Evaluation of the “Dip Effect” Phenomenon in Antifungal Susceptibility Testing of Candida spp. against Echinocandins by Use of Gradient Concentration Strips

Abstract

The “dip effect” phenomenon complicates antifungal susceptibility testing with gradient concentration strips. Of 60 Candida isolates tested with the three echinocandins, this phenomenon was observed only for caspofungin with most (>90%) Candida albicans, Candida glabrata, and Candida tropicalis isolates and for isolates with CLSI MICs of ≤0.25 mg/liter. In order to facilitate MIC determination, a practical approach was developed using the inhibition zones at 32, 8, 2, and 1 mg/liter, increasing the agreement with the CLSI method >86%.

TEXT

Standardized broth microdilution methods of the Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) organizations are universally accepted as reference procedures for antifungal susceptibility testing of yeast (1, 2). However, they are labor-intensive and time-consuming and not easily incorporated into busy clinical settings. Although products for traditional broth microdilution MIC testing are commercially available, there are alternative agar-based assays that appear to be advantageous in terms of use, convenience, and flexibility. Gradient concentration diffusion susceptibility testing exploits the properties of drug diffusion on an inoculated agar surface by utilizing a predefined concentration gradient of a single antifungal agent stabilized on a strip to directly generate MIC values that correspond to the growth inhibition in an elliptical zone. Regardless of the widespread use of this type of testing in clinical laboratories, technical difficulties concerning reading and interpreting the MIC values with these tests are still encountered. In particular, the “dip effect” phenomenon, which is defined as a narrow inhibition zone at low concentrations next to the strip, may hinder objective determination of MICs of echinocandins against yeast isolates, requiring experience and guidance by the manuals provided by the manufacturer. Although this phenomenon is described in the manufacturer's instructions for gradient concentration strip methods for bacteria and fungi, it has not been investigated so far in the literature (3, 4). Therefore, the aim of the present study was to assess the performance of gradient concentration strips versus the broth microdilution method for in vitro susceptibility testing of echinocandins against Candida spp. and to develop a practical approach that improves MIC determination when the dip effect occurs.

A total of 60 clinical bloodstream isolates of Candida spp., including 12 C. albicans, 10 C. glabrata, 12 C. parapsilosis, 11 C. tropicalis, 4 C. krusei, and 4 C. dubliniensis isolates and 5 belonging to rare species (1 each of C. inconspicua, C. kefyr, C. lipolytica, C. lusitaniae, and C. rugosa) and the 2 quality control isolates C. krusei ATCC 6258 and C. parapsilosis ATCC 22019, were tested against anidulafungin, caspofungin, and micafungin. All isolates were identified with the Vitek system, verified with the Auxacolor system, and stored in normal saline with 10% glycerol at −70°C until the study was performed. The isolates were chosen in order to encompass various species and different MICs of echinocandins as determined with the Sensititre YeastOne (SYO) microdilution method. Although biochemical identification of rare species is problematic, the 5 isolates belonging to rare species were chosen on the basis of the highest level of certainty with both biochemical tests and particularly of their in vitro susceptibility to echinocandins. Prior to testing, each isolate was revived by subculturing it twice onto Sabouraud dextrose agar plates with gentamicin and chloramphenicol (SGC2; bioMérieux) at 30°C for 24 h to ensure purity and viability. The MICs of caspofungin, micafungin and anidulafungin were determined with the Sensititre YeastOne panels (SYO; Trek Diagnostic Systems, Cleveland, OH, USA) and the gradient concentration MIC test strips (MTS; Liofilchem, Roseto degli Abruzzi, Italy). For drugs demonstrating the dip effect phenomenon, the MICs were also determined with the Etest gradient concentration strips (bioMérieux, Marcy l'Etoile, France) and the CLSI broth microdilution reference method.

For the gradient concentration strip methods, inoculum suspensions equivalent to a 0.5 McFarland standard were prepared in normal saline and solidified RPMI (2% glucose, buffered with morpholinepropanesulfonic acid [MOPS]) agar plates (Liofilchem, Roseto degli Abruzzi, Italy) were inoculated via pouring a 1:5 dilution of the standardized yeast suspension onto the agar according to the manufacturer's recommendations. After allowing 1 to 2 min for the suspension to achieve a uniform distribution, excess moisture was absorbed into the agar, the surface was left to dry completely (15 to 20 min at 37°C), and the gradient concentration strips (MTS or Etest) were applied to the center of each inoculated plate. For the SYO method, a suspension of yeast in demineralized water was adjusted to match the turbidity of a 0.5 McFarland standard; 20 μl of this was transferred into an 11-ml tube of SYO inoculum broth, which was then used to inoculate the wells of the SYO plates. For the broth microdilution reference method, the CLSI M27-A3 protocol (2) was followed, using a final inoculum of 0.5 × 10³ to 2.5 × 10³ CFU/ml in 2-fold serial dilutions ranging from 0.005 mg/liter to 8 mg/liter in RPMI 1640 medium (with l-glutamine, without bicarbonate) (AppliChem, Darmstadt, Germany) buffered to pH 7.0 with 0.165 M MOPS (morpholinepropanesulfonic acid) (AppliChem, Darmstadt, Germany). CFU counts were affirmed each time by spread plate counts on SGC2 plates.

For each method, the echinocandin MICs were determined in accordance with the respective instructions after 24 h of incubation at 35°C. The CLSI MIC was read visually as the lowest concentration of drug corresponding to a prominent decrease in turbidity (50% reduction in growth) relative to that of the drug-free control, and the SYO MIC was recorded as the lowest concentration of antifungal agent preventing the development of a red color (first blue or purple well), whereas the MTS and Etest MIC was determined as the lowest drug concentration at which the border of the elliptical inhibition intercepted the strip scale, ignoring trailing growth. In the presence of the dip effect, the MICs were determined as (i) the drug concentration where the inhibition zone intersected the strip at the bottom of the dip, according to the recommendations of the companies' reading guides (3, 4); (ii) the drug concentration where the inhibition zone intersected the strip, extrapolating visually the elliptic inhibition zone at the top of the dip; and (iii) with a new approach consisting of a linear regression analysis of inhibition zones at supra-MICs (y scale) over the corresponding log₂ concentrations (x scale) after subtracting the inhibition zone at the bottom of the dip effect. When no dip effect was present, the width of the strip was subtracted from the inhibition zones at supra-MIC concentrations. Preliminary analysis with 3, 4, and 5 points at 2- and 4-fold concentration ranges indicated that log-linear regression analysis of inhibition zones at 32-, 8-, and 2-mg/liter concentrations provides the best results with minimal points for all Candida species except C. albicans, for which the inhibition zone at 1 mg/liter was also used. The MIC was then defined as the x intercept of the linear regression analysis rounded up to the next 2-fold concentration (Fig. 1).

FIG 1.

The dip effect as a narrow inhibition zone d at sub-MICs parallel to the strip and MIC determination by log-linear regression analysis of inhibition zones A, B, and C at supra-MICs (32, 8, and 2 mg/liter) in a C. tropicalis 261 isolate. The MTS MIC of caspofungin was <0.002 mg/liter according to the manufacturer's instructions (bottom of the dip), 0.125 mg/liter when determined visually at the top of the dip, and 0.06 mg/liter using the new approach developed in the present study (right graph). The Etest MIC was 0.125 mg/liter and, using the new approach, 0.06 mg/liter without the isolate, demonstrating an obvious dip effect (Fig. 2B). The MIC of the isolate was also 0.06 mg/liter with both CLSI and Sensititre YeastOne microdilution methods.

The number of isolates with the dip effect was calculated for each species and associated with MICs using Fisher's exact test. The agreement within ±1 and ±2 2-fold dilutions and the 2-fold differences between the MTS and the Etest compared to the SYO and the CLSI MICs were calculated after rounding up the MTS and the Etest MICs to the next 2-fold concentration for isolates with and without the dip effect phenomenon.

No dip effect was found with the MTS of anidulafungin and micafungin, whereas the dip effect was observed with caspofungin MTS for 12/12 C. albicans isolates, 9/10 C. glabrata isolates, 11/11 C. tropicalis isolates, 0/5 C. krusei isolates, 2/13 C. parapsilosis isolates, 4/4 C. dubliniensis isolates, 1/1 C. inconspicua isolate, 1/1 C. kefyr isolate, 0/1 C. lipolytica isolate, 1/1 C. lusitaniae isolate, and 1/1 C. rugosa isolate. A significant correlation was found between MIC values and percentages of isolates with the dip effect (P < 0.0001). The dip effect phenomenon was found for 0/17 isolates with CLSI MICs of ≥0.5 mg/liter and 42/43 isolates with CLSI MICs of ≤0.25 mg/liter. Isolates demonstrating the dip effect phenomenon had MTS MICs (determined as the top of the dip effect) ranging from 0.125 mg/liter to 1 mg/liter (median, 0.5 mg/liter). For these isolates, a short or no obvious dip effect was observed with caspofungin Etest strips (Fig. 2A and B). Notably for these isolates, a more elongated elliptical inhibition zone and a smaller intersection angle were observed compared to isolates without a dip effect (Fig. 2A and B versus C). The Etest MICs of the isolates with a dip effect were higher than the Etest MICs of the isolates without the dip effect (median of 0.125 and range of 0.03 to 0.5 mg/liter versus median of 0.5 and range of 0.25 to 2 mg/liter, respectively; P < 0.0001). When the new approach was implemented, the fitness of log-linear regression analysis was excellent for all isolates (median R², 0.96, range R², 0.89 to 1).

FIG 2.

Etest photos for C. tropicalis 661 isolate demonstrating a short dip effect using the Etest strips and a pronounced dip effect using the MTS (A), C. tropicalis 261 isolate without an obvious dip effect using the Etest strips and a pronounced dip effect using MTS (Fig. 1) (B), and C. parapsilosis 397 isolate with no obvious dip effect using the Etest and the MTS (C). Note the more elongated elliptical inhibition zones and the smaller intersection angles with the strips of the isolates with a dip effect (A and B) compared to the isolate without a dip effect (C).

The agreement within ±1 and ±2 log₂ dilutions between MTS and SYO MICs was 90% and 97%, respectively, for anidulafungin with a median (range) difference of 0 (−1 to 4) 2-fold dilutions and 73% and 97%, respectively, for micafungin with a difference of 1 (−1 to 3) 2-fold dilutions. For caspofungin, the agreement within ±1 and ±2 log₂ dilutions between MTS and SYO methods was 94% and 100%, respectively, for isolates with no dip effect with a median (range) 2-fold difference of 0 (−1 to 2) 2-fold dilutions. For isolates demonstrating the dip effect phenomenon, the corresponding levels of agreement for caspofungin were 0% and 0%, respectively, when MTS MICs were determined at the bottom of the dip effect according to the manufacturer's instructions (all MTS MICs were ≤0.006 mg/liter since inhibition zones intersected at very low concentrations) and 24% and 56%, respectively, when MTS MICs were determined at the top of the dip effect as a result of MTS MICs being significantly higher, by 2 (range, 0 to 4) 2-fold dilutions, than those of the SYO method. The agreement between these two methods increased to 73% and 90%, respectively, with 0 (−3 to 3) 2-fold differences when MTS MICs were determined with the new approach.

The findings for caspofungin where the dip effect occurred were verified using the CLSI reference method and another gradient concertation strip method, the Etest. The agreement within ±1 and ±2 log₂ dilutions between MTS and CLSI was 71% and 100%, respectively, for isolates without the dip effect with a median (range) −1 (−2 to 1) 2-fold difference. For isolates demonstrating the dip effect phenomenon, there was 0% agreement between MTS and CLSI when caspofungin MICs were determined at the bottom of the dip according to the MTS manufacturer's guidelines since all MTS MICs were ≤0.006 mg/liter. When the MTS MICs were determined visually at the top of the dip, the corresponding levels of agreement between MTS and CLSI were 37% and 73%, respectively, as a result of MTS MICs being significantly higher, by 2 (range, 0 to 3) 2-fold differences. The agreement between these two methods increased to 88% and 100%, respectively, with 0 (−2 to 2) 2-fold difference when MTS MICs were determined with the new approach (Table 1).

TABLE 1.

Agreement between MTS gradient concentration strips and CLSI broth microdilution reference method for caspofungin MIC determination for isolates with a dip effect

Species (n with dip effect/all isolates)	Median (range) MIC (mg/liter) by method:			MTS
				Top of the dip			New approach
				Median (range) MIC (mg/liter)	% agreement at no. of 2-fold dilutions:		Median (range) MIC (mg/liter)	% agreement at no. of 2-fold dilutions:
	Etest	Sensititre	CLSI		±1	±2		±1	±2
C. albicans (12/12)	0.125 (0.03–0.5)	0.06 (0.008–0.25)	0.125 (0.016–0.25)	0.25 (0.125–0.50)	58	83	0.125 (0.016–0.5)	75	92
C. glabrata (9/10)	0.25 (0.03–0.5)	0.125 (0.06–0.25)	0.125 (0.06–0.25)	0.50 (0.25–0.50)	22	89	0.125 (0.03–0.25)	100	100
C. parapsilosis (2/13)	0.4 (0.25–0.5)	0.19 (0.125–0.25)	0.125 (0.125–0.125)	0.50 (0.50–0.50)	0	100	0.09 (0.06–0.25)	100	100
C. tropicalis (11/11)	0.125 (0.06–0.5)	0.06 (0.016–0.25)	0.06 (0.016–0.25)	0.50 (0.125–0.50)	27	55	0.06 (0.03–0.25)	100	100
Rare Candida spp. (7/12)	0.25 (0.06–0.5)	0.125 (0.03–0.25)	0.125 (0.06–0.25)	0.50 (0.25–1.0)	43	57	0.06 (0.03–0.25)	86	100
All isolates (41/60)	0.125 (0.03–0.5)	0.06 (0.008–0.25)	0.125 (0.016–0.25)	0.50 (0.125–1.0)	37	73	0.125 (0.016–0.5)	88	100

The agreement within ±1 and ±2 log₂ dilutions between Etest and CLSI was 63% and 93%, respectively, for the isolates with the dip effect (Etest MICs determined at the bottom of the dip effect whenever present were 1 [−2 to 4] dilution higher than CLSI MICs) and 94% and 100%, respectively, for isolates without the dip effect (Etest MICs were 0 [−2 to 1] dilution different from CLSI MICs). Of note, when the new approach was applied to the isolates with the dip effect, the agreement within ±1 and ±2 log₂ dilutions between Etest and the CLSI increased to 86% and 100%, respectively, with a median (range) 0 (−2 to 2) 2-fold difference. When the new approach was applied to the isolates without the dip effect, the agreement remained high (94% and 100% within ±1 and ±2 log₂ dilutions, respectively).

The present study investigated for the first time in the literature the impact of the dip effect phenomenon on in vitro susceptibility testing of echinocandins against Candida spp. using gradient concentration strips. Our findings indicate a drug and MIC dependence of the phenomenon, as it was observed only for caspofungin and its occurrence was inversely related to the MIC of the isolate. In the context of its impact on the poor agreement between broth dilution and gradient concentration strips, we developed a practical approach based on linear regression analysis which significantly improved the percentages of agreement within ±1/±2 log₂ dilutions to 88%/100% from 0%/0% (bottom of the dip) and 37%/73% (top of the dip) for the MTS and to 86%/100% from 63%/93% for the Etest even for isolates with a short or no obvious dip effect with the latter strips.

Caspofungin susceptibility testing using the reference standard methods of CLSI and EUCAST is challenging. Espinel-Ingroff et al. (5) recently documented a significant interlaboratory variation in CLSI/EUCAST caspofungin MICs for Candida spp. As variable CLSI and EUCAST modes were observed (spanning 5 to 6 dilution steps among the most common Candida spp.), Espinel-Ingroff et al. declared that this may potentially result in incorrect categorization of the results (5). Furthermore, Dannaoui et al. compared the results that were obtained in a reference laboratory using EUCAST techniques and on a routine basis in nine hospital laboratories using gradient concentration strips and found that there was a 3-fold difference in caspofungin median MICs obtained by the two methods (6). On the other hand, Sensititre YeastOne is a commercially prepared 96-well broth microdilution system comprising increasing concentrations of antifungals and relies on the colorimetric change of alamarBlue as an indicator of fungal growth. In the first multicenter, retrospective study, Eschenauer et al. showed that the Sensititre YeastOne assay may reduce the variability in caspofungin MICs against Candida spp. that is observed between reference laboratories using CLSI methods (7). Moreover, the Sensititre YeastOne antifungal panel has been evaluated for susceptibility testing of Candida spp. against anidulafungin and micafungin compared with the reference microdilution method, and excellent agreement was concluded (8, 9). In the present study, routine isolates were screened for their susceptibility to echinocandins using the Sensititre YeastOne broth microdilution method and the susceptibility results were verified using the CLSI reference method for isolates demonstrating the dip effect phenomenon.

A high agreement (>73%) between the MTS and the Sensititre YeastOne method was found in the present study for micafungin and anidulafungin and for caspofungin only for isolates without the dip effect phenomenon. The performance of the MTS for caspofungin was severely impacted by the latter phenomenon, since the inhibition zone did not intersect with the strip, resulting in offscale MICs and 0% agreement with broth microdilution methods for the isolates demonstrating this phenomenon. MTS MIC determination using the top of the dip effect improved the agreement with the microdilution reference method (37%/73% within ±1/±2 log₂ dilutions). However, visual determination of the edge at the top of the dip effect is quite challenging and not clear-cut and may be variable depending on the inoculum and incubation conditions as well as the MIC for the isolate. Extrapolating the elliptical inhibition zone observed at the top of the dip is also difficult, since the curvature of this zone is different for different isolates and it is impossible to capture it precisely by eye in order to make the correct extrapolation to the strip. In contrast, the new approach developed in the present study utilizes the inhibition zones at high concentrations, which are more easily determined and less variable. The new approach is amenable to automation and provides a unique MIC since the entire inhibition zone is measured. More importantly, the performance of both Etest and MTS strips for caspofungin susceptibility testing of isolates demonstrating this phenomenon was remarkably improved, since the agreement with the CLSI method increased to levels comparable to those observed with the other echinocandins (>86%). Furthermore, the new approach can be applied also to isolates with no dip effect without comprising the agreement with the reference methodology.

The echinocandins represent a new class of antifungals with a novel mechanism of action, targeting the cell wall synthesis by inhibiting 1,3-beta-d-glucan synthase. They are large semisynthetic lipopeptide molecules derived via chemical modification of natural products of filamentous fungi. Namely, they are amphiphilic cyclic hexapeptides with N-linked acyl lipid side chains which intercalate with the phosphor-lipid bilayer of the fungal cell membrane. The echinocandins have significant structural differences concerning the type of side chain, i.e., anidulafungin has an alkoxytriphenyl side chain, caspofungin has a fatty acid side chain, and micafungin has a complex aromatic side chain (3,5-diphenyl-substituted isoxazole) (10). This parameter differentiates them and is speculated to be the basis for the variations in their solubility, potency, and pharmacokinetic disposition. In addition, caspofungin is the antifungal compound with the lowest protein binding rate and molecular weight (slightly) among the three echinocandins. The principle of agar-based assays is that the rate of diffusion is dependent on the solubility properties of the drug in the agar and the molecular weight of the chemical. Lower-molecular-weight water-soluble substances will diffuse more rapidly, spreading farther through the agar in the same period. Differences in chemical properties among the three echinocandins may explain why the dip effect occurs only with caspofungin. The fact that the phenomenon was observed for isolates with CLSI MICs of ≤0.25 mg/liter may be due to diffusion problems of caspofungin from the strips into agar, which affect isolates with MICs lower than these concentrations but not isolates with MICs higher than these concentrations.

A short or no obvious dip effect was observed with the plastic Etest strips, indicating that the nature of strips may also be important in addition to other factors. Of note, the agreement for isolates demonstrating the dip effect phenomenon (63%/93% within ±1/±2 log₂ dilutions) was lower than for isolates without the dip effect (94%/100% within ±1/±2 log₂ dilutions) with the Etest MICs being 1 2-fold dilution higher than the CLSI MICs. This was also previously reported for most Candida spp., with a modal MIC of 0.03 to 0.06 mg/liter for caspofungin but not for the other two echinocandins and not for the other Candida spp., with a modal MIC of 1 to 2 mg/liter for which the modal Etest MICs were equal to or lower than the modal CLSI MICs (11). The Etest MICs of the isolates with the dip effect were lower than the Etest MICs of the isolates without the dip effect in the present study. Furthermore, when we determined the Etest MICs of the isolates with the dip effect using the new approach, the agreement within ±1 log₂ dilution increased from 63% to 86%. This indicates that diffusion problems may affect results of gradient concentration strips in different ways, with the most extreme situation being a long dip effect where the inhibition zone does not intersect the strip. Short dip effects with strip intersection at low concentrations have been described for the Etest and the MTS strips (3, 4). Interestingly, the photos provided by the two manufacturers to demonstrate the dip effect refer to a C. albicans isolate with a caspofungin MIC of <0.125 mg/liter, which is a typical isolate to demonstrate this phenomenon as found in the present study. However, the effect of this phenomenon on the MIC was not investigated in comparison to a reference microdilution method. Different degrees of dip effect may affect the MICs of gradient concentration strips, particularly for isolates with CLSI MICs of ≤0.25 mg/liter, as found in the present study in a greater degree with the MTS strips and a lesser degree with the Etest strips. Because the dip effect occurred for isolates with MICs close to the CLSI susceptibility breakpoints for caspofungin, namely, 0.25 mg/liter for C. albicans, C. tropicalis, and C. krusei and 0.125 mg/liter for C. glabrata, correct MIC determination is very important in order to avoid classification errors in the susceptibility of isolates as previously reported for caspofungin (12).

The dip effect phenomenon has also been reported in antibacterial susceptibility testing using the Etest and MTS gradient concentration strips for glycopeptides, polypeptides, and ticarcillin-clavulanate combination (3, 4). Although a thorough evaluation of this phenomenon and its impact on MIC is lacking, Etest MICs higher than broth microdilution MICs have been reported for glycopeptides and polypeptides as found for caspofungin in the present study (13, 14). It would be interesting to apply the approach described in the present study for these drugs and compare the results with the reference broth microdilution method.

In summary, the dip effect was observed only in caspofungin MIC testing of different Candida species and its incidence was inversely related to the MIC of the isolate, as no dip effect was found for strains with CLSI MICs of ≥0.5 mg/liter. The proposed formula implemented in the present study is easy to perform and less subject to errors and may prove useful for MIC determination with gradient concentration strips. Since the new approach may not be practical in daily routine work and considering that the MTS MICs determined at the top of the dip effect were 2 (0 to 3) dilutions higher than the CLSI MICs and that the Etest MICs determined at the bottom of the dip were 1 (−2 to 4) dilution higher than the CLSI MICs, the new approach could be used particularly for all isolates with MTS MICs of 0.25 to 1 mg/liter and Etest MICs of 0.03 to 0.5 mg/liter. These isolates may be wrongly classified as nonsusceptible, resulting in major errors even when a dip effect phenomenon is not obvious. Isolates with lower MICs would be susceptible even after correct MIC determination by applying the new approach, whereas isolates with higher MICs would not be affected by the dip effect phenomenon as found in the present study. Thus, the application of the new approach will be particularly important for isolates with gradient concentration MICs of 0.03 to 1 mg/liter to avoid misclassification errors.