concentration of 10 to 100 CFU/ml. One milliliter of this suspension was inoculated into a BACTEC 92410 Aerobic blood culture vial containing approximately 10 ml of a clinical specimen of human blood that had already been incubated for 5 days without microbial growth. Seeded vials were incubated until identified as positive by the BACTEC detection system. The fluid was then Gram stained to verify the presence of yeast forms and processed according to the protocol described for the prospective isolates.
Statistics.
Identification agreement rates between the AUX and the API20C were calculated by using an approximate binomial confidence interval calculation, and differences in time for species level identification were tested by the Wilcoxon sign rank test (5).
Results.
Forty-four episodes of fungemia, representing seven yeast species, were enrolled (Table 1). Of these, 24 isolates were detected in BACTEC Plus Aerobic vials, 13 in BacT-Alert Aerobic vials, 3 in BacT-Alert Aerobic vials with charcoal, and 4 in the BACTEC Peds Plus vials.
TABLE 1.
Number of isolates tested by direct inoculation protocol for both prospective isolates and seeded blood culture bottles
| Species | No. of isolates
|
||
|---|---|---|---|
| Prospectively collected cultures | Seeded cultures | Total | |
| Candida albicans | 23 | 0 | 23 |
| Candida glabrata | 9 | 0 | 9 |
| Candida parapsilosis | 6 | 4 | 10 |
| Candida tropicalis | 3 | 4 | 7 |
| Cryptococcus neoformans | 1 | 3 | 4 |
| Candida guilliermondii | 1 | 2 | 3 |
| Rhodotorula glutinis | 1 | 0 | 1 |
| Rhodotorula rubra | 0 | 1 | 1 |
| Candida inconspicua | 0 | 1 | 1 |
| Saccharomyces cerevisiae | 0 | 3 | 3 |
| Candida krusei | 0 | 4 | 4 |
| Candida lusitaniae | 0 | 3 | 3 |
| Total | 44 | 25 | 69 |
Of the prospective isolates, 43 of 44 (97.7%; 95% confidence interval, 93.3 to 100%) were correctly identified by direct inoculation of the AUX. One strain of Rhodotorula glutinis could not be identified by the AUX, generating a code not found in the AUX database. Repeat testing using an isolated colony from the subculture generated an identical code. Twenty-one of 25 (84.0%; 95% confidence interval, 70.4 to 98.4%) of the seeded vials were correctly identified by the AUX. Three isolates of Saccharomyces cerevisiae and one Candida lusitaniae isolate could not be identified, generating codes not found in the AUX database. Repeat testing with isolated colonies from subcultures generated an identical code for the C. lusitaniae isolate, but for all three strains of S. cerevisiae, the test well for raffinose was positive (originally it was negative when inoculated directly from the vial), generating the correct code. All isolates were identified correctly by the API20C.
Direct inoculation significantly reduced the time for species level identification of all groups, with a more marked difference seen with non-C. albicans isolates (Table 2).
TABLE 2.
Comparison of the mean times to correct identification of isolates from prospective clinical specimens and seeded bottles
| Group | No. of isolates | Time (days) to identification with:
|
|
|---|---|---|---|
| AUX | API20C | ||
| Prospective specimens | |||
| C. albicans | 23 | 1.91 | 3.35*a |
| Non-C. albicans | 20 | 1.45 | 4.05* |
| Seeded vials (non-C. albicans) | 21 | 1.43 | 3.40* |
| All isolates | 64 | 1.70 | 3.70* |
*, statistically significant difference (P < 0.001).
Discussion.
Several methods for rapid identification of yeast isolates have been described previously (2, 7, 8). Unfortunately, most require either expensive automation or specialized expertise. This study reports the preliminary results of an inexpensive, easily interpreted strategy that significantly reduced the time required for identification of yeast isolates from blood culture without these disadvantages.
The strategy of direct inoculation of the AUX yielded accurate results. The only consistent problem identified was a false-negative raffinose reaction noted when S. cerevisiae isolates were inoculated directly from blood culture vials. The reason for this discrepancy is unclear. S. cerevisiae may express a different auxotrophic phenotype in blood culture media than on Sabouraud dextrose agar. While a substance in the blood culture bottles could be interfering with the reaction, one would expect similar errors with other species. All other isolates had the expected reactions with respect to raffinose, and the two other isolates that were not identifiable (R. glutinis and C. lusitaniae) did not display this variable phenotype. These latter two errors likely represent inadequacies in the AUX database—a phenomenon described previously (1).
Identification of C. albicans by direct inoculation of the AUX, while more rapid than identification by the API20C, required almost 48 h in most cases. This does not represent an improvement over the germ tube test performed on a sample from a 24-h subculture. Combining direct inoculation of the AUX with a germ tube test is an attractive option that should allow identification of the majority of isolates within 24 h of detection of a positive vial. Our experience suggests that the germ tube test can be performed directly on a sample from the blood culture vial, allowing one to reduce costs by using direct inoculation of the AUX to identify only non-C. albicans isolates.
Acknowledgments
We are indebted to Sanofi Diagnostics Pasteur for the gift of the AUX systems, to Guy St.-Germain for providing some of the yeast strains, and to Steven J. Reynolds for technical assistance.
REFERENCES
- 1.Davey K G, Chant P M, Downer C S, Campbell C K, Warnock D W. Evaluation of the Auxacolor system, a new method of clinical yeast identification. J Clin Pathol. 1995;48:807–809. doi: 10.1136/jcp.48.9.807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fenn J P, Segal H, Barland B, Denton D, Whisenant J, Chun H, Christofferson K, Hamilton L, Carroll K. Comparison of updated Vitek Yeast Biochemical Card and API 20C yeast identification systems. J Clin Microbiol. 1994;32:1184–1187. doi: 10.1128/jcm.32.5.1184-1187.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Nguyen M H, Peacock J E, Morris A J, et al. The changing face of candidemia: emergence of non-Candida albicans species and antifungal resistance. Am J Med. 1996;100:617–623. doi: 10.1016/s0002-9343(95)00010-0. [DOI] [PubMed] [Google Scholar]
- 4.Pfaller M A, Jones R N, Messner S A, Edmond M B, Wenzel R P. National surveillance of nosocomial blood stream infection due to species of Candida other than Candida albicans: frequency of occurrence and antifungal susceptibility in the SCOPE program. Diagn Microbiol Infect Dis. 1998;30:1–8. doi: 10.1016/s0732-8893(97)00240-x. [DOI] [PubMed] [Google Scholar]
- 5.Rosner B. Fundamentals of biostatistics. 4th ed. 1995. pp. 82–88. , 560–562. Duxbury Press, Belmont, Calif. [Google Scholar]
- 6.Sheppard D C, deSouza E, Hashmi Z, Robson H G, René P. Evaluation of the Auxacolor system for biochemical identification of medically important yeasts. J Clin Microbiol. 1998;36:3726–3727. doi: 10.1128/jcm.36.12.3726-3727.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Shin J H, Nolte F S, Morrison C J. Rapid identification of Candida species in blood cultures by a clinically useful PCR method. J Clin Microbiol. 1997;35:1454–1459. doi: 10.1128/jcm.35.6.1454-1459.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.St.-Germain G, Beauchesne D. Evaluation of the MicroScan Rapid Yeast Identification panel. J Clin Microbiol. 1991;29:2296–2299. doi: 10.1128/jcm.29.10.2296-2299.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Warren N G, Hazen K C. Candida, Cryptococcus, and other yeasts of medical importance. In: Murray P R, Baron E J, Pfaller M A, Tenover F C, Yolken R H, editors. Manual of clinical microbiology. 6th ed. Washington, D.C: American Society for Microbiology; 1995. pp. 723–737. [Google Scholar]