|Home | About | Journals | Submit | Contact Us | Français|
A multicenter study was conducted to assess the accuracy of the ISO standard 20776-1 and the serial 2-fold dilution procedures for antifungal susceptibility testing. Fluconazole trays can be accurately prepared by following ISO and serial dilution schemes. However, itraconazole trays showed a significant lack of reproducibility that was independent of which method was followed.
Susceptibility testing of fungi has been standardized since 1997 with recommendations for broth microdilution MIC determinations from both the Clinical Laboratory Standards Institute (CLSI) and the European Committee for Antimicrobial Susceptibility Testing (EUCAST) antifungal subcommittees (3, 4, 16, 17). According to those standards and the ISO 20776-1 recommendations, the 2-fold dilutions of compounds should be performed in a multistep, rather complex, and time-consuming fashion, avoiding serial dilutions in order to prevent the risk of a systematic error being induced, for instance, by an imprecise pipette (9). While this is the gold standard for preparation of dilution series, the question has risen if modern pipettes are sufficiently precise to allow dilution series to be prepared as serial dilutions without a loss of accuracy. In this multicenter study we address this issue by comparing the concentrations of hydrophilic and hydrophobic antifungal agents, represented by fluconazole (FLC) and itraconazole (ITC), respectively, in broth microdilution plates prepared according to the ISO standard as well as by serial dilution procedures.
Stock solutions of FLC (Pfizer SA, Madrid, Spain) and ITC (Janssen SA, Madrid, Spain) were prepared centrally in 1-ml volumes of dimethyl sulfoxide (DMSO; Sigma-Aldrich, Madrid, Spain) at 12,800 μg/ml and 1,600 μg/ml, respectively, and distributed to the following participating laboratories: Statens Serum Institut, Copenhagen, Denmark (SSI Denmark); Innsbruck Medical University, Innsbruck, Austria (Inn Austria); and Servicio de Micología of the Spanish National Center for Microbiology, Madrid, Spain (CNM Spain). Stocks were kept at −70°C until use. Microtiter plates were prepared according to the ISO 20776-1 standard and by serial dilution procedures recommended for preparation of trays of both hydrophilic and hydrophobic drugs (9, 16). Trays were done in duplicate on separate days at each center and then shipped frozen by special delivery to Spain for further chromatographic concentration determinations.
Stock solutions and microdilution wells were tested for concentration of antifungal agent using high-performance liquid chromatography (HPLC) methods previously described (1, 2, 5). Samples for analysis were prepared by pipetting 75 μl of the content of selected wells and mixing them with the same volume of distilled water (diluted 1/2). Fifty microliters of the resultant mix was injected directly into the HPLC system.
The presence and concentration of FLC or ITC in each well was determined by comparing the chromatographic signal (peak area, retention time, and a UV spectral pattern of between 210 and 300 nm) from the sample and from standard solutions.
FLC eluted at 4.2 min and ITC eluted at 4.1 min. The limit of detection (LOD), defined as the smallest amount of antifungal reliably differentiated from background noise (18), was found to be 1 μg/ml for FLC and 0.5 μg/ml for ITC. A linear calibration curve was obtained in the range of 1 to 128 μg/ml for FLC and 0.5 to 16 μg/ml for ITC using peak area. The method was validated by performing three calibration curves on consecutive days. Comparisons between results obtained by the tray preparation method were done by using analysis of variance (ANOVA). A P value of <0.05 was taken as statistically significant. All statistical analyses were done with the statistical package PAWS statistics 18.0 (SPSS S.L., Madrid, Spain).
In order to determine and compare the accuracy of each method described for microdilution, two consecutive HPLC determinations of the azole concentration were calculated from different wells of trays prepared according ISO recommendations, serial dilutions for hydrophobic drugs, and serial dilutions for hydrophilic compounds. Results are shown in Tables Tables11 (FLC) and and22 (ITC).
Taking into account data from FLC plates, precision and accuracy studies revealed that average coefficients of variation (CV) (calculated as % CV = [standard deviation/mean] × 100) and relative errors (E) (calculated as % E = [(theoretical value − calculated value)/theoretical value] × 100) of HPLC determinations ranging between 0.2% and 39.5% and between 0.3% and 58.4%, respectively. HPLC is a sensitive and reproducible method for determining levels of azole components in a great variety of matrices, as has been confirmed in previous reports (1, 2, 5, 10-12, 15, 19-21, 22). The detection limit of our HPLC-based technique (1 μg/ml) did not permit us to know what happened for lower concentrations in the trays.
Overall, no significant statistical differences (P > 0.01) were found between the three methods for FLC quantification. Notably, although relative errors were greater for the lowest concentrations, we did not observe an increasing inaccuracy when performing serial dilutions compared to dilutions performed following the ISO 20776-1 procedure as would have been expected if the precision of modern pipettes had been insufficient. Thus, serial dilution, which is attractive due to its simplicity, may be appropriate. Regarding serial 2-fold dilution procedures, better results were detected when FLC plates were prepared according to the hydrophilic rather than the hydrophobic way.
Our data showed similar precision and accuracy within each laboratory of the three procedures used for FLC microplate preparation. However, for itraconazole, significant differences were found per center, number of repetitions, and method by the ANOVA (P < 0.05). The greatest discrepancy was observed for the lowest concentrations and for the dilutions prepared outside Spain and shipped for concentration determinations. Thus, ITC data obtained from trays prepared by the Inn Austria participant were excluded from the analysis since tray wrappings showed evidence of defrosting and spillage; also, concentrations of dilutions prepared in Denmark were unacceptably different from the target concentrations. ITC precipitation after thaw-freeze cycles during transport is the most probable explanation. Furthermore, previous studies have demonstrated a much higher interlaboratory variation in MIC results for quality control strains for itraconazole than for flucytosine and fluconazole (6-8). Considering only data for dilution series prepared in Spain, no systematic difference was detected between ISO dilution and serial dilution.
ITC is a weakly basic drug (pKa ca. 3.7) with very low water solubility (RPMI is a hydrophilic medium) (14). This azole compound can only be ionized and solubilized in water at a very low pH, and also earlier studies have indicated that poor solubilization can have a significant influence on generated MIC values (13). Since the hydrophilic growth medium RPMI-G at pH 7 has been selected for testing in vitro antifungal susceptibility, it was important to describe the lack of proper solubility of this compound.
In summary, our data suggest that microdilution trays for hydrophilic drugs such as FLC can be accurately prepared following the ISO 20776-1 recommendations or the serial 2-fold dilution schemes. However, preparation of ITC trays for antimicrobial susceptibility testing (AST) microdilution tests is associated with a significant lack of reproducibility, particularly after freezing and shipping, independent of which dilution method was followed. Quality control procedures of AST should be implemented in order to minimize the inaccuracy of dilution methods and lack of reproducibility when hydrophobic compounds such as ITC are tested.
We have no conflict of interest to declare.
Published ahead of print on 10 March 2010.
†The authors have paid a fee to allow immediate free access to this article.