Comparative Performance of the RapID Yeast Plus System and the API 20C AUX Clinical Yeast System

Abstract

The performance of the RapID Yeast Plus System (Innovative Diagnostic Systems, Norcross, Ga.), a 4-h micropanel using single-substrate enzymatic test reactions, was compared with that of the API 20C AUX Clinical Yeast System (bioMerieux Vitek, Hazelwood, Mo.), a 48- to 72-h carbohydrate assimilation panel. Two hundred twenty-five yeasts, yeast-like fungi, and algae, comprising 28 species and including 30 isolates of Cryptococcus neoformans, an important pathogen not tested in appreciable numbers in other comparisons, were tested by both methods. On initial testing, 196 (87.1%) and 215 (95.6%) isolates were correctly identified by the RapID and API systems, respectively. Upon repeat testing, the number of correctly identified isolates increased to 220 (97.8%) for the RapID system and 223 (99.1%) for the API system. Reducing the turbidity of the test inoculum to that of a no. 3 McFarland turbidity standard, which is below that recommended by the manufacturer, resulted in the correct identification of most of the isolates initially misidentified by the RapID system, including 10 of 30 C. neoformans isolates. Concordance between the RapID and API results after repeat testing was 97.3%.

The RapID Yeast Plus System (RYP) (Innovative Diagnostic Systems, Norcross, Ga.) is a 4-h micropanel system that utilizes single-substrate test reactions based on preformed enzymes in tested isolates to identify yeasts. Reactions are determined by color changes of chromogenic substrates within the sample wells, some of which require addition of reagents. Test reactions are grouped in triads and scored on the basis of a positive or negative reaction, resulting in a 6-digit microcode which is matched to an empirically derived database of commonly occurring microcodes for identification of the isolate. Previous evaluations have demonstrated accuracies of 94 to 98% in identifying yeasts and a few yeast-like fungi to the species level (1, 2). While the range of species in the evaluation by Kitch et al. included some of the less-common yeasts, 238 of 304 fungi (78%) evaluated were Candida albicans, Candida parapsilosis, Candida tropicalis, or Candida glabrata, and 106 of 125 (85%) in the study by Heelan et al. were C. albicans, C. tropicalis, or C. glabrata (1, 2). Additionally, Kitch et al. tested only eight isolates and Heelan et al. tested only one isolate of Cryptococcus neoformans (1, 2). Cryptococcus neoformans, an important pathogen, causes life-threatening infections in approximately 6 to 10% of patients with AIDS in the United States and is a frequently isolated yeast in medical facilities serving this patient population (5). The purpose of the present study was twofold: first, to evaluate the RYP with an expanded battery of yeasts, yeast-like fungi, and yeast-like organisms; and second, to evaluate its performance in identifying an increased number of Cryptococcus neoformans isolates.

A total of 225 isolates of yeasts, yeast-like fungi, and algae encompassing 28 species were tested (Table 1). Isolates were obtained from the culture collection maintained at the Medical Mycology Research Center at the University of Texas Medical Branch, the University of Iowa, and the University of Texas Health Science Center at San Antonio. Cultures were subcultured at least twice onto Sabouraud-Emmons dextrose agar plates (Remel, Lenexa, Kans.) and incubated for at least 24 h but not longer than 48 h at 30°C prior to testing. Testing of each isolate was performed with both the RYP and the API 20C AUX system (bioMerieux Vitek, Hazelwood, Mo.) simultaneously. Identification procedures for each system were performed according to the manufacturer’s instructions, as described elsewhere (2, 3). With the API 20C AUX system, isolates were identified by using the second edition of the Analytical Profile Index and results were correlated with morphology on cornmeal-Tween 80 agar plates inoculated by the Dalmau method as recommended by the manufacturer. As no such recommendation is made by the manufacturer of the RYP, the RYP was evaluated as a stand-alone system. For this reason, presumptive or probability overlap identifications were considered inadequate for the purpose of this study. Discrepancies between the RYP and API 20C AUX results were resolved by confirming the purity of the isolate and by repeat testing with both systems. If results remained discrepant, disagreements were resolved by morphological examination or traditional biochemical tests, depending on the isolate (4).

TABLE 1.

Identification of yeasts and yeast-like organisms by the RYP and API systems

Species (n)	No. identified correctly by:
	RYP		API
	Initially	On repeat	Initially	On repeat
Candida albicans (20)	20	20	20	20
Candida krusei (14)	13	14	14	14
Candida lusitaniae (14)	12	14	12	14
Candida parapsilosis (20)	19	20	20	20
Candida tropicalis (22)	22	22	21	22
Candida glabrata (20)	19	20	20	20
Candida guilliermondii (9)	7	8	7	8
Candida stellatoidea (5)	5	5	5	5
Yarrowia lipolytica (2)	2	2	2	2
Candida kefyr (5)	5	5	5	5
Candida ciferrii (2)	2	2	2	2
Candida humicola (1)	0	0	1	1
Candida zeylanoides (3)	0	3	0	3
Cryptococcus neoformans (30)	20	30	30	30
Cryptococcus laurentii (5)	4	5	5	5
Cryptococcus albidus (1)	1	1	1	1
Saccharomyces cerevisiae (7)	6	7	7	7
Rhodotorula rubra (3)	2	3	2	3
Rhodotorula glutinis (1)	0	0	1	1
Rhodotorula minuta (2)	1	1	2	2
Blastoschizomyces capitatus (9)	9	9	9	9
Trichosporon beigelii (13)	12	13	13	13
Hansenula anomala (5)	5	5	5	5
Hanseniaspora uvarum (1)	0	1	1	1
Sporobolomyces salmonicolor (1)	1	1	0	0
Torulopsis candida (6)	5	5	6	6
Prototheca zopfii (2)	2	2	2	2
Prototheca wickerhamii (2)	2	2	2
Total (225)	196	220	215	223
Accuracy	87.1%	97.8%	95.6%	99.1%

On initial testing of isolates, the accuracy of the RYP was 87.1%, with 196 of 225 isolates being identified correctly (Table 1). Of the 29 isolates incorrectly or inadequately identified, one C. krusei isolate was identified as presumptive while the remainder of the isolates coded as either no identification or an incorrect species. The three most frequently misidentified isolates were Cryptococcus neoformans (10 of 30 isolates incorrectly identified), Candida zeylanoides (3 of 3 isolates misidentified), and Candida lusitaniae (2 of 14 isolates incorrectly identified). Of the 10 incorrectly identified Cryptococcus neoformans isolates, 2 were identified as Rhodotorula minuta; 2 coded as probability overlap between Cryptococcus neoformans and R. minuta; 4 coded as no identification; 1 coded as probability overlap between Cryptococcus neoformans, Saccharomyces cerevisiae, Candida kefyr (C. pseudotropicalis), and Candida krusei; and 1 was identified as S. cerevisiae. Two of the C. zeylanoides isolates were identified as C. parapsilosis, and one coded as probability overlap among Rhodotorula glutinis, Rhodotorula rubra, C. parapsilosis, and C. zeylanoides. Of the two C. lusitaniae isolates, one coded as no identification and 1 coded as probability overlap between Torulopsis candida and C. lusitaniae. Upon repeat testing, the RYP correctly identified 24 of the 29 isolates, including all Cryptococcus neoformans isolates. The five isolates not identified correctly, followed in parentheses by the identification(s) provided by the RYP (initial and repeat), included one Candida guilliermondii isolate (no identification; C. parapsilosis), one Candida humicola isolate (Cryptococcus terreus; Trichosporan beigelii), one R. glutinis isolate (C. guilliermondii; R. minuta), one R. minuta isolate (no identification both times), and one Torulopsis candida isolate (Cryptococcus neoformans both times). After repeat testing, the overall accuracy of the RYP was 97.8%.

By way of comparison, the API 20C AUX system showed an accuracy of 95.6% on initial testing. Ten isolates were not correctly identified. The most frequently misidentified species were two C. lusitaniae isolates (no identification; C. albicans) and three C. zeylanoides strains (C. paratropicalis; no identification on two isolates). On retesting, only two isolates remained incorrectly identified: one Sporobolomyces salmonicolor isolate (no identification twice) and one C. guilliermondii isolate (Candida stellatoidea; C. parapsilosis). After repeat testing, the overall accuracy of the API panel was 99.1%.

On initial testing, the RYP identified only two-thirds (20 of 30) of Cryptococcus neoformans isolates. A similar 4-h identification system based on preformed enzymes, the Microscan Yeast Identification Panel (Baxter-Microscan, W. Sacramento, Calif.), showed similar problems in identifying this organism in one evaluation (3). The authors believed that the problem might have been related to an insufficient database at the time the study was done, and in a subsequent study at a different institution, the Microscan panel correctly identified all isolates of Cryptococcus neoformans (6). In previous evaluations of the RYP, problems in identifying Cryptococcus neoformans were not experienced; however, only eight isolates were included in the study by Kitch et al., and only one isolate was tested by Heelan et al. (1, 2). Aberrant reactions resulting in improper or inadequate identifications in this study were felt to be related to inoculum strength. The manufacturer’s directions state that colonies of the test isolate should be added to the RapID inoculation fluid until the black lines on the provided inoculation card are no longer visible. When this was done, subsequent reactions in the test wells were often aberrant. When the test was repeated on these isolates with an inoculum of a density corresponding to a no. 3 McFarland turbidity standard (with which the black lines on the inoculation card are still discernible), an accuracy of 97.8% was obtained. Similar problems with frequent aberrant reactions were not experienced in the study by Kitch et al.; however, a no. 3 McFarland turbidity standard was utilized to prepare inocula in their study, rather than the manufacturer’s recommended method (2). In the evaluation by Heelan et al., the manufacturer’s directions were followed, and of the five yeast species misidentified by the system, four (C. lusitaniae, C. parapsilosis, R. rubra, and Trichosporan beigelii) were also among the isolates initially incorrectly identified in our study, with the correct identification obtained upon reducing the inoculum density to that of a no. 3 McFarland turbidity standard (1).

In summary, the RYP results correlate well with those obtained by using the API 20C AUX system. Concordance between the two systems in this study was 97.3%. The RYP is inoculum sensitive, however, and some species, particularly Cryptococcus neoformans, appear more susceptible to aberrant reactions with overinoculation than others. Utilization of an inoculum density familiar to microbiologists, corresponding to the no. 3 McFarland turbidity standard, rather than the procedure described by the manufacturer appears to reduce misidentifications due to overinoculation. With this modification, the RYP is useful for the identification of yeasts, some yeast-like fungi, and certain achlorophyllous algae from culture in 4 h.