Evaluation of the modified API 20C system for identification of clinically important yeasts

JOURNAL OF CLINICALMICROBIOLOGY, May1979,p.565-569 0095-1137/79/05-0565/05$02.00/0

Vol. 9, No. 5

Evaluations of the Modified API 20C

System for Identification

of Clinically

Important Yeasts

WILLIAM J. BUESCHING, KATHRYN KUREK, ANDGLENN D. ROBERTS*

Sectionof Clinical Microbiology,MayoClinicand Mayo Foundation, Rochester,Minnesota55901

Receivedfor publication9February1979

The modified API 20Csystem (Analytab Products, Inc.) containing 19

carbo-hydrate assimilation testswasusedtoidentify stock culturesof clinical isolates

and routine clinical isolates from the Mayo Clinic mycology laboratory. The

systemprovided correctidentificationsfor 96% of the505organisms tested. The

API 20C represents a commercial system useful for the identification ofyeasts

fromclinical specimens. Although reliable, it isnotacompletesystemandmust

be used in conjunction with microscopic morphological features for definitive

identification. Since thesystemrequires 72 h for identification, it isnotdesigned

for the rapid presumptive identification of such organisms as Cryptococcus

neoformans; other biochemicaltestsmustbe usedfor thispurpose.

In recent years, the number of serious yeast

infections, particularly in the

immunocompro-mised patient (5), has increased significantly.

This situationhasnecessitatedthat clinical lab-oratoriesbecome moreproficientintheirabiity

to isolate and rapidly identify yeasts of medical

importance. Traditional procedures for yeast

identification (9) are considered bymany tobe technicallycomplexand time consuming. Also,

many reagentsrequired for identificationof the

yeasts tospeciesare notcommerciallyavailable.

In1975, the API20C microtube system, a

com-mercial product (Analytab Products, Inc.) for

the identification of yeasts,wasintroduced. This

system miniaturized the traditional procedures

of carbohydrate fermentations and assimila-tions, permitting identification ofyeasts within 72h after inoculation (2, 8, 11). Recently, this

system has

undergone

extensive modification. Thecarbohydratefermentationtestshavebeen eliminatedandreplaced byadditional

assimila-tion tests. In addition, a numerical approach

witha computer-assistedsystem for the

identi-fication ofonly clinically important yeasts has beenintroduced.Itwasthepurpose of this

study

to determine whether the modified API 20C

systemprovidesareliable andaccuratemethod

for the identification of medically important

yeasts.

MATERIALS AND METHODS

SincetheAPI 20Csystemwasdevelopedtoprovide identifications foronlyclinicallyimportantyeasts, 505 isolates ofyeastsandyeastlike organismsrecovered from clinicalspecimenswereusedin thisstudy.Two hundred fifty-one isolates were selected from the MayoClinic stock culturecollectionand codedsothat theidentitywasunknownuntilalldatawereanalyzed.

Theremaining254 isolateswererecovered from res-piratory secretions by the MayoClinic mycology lab-oratory and all were identified within 2 to 10 days. The identification of eachorganism wasmade using methods previously described by Roberts (6) and Konemanet al. (4). All organisms used in the study were identified by the conventional method used in thislaboratory and by the API 20C yeast systemso

thatacomparison could be made between the identi-fications given by both methods. No attempt was madetocomparethemethodsonasubstrateto sub-stratebasis.

Conventional method. The conventional method used by the Mayo Clinicmycology laboratory is de-scribedindetailby Roberts (6) and by Konemanetal. (4).Identificationswerebasedongerm tubeformation,

microscopic morphologyoncornmeal-Tween80agar,

ureaseproduction,nitratereduction,temperature tol-erance, pigment production on niger seed agar, the

caffeicacid-ferric citrate reaction, and carbohydrate

assimilation and fermentation.

Anauxanographic method (4) was used to

deter-mineassimilation ofglucose,maltose,sucrose,lactose, raffmiose,trehalose,galactose,melibiose,and inositol. Asuspension ofyeast insterile salineequivalenttoa McFarland standardno. 4 wasusedtoflood thesurface ofa100-by15-mmpetridishcontainingyeastnitrogen

base (Difco) and 0.4% bromocresol purple (Difco).

Carbohydrate disks (Difco) 6 or 13mm indiameter were spaced evenly on the agarsurface. The plates wereincubatedat30°C for24to48hand examined for a colorchangefrompurpletoyelloworforgrowth around thedisks,whichindicatedassimilation of the carbohydrate substrates.

Fermentation ofglucose,maltose,sucrose, and lac-tosewasdeterminedbyinoculatingcarbohydrate fer-mentation tubes(Difco) with 5drops (0.2 ml) ofthe

yeast-salinesuspension.Thetubeswereincubatedat 37°C and read after 24 and 48 h andagain after 10 daysfor thepresence of gas in the inverted Durham tubes.

API20C method.

Ail

testswereperformed accord-565

on February 7, 2020 by guest

http://jcm.asm.org/

ing to the manufacturer's instructions. Kits were storedat4°C until used.Ampoulescontaining the API 20C basal medium wereplacedinaboiling-water bath andallowedtoremainfor5minaftermeltingtoensure completeliquification. Eachampoulewasremovedto a 50°C water bath andheld there until inoculation. Oncemelted, theampouleswereused within 2h. The inoculumwaspreparedby lightlytouchinganisolated yeastcolony with thetipofasterile woodenapplicator stick and thengrinding thetipagainstthe bottom of an ampoule containing the molten basal medium. A uniform suspension was achieved by gentle stirring, with carebeing takentoavoid the excessive produc-tion of air bubbles. Thedensity ofthe inoculumwas adjustedtojust below 1+when heldagainsta Wick-erham card. With asterile Pasteur pipette, the sus-pension was inoculated into each ofthe 20 plastic cupuleson the APIstrip. The first cupulecontained only basal medium and served asazero-growth con-trol. The remaining cupules contained dehydrated substrates usedtodetect assimilation of the carbohy-drates glucose, glycerol, 2-keto-D-gluconate, L-arabi-nose,xylose, adonitol, xylitol, galactose, inositol, sor-bitol, methyl-D-glucoside, N-acetyl-D-glucosamine, cellobiose, lactose,maltose, sucrose, trehalose, mele-zitose, andraffinose. The strips containing the cupules wereplacedinplastictrayscontaining5.0ml of water and wereincubated at30°C for 72 h. Readingswere made after24, 48,and 72 hof incubation. Turbidity greaterthan thatofthegrowthcontrol wasusedas an indication ofassimilation. After the 72-h reading, a seven-digit profile number based on the organism's assimilation reactions wasobtained according to in-structionsprovided bythemanufacturer. The identi-fication of organisms was based on the numerical ProfileIndexenclosedwith thecommercialproductin conjunction with the microscopicmorphologyon corn-meal-Tween80agar(4).Whenaprofilenumber was encounteredwhich wasnotlistedinthe ProfileIndex, the APIcomputerservicewasconsulted.

Criteria used for the identification oforganismsby the modified API 20C system were as follows. (i) A correct identification was obtained if a seven-digit profile number obtained from the assimilation reac-tions ofan organism was listed inthe Profile Index supplied by the manufacturer. (ii)A correct presump-tiveidentificationwasobtained ifanassimilation pro-file number ofanorganismwas notlisted in the Profile Index and the identification requiredconsultation of theAPI computerservice. In this instance the com-puterliststhreepossible choices, indecreasingorder of likelihood, for the identification of an unknown organism. If the first choice listed was correct and described as an excellent, very good, or acceptable identification,theidentificationwas considered as pre-sumptively correct. If the correct identification was listedamong the three choices and could be

differen-tiatedonthe basis of morphology on cornmeal-Tween 80agar, theidentificationwasalso considered as pre-sumptivelycorrect.

One hundred fortystockculture isolates of yeast and yeastlikeorganisms weresubcultured onto Em-mon's modification ofSabouraud dextrose agar and inhibitorymoldagarcontaining 5 ,g of gentamicin and 125,igofchloramphenicolper ml. The API20C yeast

systemwasinoculated withorganisms from each me-dium to determine theeffectofantibioticsonthefinal identificationof theorganismstested.

RESULTS

The modified API 20C system provided

cor-rect identifications for 486 or 96% of the 505

isolates tested (correct plus presumptively

cor-rectidentifications; incorrect identifications, 19

or4%). Of the 251 stock cultures tested, 236or

94.1% were identified correctly, including 25

(10%) which required consultation of the API

computerservice, whereas 15 (5.9%) were

incor-rectly identified (Table 1).

The modified API 20C system provided

cor-rect identifications for 98.4% of the 254 fresh

clinical isolates examined, including 19 (7.5%)

which required consultation of the API

com-puter service (Table 2), whereas 4 (1.6%) were

incorrectly identified.

The conventional method provided correct

identificationsfor all fresh clinical isolates with

one exception, a weakly pigmented strain of Rhodotorula.

One hundredforty-twoorganisms maintained

as stock cultures were subcultured onto both

Sabouraud dextrose agar (Emmon's

modifica-tion) and inhibitory mold agarcontaining5

lig

ofgentamicin and125

,ug

ofchloramphenicolper

mlbefore inoculation of the API 20Cstrips.The

modified API system provided correct

identifi-cations for 95% of the organisms tested from

both media, whichsuggested that the

identifi-cationswere notinfluenced by the type of

me-diumusedbeforetesting.

Theoptimaltimerequired for the

identifica-tion oftheorganismstestedwas 72h. However,

82.5% of the Torulopsis glabrata isolates were

identified after 24 h, and identification of all

isolates of thisspecieswascomplete after48h. Two of seven isolates ofCryptococcus terreus

required incubation for96hbeforeidentification

was made due to delayed inositol assimilation

reactions. A majority of the cryptococci tested

required 72 h of incubation before assimilation

ofinositol could be detected.

Table 3 lists the organisms misidentified by

the modified API system and the reasons for

misidentification. At present, the assimilation

patterns forCandidalambica and Cryptococcus

luteolus are not included in the API computer databaseand,therefore,cannot beidentified by this system. The most frequently encountered

probleminvolved the identification of Candida

albicans and Candidatropicalis.Threeisolates

of C. albicans were identified by the API 20C

system as C. tropicalisdue to thefalse-positive

assimilation of melezitose. One C.tropicalis

on February 7, 2020 by guest

http://jcm.asm.org/

MODIFIED API 20C FOR IDENTIFICATION OF YEASTS 567 TABLE 1. Identification of stock cultures of clinical isolates bythe modified API 20C

Modified API 20C

Organism No.tested No. correct,

presump-No.correct M) No.

corrective

(M% No. incorrect('

Candidaalbicans 21 18 (85.7) 2(9.5) 1 (4.8)

C.guilliermondii 9 7 (77.8) 1 (11.1) 1(11.1)

C. krusei 20 19 (95) 1 (5) 0

C.lambicaa 1 0 0 1 (100)

C.parapsilosis 25 22 (88) 2(8) 1(4)

C.pseudotropicalis 10 10 (100) 0 0

C. rugosa 1 0 1 (100) 0

C.tropicalis 24 16 (66.7) 7(29.2) 1(4.1)

Cryptococcus albidus 24 23 (95.8) 1(4.2) 0

C.laurentii 9 9 (100) 0 O

C.luteolus' 5 0 0 5(100)

C.neoformans 38 33 (86.8) 2 (5.2) 3(8)

C.terreus 7 4 (57.1) 1 (14.3) 2(28.6)

C.uniguttulatus 3 3 (100) 0 O

Geotrichum candidum 1 1 (100) 0 0

Rhodotorula glutinis 7 0 7 (100) 0

R.pilimanae 2 0 2 (100) 0

R. rubra 1 0 1 (100) 0

Saccharomyces cerevisiae 10 10(100) 0 O

Torulopsis glabrata 22 22(100) 0 0

Trichosporon beigelii il 6(54.5) 5 (45.5) 0

Total 251 211(84.1) 25 (10) 15 (5.9)

Presently

notincluded

in

computerdatabase of

modifed

API 20Csystem.

TABLE 2. Identificationof fresh clinical isolates bythemodified API 20C

Organism No.tested No. correct( No. correct,presumptive No.incorrect (<)

Candidaalbicans 28 24(85.7) 2(7.15) 2(7.15)

C.guilliermondii 5 5(100) 0 O

C. krusei 13 13 (100) 0 0

C.parapsilosis 37 30(81.1) 7(18.9) 0

C.pseudotropicalis 3 3(100) 0 0

C.tropicalis 34 34 (100) 0 0

C.zeylanoides 2 2(100) 0 0

Cryptococcus albidus 15 12 (80) 3(20) 0

C.laurentii 1 0 1(100) 0

C.luteolus 1 0 0 1(100)

C.neuformans 16 14 (87.5) 2 (12.5) 0

Rhodotorula sp. 22 17(77.3) 4 (18.2) 1(4.5)

Saccharomycescerevisiae 9 9 (100) 0 0

Torulopsisglabrata 64 64 (100) 0 0

Trichosporonbeigelii 4 4(100) 0 O

Total 254 231(90.9) 19 (7.5) 4(1.6)

late wasidentified as C. albicans because of a

false-negative melezitose reaction.

DISCUSSION

The modified API 20C system provided

cor-rect identifications for 96% of the organisms

tested. This is comparableto theaccuracyof the

originalsystem as determined in previous

stud-ies (2, 8, 11).

The modified API 20C systemrequiredmuch

less technical proficiency than did the

original

system. With the elimination of

carbohydrate

fermentation tests,

dehydrated

substrateis now

contained in 20 plastic

cupules

rather than in

tubules. This modification has

simplified

filling

with theagar-inoculum

suspension

and has

elim-inated concern over

entrapped

air spaces and

theuse ofVasparseals.

VOL. 9, 1979

on February 7, 2020 by guest

http://jcm.asm.org/

568 ROBERTS TABLE 3. Analysis of incorrectly identified

organisms by the API 20Csystem

Organism No. Explanationofincorrect iden-tification

Candida albicans 3 IdentifiedasC.tropicalis due to positive assimilation of melezitose

C.guilliernondii 1 Arabinose, xylose, xylitol, and melezitose not assimilated after 72 h

C.lambica 1 Profile not included in com-puter data base

C.parapsilosis 1 Arabinose not assimilated after 72h; false-positive assimila-tion of xylitol andcellobiose C.tropicalis 1 IdentifiedasC. albicans due to negative assimilation of me-lezitose

Cryptococcus luteo- 6 Profile not included in

com-lus puterdata base

C.neoformnans 3 One did not assimilate adoni-tol,N-acetyl-D-glucosamine, trehalose,orraffinose after

72h; one gavefalse-positive assimilationofglycerol; one did not assimilate inositol, xylitol, arabinose, or raffi-nose after 72 h.

Rhodotorula sp. 1 Did not assimilate sucrose, maltose, melezitose, or raffi-nose after 72 h.

The properinoculation density was ofsome

importance. A too dense inoculum resulted in

apparent false-positive assimilation reactions

and ina"carryover" phenomenon wheregrowth

wasobserved in the zero-growth control cupule.

This shouldprovide little problem if the

inocu-lum is adjustedtotheproperdensity byuse of

theWickerham card and if the readercompares

the individual carbohydrate assimilation tests

with thezero-growth control cupule.

The manufacturer recommends that the API

strips be incubated at 30°C and read after 24,

48, and 72 h. These instructionsareappropriate

and should be followed sincemost cryptococci

studied required 72 h for identification. If the

reactions appearincomplete after 72h and

mi-croscopic morphology resembles Cryptococcus,

we recommend that the strips be heldat30°C

foran additional 24 h. Some strains of

crypto-coccimayrequire the additional incubationtime

before inositol assimilationcanbe detected.

In the instructions, the manufacturer states

that well-isolated colonies on Sabouraud

dex-trose agar serve as the inoculum source. Our

results indicatethat the inoculum mayalso be

prepared from organisms grown on inhibitory

moldagarwhich contains chloramphenicol(125

,ug/ml) and gentamicin (5 ,g/ml).Thepresence

ofthese antibiotics didnot affect the

assimila-tion patterns or the final identifications ofthe

organisms tested. However,

antibiotic-contain-ing media should be used with caution. API strips inoculated withyeasttaken fromselective media are subject tocontamination due tothe

presence of bacteria whose growth was

sup-pressed bytheantibiotics in the media.

The identification ofyeasts based solely on

the assimilationpatternsprovided by the modi-fied API 20C systemisnotreliable unless used

in combinationwithmicroscopic morphological features (presence of

arthroconidia,

blastoconi-dia, hyphae, or

pseudohyphae)

on

cornmeal-Tween 80 agar as recommended

by

the

manu-facturer. Crytococcus

laurentii,

C. terreus, and

Candidaguilliermondii

can

give

anassimilation

pattern identical to that ofTrichosporon

beige-lii. Inaddition,thefollowing pairsoforganisms

canexhibitidentical assimilationreactions:

Tri-chosporon

capitatum

and Candida

krusei,

To-rulopsismaris and Candida rugosa, T. maris andTrichosporon

penicillatum,

and

Torulopsis

candida and C.

guilliermondii.

The API 20C

yeast systemusesmelezitose assimilationasthe

only reaction

differentiating

C.

tropicalis from

C.albicans;however,this reactionisnot defin-itive since both

organisms

arevariable in their

ability to utilize melezitose

(1). Cryptococcus

neoformans and C.

tropicalis

cannotbe differ-entiatedif a

false-negative

inositol reaction

oc-curs. The organisms with identical assimilation

reactionscan, however, be

distinguished

onthe basis of their

microscopic morphological

fea-tures.

Therefore,

the API 20C systemmustbe

used inconjunction with

microscopic

morphol-ogy on cornmeal-Tween 80 agar. We also

rec-ommend that germ tube formation and chla-mydospore

production

beusedasadditional cri-teria when C.

albicans

isdifferentiated from C. tropicalis.

TheAPIsystem is not designedtoprovide a

rapid

presumptive

identification of C.

neofor-mansdue to the slowgrowth ofthis organism. Suchan identification may beobtained byuse

ofpreviously evaluated rapid techniques includ-ing the rapid nitratetest (3),rapid urease test

(7), and thecaffeic acid-ferric citrate test (10).

Wefeelthatthe final identification of C.

neofor-mansshouldalways be based on results of the

nitratetest,carbohydrateassimilation tests, and

pigment on niger seed agar.

Rapid,cost-effective identification of C.

albi-cans is possible by use of the germ tube test.

Therefore,inoculation of API stripswith germ

tube-positive

organisms

doesnotappear neces-sary. However, the API 20C should be useful in

differentiating

C. stellatoidea from C. albicans

and in identifying

germ

tube-negative variants

ofC. albicans.

Atpresent, C.

lambica

andC.luteolusare not

includedin the data base of the API20C system.

on February 7, 2020 by guest

http://jcm.asm.org/

MODIFIED API 20C FOR IDENTIFICATION OF YEASTS 569

However, the manufacturer stated that the

as-similationprofiles of these twoorganismswill be

addedtothesystem once asufficientnumber of

isolates havebeen obtained andcharacterized.

Inconclusion, we feel that themodifiedAPI

20Cis acommercialsystem useful for the

iden-tification ofyeastsfrom clinicalspecimens. The

system,although reliable, is not complete and is

designedby themanufacturertoalwaysbe used in conjunction with microscopic morphological features for definitive identification.Inaddition,

it is not designed for the rapid presumptive identification of such organisms as C.

neofor-mans. Other biochemicaltestsmay be used for

this purpose. The modified API 20C system is

technically lesscomplex than the original, does notrequireanymediumpreparationin the

lab-oratory,is stable at4°Cfor 1 year from the date

ofmanufacture, and requireslittlestorage area.

These features should increase the ability of evensmall laboratoriestoidentifyyeasts of

med-ical importance. Acquisition and characteriza-tionofnewandadditionalisolates by the

man-ufacturershould result in improvementand

ex-pansion of thesystem.

ACKNOWLEDGMENTS

Thiswork wassupported inpart byAnalytab Products, Inc., DivisionofAyherst Laboratories, Plainview, N.Y.

LITERATURE CITED

1. Ahearn, D. G.,S. A. Meyer, G. Mitchell,M. A. Ni-cholson, and A. I. Ibrahim. 1977. Sucrose-negative variants of Candida tropicalis.J. Clin. Microbiol. 5: 494-496.

2. Bowman, P.I., and D. G. Ahearn.1976.Evaluation of commercial systems for the identification of clinical yeastisolates. J. Clin. Microbiol.4:49-53.

3. Hopkins,J.M., and G. A. Land. 1977.Rapidmethod for determining nitrateutilization by yeasts.J. Clin. Microbiol. 5:497-500.

4. Koneman, E.W.,G.D. Roberts, and S. E.Wright. 1978.Practical laboratorymycology,2nd ed. The Wil-liams& Wilkins Co., Baltimore.

5. Land,G. A.1976.Opportunisticmycotic infections and malignancies. J. Clin. Hematol. Oncol.6:113-139.

6. Roberts,G.D. 1976.Laboratorydiagnosis offungal

in-fections. Hum.Pathol.7:161-168.

7. Roberts, G.D., C. D.Horstmeier,G. A.Land,andJ.

H.Foxworth. 1978.Rapidureabrothtestforyeasts. J. Clin.Microbiol.7:584-588.

8. Roberts,G.D.,H. S.Wang,and G.E.Hollick. 1976.

Evaluation of the API20C microtube systemforthe identification ofclinically important yeasts. J. Clin. Microbiol. 3:302-305.

9. vanderWalt,J.P.1970.Criteria andmethods used in classification, p.34-113.InJ. Lodder(ed.),Theyeasts.

Ataxonomicstudy, 2nd ed. North-HollandPublishing Co.,Amsterdam.

10. Wang, H.S.,R.T.Zeimis,andG. D. Roberts.1977.

Evaluation ofacaffeic acid-ferric citratetestforrapid identification of Cryptococcusneoformans.J. Clin. Mi-crobiol.6:445-449.

11.Zwadyk, P., R. A.Talton,and A. Proctor.1976.

Eval-uation of API20Cforidentificationof yeasts.Am.J. Clin. Pathol.67:269-271.

VOL. 9,1979

on February 7, 2020 by guest

http://jcm.asm.org/