Correlation between Microdilution, E-test, and Disk Diffusion Methods for Antifungal Susceptibility Testing of Posaconazole against Candida spp.

Agar-based antifungal susceptibility testing is an attractivealternative to the microdilution method. We examined the correlationbetween the microdilution, E-test, and disk diffusion methodsfor posaconazole against Candida spp. A total of 270 bloodstreamisolates of Candida spp. with a broad range of posaconazoleMICs were tested using the CLSI M27-A2 method for microdilution,as well as the M-44A method and E-test methods for agar-basedtesting on Mueller-Hinton agar supplemented with 2% glucoseand 0.5 µg of methylene blue. MICs and inhibitory zonediameters at the prominent growth reduction endpoint were recordedat 24 and 48 h. The Candida isolates included Candida albicans(n = 124), C. parapsilosis (n = 44), C. tropicalis (n = 41),C. glabrata (n = 36), C. krusei (n = 20), C. lusitaniae (n =3), and C. dubliniensis (n = 2). The overall concordance (i.e.,the percentage of isolates within two dilutions) between theE-test and microdilution was 64.8% at 24 h and 82.6% at 48 h.When we considered an arbitrary breakpoint of $≤$ 1 µg/ml,the agreement between the E-test and microdilution methods was87.8% at 24 h and 93.0% at 48 h. The correlation of MICs withdisk diffusion zone diameters was better for the E-test thanthe microdilution method. Zone correlation for diameters producedby the disks of two manufacturers was high, with a Pearson testvalue of 0.941 at 24 h. The E-test and microdilution MICs showgood concordance and interpretative agreement. The disk diffusionzone diameters are highly reproducible and correlate well withboth the E-test and the microdilution method, making agar-basedmethods a viable alternative to microdilution for posaconazolesusceptibility testing.

INTRODUCTION

Top

Introduction

Standardization of antifungal susceptibility testing is an areaof active research since the availability of reference methodsfor the testing of yeasts (Clinical and Laboratory StandardsInstitute [CLSI] M27-A2 method) and molds (CLSI M38-A method)(17). The emergence of fluconazole-resistant Candida albicansand selection for inherently fluconazole-resistant Candida spp.has prompted the use of alternative agents for the treatmentof invasive Candida infections (7, 8). The alternatives includethe echinocandins and the newer azoles, voriconazole, ravuconazole,and posaconazole. The azoles are inhibitors of the sterol 14-alpha-demethylaseenzyme, blocking the production of the ergosterol componentof the fungal cell membrane. Posaconazole, a triazole agentcurrently in clinical trials, is a more potent inhibitor ofthis enzyme than itraconazole and voriconazole in Aspergillusspecies and retains activity against the mutated enzyme responsiblefor resistance to fluconazole, itraconazole, and voriconazolein Candida (11). It has shown activity superior to fluconazoleand itraconazole against Candida spp. in previous in vitro surveysusing the broth microdilution (MD) technique according to theCLSI method (15). There has been much research interest in agar-basedantifungal susceptibility via E-test (ET) and disk diffusion(DD) methods due to their relative ease and the lack of needfor specialized equipment (6, 17). The MD susceptibility resultsfor posaconazole have been compared to E-test results with goodcorrelation (16); however, no correlation has been made withdisk diffusion methods for posaconazole.

In the present study, we compared the MD, ET, and DD methodsfor the determination of susceptibilities to posaconazole against270 Candida spp. isolates. The ET and DD methods are well studiedwith respect to fluconazole and voriconazole (2, 9, 13), andthe present study expands that knowledge base to include posaconazole.

MATERIALS AND METHODS

Top

Materials and Methods

Isolates. A total of 270 bloodstream isolates of Candida spp. identifiedwith the API 20C AUX system (Biomerieux Vitek, Hazelwood, MO)were randomly selected from a collection from the United Statesfor testing. The isolates included 124 isolates of Candida albicans(46.3%), 44 isolates of C. parapsilosis (15.9%), 41 isolatesof C. tropicalis (15.2%), 36 isolates of C. glabrata (13.3%),20 isolates of C. krusei (7.4%), 3 isolates of C. lusitaniae(1.1%), and 2 isolates of C. dubliniensis (0.8%). The isolateswere stored in distilled water at room temperature or in 30%glycerin at –80°C. Each isolate was subcultured atleast twice on Sabouraud dextrose agar and incubated at 35°Cto ensure purity and optimal growth. All isolates were subculturedagain 24 h prior to testing. Quality control isolates includedC. albicans ATCC 90028, C. parapsilosis ATCC 22019, and C. kruseiATCC 6258.

Inoculum suspensions. Yeast inoculum suspensions were prepared as described in CLSIM27-A2 (5). Using spectrophotometry at 530 nm, turbidity wasmeasured and adjusted to match a 0.5 McFarland density standardresulting in an inoculum containing 1 x 10⁶ to 5 x 10⁶ yeastcells/ml. This suspension was used to directly inoculate agarplates for the ET and DD procedures or was diluted as directedby CLSI M27-A2 for the MD procedure.

Antifungal agents. Posaconazole research powder was obtained from Schering-Plough(Kenilworth, NJ) and was stored at –20°C until reconstitutedfor the MD procedure. Posaconazole and fluconazole E-test stripswere obtained from AB Biodisk (Solna, Sweden) with concentrationranges of 0.002 to 32 µg/ml and 0.016 to 256 µg/ml,respectively. Two varieties of paper disks containing posaconazoleat 5 µg were obtained from Oxoid, Ltd. (Basingstoke, England),and BD (Maryland).

Susceptibility testing methods. Broth MD testing was performed as described in the CLSI M27-A2using RPMI 1640 broth buffered with morpholinepropanesulfonicacid (Sigma Chemical Co., St. Louis, MO), and the pH was adjustedto 7.0. Posaconazole was serially diluted (twofold) in media,and 100-µl aliquots were dispensed into microdilutionplates for a final drug concentration range of 0.03125 to 16µg/ml. Then, 100 µl of yeast inoculum containing1 x 10³ to 5 x 10³ yeast cells/ml was added to each well. Theplates were incubated at 35°C and read at 24 and 48 h byvisual inspection and spectrophotometry at 490 nm. The MIC wasdefined as the lowest drug concentration that reduced growthby 50% compared to drug-free controls.

The ET and DD testing methods were performed according to theCLSI M44-A (4) method on Mueller-Hinton agar plates (10 cm,with 60 ml of agar) supplemented with 2% glucose and 0.5 µgof methylene blue (Hardy Diagnostics, Santa Maria, CA). Foreach isolate, duplicate plates were inoculated by dipping twosterile cotton swabs into the inoculum and evenly streakingthe entire surface of the plates in three directions. Afterdrying for 15 min, an E-test strip was applied to one plate,and two disks were applied to the other plate. The plates wereincubated at 35°C. In addition to the posaconazole tests,quality control specimens were tested with a fluconazole E-test.The plates were read at 24 and 48 h. The MIC for the E-testand the inhibitory zone diameter for the disks were measuredat the transition point where growth abruptly decreased as determinedby a marked reduction in colony size, number, and density.

RESULTS

Top

Results

Table 1 summarizes the drug MICs measured by the MD and ET methodsafter 24 and 48 h of incubation. Table 2 summarizes the concordancebetween the ET and MD methods measured at 24 and 48 h. Valueswere considered concordant if the measured MICs by each methodwere within two or fewer dilutions of each other. Overall, concordancewas 64.8% at 24 h and 82.6% at 48 h. Table 2 also summarizesthe interpretive agreement for the two methods based on an arbitrarybreakpoint of $≤$ 1 µg/ml (14) for determining a susceptibleisolate. Overall, the agreement between MD and ET was 87.8%at 24 h and 93.0% at 48 h. Tables 3 and 4 demonstrate the correlationbetween MD and ET with Spearman's rho values of 0.635 at 24h and 0.648 at 48 h. The reading time specified for MD by CLSI27-A2 and ET by the manufacturer is 48 h.

View this table:
[in this window]
[in a new window]
TABLE 1. Summary of MIC₅₀ and MIC₉₀ values determined by the MD and ET methods at 24 and 48 h

View this table:
[in this window]
[in a new window]
TABLE 2. Concordance and agreement of MIC data measured by the MD and ET methods at 24 and 48 h

View this table:
[in this window]
[in a new window]
TABLE 3. Plot of posaconazole ET MIC versus MD MIC at 24 h^a

View this table:
[in this window]
[in a new window]
TABLE 4. Plot of posaconazole ET MIC versus MD MIC at 48 h^a

Comparison of MD MIC with DD inhibitory zone exhibited goodcorrelation, as evidenced by a Spearman's rho values of –0.580and –0.605 for the two disks at 24 h and Spearman's rhovalues of –0.551 and –0.571 at 48 h. Table 5 demonstratesthe correlation between the MICs obtained by MD with the DDinhibitory zone diameters (for the BD disk) compared at thereading times specified by CLSI M27-A2 and CLSI M44-A. Thiscomparison showed slightly lower correlation with a Spearman'srho value of –0.470. The results were similar for theOxoid disk, and zone diameters correlated well between the twodisk types, as shown by Pearson test values of 0.941 at 24 hand 0.960 at 48 h.

View this table:
[in this window]
[in a new window]
TABLE 5. Plot of 24-h posaconazole disk inhibition zone diameter versus 48-h MD MIC^a

Comparison of the ET MIC with the DD inhibitory zone exhibitedeven better correlation, as evidenced by Spearman's rho valuesof –0.793 and –0.791 for the two disks at 24 h andby Spearman's rho values of –0.827 and –0.839 at48 h.

DISCUSSION

Top

Discussion

MD and ET MICs show good concordance and interpretive agreement.The ET MIC was generally higher than the MIC, as determinedby the MD method, at both 24 and 48 h. The ET MIC was generallyone dilution higher at 24 h than at 48 h, in contrast to theMD method in which the 24-h MIC was usually lower. This phenomenonis explained by the fact that, for most Candida spp., therewas little or no distinction between the ET zone of markedlydecreased growth and the zone of no growth at 24 h. At 48 hand further growth, the distinction between the zones of markedlyreduced growth and no growth became more prominent. In addition,the differences may relate to the use of glucose-supplementedMueller-Hinton agar for ET as opposed to the manufacturer-recommendedglucose-supplemented RPMI agar or the CLSI M27-A-recommendedmedia used for MD. This medium was selected since it is specificallyrecommended by CLSI M44-A for use in routine agar-based antifungalsusceptibility testing. This phenomenon was not seen in themore resistant species (C. glabrata and krusei), resulting insimilar results using the ET and MD methods.

Concordance was higher at 48 h for all species (except C. glabrata)and was >85% for C. albicans, C. dubliniensis, C. krusei,C. lusitaniae, and C. parapsilosis. C. albicans disconcordancewas mainly due to the phenomenon discussed above. C. tropicalisdisconcordance was due to the well-described trailing growthphenomenon (1, 12). C. glabrata isolates showed the lowest ratesof concordance with a much higher MIC measured by the ET thanthe MD method. The isolates with high MIC determined by ET methodwere the same isolates with high MICs as determined by the MDmethod; however, the ET measurements were several dilutionshigher.

There are no established clinical breakpoints for posaconazole,so it is unclear whether these higher MICs determined by ETwould change the "resistant" or "susceptible" interpretive categoryof these isolates compared to the MICs measured by MD method.However, when an arbitrary MIC breakpoint of $≤$ 1 µg/ml wasused to identify potentially susceptible isolates (14), allspecies except C. glabrata showed an agreement of >96% at24 h. At 48 h, the MD and ET agreement for C. glabrata increasedto >75% and increased for all species except C. tropicalis.

DD inhibitory zone diameters are highly reproducible and correlatewell with ET and MD MICs. Correlation may have been decreasedby the fact that MD MICs of <0.03 µg/ml were groupedand by trailing growth at 48 h.

Using the data in Tables 3 to 5, comparisons of the MD methodwith ET and DD using the error-rate-bounded method (10) andthe arbitrary MIC breakpoint of $≤$ 1 µg/ml are shown in Table6. The numbers of major errors and very major errors are similarto the acceptable error rates recommended by the CLSI M23-A2(3) for this method of analysis. Of the four isolates causingvery major errors, three are C. tropicalis.

View this table:
[in this window]
[in a new window]
TABLE 6. Comparison of MD, ET, and DD categorical agreement and error rates

The results of the present study are similar to those of previousstudies comparing the MD, ET, and DD methods of susceptibilitytesting for other antifungal drugs (2, 9). The agar-based methodscompare well to the reference method, especially for the moresusceptible species of Candida. In the present study, C. glabrataaccounted for most of the outlying results, although previouswork with other antifungal drugs has shown better concordancebetween the MD and ET methods for this species (13). Agar-basedmethods are a valid alternative to MD methods for posaconazolesusceptibility testing in Candida spp., although more clinicalexperience is needed prior to true breakpoint determination.

FOOTNOTES

* Corresponding author. Mailing address: Laboratory of Mycology Research, Division of Infectious Diseases, University of Texas-Houston Medical School, 6431 Fannin MSB 2.112, Houston, TX 77030. Phone: (713) 500-6734. Fax: (713) 500-5495. E-mail: Charles.R.Sims{at}uth.tmc.edu.

REFERENCES

Top