MYCOSESMYCOSES 44, 37±45 (2001) ACCEPTEDCCEPTED: NOVEMBEROVEMBER 18, 1999

Impact of endpoint de®nition on the outcome of antifungalsusceptibility tests with Candida species: 24- versus 48-hincubation and 50 versus 80% reduction in growth

Zur Endpunktde®nition bei antifungalen Emp®ndlichkeitspruÈfungen:24- gegen 48-h-Inkubation und 50%- gegen80%-Wachstumsreduktion

G. St-Germain

Key words. Candida, susceptibility testing, endpoint.

SchluÈsselwoÈrter. Candida, Emp®ndlichkeitspruÈfung, Endpunkt.

Summary. The growth inhibition patterns of 764clinical yeast isolates, in response to amphoter-icin B, ¯ucytosine, itraconazole and ¯uconazole,were studied in order to determine the frequencyof trailing growth and any impact this, as wellas 24- or 48-h incubation periods, may haveon minimum inhibitory concentration (MIC)results. A broth microdilution method followingNational Committee for Clinical LaboratoryStandard No. M27A recommendations was used.Trailing growth was observed mainly withazoles. Furthermore, over 98% of isolates exhi-biting a trailing effect at 24 h with ¯uconazoleand itraconazole were either Candida albicans orCandida tropicalis. When comparing 24- and 48-hIC50 values, discrepancies were observed withitraconazole and ¯uconazole, respectively, in 18and 11% of C. albicans and 24 and 30% of C.tropicalis. When comparing IC50 and IC80 valuesat 24 h, discrepancies were again essentially seenwith itraconazole and ¯uconazole, respectively,in 11 and 10% of C. albicans, and 17 and 27% ofC. tropicalis. In susceptibility tests performed with

a microdilution method and read spectrophoto-metrically, 48-h IC80 values result in an unlikelyhigh number of resistant isolates, indicatingthat a 24-h incubation and a 50% reduction inoptical density may correlate better with clinicaloutcome.

Zusammenfassung. Die Emp®ndlichkeits-muster von 764 klinischen Hefeisolaten fuÈrAmphotericin B, Flucytosin, Itraconazol undFluconazol wurden bestimmt mit besondererBeruÈcksichtigung des Ein¯usses von Trailing-effekten auf die MHK-Werte bei 24 oder 48 hInkubation. Hierzu wurde ein Mikrodilutionstestnach NCCLS M27A eingesetzt. Trailing wurdebesonders bei Azolen beobachtet. Mehr als98% der Isolate, die Trailing nach 24 h beiFluconazol und Itraconazol zeigten, warenCandida albicans oder Candida tropicalis. DiskrepanteErgebnisse zwischen MHK50 nach 24 bzw. 48 hInkubation fuÈr Itraconazol und Fluconazolzeigten 18 bzw. 11% von C. albicans und 24 bzw.30% von C. tropicalis. Diskrepante Ergebnissezwischen MHK50 und MHK80 wurden beiItraconazol und Fluconazol an 11 bzw. 10% beiC. albicans und an 17 bzw. 27% bei C. tropicalisgesehen. Bei spektrophotometrischer Ablesungvon MHK80 nach 48 h fand sich eine unglaub-wuÈrdig hohe Zahl resistenter Isolate. Eine50%-Reduktion der OD nach 24 h Inkubationscheint besser mit dem klinischen Therapieerfolgkorreliert zu sein.

Correspondence: G. St-Germain, Laboratoire de Sante Publiquedu QueÂbec, 20045 chemin Sainte-Marie, Sainte-Anne-de-Bellevue, QueÂbec Canada H9X 3R5. Tel: +1-514-457-2070Fax: +1-514-457-6346 E-mail: ggermain@lspq.org

This work was presented in part at the 98th General Meetingof the American Society for Microbiology, Atlanta, Georgia,USA, 17±21 May, 1998.

Laboratoire de Sante Publique du QueÂbec, Canada.

Introduction

Over the last two decades, the increase in thefrequency of serious fungal infections, the avail-ability of new antifungal drugs, and the emergenceof resistance to antifungal drugs has spurred inter-est in the development of standardized, reproduci-ble in vitro antifungal susceptibility testing methods.This has led to the development, by the NationalCommittee for Clinical Laboratory Standards(NCCLS), of a broth macrodilution method and itsmicrodilution adaptation for yeast testing [1].However the selection of important testing vari-ables such as the duration of incubation and thepercentage of inhibition required for azole end-point determination, is still being questioned. Ithas long been thought that a 24-h incubation per-iod leads to an underestimation of the minimuminhibitory concentration (MIC) for ¯uconazole insome isolates. In fact, in vitro±in vivo correlation stu-dies may demonstrate that a 48-h incubation over-estimates MICs [2±5]. It has also been suggestedthat a 50% rather than an 80% reduction ingrowth may afford a better correlation between invitro data and clinical outcome when using themicrodilution method [4]. The study of these vari-ables requires that spectrophotometric determina-tion of endpoints be used in order to avoidvariations in interpretation inherent to visual read-ing and to allow quantitative analysis of growthvariables. The following is a study of the growthinhibition patterns of seven NCCLS referenceisolates and of 764 clinical isolates comprisingsix Candida species, demonstrating the effects ofduration of incubation and endpoint inhibitoryconcentration on the outcome of antifungal sus-ceptibility tests for amphotericin B, ¯ucytosine,itraconazole and ¯uconazole [6].

Materials and methods

Clinical isolates

Seven hundred and sixty-four strains of Candidawere tested, all of which were isolated frompatients during 1997±98 in Quebec Hospitallaboratories. Fifty-two per cent of these werefrom blood, 13% from biopsies and normallysterile ¯uids, and 12% from patients with oro-pharyngeal infection. The distribution of speciesis shown in Table 2. Organisms were identi®edby germ tube analyses or, when necessary,with carbohydrate assimilations using API 20CAUX strips (bioMeÂrieux Vitek, Inc., Hazelwood,MO, USA) supplemented with a urease test anda morphology evaluation on cornmeal Tween80 agar.

Reference isolates

Two NCCLS quality control strains, Candidaparapsilosis ATCC 90018 and Candida kruseiATCC 6258, were tested with each set of clini-cal isolates. Other NCCLS reference isolates,Candida albicans ATCC 90028 and ATCC 24433,Candida glabrata ATCC 90030, C. parapsilosis ATCC22019, Candida tropicalis ATCC 750 [1], andan isolate of Candida lusitaniae 5W31 kindly pro-vided by John Rex (University of Texas MedicalSchool at Houston) were tested repeatedly.

Microdilution method

Testing was performed by a microdilutionmethod following NCCLS recommendations[1]. The culture media used were RPMI 1640(Gibco, Gaithersburg, MD, USA) with glutamine,without bicarbonate and phenol red, bufferedwith morpholinopropanesulfonic acid (MOPS)(0.165 MM, pH 7.0) for 5-¯uorocytosine and theazoles and M3 broth with 2% glucose(BBL, Becton Dickinson Microbiology Systems,Cockeysville, MD, USA) for amphotericin B.Two-fold serial dilutions of amphotericin B(0.008±8 mg mlx1) (Sigma Chemical Co., St-Louis,MO, USA), ¯ucytosine (0.06±64 mg mlx1) (SigmaChemical Co.), itraconazole (0.03±8 mg mlx1)( Janssen Pharmaceutica, Mississauga, Ontario,Canada) and ¯uconazole (0.25±256 mg mlx1)(P®zer Pharmaceuticals, Kirkland, Quebec,Canada) were prepared and dispensed in 50 mlaliquot, in ¯at-bottom 96-well assay plateswhich were kept frozen at x70 uC in sealedplastic bags until used. The inoculum was pre-pared spectrophotometrically and standardizedto a concentration of 1.0±5.0 103 CFU per ml.A 50 ml volume of this suspension was used toinoculate each well containing 50 ml of thedouble concentration of the drug to be tested.Once inoculated, each well therefore contained100 ml of broth. The 100 ml ®nal volume wasfavoured over 200 ml to facilitate the agitation ofthe plates prior to spectrophotometric reading.

After an incubation period of 24 and 48 hat 35 uC, the plates were agitated for 3 min at900 r.p.m. with a shaker (Model EAS 2/4; SLTLaboratory Instruments, GroÈdig, Austria) andthe optical density of the growth in each wellwas determined with the use of an automaticplate reader set at 495 nm (Pasteur DiagnosticLP400; Adil Instruments, Strasbourg, France).The data was transferred to an Excel spreadsheet (Excel, Microsoft Corporation, Roselle, IL,USA) where the optical density of the mediumcontrol was subtracted from all wells.

mycoses 44, 37±45 (2001)

38 G. STT-GERMAINERMAIN

Endpoint determination

For amphotericin B, both the 100 and 50%inhibitory concentrations (IC100 and IC50) werecomputed mathematically. These correspondedto the lowest drug concentration at which theoptical density (OD) was reduced to 0 and j.50% of that of the growth control well. How-ever, as an actual 100% decrease in turbidity istoo stringent an endpoint, considering the occur-rence of plate imperfections and equipmentreadout variations, the IC100 was in fact deter-mined as the ®rst drug concentration with anOD below 0.01. For the other drugs, both IC50

and IC80 values were determined (Fig. 1).

Trailing effect

Isolates were reported as producing a trailingeffect when the median OD measured in thewells following that corresponding to the IC50

value was greater than 0.01 (Fig. 1). The turbid-ity associated with an OD of 0.01 can still beseen as very faint growth with the unaided eye.

Interpretive guidelines

The guidelines used for the interpretation ofantifungal tests for ¯ucytosine, itraconazole and¯uconazole are those provided by the NCCLS[1]. Presently, no guidelines are available for

amphotericin B. In our study, isolates with inhib-itory concentration (IC) values i2 mg mlx1 wereconsidered resistant and those with IC valuesj0.25 mg mlx1, sensitive. Signi®cant discrepan-cies reported when comparing results obtainedwith two different endpoint variables refer tosituations were one result is interpreted assensitive and the other as resistant.

Results

Reference isolates

Table 1 summarizes the in vitro susceptibilities ofeight reference Candida strains. After 48 h incu-bation, modal IC100 values of NCCLS referenceisolates were all within acceptable NCCLSranges for amphotericin B. Modal IC50 valuesfor the other antifungals were also within rangewith the exception of C. albicans ATCC 24433for ¯ucytosine [1, 7]. Notably, the IC50 valuesfor ¯uconazole and itraconazole that were readat 24 h were also within acceptable limits.

Growth of clinical and reference isolates

Overall, the growth of NCCLS selected refer-ence isolates was typical of that of clinical iso-lates presented in Table 2. Due to the additionof 2% glucose, growth was much faster in M3broth than in RPMI. Noticeably, isolates of C.parapsilosis exhibited slower growth within the®rst 24 h followed by increased growth at 48 h,resulting in ODs that surpassed those of otherCandida species. Only four C. albicans isolates, oneC. glabrata, two C. parapsilosis and one C. tropicalisdid not produce suf®cient growth (ODj0.01)for IC determination after a 24-h incubationperiod in RPMI or M3 broths.

Trailing growth in clinical and reference isolates

As previously observed, trailing growth is rarelyencountered with amphotericin B [1]. Thatwhich is reported in Table 3 was due in greatpart to spurious growth only. Trailing growthwas only occasionally seen with ¯ucytosine.Reference isolate C. albicans ATCC 24433 aswell as 16 clinical isolates including 14 C. albicansand two C. tropicalis produced some trailing with¯ucytosine at 24 h. This characteristic appearsto be a good predictor of resistance as the refer-ence isolate as well as 12 of the C. albicans clini-cal isolates and both C. tropicalis were foundresistant (i64 mg mlx1) to ¯ucytosine at 48 h(data not shown). With azoles, trailing growth at24 h was frequent and associated in more than

Figure 1. Growth curves with ¯uconazole. Candida albicans MY-52638 was tested by the microdilution version of the M27-A methodand the growth at various ¯uconazole concentrations recordedspectrophotometrically at 24 h (dotted line) and 48 h (solid line). Thelowest drug concentrations corresponding to a decrease in turbidity of50 and 80% compared with that of the growth control are indicatedalong the curves (IC50, IC80). Isolates were reported as producing atrailing effect when the median OD measured in the wells following thatcorresponding to the IC50 was greater than 0.01. For isolate MY-52638,trailing growth is present at both 24 and 48 h.

mycoses 44, 37±45 (2001)

ENDPOINT IN SUSCEPTIBILITY TESTING OF YEASTSNDPOINT IN SUSCEPTIBILITY TESTING OF YEASTS 3939

Table 1. Candida reference isolates; modal ICs (mg mlx1) recorded after 24 and 48 h of incubation with variable endpoints

Strain (no. times tested) Amphotericin B Flucytosine Itraconazole Fluconazole

NCCLSa

48 h

24 h

IC100

48 h

IC100

NCCLS

48 h

24 h 48 h NCCLS

48 h

24 h 48 h NCCLS

48 h

24 h 48 h

IC50 IC80 IC50 IC80 IC50 IC80 IC50 IC80 IC50 IC80 IC50 IC80

C. albicans 1 0.25 0.5 1 0.12 0.25 0.5 0.5 NG 0.03 0.03 0.03 >8 0.5 j0.25 0.5 j0.25 4

ATCC 90028 (11)

C. albicans 0.5 0.25 0.5 2 0.5 1 >64 >64 NG 0.03 0.03 >8 >8 0.5 j0.25 0.5 j0.25 >256

ATCC 24433 (31)

C. glabrata NG 1 2 NG j0.06 j0.06 j0.06 j0.06 NG 0.5 0.25 1 0.5 NG 4 8 8 16

ATCC 90030 (12)

C. parapsilosis 0.5 0.12 1 0.25 j0.06 0.12 0.25 0.25 0.12 0.12 0.12 0.25 0.5 4 2 2 2 4

ATCC 22019 (103)

C. parapsilosis 1 0.12 1 0.12 j0.06 j0.06 j0.06 j0.06 NG 0.03 0.06 0.06 0.12 0.5 0.5 0.5 0.5 1

ATCC 90018 (11)

C. tropicalis 1 0.25 1 0.12 j0.06 j0.06 j0.06 j0.06 NG 0.06 0.12 0.25 >8 1 0.5 1 2 >256

ATCC 750 (11)

C. krusei 1 0.5 1 8 8 16 16 16 0.25 0.25 0.25 0.5 0.5 32 16 32 32 32

ATCC 6258 (102)

C. lusitaniae NG 1 8 NG >64 >64 >64 >64 NG 0.12 0.25 0.25 0.25 NG 1 2 1 2

5W31 (8)

aExpected MIC as per NCCLS method M27A [1]. NG, not given.myco

ses4

4,

37±45

(2001)

40

G.

STT

-GE

RM

AIN

ER

MA

IN

98% of cases to C. albicans and C. tropicalis. Inreference isolates, residual growth was essentiallyobserved in C. albicans isolates ATCC 24433and ATCC 90028 and C. tropicalis ATCC 750.With these organisms, reproducible results aremore dif®cult to obtain at 48 h due to a substan-tial increase in trailing growth. Over a series ofrepetitions, 24-h readings repeatedly producedIC50 values j0.25 mg mlx1 whereas 48-h resultsvaried two or three dilutions. Similarly, three to®ve dilution variations were observed with IC80

values at 48 h.

Impact of duration of incubation and endpoint turbidityde®nition on values of IC

Table 1 summarizes the in vitro susceptibilities ofeight reference isolates read at 24 and 48 h aswell as when using different turbidity endpoints.Most reference isolates were not signi®cantlyaffected by these variables. Discrepancies ofgreater than two two-fold dilutions were observedin C. parapsilosis ATCC 90018 and ATCC 22019and C. lusitaniae 5W31 for amphotericin B, in C.albicans ATCC 24433 for ¯ucytosine, and in C.albicans ATCC 90028 and ATCC 24433, and

C. tropicalis ATCC 750 for the azoles. These weremainly associated to results read as IC80 valuesat 48 h.

In clinical isolates, few discrepancies werenoted with amphotericin B when comparing24- and 48-h incubation results. Some variationin outcome was seen, more speci®cally withIC100 values and clinical as well as referenceisolates of C. parapsilosis. Seventy-four per cent ofC. parapsilosis clinical isolates exhibited values ofIC100 greater than two two-fold dilutions after48 as compared to 24 h of incubation (data notshown). However these IC values were low andthis variation resulted in a change in interpreta-tion from susceptible at 24 h to resistant at 48 hin only 11% of isolates (Table 4). Overall, theresults summarized in Table 5 indicate that IC50

or IC100 values for amphotericin B did not varysigni®cantly for any of the species tested. With¯ucytosine, an average of 2% of isolates wasaffected by the duration of incubation resultingin a change in interpretation for some isolates ofC. albicans, C. tropicalis and C. lusitaniae (Table 4).Resistance in C. albicans was not detected at24 h when using the NCCLS guideline ofi32 mg mlx1. However in 15 of 17 C. albicans

Table 2. Average growth (optical densityt SD) of Candida clinical isolates in RPMI 1640 and M3 broths at 24 and 48 h incubation

Organism (no. of strains) RPMI 1640 M-3

24 h 48 h 24 h 48 h

C. albicans (448) 0.13t0.02 0.20t0.03 0.75t0.12 0.74t0.10

C. glabrata (117) 0.13t0.03 0.20t0.04 0.85t0.16 0.84t0.18

C. parapsilosis (79) 0.10t0.03 0.29t0.10 0.22t0.14 0.91t0.26

C. tropicalis (71) 0.11t0.03 0.27t0.08 0.63t0.15 0.68t0.05

C. lusitaniae (26) 0.15t0.05 0.25t0.07 0.72t0.32 0.82t0.15

C. krusei (23) 0.13t0.03 0.19t0.04 0.64t0.16 0.77t0.04

Table 3. Trailinga growth in Candida clinical isolates

Organism (no. of strains) % Frequency of occurrence with following antifungal agents and incubation periods:

Amphotericin B Flucytosine Itraconazole Fluconazole

24 h 48 h 24 h 48 h 24 h 48 h 24 h 48 h

C. albicans (448) 0 0 3 2 54 72 53 72

C. glabrata (117) 1 1 0 0 3 46 1 20

C. parapsilosis (79) 0 1 0 1 0 1 1 3

C. tropicalis (71) 0 0 3 0 54 69 46 60

C. lusitaniae (26) 0 4 0 0 4 4 0 10

C. krusei (23) 0 4 0 4 0 0 0 5

Total (764) 0 0 2 1 37 56 36 49

aIsolates were reported as producing a trailing effect when the median OD measured in the wells following that corresponding to the IC50

was greater than 0.01.

mycoses 44, 37±45 (2001)

ENDPOINT IN SUSCEPTIBILITY TESTING OF YEASTSNDPOINT IN SUSCEPTIBILITY TESTING OF YEASTS 4141

isolates resistant at 48 h, the values of IC50 at24 h were i1 mg mlx1. Among the 448 clinicalisolates of C. albicans, only three others withinthis category would not be considered resistantat 48 h with IC50 values=4 or 16 mg mlx1. Over-all, the 50 or 80% decrease in growth endpointshad little effect on the outcome of tests with¯ucytosine. With respect to itraconazole and¯uconazole, as summarized in Tables 4 and 5,the duration of incubation and endpoint de®ni-tion both strongly in¯uenced IC results and their®nal interpretation, especially with C. albicansand C. tropicalis. This is consistent with thepredominance of trailing growth in these twospecies. The percentage of resistant isolatesto all four antifungals, given various endpointde®nitions, is presented in Table 6.

Discussion

Spectrophotometric turbidity readings of brothmicrodilution antifungal susceptibility tests havemany advantages over visual endpoint determi-nations [5, 8, 9]. By making data analysis andendpoint reading truly quantitative, not only dothey eliminate reader bias but they also allow forthe quantitative study of growth characteristicsat various incubation periods as well as from thestandpoint of individual yeast species.

Amphotericin B

The NCCLS reference method stipulates thatamphotericin B MICs be read as the lowestdrug concentration that prevents any discernible

Table 4. Effect of 24 versus 48 h incubation on the outcome of susceptibility tests with Candida clinical isolates

Organism (no. of strains) % Isolates with IC interpretationa changing from susceptible to resistant for the following antifungal

agents and endpoint criteria:

Amphotericin B Flucytosine Itraconazole Fluconazole

IC50 IC100 IC50 IC80 IC50 IC80 IC50 IC80

C. albicans (448) 1 1 4 3 18 54 11 46

C. glabrata (117) 9 3 0 0 3 3 0 3

C. parapsilosis (79) 0 11 0 0 1 3 1 1

C. tropicalis (71) 1 3 1 3 24 20 30 35

C. lusitaniae (26) 4 4 0 4 0 0 0 0

C. krusei (23) 4 0 0 0 0 0 0 0

Total (764) 2 3 2 2 13 34 9 31

aSusceptibility (S) and resistance (R) de®ned as follows: amphotericin B, S: j0.25 mg mlx1, R: i2 mg mlx1, ¯ucytosine, S:

j4 mg mlx1, R: i32 mg mlx1, itraconazole, S: j0.125 mg mlx1, R: i1 mg mlx1 and ¯uconazole, S: j8 mg mlx1, R:

i64 mg mlx1.

Table 5. Effect of turbidity decrease endpoint de®nition on the outcome of susceptibility tests with Candida clinical isolates

Organism (no. of strains) % Isolates with IC interpretationa changing from susceptible to resistant for the following antifungal

agents and endpoint criteria:

Amphotericin B

IC50 versus IC100

Flucytosine

IC50 versus IC80

Itraconazole

IC50 versus IC80

Fluconazole

IC50 versus IC80

24 h 48 h 24 h 48 h 24 h 48 h 24 h 48 h

C. albicans (448) 0 0 0 0 11 48 10 46

C. glabrata (117) 0 0 2 0 0 1 0 2

C. parapsilosis (79) 0 1 0 0 0 0 0 0

C. tropicalis (71) 0 0 1 0 17 30 27 30

C. lusitaniae (26) 0 0 0 0 0 0 0 0

C. krusei (23) 0 0 0 0 0 0 0 0

Total (764) 0 0 1 0 8 31 8 0

aSusceptibility (S) and resistance (R) de®ned as follows: amphotericin B, S: j0.25 mg mlx1, R: i2 mg mlx1, ¯ucytosine,

S: j4 mg mlx1, R: i32 mg mlx1, itraconazole, S: j0.125 mg mlx1, R: i1 mg mlx1 and ¯uconazole, S: j8 mg mlx1,

R: i64 mg mlx1.

mycoses 44, 37±45 (2001)

42 G. STT-GERMAINERMAIN

growth [1]. The IC100 values measured spectro-photometrically as described in the methodssection conform to this recommendation. How-ever, our data show no signi®cant discrepanciesbetween IC50 and IC100 values for all species(Table 5). This is not surprising given the rarityor spurious nature of trailing growth seen withthis antifungal agent. Also, when examining resultinterpretations read at 24 versus 48 h, few iso-lates show a conversion from susceptible toresistant. Candida parapsilosis stands out with 11%of strains presenting this characteristic (Table 4).This may be explained by its slower growth inthe ®rst 24 h of incubation (Table 2). However,at 48 h, an increase in the number of resis-tant isolates was noticed for all species, andmost importantly with C. glabrata, C. parapsilosis,C. lusitaniae and C. krusei (Table 6). The ICvalues in these species were often borderline tothe i2 mg mlx1 resistance cut-off point whichexplains the important increase in number ofresistant isolates at 48 h. It remains to bedemonstrated whether the high frequencies ofresistance associated with these species are com-patible with clinical outcome. Pfaller et al. [10],using a similar panel of yeast species, reported3.6% isolates inhibited by i2 mg mlx1 ofamphotericin B as determined by the referencemicrodilution method read at 48 h compared to17.3% by Etest. Interestingly, based on the sameinterpretative guideline, our spectrophotometricIC100 24-h result for all yeasts species (2%) issimilar to their 48-h visual result, whereas our48-h result (15%) compares with their Etest result.In an attempt to correlate in vitro susceptibilitydata and treatment failure, Nguyen et al. [11]reported the following percentages of resistanceusing a macrodilution method and a breakpointMIC of i1 mg mlx1 at 48 h: C. albicans (2%),

C. glabrata (10%), C. parapsilosis (10%), C. tropicalis(0%), C. lusitaniae (33%) and C. krusei (0%).Although clinical data are not available for ourisolates, the percentage of putative resistant iso-lates at 48 h is similar. For some time, theNCCLS methodology for amphotericin B hasbeen known to suffer from the narrow range ofMICs obtained, complicating the discriminationbetween susceptible and resistant isolates [12].Our IC values ranged from 0.008 to 2 mg mlx1

at 24 h and from 0.008 to 8 mg mlx1 at 48 h.Using M3 broth, we found the same limitationswith 95% of isolates exhibiting IC values fallingwithin an eight-fold range of dilutions at both24 and 48 h.

Flucytosine

None of our clinical isolates of C. glabrata, C.parapsilosis and C. lusitaniae were found resistantto ¯ucytosine using IC50 values at 24 or 48 h(Table 6). However, 17 C. albicans isolates werefound resistant at 48 h compared to none at24 h. As observed by To et al. [12] it is apparentthat, using the present NCCLS methodologyand breakpoint guideline for resistance to ¯ucy-tosine (i32 mg mlx1), it may not be possible todetect resistance to this drug in C. albicans at24 h. In this study, when using a resistancebreakpoint de®ned as an IC50i1 mg mlx1 at24 h, a correlation with IC50 values at 48 hi32 mg mlx1 was obtained for 15 of 17 isolatesof C. albicans. At 24 h the distinction betweenthis group of isolates and susceptible isolates wassigni®cant as 95% of all C. albicans isolates exhib-ited values IC50 j0.25 mg mlx1. In C. tropicalis,six isolates were already resistant at 24 h out ofa group of eight resistant at 48 h. With the1 mg mlx1 guideline, resistance could be detected

Table 6. Resistance in Candida clinical isolates

Organism (no. of strains) % Isolates resistanta to the following antifungal agents:

Amphotericin B

IC50/IC100

Flucytosine

IC50/IC80

Itraconazole

IC50/IC80

Fluconazole

IC50/IC80

24 h 48 h 24 h 48 h 24 h 48 h 24 h 48 h

C. albicans (448) 0/0 2/3 0/1 4/4 3/16 24/78 5/14 17/62

C. glabrata (117) 0/4 30/50 0/2 0/0 11/11 57/54 10/7 11/11

C. parapsilosis (79) 0/0 0/15 0/0 0/0 0/0 1/3 0/0 1/1

C. tropicalis (71) 0/0 3/4 9/11 11/11 6/30 36/69 3/32 30/61

C. lusitaniae (26) 0/8 35/62 0/0 0/4 0/0 0/0 0/0 0/0

C. krusei (23) 0/13 44/57 0/0 9/13 9/9 35/61 13/17 48/44

Total (764) 0/2 9/15 1/2 4/4 5/14 27/62 5/13 16/45

aResistance de®ned as follows: amphotericin B, i2 mg mlx1, ¯ucytosine, i32 mg mlx1, itraconazole, i1 mg mlx1 and ¯uconazole,

i64 mg mlx1.

mycoses 44, 37±45 (2001)

ENDPOINT IN SUSCEPTIBILITY TESTING OF YEASTSNDPOINT IN SUSCEPTIBILITY TESTING OF YEASTS 4343

in seven of eight isolates at 24 h. However,this rule did not hold for C. krusei, as the IC50

values for this species were i1 mg mlx1 forall 23 isolates at 24 h but only two correlatedwith IC50 values i32 mg mlx1 at 48 h.

Azoles

It has been observed that trailing growth ismainly associated to azoles, these drugs beingfungistatic rather than fungicidal [8, 13]. In thiscontext, the lack of a clear-cut endpoint compli-cates the interpretation of the test. The NCCLSmethod description is not as explicit with regardto the decrease in turbidity required for end-point determination with the microdilutionmethod in comparison with the macrobrothmethod where an 80% decrease in growth isspeci®ed. Instead, the endpoint is described asthe lowest concentration exhibiting a `prominentdecrease in turbidity' when compared with thegrowth control well [1]. For those tests readspectrophotometrically, it is likely that thiscorresponds better to a 50% rather than an 80%decrease in OD. Many have observed thata 48-h incubation time increases the likelihoodthat an isolate will demonstrate a trailing effectand a higher MIC [2, 3, 8]. Furthermore, thedata presented here show that this phenomenonappears to be species speci®c (Tables 3, 4, 6).Among the major species of clinically importantyeasts, only isolates of C. albicans and C. tropicaliscommonly produce trailing growth. The tran-sient tolerance (adaptation) of a portion ofthe population of the inoculum cells to high con-centrations of azoles may be due to an abilityto induce detoxi®cation mechanisms such asaltered membrane permeability to azoles and/orincreased expression of drug ef¯ux transporters[14]. Recently, Marr et al. have shown the trail-ing endpoint phenomenon to be pH dependentin C. albicans and C. tropicalis [15]. In our study,little or no trailing growth was observed at24 h in C. glabrata, C. parapsilosis, C. krusei, and C.lusitaniae and, indeed, the IC values for thesespecies did not tend to vary signi®cantly whenread spectrophotometrically at 24 or 48 h or asIC50 or IC80 values (Tables 4 and 5). For thesespecies, the higher percentage of resistant iso-lates seen in Table 6 at 48 h is due to the highernumber of isolates falling in the intermediatecategory at 24 h. However, our data show a per-centage of isolates with ¯uconazole IC inter-pretations changing from susceptible to resistantequal to 11% in C. albicans and 30% in C.tropicalis when comparing IC50 values read at 24and 48 h. With IC80 values, the percentages

increase to 46 and 35%, respectively. Overall,5% of our C. albicans isolates were resistant whentaking into account 24 h IC50 values, and 62%with 48 h IC80 values. Undoubtedly, the 48 hresults lead to an unlikely level of resistance andsome studies with in vitro versus in vivo correlationdata are now suggesting that a 48-h incubationtime may lead to false in vitro resistance foragents such as ¯uconazole [2±4]. Furthermore, itmust be kept in mind that with species prone totrailing such as C. albicans and C. tropicalis, sub-stantially less variation in results will occur withvalues of IC50 compared with those for IC80, asthe endpoint for the IC50 values is recorded inthe steep part of the decreasing growth curveinstead of the trailing section [8, 9, 16]. Readingplates at 24 h instead of 48 h results in relativelylittle increase in IC values with C. glabrata, C.parapsilosis, C. krusei and C. lusitaniae as shown inTable 4. Overall, the above-mentioned charac-teristics observed in clinical strains apply toNCCLS recommended reference isolates. Trail-ing growth with azoles was essentially observedin C. albicans isolates ATCC 24433 and ATCC90028 and C. tropicalis ATCC 750. As trailinggrowth appears to be virtually absent in C. kruseiand C. parapsilosis, these species are likely to beless sensitive indicators of variation in inoculumconcentration and endpoint interpretation forthe daily control of tests. Although the two qual-ity control isolates (C. krusei ATCC 6258 and C.parapsilosis ATCC 22019) suggested for routineuse by the NCCLS are useful, especially in detect-ing drug dilution errors, replacing one of thesestrains with an isolate of C. albicans exhibitinga trailing growth should be considered.

Our study emphasizes that when perform-ing susceptibility tests with ¯uconazole and itra-conazole, trailing growth occurs mainly withC. albicans and C. tropicalis. Subsequently, majordiscrepancies between results obtained after 24-and 48-h incubation periods or read at a 50 or80% decrease in OD were found to occurmore speci®cally with these two species. Althoughstudies are still needed to determine which ofthese endpoint variables correlate better withclinical outcome, a 24-h incubation period and50% decrease in turbidity breakpoint result ina more likely percentage of resistant isolates.

Acknowledgements

I thank Christiane Dion for expert technicalassistance and Martine Raymond for helpfulcomments on the manuscript.

mycoses 44, 37±45 (2001)

44 G. STT-GERMAINERMAIN

References

1 National Committee for Clinical Laboratory Standards.(1997) Reference Method for Broth Dilution Antifungal SusceptibilityTesting of Yeasts. Approved standard M27-A, Vol. 17, No. 9.Villanova, PA: National Committee for Clinical LaboratoryStandards.

2 Tornatore, A. M., Noskin, G. A., Hacek, D. M.,Obias, A. A. & Peterson, L. R. (1997) Effects of incubationtime and buffer concentration on in vitro activities ofantifungal agents against Candida albicans. J. Clin. Microbiol.35, 1473±1476.

3 Revankar, S. G., Kirkpatrick, W. R., McAtee, R. K., et al.(1998) Interpretation of trailing endpoints in antifungalsusceptibility testing by the National Committee forClinical Laboratory Standards Method. J. Clin. Microbiol.36, 153±156.

4 Rex, J. H., Nelson, P. W., Paetznick, V. L., Lozano-Chiu,M., Espinel-Ingroff, A. & Anaissie, E. J. (1998) Optimisingthe correlation between results of testing in vitro andtherapeutic outcome in vivo for ¯uconazole by testing criticalisolates in a murine model of invasive candidiasis. J. Clin.Microbiol. 42, 129±134.

5 Nguyen, M. H. & Yu, C. Y. (1999) In¯uence of incuba-tion time, inoculum size, and glucose concentrations onspectrophotometric endpoint determinations for ampho-tericin B, ¯uconazole, and itraconazole. J. Clin. Microbiol.37, 141±145.

6 St-Germain, G. & Dion, C. (1998) Fluconazole suscep-tibility testing of Candida species: growth inhibition patternsat 24 and 48 h. In: Abstracts of the 98th General Meeting of theAmerican Society for Microbiology. Washington, DC: AmericanSociety for Microbiology. Abstract. F±92, p. 269.

7 Pfaller, M. A., Bale, M., Buschelman, B., et al. (1995)Quality control guidelines for National Committee for

Laboratory Clinical Standards recommended brothmacrodilution testing of amphotericin B, ¯uconazole and¯ucytosine. J. Clin. Microbiol. 33, 1104±1107.

8 Rex, J. H., Pfaller, M. A., Rinaldi, M. G., Polak, A. &Galgiani, J. N. (1993) Antifungal susceptibility testing. Clin.Microbiol. Rev. 6, 367±381.

9 Odds, F. C., Vranckx, L. & Woestenborghs, F. (1995)Antifungal susceptibility testing of yeasts: evaluation oftechnical variables for test automation. J. Clin. Microbiol.39, 2051±2060.

10 Pfaller, M. A., Messer, S. A. & BolmstroÈm, A. (1999)Evaluation of Etest for determining in vitro susceptibilityof yeast isolates to amphotericin B. Diagn. Microbiol. Infect.Dis. 32, 223±227.

11 Nguyen, M. H., Clancy, C. J., Yu, V. L., et al. (1998) Doin vitro susceptibility data predict the microbiologic responseto amphotericin B? J. Infect. Dis. 177, 425±430.

12 To, W.-K., Fothergill, A. W. & Rinaldi, M. G. (1995)Comparative evaluation of macrodilution and alamarcolorimetric microdilution broth methods for antifungalsusceptibility testing of yeast isolates. J. Clin. Microbiol. 33,2660±2664.

13 Sheehan, D. J., Espinel-Ingroff, A., Moore, L. S. &Webb, C. D. (1993) Antifungal susceptibility of yeasts:a brief overview. Clin. Infect. Dis. 17 (Suppl. 2), 494±500.

14 Marr, K. A., Rustad, T. R., Rex, J. H. & White, T. C.(1999) The trailing endpoint phenotype in antifungalsusceptibility testing is pH dependent. Antimicrob AgentsChemother. 43, 1383±1386.

15 Marr, K. A., Lyons, C. N., Rustad, T. R., Bowden, R. A. &White, T. C. (1998) Rapid, transient ¯uconazole resistancein Candida albicans is associated with increased MRNAlevels of CDR. Antimicrob Agents Chemother. 42, 2584±2589.

16 Denning, D. W., Baily, G. G. & Hood, S. V. (1997) Azoleresistance in Candida. Eur. J. Clin. Microbiol. Infect. Dis. 16,261±280.

mycoses 44, 37±45 (2001)

ENDPOINT IN SUSCEPTIBILITY TESTING OF YEASTSNDPOINT IN SUSCEPTIBILITY TESTING OF YEASTS 4545