ABSTRACT
The antifungal susceptibility tests used in clinical laboratories have several limitations. We developed a new test, SensiFONG, based on the detection of chitin levels after exposure to antifungal drugs. The optimal culture conditions were 30°C for 6 h for yeast strains and 26°C for 16 h for molds. The strains were exposed to a range of echinocandin or azole concentrations. Chitin was stained with calcofluor white. The percentage of fungal cells with high chitin levels was determined with an automatic epifluorescence microscope. The SensiFONG results were compared to those with the EUCAST method. Image acquisition and analysis were performed with ScanR software. Fifty-nine strains (28 Candida albicans, 17 Candida glabrata, and 14 Aspergillus fumigatus) were analyzed. Thresholds for the classification of strains as resistant or susceptible were determined for each fungal species. The strains displaying an increase in chitin content of ≥32% for C. albicans, ≥6% for C. glabrata, and ≥17% for A. fumigatus were considered susceptible. The application of these thresholds to all 59 strains resulted in a sensitivity of 0.87, 0.93, and 1.00 and a specificity of 0.93, 0.84, and 0.82 for C. albicans, C. glabrata, and A. fumigatus, respectively. The correlation between the results obtained in the SensiFONG and EUCAST assays was excellent. We developed a new test, SensiFONG, based on a new concept. While current assays assess growth inhibition, our test detects changes in chitin levels after exposure to antifungal drugs. Here, we present preliminary results and we propose a proof of concept of this methodology.
INTRODUCTION
Invasive fungal infections (IFIs) in immunocompromised patients, such as solid organ or stem cell transplant recipients and AIDS patients, has become a major public health problem. Patients in intensive care units are also at high risk of IFIs due to a number of factors, including the use of invasive medical devices and long-term antibiotic treatments. IFIs are frequently associated with high overall mortality, prolonged hospital stay, and excess costs (1–5). New low-toxicity antifungal drugs, such as the new azoles and echinocandins, have been licensed in the last few decades and are now widely used for curative, prophylactic, or empirical antifungal treatments. However, the increasing use of these agents has increased selective pressure, resulting in an increase in antifungal resistance (6–9). Rapid clinical failures have been reported, particularly during the management of invasive candidiasis and aspergillosis (10–14).
The early initiation of efficient antifungal therapy is crucial to improve the prognosis. Treatment must also be adapted to the fungal strains and their susceptibility profiles (11, 12). Antifungal susceptibility tests (AFSTs) have been developed and standardized by the international expert committees of EUCAST and the CLSI. Two reference methods have been described, both based on broth microdilution (15–20). Various commercial systems, such as Etest (bioMérieux, Marcy-l’Etoile, France), Sensititre (Trek Diagnostic System Ltd., East Grinstead, UK), and Fungitest (Bio-Rad, France), are routinely used in clinical mycology laboratories. These assays determine the MIC of the antifungal drug, which is then used to classify the strain as susceptible (S), intermediate (I), or resistant (R) according to the clinical breakpoints (CBPs) recommended by the CLSI (21) or EUCAST (22) or to epidemiological cutoff values (ECOFFs). Other methods based on disk diffusion classify isolates directly as S, I, or R according to the diameter of the inhibition zone. All these methods, whether giving an MIC evaluation or direct S/I/R classification, are based on evaluations of the inhibition of fungal growth upon exposure to the antifungal agent. As such, their reading and interpretation are subject to the so-called “trailing effect” or “trailing endpoint.” This effect results from incomplete inhibition with residual growth over a broad range of MIC values specifically encountered with fungistatic drugs, and it leads to poorly reproducible operator-dependent results (23). This phenomenon leads to a risk of variation in the MIC results obtained, with random misclassification errors (24). Furthermore, as a long time period is required to detect the inhibition of fungal growth, current methods take at least 24 and sometimes 48 h to deliver results, which might delay optimal treatment. There is, therefore, a need to develop a more rapid method to optimize clinical treatment and patient outcome.
The major classes of antifungal agents currently used to treat IFIs are azoles, echinocandins, and polyenes. These drugs mostly target the cell wall and/or the plasma membrane (25). Previous studies have reported the upregulation of several stress-activated signaling pathways, such as the protein kinase C (PKC), mitogen-activated protein (MAP) kinase, high osmolarity glycerol (HOG), and Ca2+ cascades, to counteract cell wall damage, resulting in an upregulation of chitin synthesis (26–28). The novel AFST method presented here is based on the detection of this compensatory mechanism. We have developed an image cytometry assay (SensiFONG) for screening changes in the cell wall. Here, we describe this new methodology and present preliminary results. We have compared SensiFONG results with those obtained with the gold standard method, the EUCAST assay, for susceptible and resistant reference and clinical strains (18–20).
RESULTS
Assay optimization.The SensiFONG assay was developed with synthetic complete (SC) medium, which is suitable for Candida albicans and Aspergillus fumigatus but not for Candida glabrata. Indeed, no significant fluorescent signal was detected in C. glabrata after the 6 h incubation period or after an increased exposure time. The use of RPMI 1640 medium at 37°C rendered the calcofluor white (CFW) staining more obvious at 6 h. Chitin response was obtained in 6 h for A. fumigatus; however, hyphal germination was not optimal after 6 h of incubation in SC medium, and we therefore used an overnight incubation (16 h).
Repeatability and reproducibility of the SensiFONG assay.The repeatability of chitin content determinations and cell counts was evaluated. We analyzed C. albicans strain SC5314 ten times in the same experiment with fluconazole. The intraplate coefficients of variation (CVs) were 10.4% for chitin content and 1.9% for yeast counts. For evaluation of the reproducibility of the SensiFONG assay, C. albicans strain SC5314 exposed to fluconazole was analyzed by three operators of the same laboratory on three different days. Interplate CVs were 13% for chitin content and 5.9% for yeast counts. The repeatability and reproducibility of the SensiFONG assay were, therefore, satisfactory.
Antifungal susceptibility profile determination with the SensiFONG assay.The determination of the antifungal drug susceptibility profile with the SensiFONG assay is based on evaluations of the proportion of Candida sp. or Aspergillus sp. cells with a high fluorescence and the number of cells or the length of hyphae. We used both reference and well-characterized resistant clinical strains to establish the experimental conditions for our assay (Table 1). Figures 1 and 2 show examples of susceptible profiles for yeast and filamentous strains, respectively. In S strains, the percentage of cells with a high fluorescence increased, and the antifungal agent significantly inhibited cell proliferation (smaller cell number or hyphal length). Conversely, in R strains, the proportion of cells with a high fluorescence did not increase following exposure to the antifungal agent, even at high concentrations (Fig. 1 and 2).
SensiFONG antifungal susceptibility profiles in Candida spp. Susceptibility profiles obtained by the SensiFONG assay. An evident increase of chitin content and significant inhibition of cell proliferation (cell number) were observed in susceptible yeast strains as determined by EUCAST. No change in chitin content nor in cell proliferation was observed in resistant yeast strains as determined by EUCAST. The red curves illustrate the proportions of cells with elevated chitin contents detected by calcofluor white as a function of antifungal concentration. The green curves illustrate the cell counts. Candida albicans: susceptible strain, SC5314; resistant strains, DSY296 (top) and TOP (bottom). C. glabrata: susceptible strain, ATCC 2001; resistant strain, Tg5.
SensiFONG antifungal susceptibility profiles in A. fumigatus. Susceptible strain, IHEM15656; resistant strains, Ne1 (top) and DPL1035 (bottom). The red curves illustrate the proportions of cells with elevated chitin contents detected by calcofluor white as a function of antifungal concentration. The green curves illustrate the lengths of the germinated tubes as a function of antifungal concentration.
We defined the optimal thresholds for distinguishing between susceptible and resistant profiles by applying the SensiFONG assay to a series of clinical strains. The accuracy of the SensiFONG assay, assessed by calculating the area under the concentration-time curve (AUC) for the receiver operating characteristic (ROC), was good for all species considered together (AUCchitin-all species, 0.881, 95% confidence interval [CI], 0.838 to 0.923; AUCgrowth-all species, 0.800 [0.723 to 0.878]). Similar results were obtained when the Candida spp. were grouped and analyzed separately from Aspergillus fumigatus (AUCchitin-all Candida species, 0.826 [0.767 to 0.886]; AUCgrowth-all Candida species, 0.902 [0.857 to 0.947]). Species-specific analyses improved the results significantly for analyses based on chitin levels alone: AUCchitin 0.957 [0.926 to 0.985]; 0.925 [0.816 to 1.00], and 0.996 [0.986 to 1.00] for C. albicans, C. glabrata, and A. fumigatus, respectively (Fig. 3A and C). In contrast, no improvement in growth inhibition performance criterion was observed when the fungal species were analyzed separately (Fig. 3B and C). We therefore used the change in chitin content as the sole criterion for the interpretation of antifungal drug susceptibility, with thresholds for the increase in chitin content of 32.675% for C. albicans, 6.325% for C. glabrata, and 17.05% for A. fumigatus (Fig. 3). If the proportion of fungal cells with a high fluorescence was at or above this threshold, the strain was considered to be an S strain; by contrast, if the proportion of fungal cells with a high fluorescence was below this threshold, it was considered to be an R strain.
ROC curves for chitin content increasing (A) and cell growth (B) obtained with SensiFONG assay after the incubation of fungi with azoles and echinocandins. ROCs were created for C. albicans strains (n = 25), C. glabrata (n = 15), and A. fumigatus (n = 11). (C) Values corresponding to ROCs in panels A and B.
SensiFONG performance.We interpreted the results for the 59 strains and analyzed the concordance between EUCAST and SensiFONG assays using the following simplified thresholds: 32% for C. albicans, 6% for C. glabrata, and 17% for A. fumigatus. We obtained a sensitivity (Se) of 0.87 (0.77 to 0.94) and a specificity (Sp) of 0.93 (0.84 to 1.00) for C. albicans, Se of 0.93 (0.85 to 0.98) and Sp of 0.84 (0.61 to 1.00) for C. glabrata, and an Se of 1.00 (1.00 to 1.00) and Sp of 0.82 (0.55 to 1.00) for A. fumigatus. A high degree of concordance was observed between the results of the SensiFONG assay and those of the EUCAST reference method (Table 2), with Cohen’s kappa coefficients of at least 0.8 for C. albicans and A. fumigatus and 0.74 for C. glabrata. Categorical agreement was assessed, defined as the percentage of isolates classified in the same category (i.e., as susceptible, intermediate, or resistant isolates) by both techniques (29), as shown in Table 3. A very major error was defined as a strain found resistant by EUCAST and susceptible by SensiFONG. A major error was defined as a strain found susceptible by EUCAST and resistant by SensiFONG. The other disagreements between EUCAST and SensiFONG were defined as minor errors. The highest (100%) categorical agreement was found for isavuconazole. The lowest (64.4%) categorical agreement was found for fluconazole. Very major errors were observed in one C. albicans isolate and voriconazole, two C. glabrata isolates and micafungin, and one C. glabrata isolate and anidulafungin. We observed 11 (9.8%), 2 (2.9%), and 2 (3.5%) major errors with C. albicans, C. glabrata, and A. fumigatus, respectively.
Concordance between the results of the EUCAST method and the SensiFONG assay
Categorical agreement between the EUCAST and SensiFONG methods for in vitro antifungal susceptibility testing of major pathogenic fungi
DISCUSSION
We report here the development of a novel AFST based on the detection of a mechanism compensating for the cell wall damage caused by antifungal drugs. Proof of concept was demonstrated with the most common pathogenic fungal species and the major antifungal agents used in clinical practice. There were differences between species, but the SensiFONG assay was able to discriminate between susceptible and resistant strains in a rapid and reliable manner. The change in chitin levels in the cell wall was detected and quantified in an automated image cytometry assay, and so the interpretation of the results was objective and not operator dependent.
Triazoles and echinocandins are the two main classes of antifungal drugs used for the clinical treatment of invasive fungal infections. The triazoles in current use, fluconazole, voriconazole, itraconazole, posaconazole, and isavuconazole, block ergosterol synthesis, thereby altering the fungal membrane. The echinocandins used, anidulafungin, micafungin, and caspofungin, block (1,3)-β-d-glucan synthase, thereby causing cell wall synthesis defects (25). Stresses related to the depletion of ergosterol from the cell membrane or glucan from the cell wall can be detected by sensors that activate signaling pathways, increasing the expression of chitin synthase genes (30). The induction of the mechanism compensating for cell wall damage and the resulting increase in the chitin content of the cell wall can be seen as evidence of the activity of azoles and echinocandins in susceptible fungal strains. Conversely, in resistant strains, in which the membrane and cell wall are not damaged, this repair response is not activated.
Calcofluor white is a fluorescent stain that binds strongly to structures containing chitin and cellulose. It is widely used to sharply delineate fungal elements in clinical specimens (31). We used this stain to label the chitin in the fungal cell wall. The chitin content of the cell wall was variable and depended on the species of Candida or Aspergillus. The exposure time required for the measurement of fluorescence was longer for C. glabrata than for other Candida spp. (20 ms versus 3 to 5 ms), because the fluorescence intensity was lower in this species, consistent with the flow cytometry findings of Costa-de-Oliveira et al. (32). We therefore used different culture conditions for different species, with the rich RPMI 1640 medium and a culture temperature of 37°C for C. glabrata to favor the compensatory phenomenon. However, the conditions used for C. glabrata could not be applied to all Candida spp., because they induce the yeast-to-hyphae transition, impairing the gating of the yeast forms.
The antifungal drug susceptibility profiles predicted by the SensiFONG assay are based on evaluations of the proportion of fungal cells with a high chitin content after antifungal drug treatment relative to the proportion of such cells in the absence of the antifungal drug. Chitin levels were evaluated by image cytometry, which can be used not only to analyze the chitin content per cell but also to relate the fluorescence level of each cell to its area, potentially preventing interpretation biases due to differences in the sizes of fungal cells. We also analyzed growth inhibition, the basis of all the other AFSTs currently available, but our data indicated that the detection of growth inhibition by the SensiFONG assay was less sensitive and less specific than that for the change in chitin levels for interpreting the phenotype of the strains. Indeed, in growth inhibition tests, the partial growth detected as microcolonies on solid media or as mild turbidity in liquid media should be disregarded when the results of the test are read. Such features are not very evident, and difficulties in their interpretation often lead to operator-dependent results. As SensiFONG is not based on growth inhibition, we can reasonably predict that the reading and interpretation of this chitin-based assay is not subjected to the trailing effect. However, additional studies addressing specifically this feature are needed.
In conclusion, we have developed an innovative AFST based on the detection of chitin fluorescence for rapid determination of the antifungal drug susceptibility profiles of fungal strains. Further studies, with more fungal species and more clinical isolates, are required to confirm the reliability of the SensiFONG assay and to adjust the chitin thresholds, but the SensiFONG assay can already be considered as an objective assay with automated reading and interpretation that can be performed more rapidly than conventional diagnostic assays.
MATERIALS AND METHODS
Strains.Yeast and filamentous fungal species of clinical importance, such as Candida albicans, C. glabrata, and Aspergillus fumigatus, were used in this study. Eight susceptible reference strains and clinical strains harboring mutations associated with antifungal drug resistance were used to optimize the experimental conditions (Table 1). We then used 51 clinical strains, including 25 C. albicans, 15 C. glabrata, and 11 A. fumigatus strains, obtained from the clinical mycological laboratory of Grenoble University Hospital, France, or kindly given by colleagues (see Acknowledgments), to evaluate the performance of the SensiFONG test and to provide proof of concept for this new assay (Table 4). Fungi stored at −80°C were cultured on yeast extract-peptone dextrose (YPD) agar plates at 30°C for 48 h for Candida. Aspergillus strains were incubated on chloramphenicol-Sabouraud agar dishes (Becton, Dickinson, France) at 35°C for 48 h. The colonies were then used for EUCAST or SensiFONG assays.
EUCAST assay.The EUCAST reference method was used to establish the susceptibility profiles of the strains (Tables 1 and 4) and as the gold standard for this study. The protocol used conformed to EUCAST specifications for determination of the MICs for yeasts and filamentous fungi (19, 20). The Candida parapsilosis ATCC 22019 strain was used for quality control. Two classes of antifungal drugs, azoles (fluconazole [FLC] and voriconazole [VRC] for Candida spp., isavuconazole [ISA] and VRC for Aspergillus spp.) and echinocandins (anidulafungin [ANI] and micafungin [MICA] for both Candida spp. and Aspergillus spp.), were tested. All the antifungal drugs were provided in powder form. They were dissolved in dimethyl sulfoxide (DMSO) and stored at −80°C. The ranges of tested concentrations are indicated in Table 5. EUCAST MICs were determined at endpoint, following optical density (OD) measurement by spectrophotometry.
Ranges of antifungal agent concentrations used in this study
SensiFONG assay.C. albicans cells were grown in liquid synthetic complete (SC) medium (2% glucose, 0.5% ammonium sulfate [Sigma, France], 0.17% yeast nitrogen base [Sigma, France], 0.2% synthetic complete mix [Bufferad, USA]) buffered at pH 7 with HEPES (Euromedex, France) at 30°C with shaking at 200 rpm for 6 h with or without antifungals tested at the concentrations described in Table 5. For C. glabrata, we used RPMI 1640 (Thermo Fisher Scientific, France) supplemented with 1.8% glucose and buffered at pH 7 with N-morpholino propanesulfonic acid (MOPS) (Sigma, France) at 37°C with shaking at 200 rpm for 6 h. A. fumigatus strains were cultured in SC medium (pH 7) at 26°C for 16 h without shaking. All antifungal drug susceptibility tests were performed in 96-well microplates in a final volume of 200 μl/well. We used inoculum concentrations of 2 × 105 CFU/ml for C. glabrata and 5 × 104 CFU/ml for C. albicans and A. fumigatus. The response of each strain to each concentration of antifungal agent was evaluated in duplicates. Cells cultured without antifungal reagents were used as controls. At the end of the incubation period, 2 μl of calcofluor white (CFW) (Sigma) at 10 μg/ml was added. CFW binds chitin, chitosan, and cellulose and was previously used in flow cytometry to detect chitin content in yeasts (32, 33). The plates were incubated at room temperature in the dark for 10 to 15 min and then centrifuged at 100 × g for 1 min to pellet the fungal cells.
Image cytometry.Image acquisition, processing, and analysis were performed with an epifluorescence microscope and ScanR analysis software, according to the manufacturer’s instructions (Olympus, France). Images were obtained with a 20× objective (numerical aperture [NA], 0.45) in the fluorescence channel for CFW (excitation at 360 to 370 nm, emission at 420 to 460 nm). We imaged 16 fields per well to ensure the capture of sufficient numbers of cells to obtain statistically significant results. Image processing was performed as follows. Only particles tagged by CFW are visible under a fluorescence microscope (Fig. 4Aa). Image processing began with the application of background subtraction to all the images to reduce noise (Fig. 4Ab). An outline segmentation algorithm with a minimum object size of 10 pixels and a maximum object size of 30 pixels was then applied to target individual fungal cells (Fig. 4Ac). Finally, a watershed algorithm was applied in Candida spp. to separate individual yeast cells from each other in cell aggregates, to ensure that the cells were well detected as separate circular or ovoid fluorescent objects (Fig. 4Ad); only the first three steps of this algorithm were applied to A. fumigatus (Fig. 4Ba to c).
SensiFONG image processing. Candida sp. (A) and Aspergillus sp. (B) cells as processed by the SensiFONG assay ScanR Olympus software. (a) Raw images. (b) After manual adjustment of the brightness and the contrast of filtered images. (c) Selection of fungal cells with an edge segmentation algorithm with a minimum object size of 10 pixels and a maximum object size of 30 pixels. (d) The masks obtained show the fungal cells that were selected. Data displayed as scatter plots and histograms. (C) Target cells (R01 in green) were gated in the first scatter plot (area versus circularity) to exclude aggregates of Candida spp. and debris. Mean fluorescence intensity was measured for events gated on R01. (D) The threshold to discriminate the normal chitin content was utilized for cells without antifungals (R02 in blue). (E) The fungal cells showing increased chitin content were gated in R03 (in red). The mean length of germinated tube in Aspergillus was measured for events gated on R01 without (F) or with (G) antifungal.
In the first scatter plot (Fig. 4C), area-versus-circularity gating (R01) was applied to gate the fungal cells (Fig. 4C) and exclude cell aggregates. ScanR analysis software was used to determine the mean fluorescence intensity of each gated fungal cell in R01 (Fig. 4C). In Fig. 4D, R02 was the gate used to identify cells with a low fluorescence, and R03 was the gate used to identify cells with a high fluorescence content. The R02 and R03 gates were set on the basis of the control without antifungal reagent, so as to minimize the number of cells in the R03 gate under these conditions (Fig. 4D). When cells were incubated with antifungal agents, the number of cells in the R03 gate increased (Fig. 4E). Besides fluorescence, the number of cells in the R01 gate for each well (Fig. 4C) was used for yeast. For Aspergillus spp., the length of the germination tube was determined as the maximum Feret diameter (in microns) with ScanR software (Fig. 4F and G).
Prism 6.0d software (GraphPad Software, Inc.) was used to illustrate the results.
Interpretation of the results.The results are reported as a classification into susceptible (S)/intermediate (I)/resistant (R) strains for the EUCAST assay and S/R for the SensiFONG assay. For the EUCAST method, the MIC was interpreted according to the clinical breakpoints (CBPs) or epidemiological cutoff values (ECOFFs) of the EUCAST Committee (19, 20). For the SensiFONG assay, the percentage of cells with higher fluorescence levels than control without antifungal was used to distinguish between S and R strains.
Statistical methods.Optimal thresholds were identified by obtaining predicted values for resistant profiles relative to S/I profiles for the EUCAST method with logistic regression models for all species (Candida spp. and A. fumigatus), all Candida species (C. albicans and C. glabrata), and separately for A. fumigatus, C. albicans, and C. glabrata. These values were used to construct ROC curves and to calculate the AUC and its 95% confidence interval. The optimal thresholds obtained were rounded for clinical use, and their sensitivity and specificity were assessed and 95% confidence intervals calculated by the bootstrap method. The R package pROC was used. The concordance of SensiFONG and EUCAST results was assessed by calculating the Cohen kappa coefficient. The scale used to assess the degree of agreement was as follows: κ of ≤0.2, slight; 0.21 to 0.40, fair; 0.41 to 0.60, moderate; 0.61 to 0.80, substantial; and 0.81 to 1, almost perfect agreement (34).
ACKNOWLEDGMENTS
We thank D. Sanglard, P. Lepape, C. d’Enfert, ME Bougnoux, A. Fekkar, L. Cowen, A. Paugam, T. Noël, and D. Perlin for the kind gift of strains used in this study.
This study was supported by SATT Linksium Grenoble, University Grenoble Alpes, France, and Centre National de la recherche scientifique, France. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
FOOTNOTES
- Received 3 June 2019.
- Returned for modification 25 June 2019.
- Accepted 11 October 2019.
- Accepted manuscript posted online 28 October 2019.
- Copyright © 2019 American Society for Microbiology.
