Evaluation of the New API 20C Strip for Yeast Identification

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Sept. 1979, p /79/ /08$02.00/0 Vol. 10, No. 3 Evaluation of the New API 20C Strip for Yeast Identification Against a Conventional Method G. A. LAND,t* B. A. HARRISON,' K. L. HULME,1 B. H. COOPER,2 AND J. C. BYRD3 Department ofmicrobiology, Wadley Institutes ofmolecular Medicine, Dallas, Texas ; Department of Microbiology, Baylor University Medical Center, Dallas, Texas ; and Department ofmicrobiology, St. Paul Hospital, Dallas, Texas Received for publication 1 July 1979 The new API 20C yeast identification system together with appropriate microscopic morphology determinations achieved a 97% correlation with a rapid conventional method. Whereas a group composed of Candida, Torulopsis, Saccharomyces, and Rhodotorula was identified with ease (98% overall correlation), a second group, containing Cryptococcus, Trichosporon, and Geotrichum species, appeared to give the system the most difficulty (90% correlation). Within this group particular difficulty was encountered in identifying varieties of Cryptococcus albidus, C. terreus, C. laurentii, Trichosporon beigelli, and Geotrichum spp. as to species. The API 20C system should be incubated the full 72 h prescribed by the manufacturer. However, when used in conjunction with appropriate morphological tests, presumptive identifications of some Candida and Torulopsis species may be made at 24 to 48 h. To facilitate identifications of the more difficult group of yeasts, ancillary tests for determining nitrate reductase, urease, and phenol oxidase activities should be considered as additions to the strip. Incorporating the phenol oxidase test would be especially important for identification of Cryptococcus neoformans, a yeast which should be identified as quickly and as accurately as possible. The API 20C system with computer assistance has proved to be an easy-to-inoculate, versatile, and fairly rapid method of yeast identification, giving results comparable to those obtained by conventional methodologies. The availability within the past several years that accompany their products. As reported in of commercial products which aid in the acquisition and interpretation of data for identifica- with the use of these commercial systems to recent studies (4, 5, 7, 15, 20), it is now possible tion of medically important yeasts has rendered reliably identify most medically important the task of obtaining this information much less yeasts within 48 to 72 h from the time biochemical tests are inoculated. In contrast, the more demanding than it once was (7). The majority of the commercial products currently available traditional methods require a maximum of 14 provide carbohydrate assimilation tests in a convenient plate or strip form. Some products in- The original API 20C yeast identification strip days for completion (1, 3, 9). corporate carbohydrate assimilation as well as (Analytab Products, Div. ofayerst Laboratories, other biochemical tests, and these products Plainview, N.Y.), which was one of the first eliminate the necessity for preparing test media commercial products to be introduced for testing and simplify the storage of the large variety of medically important yeasts, provided media in media required for identifying yeast isolates (15). dehydrated form for testing both carbohydrate Although based on traditional methodology, the fermentation and carbohydrate assimilation. miniaturization of biochemical tests in these These properties, when used in combination commercial kits permits the reading of results with morphological characteristics, permitted after a shorter period of incubation than was identification of many yeast species with a respectable level of reliability (5, 15, 20). However, feasible with the earlier conventional methods (2, 13, 18, 19). Most manufacturers recognize the certain aspects of the system were less convenient than was desired, and the data base which necessity for conducting morphological examinations along with biochemical testing and recommend such procedures in the instructions including only a few isolates of certain species of had been used to develop the system was limited, t Present address: Division of Laboratory Medicine, University of Cincinnati Medical Center, Cincinnati, OH ratories. The main technical difficulty in using yeasts commonly encountered in clinical labo- 357

2 358 LAND ET AL. the system involved the complexities of filling the microtubules without trapping air bubbles in the agar, which subsequently led to misinterpretation of fermentation tests (5). To circumvent these relatively minor disadvantages and to improve the overall capabilities of the system, a new generation of the API 20C was introduced in early The new design incorporated the following modifications of the original system: (i) the use of assimilation tests only; (ii) larger cupules for easier filling; (iii) addition of new substrates and elimination of those that had been shown after extensive development evaluation to provide imprecise differentiations of species based on computer-assisted interpretation of test data (14); and, most importantly, (iv) clinical evaluation of the system with an extensive data base consisting of at least 20, and in some cases over 100, isolates of each of 16 yeast taxa. With this new system, a binomial profile number can be derived from growth patterns on 19 substrates. This profile number provides a convenient method for comparing unknown yeast isolates with those numerical profiles in the data base in order to derive an identification of the unknown (14). For biotypes that are not found in the profile index, a phone-in computer service is available for determining a likely identification of unusual isolates. The following is a report of an evaluation of the revised API 20C, using clinical isolates derived from the diagnostic mycology laboratories of three hospitals along with stock isolates of selected species. The purpose of the study was to ascertain the level of reliability of the API 20C system when compared with a rapid conventional method (RCM) for yeast identification (11). MATERIALS AND METHODS Microorganisms. All of the yeasts used in this study were either clinical laboratory isolates or stock cultures from the following sources: the Wadley Institutes of Molecular Medicine, the Baylor University Medical Center, and the St. Paul Hospital, Dallas, Tex., and the University of Oklahoma at Norman. The following American Type Culture Collection isolates were also used: ATCC 26310, Candida albicans; ATCC 24064, Cryptococcus neoformans; ATCC 16725, Rhodotorula glutinis; ATCC 10663, Trichosporon capitatum; and ATCC 9331, Trichosporon pullulans. Other isolates serving as known positive controls for identification were proficiency testing samples from the New York City Department of Health, College of American Pathologists, Skokie, Ill., and center for Disease Control (Atlanta, Ga.) proficiency testing programs. All clinical isolates were identified by one laboratory and then coded, randomized, and distributed to the other participating laboratories for identification and comparison. To complete the double-blind J. CLIN. MICROBIOL. study, all stock cultures and other strains serving as positive controls were treated in the same manner. API 20C yeast identification system. The tip of a wooden applicator stick was used to pick up a portion of a small colony of each unknown yeast to inoculate molten (42 C) API 20C basal medium, and cupules were filled as per the manufacturer's directions. Also in accordance with the manufacturer's directions, yeasts were inoculated onto cornmeal agar plates via the Dalmau culture technique for morphological determinations. Both morphology and assimilation tests were incubated at 30 C, and results were recorded in the manner suggested by the API package insert. Upon the final observation of the API system, a profile number was assigned to each isolate and compared with profile numbers listed in the provided quick index. Occasionally, one numerical profile was assigned to two or even three species, in which case both morphology and ancillary biochemical tests (i.e., nitrate reductase, fermentations, and urease production) suggested by API were used to determine the final identification. In those cases where an organism produced a profile that was not listed in the API profile register, the API supplementary computer identification system was utilized. Computer-assisted presumptive identification was dependent upon the calculation of likelihood of occurrence between the unknown yeast's biochemical characteristics and isolates with similar characteristics present in the computer's data bank. RCM. The RCM system of yeast identification, as previously described (11), consisted of four biochemical tests: a dye pour plate auxanogram, Tween-80- oxgall-caffeic acid (TOC) medium, a 10-min swab nitrate test, and a 4-h urease test with urea R broth (Difco Laboratories, Detroit, Mich.). Briefly, the tests were done as follows: a modified DPPA medium (10-12, 16), for determining yeast assimilation patterns on 14 different carbohydrates, was composed of (per liter of water): 20 g of agar, 0.67 g of yeast nitrogen base, and 20 mg of bromocresol purple. Ingredients were then solubilized by heating, and the ph of the molten solution was adjusted to 7.2 and sterilized. Sterile, molten dye agar medium was aseptically dispensed in 60-ml portions into sterile prescription bottles and stored at 4 C until required. For assimilation testing, dye pour plate auxanogram medium in two bottles was melted, cooled to 40 to 43 C, inoculated with 5 ml of a MacFarland no. 5 suspension of yeasts, poured into petri plates (150 by 15 mm), and cooled. Individual stock solutions of 14 carbohydrates were made in normal saline (ph 7) at concentrations which would dispense in 0.1 ml that amount of carbohydrate necessary for its optimum assimilation by yeasts. Stock solutions were filter (0.45 [Lm; Millipore Corp., Bedford, Mass.) sterilized, and 0.1 ml of each was placed individually on a sterile 0.5-inch (ca. 1.3-cm) concentration disk. The following carbohydrates were arranged individually (7 per petri plate) on inoculated and solidified dye medium as follows: plate 1, dextrose, galactose, sorbose, sucrose, maltose, cellobiose, and trehalose; plate 2, lactose, melibiose, raffinose, melezitose, xylose, dulcitol, and inositol. Observations were made after 24 and 48 h of incubation at 25'C, with a positive test recorded as either a reduction of the

3 VOL. 10, 1979 purple dye to yellow or, if the plate had completely reduced to yellow, appearance of growth around the carbohydrate disk. Microscopic observation of pseudohyphae, hyphae, arthrospores, and blastospores were made on TOC plates. TOC medium has also been reported to promote germ tubes and chlamydospores for appropriate Candida species, in addition to the above morphological characteristics (8). Caffeic acid added to the medium served as a substrate for the phenol oxidase reaction, an enzymatic reaction presumably characteristic of C. neoformans (17). The TOC medium contained 10 g of oxgall (Difco), 20 g of Davis agar, 0.3 g of caffeic acid, and 1 ml of Tween-80 brought to a boil in 1,000 ml of distilled water. The solubilized components were then autoclaved, and plates containing 30 ml of medium were poured. Dried plates were streaked with a sterile swab, depositing a heavy inoculum on one corner of the plate and then continuing lightly with the characteristic Dalmau technique, finally overlaying the lightest streak with a sterile cover slip. The heaviest streak was used to rapidly detect the characteristic brown pigment of C. neoformans, whereas the lighter streak was used to monitor specific morphological changes in the yeasts as they grew. Once inoculated, plates were incubated at 37 C for 3 h, after which they were inspected for germ tube production and pigmentation, whereupon they were incubated further at 25 C. The TOC plates were subsequently observed at 6, 24, and 48 h for either brown pigmentation, chlamydospore formation, or other morphological changes. The presence of nitrate reductase in yeasts was determined by swabs saturated with a fivefold-concentrated liquid medium (ph 5.8 to 6.0) containing 2 g of KNO3, 11.7 g of NaH2PO4, 1.14 g of Na2HPO4, and 1.2 ml of a 17% solution of Zephiran chloride in 200 ml of water (11). Dried swabs were inoculated by sweeping them across several colonies of a plate and then swirling them against the bottom of an empty test tube (13 by 150 mm) to ensure contact between organisms and substrate. The tube and swab were incubated for 10 min at 45 C, and the swab was then placed in a second tube containing two drops each of 0.5% ca-naphthyla- API 20C STRIP EVALUATION 359 mine and 0.8% sulfanilic acid, each in 5 N acetic acid. A positive test was indicated by the swab tip turning bright cherry red. For the urease test, one vial of urea R broth was reconstituted with distilled water on the day it was to be used. A 0.2-ml amount of the resultant liquid was dispensed into each well of a 96-well microtiter test plate (Microtiter II; Falcon Plastics, Oxnard, Calif.). A heavy inoculum (3 to 4 small colonies) of the unknown yeast was transferred via a wooden applicator tip to the microtiter wells containing urea R broth. Wells were sealed with clear sealing tape and incubated for 4 h at 37 C. Observations were made hourly, with any change of the straw-colored medium to pink considered a positive test (11). RESULTS The overall results obtained with the new API 20C correlated very closely with the RCM. There were 1,063 positive identifications out of 1,093 total cultures (97%), using the API biochemical tests in conjunction with morphology and a nitrate reductase test (Table 1). Ninetytwo percent of these yeasts were compatible with profiles appearing in the system's profile index; the remaining profiles (5%) were similar to profiles stored in the computer's data base. Using the biochemical tests alone, 75% of the isolates could be correctly identified with the commercial system. The average time to positivity for Candida, Torulopsis, Saccharomyces, and Rhodotorula species (group 1 yeasts) was 31 h, as opposed to 72 h for a more difficult group (group 2), composed of Cryptococcus, Trichosporon, and Geotrichum species. C. albicans and Torulopsis glabrata could routinely be identified within 24 h by using the API strip in conjunction with appropriate morphological changes. Approximately 5% of the yeasts surveyed required computer assistance for identification, and 3% were not identifiable by API 20C. TABLE 1. Correlation ofapi 20C with RCM in the identification of 1,093 yeasts from six medically important genera Correlation' Avg time to positive Identification method No. of isolates B B/M/NO3 API (h) RCM (h) Quick index Group 1b (77) 677 (96) Group 2C (71) 337 (87) Total 1, (75) 1,014 (92) Computer aided Group 1b 21 (2.0) Group 2C 28 (2.6) Overall correlation 1, (75) 1,063 (97) a The numerical and percent (in parentheses) correlations of API 20C with RCM by biochemical tests only (B) or by biochemistry, morphology, and nitrate (B/M/NO3). b Isolates representing Candida, Torulopsis, Saccharomyces, and Rhodotorula species. c Isolates representing Cryptococcus, Trichosporon, and Geotrichum species.

4 360 LAND ET AL. In the clinical portion of this study, 98% of group 1 yeasts were identified by the combination of API, morphology, and nitrate reductase, whereas 86% of the organisms could be identified on the basis of assimilation results alone (Table 2). One isolate of Candida solani and four of eight isolates of Saccharomyces cerevisiae, whose profiles were not in the data base, were particularly difficult to identify with the API system. Group 2 yeasts identified with the three combined tests also had a 98% correlation rate with RCM (Table 3). Morphology was extremely TABLE 2. Clinical comparison of the API 20C yeast identification system versus RCM: group 1 (Candida, Torulopsis, Saccharomyces, and Rhodotorula) Organism No. of iso- Correlation' lates B B/M/NO3 C. albicans C. tropicalis C. parapsilosis C. krusei C. stellatoidea C. guilliermondii C. pseudotropicalis C. lipolytica C. solani 1 0 ob T. glabrata T. candida S. cerevisiae b R. glutinis R. rubra Avg apercent correlation of API 20C with RCM by biochemical tests only (B) or by biochemistry, morphology, and nitrate (B/M/NO3). b Some biotypes were not in the computer. TABLE 3. Clinical comparison of the API 20C yeast identification system versus RCM: group 2 (Cryptococcus, Trichosporon, and Geotrichum) Correlation' No. of iso- Organism lates lts B B/M/NO3 C. neoformans C. albidus var. al bidus C. albidus var diffluens C. laurentii T. beigelii T. capitatum G. candidum Avg a Percent correlation of API 20C with RCM by biochemical tests only (B) or by biochemistry, morphology, and nitrate (B/M/NO3). important in the identification of these organisms, since only 74% could be determined by biochemical tests alone. One clinical isolate of Cryptococcus laurentii could not be identified by both the combined tests and computer assistance. Results obtained in the clinical study suggested that the data obtained for the API strip were often insufficient for separating Candida krusei, Candida lipolytica, and T. capitatum as well as in identifying Cryptococcus, Trichosporon, and Geotrichum species. The stock culture portion of the study was weighted in favor of these organisms in order to provide a severe test of the capabilities of the API system. Among those yeasts in group 1 of the stock culture study, Candida stellatoidea proved to be the most difficult species for the API system to identify consistently (Table 4). Approximately 58% of the isolates utilized the API trehalose; however, according to the API percentage chart, only 1% should have reacted positively. The RCM trehalose, on the other hand, was assimilated by all isolates of C. stellatoidea. Moreover, some 13% of these were negative on API maltose, whereas according to the percent chart and RCM all isolates should have grown on the substrate. These discrepancies gave the API system a 51% overall efficiency rating for the identification of C. stellatoidea. Several organisms were found which had no correlating profile in the API data base including: 4 of 15 isolates of S. cerevisiae, 2 of 11 isolates of R. glutinis, 1 of 2 isolates of R. rubra, 3 of 80 isolates of C. TABLE 4. Comparison of identification of stock cultures by the API 20C system with RCM: group 1 (Candida, Torulopsis, Saccharomyces, and Rhodotorula) Organism No. of ioae J. CLIN. MICROBIOL. Correlation' isolates B B/M/NO3 C. albicans C. parapsilosis C. tropicalis C. krusei C. stellatoidea C. guilliermondii C. pseudotropicalis C. lipolytica T. glabrata T. candida S. cerevisiae R. glutinis R. rubra Avg apercent correlation of API 20C with RCM by biochemical tests only (B) or by biochemistry, morphology, and nitrate (B/M/NO3).

5 VOL. 10, 1979 albicans, and 3 of 44 isolates of Candida parapsilosis. These biochemically variable isolates of C. albicans and C. parapsilosis also exhibited aberrant morphologies, making their identifications by either system difficult. In addition, it was difficult to identify as to species stock cultures belonging to group 2 with the API yeast system. Seventy-one percent of this group were identified on the basis of biochemical tests alone, and 86% were identified by using the three combined tests (Table 5). Cryptococcus terreus and Geotrichum candidum were the most difficult isolates to identify. Isolates of C. terreus varied in ability to assimilate inositol, with only 57% becoming positive after 96 h of incubation. 2-Ketogluconate, a substrate utilized by the same yeasts that metabolize inositol, also demonstrated the same variability in assimilation as did inositol, further compounding the problems in identifying these yeasts. Geotrichum species did not regularly assimilate API glycerol and xylose, making their identification also difficult. The problems experienced with the API system in identifying other group 2 organisms also related to false negative assimilations of inositol or other key substrates. For example, 3 out of 164 isolates of C. neoformans API 20C STRIP EVALUATION 361 did not assimilate inositol after 96 h of incubation, and 10 of 81 Cryptococcus albidus var. albidus and 5 of 26 C. laurentii (Table 6) also failed to assimilate this carbohydrate. Twelve of TABLE 5. Comparison of the identification of stock cultures by the API 20C system with RCM: group 2 (Cryptococcus, Trichosporon, and Geotrichum) No. of Correlationa Organism isolates B R/M/NO3 C. neoformans C. albidus var. al bidus C. albidus var. dif fluens C. laurentii C. terreus C. uniguttulatus T. beigelii T. capitatum T. penicillatum G. candidum Avg apercent correlation of API 20C with RCM by biochemical tests only (B) or by biochemistry, morphology, and nitrate (B/M/NO3). TABLE 6. Comparison of selected positive assimilations in the API 20C system with RCM Assimilation (% positive) Organism No. of isolates Substratea Predicted value API RCM C. albicans 144 Mlz Gly 9 0 b Ara C. parapsilosis 72 Miz Gly Ara C. tropicalis 160 Mlz Gly Ara Cel C. stellatoidea 30 Tre C. neoformansc 164 Ins Ara Xlt C. laurentiic 26 Ins Ara Xlt C. albidus var. albidusc 81 Ins Ara Xlt a Mlz, Melezitose; Gly, glycerol; Ara, arabinose; Cel, cellobiose; Tre, trehalose; Ins, inositol; Xlt, xylitol. b_, Substrate not present in the RCM protocol. 'A number of the cryptococci had to be incubated for an additional 24 h (i.e., 96-h total) in order for the inositol reaction to become positive.

6 362 LAND ET AL. J. CLIN. MICROBIOL. TABLE 7. Yeasts identifiable by RCM which generated profile numbers not compatible with the API data base No. of Comments Organism iolate API profile no. Isolates Candida humicoli Same profile as C. laurentii, but different morphology Candida solani Candida utilis Saccharomyces chevalieri Same assimilation profile as S. cerevisiae with API, but separable by RCM Saccharomyces champagni Same as S. chevalieri Kluveromyces fragilis Same profile as Candida pseudotropicalis with API, but ascospore positive and separable by RCM Kluveromyces bulgaricus Same as Kluveromyces fragilis Kluveromyces lactis Ascospore positive Pichia ohmeri Ascospore positive Rhodotorula Rhodotorula glutinis Nitrate reductase positive Aureobasidium sp May at first be white and have same profile as C laurentii and T. beigelii, but becomes dematiaceous with characteristic morphology upon aging Ustilago sp isolates of C. albidus var. albidus, which were lactose positive by RCM, were negative on the corresponding API substrate and were for that reason inseparable from their sibling species C. albidus var. diffluens. Of those organisms which were not identified by the API 20C system, 50% were ascosporogenous yeasts (Table 7). Three isolates identified as S. cerevisiae by API 20C were termed S. chevalierii by the RCM. These organisms had similar assimilation profiles but differed in their fermentation patterns. Kluveromyces fragilis (four strains) and K. bulgaricus (three strains) had the same assimilation profiles as and similar morphology to Candida pseudotropicalis, but they differed in that they formed ascospores. Candida humicola generated profiles similar to those of both C. laurentii and Trichosporon beigellii, so identification relied heavily upon a critical evaluation of morphology, regardless of the system used. Another yeastlike organism which was confused with C. laurentii was Aureobasidium species. These isolates appeared in early culture as a white yeast with biochemical properties identical to C. laurentii or T. beigelii, but after extended culture they formed both hyaline and dematiaceous hyphae. DISCUSSION The new API 20C system had a high degree of correlation with conventional methodology, providing that adjunctive tests of morphology and nitrate reductase were used. The latter test was especially important in identifying Cryptococcus and Rhodotorula species. Pinello et al. have placed special emphasis upon showing that morphological examinations in conjunction with the API biochemical tests are necessary for complete yeast identification (14). This necessity for morphological examination in yeast identification was again underscored in our study by the fact that only 75% of these yeasts could be identified on the basis of biochemical activity alone. To emphasize this point, the data have been presented both with and without morphological characteristics being taken into consideration. The overall correlation of API with conventional methodology of 97% was in agreement with what Buesching et al. found (96%) in evaluating the system against 505 organisms (6). Paralleling our experience, Buesching et al. also found that fresh isolates appeared to grow more rapidly and to give fewer ambiguous reactions than did stock cultures. There are two clusters of medically important yeasts which are of particular importance, and, for this reason, they must be rapidly and accurately identified. The first cluster consists of C. albicans, C. parapsilosis, and Candida tropicalis, since it is possible for them to exhibit similar morphologies and assimilation patterns on traditional media and substrates. The API system is designed to separate members of this group by their respective utilizations of melezitose, glycerol, and arabinose. Melezitose, according to the API data base, is assimilated by virtually all isolates of C. tropicalis and C. parapsilosis but by 2% of C. albicans isolates. We found that the degree of melezitose positivity among C. albicans was much higher than the

7 VOL. 10, 1979 predicted percentage and would have led to an identification of C. tropicalis or C. parapsilosis, especially for those isolates which did not form chlamydospores (Table 6). Glycerol, a substrate used for delineating C. parapsilosis from C. albicans and C. tropicalis, yielded similar inconsistencies. Glycerol was assimilated by virtually all C. parapsilosis isolates, but not by C. albicans, whereas over one-third of the C. tropicalis isolates were positive, instead of the expected 11%. It appeared that arabinose assimilation was not as variable a characteristic as those above and served as an excellent means of separating C. parapsilosis from C. albicans and C. tropicalis. The addition of a germ tube test to API 20C would provide a good backup test for cornmeal- Tween-80 agar morphology and would also help to split C. albicans away from C. parapsilosis and C. tropicalis (7, 17). This additional morphological test in tandem with melezitose, glycerol, and arabinose assimilations would help to remove some of the ambiguity in relying upon assimilation entirely as a means of identification. Cellobiose assimilation has been shown to be an efficient means of separating C. tropicalis from C. albicans and C. parapsilosis (4-6, 11), and it could also augment the other key API substrates if the problem of its variable assimilation could be overcome. The assimilation of cellobiose as well as other carbohydrates by yeasts has been shown to be a function of a specific concentration range for each substrate and of the nitrogen content of the medium (12). The failure to utilize the optimum concentration for each substrate as well as the relatively high nitrogen content of the medium might also explain the failure of several other yeasts to grow on substrates that should have been assimilated. Trehalose, commonly metabolized by C. stellatoidea in other systems and used by all of these isolates on RCM, had a 50% false negative rate on API. Furthermore, inositol, a key sugar in the conventional identification of Cryptococcus species (3, 11, 13, 17), also was not utilized by 100% of cryptococci grown on API medium. Particularly noteworthy was the fact that only 26% of C. terreus isolates utilized this carbohydrate. Among the cryptococci, inositol assimilations were so weak that they had to be held an extra day to be considered positive, and 10 to 15% didn't grow at all. 2-Ketogluconate, a backup test for inositol assimilation, had a similar 10 to 15% false negative rate instead of the near 100% positive rate expected from the percentage chart. These false negative assiimilations led to some difficulty in identifying the various cryptococci, an experience also noted by Buesch- API 20C STRIP EVALUATION 363 ing et al. (6). The second cluster of yeasts of medical interest are composed of C. neoformans, C. laurentii, and C. albidus. These species appear to be fairly separable by the API system. Xylitol appears to be an adequate substrate for delineating C. laurentii from C. neoformans and C. albidus, with 100% of C. laurentii utilizing the substrate whereas only about 14% of the other cryptococci were positive. Arabinose was assimilated by 12% of the C. neoformans isolates in this study rather than the predicted 0%. This would still be a fair characteristic for identification, since 95% of the C. laurentii and 80% of the C. albidus metabolized the substrate. However, due to the medical importance of determining the presence of C. neoformans in a clinical specimen, we feel an adjunctive test for phenol oxidase activity should be provided with the kit. This test is presumably specific for the identification of C. neoformans and could easily be adapted to the API 20C (8, 11, 17). Based upon our experience with the new API 20C yeast identification system, we conclude the following: this identification system, together with the recommended morphological tests, correlates well with conventional methodology. However, some of the biochemical tests chosen by the computer as a means of separating certain taxa and the computer's use of these data differ considerably from conventional tests and dichotomies but may, with time, prove equal to conventional methods or perhaps even to be a more accurate approach to the identification of medically important yeasts. ACNOWLEDGMENT This work was supported in part by the Sammons Foundation, Dallas, Tex. LITERATURE CITED 1. Adams, E. D., Jr., and B. H. Cooper Evaluation of a modified Wickerham medium for identifying medically important yeasts. Am. J. Med. Technol. 40: Ahearn, D. G Systematics of yeasts of medical interest. Pan Am. Health Organ. Sci. Publ. 205: Ahearn, D. G Identification and ecology of yeasts of medical importance, p In J. E. Prier and H. Friedman (ed.), Opportunistic pathogens. University Park Press, Baltimore. 4. Bowman, P. I., and D. G. Ahearn Evaluation of the Uni-Yeast-Tek kit for the identification of medically important yeasts. J. Clin. Microbiol. 2: Bowman, P. I., and D. G. Ahearn Evaluation of commercial systems for the identification of clinical yeast isolates. J. Clin. Microbiol. 4: Buesching, W. J., K. Kurek, and G. D. Roberts Evaluation of the modified API 20C system for identification of clinically important yeasts. J. Clin. Microbiol. 9: Cooper, B. H., J. B. Johnson, and E. S. Thaxton Clinical evaluation of the Uni-Yeast-Tek system for

8 364 LAND ET AL. J. CLIN. MICROBIOL. rapid presumptive identification of medically important yeasts. J. Clin. Microbiol. 7: Fleming, W. H., HI, J. M. Hopkins, and G. A. Land New culture medium for the presumptive identification of Candida albicans and Cryptococcus neoformans. J. Clin. Microbiol. 5: Haley, L. D Identification of yeasts in clinical microbiology laboratories. Am. J. Med. I'echnol. 37: Huppert, M., G. Harper, S. H. Sun, and V. Delanerolle Rapid methods for identification of yeasts. J. Clin. Microbiol. 2: Land, G. A., G. L. Dorn, W. H. Fleming HI, T. A. Beadles, and J. H. Foxworth Isolation and rapid identification of yeasts from compromised hosts. Mycopathologia 65: Land, G. A., E. C. Vinton, G. B. Adcock, and J. M. Hopkins Improved auxanographic method for yeast assimilations: a comparison with other approaches. J. Clin. Microbiol. 2: Lodder, J. (ed.) The yeasts. A taxonomic study, 2nd ed. North-Holland Publishing Co., Amsterdam. 14. Pinello, C. B., P. J. Naudo, and R. F. D'Amato Development of an interpretative system for the identification of yeasts. Species 2: Roberts, G. D., H. S. Wang, and G. E. Hollick Evaluation of the API 20C microtube system for the identification of clinically important yeasts. J. Clin. Microbiol. 3: Segal, E., and L. Ajello Evaluation of a new system for the rapid identification of clinically important yeasts. J. Clin. Microbiol. 4: Silva-Hutner, M., and B. H. Cooper Medically important yeasts, p In E. H. Lennette, E. H. Sapulding, and J. P. Truant (ed.), Manual of clinical microbiology, 2nd ed. American Society for Microbiology, Washington, D.C. 18. van der Walt, J. P Criteria and methods used in classification, p In J. Lodder (ed.), The yeasts. A taxonomic study, 2nd ed. North-Holland Publishing Co., Amsterdam. 19. Wickerham, L. J Taxonomy of yeasts. Technical Bulletin no U.S. Department of Agriculture, Washington, D.C. 20. Zwadyk, P., Jr., R. A. Tarlton, and A. Proctor Evaluation of the API 20C for identification of yeasts. Am. J. Clin. Pathol. 67: Downloaded from on January 28, 2019 by guest