Overestimation of Amphotericin B Resistance in Candida auris with Sensititre YeastOne Antifungal Susceptibility Testing: a Need for Adjustment for Correct Interpretation

ABSTRACT Significant variation in minimal inhibitory concentrations (MIC) has been reported for amphotericin B (AMB) and C. auris, depending on the antifungal susceptibility testing (AFST) method. Although the Sensititre YeastOne (SYO) is widely used in routine laboratory testing, data regarding its performance for the AFST of C. auris are scarce. We tested AMB against 65 C. auris clinical isolates with the SYO and the reference methodology by the Clinical and Laboratory Standards Institute (CLSI). The essential agreement (EA, ±1 dilution) between the two methods and the categorical agreement (CA) based on the Centers for Disease Control and Prevention (CDC)’s tentative breakpoint of MIC ≥ 2 mg/L were determined. The SYO wild type upper limit value (WT-UL) was determined using the ECOFFinder. The modal (range) CLSI growth inhibitory MIC was lower than the SYO colorimetric MIC [1(0.25-1) versus 2(1-8) mg/L, respectively]). The CLSI-colorimetric SYO EA was 29% and the CA was 11% (89% major errors; MaE). MaE were reduced when the SYO WT-UL of 8 mg/L was used (0% MaE). Alternatively, the use of SYO growth inhibition endpoints of MIC-1 (75% growth inhibition) or MIC-2 (50% growth inhibition) resulted in 88% CA with 12% MaE and 97% CA with 3% MaE, respectively. In conclusion, SYO overestimated AMB resistance in C. auris isolates when colorimetric MICs, as per SYO instructions and the CDC breakpoint of 2 mg/L, were used. This can be improved either by using the method-specific WT-UL MIC of 8 mg/L for colorimetric MICs or by determining growth inhibition MIC endpoints, regardless of the color. IMPORTANCE Candida auris is an emerging and frequently multidrug-resistant fungal pathogen that accounts for life-threatening invasive infections and nosocomial outbreaks worldwide. Reliable AF is important for the choice of the optimal treatment. Commercial methods are frequently used without prior vigorous assessment. Resistance to AMB was over-reported with the commercial colorimetric method Sensititre YeastOne (SYO). SYO produced MICs that were 1 to 2 twofold dilutions higher than those of the reference CLSI method, resulting in 89% MaE. MaE were reduced using a SYO-specific colorimetric wild type upper limit MIC value of 8 mg/L (0%) or a 50% growth inhibition endpoint (3%).

Moreover, the fact that C. auris acquired resistance to different classes of antifungals warrants constant attention and further underscores the importance of accurate AFST (2)(3)(4). While the reference broth microdilution (BMD) methods of CLSI and European Committee for Antimicrobial Susceptibility Testing (EUCAST) are considered to be the gold standard for AFST, their implementation in laboratory routines is hampered by their labor-intensive and time-consuming nature. Sensititre YeastOne (SYO) is a widely used commercial colorimetric assay for the AFST of yeasts, owing to its ready-to-use nature and its ease of use (5). However, SYO was optimized for common Candida species, meaning that the testing of new species may be a challenge due to the unique characteristics of each species (growth rate, growth pattern, metabolic status). Given the differences in the MIC distributions generated using SYO and CLSI, SYO-specific epidemiological cutoff (ECOFF) values have been determined for the detection of non-wild type (WT) isolates for several species (6)(7)(8). Systematic assessments of SYO performance for C. auris AFST have not been performed (9). To date, the CDC has proposed tentative fluconazole, amphotericin B (AMB), and echinocandin MIC breakpoints for C. auris (10), and species-specific tentative CLSI and EUCAST ECOFF values have also been proposed (11). Still, no SYO ECOFF is available for C. auris versus any antifungal agent.
Currently, echinocandins are indicated as the first-line therapy for candidemia (12,13). However, in the context of the emergence of breakthrough C. auris infections as well as the therapeutic failures observed among patients treated with echinocandins (3,4,14,15) in conjunction with the high (91%) resistance rate of C. auris to fluconazole (16), treatment with liposomal AMB could be considered as an alternative therapeutic option (17). A recent meta-analysis revealed a relatively low AMB resistance rate of 12% among C. auris isolates (16), whereas resistance up to 27% and 50% was observed with the SYO assay (18) and the Etest strips (19), respectively. Despite the elevated MICs (.1 mg/L), resistance could not be confirmed molecularly, as no mutations within the ERG3, ERG5, ERG6, and/or ERG10 genes that have been associated with resistance to AMB in Candida species have been detected (20,21). Of note, the CLSI has recently suggested that AMB susceptibility results should be interpreted with caution because of the significant variability in the MIC values for C. auris and AMB across different testing methodologies (22). Therefore, we evaluated the SYO performance for the AFST of AMB against C. auris, compared to the reference CLSI BMD method in an attempt to compare the two methods and optimize SYO AFST.

DISCUSSION
Accurate AMB AFST against C. auris has recently been caught in the spotlight of attention. Herein, a comparative evaluation of SYO and the CLSI reference method showed poor correlation. The SYO color imetric endpoints were 1 to 2 twofold dilutions higher (29% EA) than the CLSI MICs, with the modal CLSI MIC of 1 mg/L being just below the CDC's breakpoint of 2 mg/L, thereby resulting in considerable interpretation errors (11% CA). Discrepancies were limited when the SYO-specific colorimetric AMB WT-UL of 8 mg/L was used to detect non-WT/resistant isolates (100% CA) and when a 50% growth inhibition endpoint was used, regardless of the color in the well (90% EA, 97% CA).
Currently, CLSI clinical breakpoints for Candida species and AMB are not available. Considering that the AMB CLSI MICs are tightly clustered between 0.25 and 1 mg/L, the MIC value of $2 mg/L is used as the resistant breakpoint for all Candida species, despite the paucity of data to support it (23). In fact, there has been little problem with this cutoff so far, as the vast majority of Candida species display MIC values of ,1 mg/L. Nevertheless, this is not the case for C. auris, as AMB resistance (CLSI MICs of .1 mg/L) typically ranges up to 30% (24,25). Of note, the tentative AMB MIC breakpoint that has been proposed by the CDC is only based on a single PK/PD study in a murine model (10). Thus, the use of the CLSI breakpoint as a reliable gold standard for C. auris raises concerns. Meanwhile, the remarkable variability in the AMB MIC distributions of C. auris across different test methods poses an additional challenge (22), thereby calling for the definition of method-specific breakpoints.
To date, studies on C. auris antifungal susceptibility profiles have mainly been conducted by testing a few isolates with MICs that have been determined using the reference methodology. While SYO is widely used in routine clinical laboratories, data on its performance for the AFST of C. auris are limited, and comparative evaluations with the CLSI/EUCAST method are scarce, thereby precluding comparisons. Only Ruiz-Gaitán et al. compared the AMB MICs of 56 Spanish C. auris strains obtained via the SYO and the EUCAST BMD methods, reporting 44.9% and 93.9% EA within 61 and 62 twofold dilutions, respectively, which is in line with our findings (29% and 88%, respectively). The SYO AMB MICs were 1 to 2 twofold dilutions higher than were the EUCAST MICs (MIC 50 value of 0.125 for EUCAST versus 0.5 mg/L for SYO), as was found in the present study for CLSI, but at higher MICs (9). However, an excellent (100%) CA was observed because all of the isolates were susceptible with both methodologies, as opposed to our results (11% CA) (9). Our CLSI AMB susceptibility data are consistent with those of previous reports (modal MIC, MIC 50 , and MIC 90 values of 1 mg/L, 1 mg/L, and 1 mg/L versus 0.5 to 1 mg/L, 0.5 to 1 mg/L, and 1 to 2 mg/L, respectively) (11,26,27), with the majority of the MIC values being gathered at one twofold dilution step lower from the proposed tentative breakpoint of 2 mg/L (22). In our study, the SYO AMB MICs were higher by 1 to 2 twofold dilutions (modal MIC, MIC 50 , and MIC 90 2 mg/L, 2 mg/L, and 4 mg/L, respectively), and they thus passed the margin of "susceptibility", thereby leading to substantial categorical disagreement. Similar disagreement has been previously described for a C. auris bloodstream isolate with a SYO MIC of 2 mg/L (resistant) and a CLSI MIC of 1 mg/L (susceptible) (28). Of note, Ruiz-Gaitán et al. provided data for isolates that were clonal in nature, gave rise to a 10-month hospital outbreak, and belonged to clade III of C. auris. Moreover, retrospective SYO data have been obtained for 42 out of 56 isolates (9). On the other hand, we included 20 strains from diverse geographic origins that spanned all five of the genetically distinct clades of C. auris as well as 45 Greek clade I strains that were collected from 10 different centers during a 22-month period so as to minimize the overrepresentation of clonal isolates that could affect the estimation of the SYO colorimetric WT-UL as well as future comprehensive performance comparisons of the commercial assay. In addition, the CLSI MICs were simultaneously tested alongside the SYO MICs on the same C. auris isolates in order to restrict variation due to inoculum sizes and incubation conditions.
Overall, unimodal CLSI AMB MIC distributions have been described implying a low rate of resistance of C. auris isolates (11,26,27), as was also shown by a recent meta-analysis of the global epidemiology of this yeast (12% AMB resistance rate) (16). However, when Maphanga et al. assessed the susceptibility of 394 South African C. auris bloodstream isolates using SYO, they reported a resistance rate of 27%, based on the CDC's tentative breakpoint (modal MIC, MIC 50 , and MIC 90 values of 1 mg/L, 1 mg/L, and 2 mg/L, respectively) (18). Intriguingly, when the AMB MICs were redetermined using gradient concentration strips (Etest), the resistance rate decreased to 6%, thereby highlighting the method variability with AMB AFST as well as the need for setting method-specific ECOFFs. Notably, the resistance rate would be 2% if the 8 mg/L SYO colorimetric MIC breakpoint that is proposed here was adopted. Given that 8 mg/L is the maximum AMB concentration contained in SYO panels, resistant isolates will be detected based on the color of a single well. Alternatively, the growth inhibition SYO MIC-2 (97% CA) could be used in order to identify resistant isolates. Since the determination of the SYO MIC-2 may be difficult, MIC-1, which is usually one twofold dilution lower than the colorimetric MIC endpoint, could also be used. However, the MaE are 12%, and resistant isolates with a MIC-1 of 2 mg/L or 4 mg/L will need to be verified via a reference method.
ECOFF values can aid in identifying isolates with elevated MICs and/or are likely to possess an acquired resistance mechanism. The estimated SYO color imetric AMB WT-UL is based mainly on MIC data for Greek isolates belonging to clade I. However, the process of establishing formal ECOFF values requires data from multiple laboratories, which are not currently available. Although clade I is the most prevalent and is characterized by a wider geographical distribution and a higher frequency of bloodstream infections, further SYO multicenter evaluations encompassing more unrelated C. auris isolates that belong to all five different phylogenetic clades are warranted, owing to the variable regional patterns of AMB resistance (16). In addition, strains with known resistance mechanisms were not included in the CLSI-SYO comparison. Nevertheless, the first mechanism contributing to clinical AMB resistance in C. auris, which is conferred by mutations in the sterol-methyltransferase gene ERG6, has only recently been identified (2). Further studies are needed to define the prevalence of the aforementioned mutations as well as other mutations among phenotypically AMB-resistant clinical isolates that could be used to optimize the current AFST methodologies for the reliable detection of AMB resistance in C. auris.
Our study demonstrated an overestimation of C. auris AMB resistance with SYO, underscoring that the SYO colorimetric AMB MICs need to be interpreted with caution if the CDC's tentative breakpoint is used to guide therapeutic decisions. Further multicenter studies are needed to verify current findings and to determine SYO-specific ECOFF values. Laboratories should be aware that C. auris isolates with a SYO colorimetric AMB MIC of #8 mg/L indicate susceptibility and that a SYO growth inhibition endpoint could be used to detect resistance.

MATERIALS AND METHODS
Isolates. A total of 45 clinical isolates, mainly bloodstream isolates (41 out of 45; 91%), were collected from individual patients who were hospitalized in 10 Greek tertiary care hospitals that are located in the Attica region from November of 2020 to August of 2022. All of the isolates were identified to the species level using MALDI-TOF mass spectrometry (Bruker Daltonics, Bremen, Germany) and clustered in clade I (South Asian) (29). In addition, 20 genetically distinct clinical strains belonging to all 5 C. auris clades and being isolated from various geographical regions were tested. These comprised of: 5 strains from clade I (South Asian; Brazil, Kuwait, Iran, India, Oman), 3 strains from clade II (East Asian; South Korea, Japan), 4 strains from clade III (African; South Africa, Spain), 4 strains from clade IV (South American; Venezuela, Colombia), and 4 strains from clade V (Iranian; Iran) (29).
Antifungal susceptibility testing. The CLSI AFST was performed, according to the M27A4 protocol guidelines, using a laboratory-grade pure AMB powder (Sigma-Aldrich, Athens, Greece). The microtiter plates were incubated at 35 6 2°C, and the MICs were defined as the lowest drug concentrations at which the total inhibition of visual growth, compared to the growth control well, was observed after 24 h (30). The Sensititre YeastOne YO10 (Thermo Fisher Scientific, Waltham, MA, USA) AFST was performed in accordance with the manufacturer's recommendations. The panels were incubated at 35 6 2°C, and the SYO color imetric MIC was recorded as the first blue well corresponding to complete growth inhibition (MIC-0) after 24 h. Alternatively, different SYO growth inhibition endpoints, defined as the lowest drug concentrations with a substantial reduction of growth (approximately 75% growth inhibition, MIC-1), a prominent reduction of growth (approximately 50% growth inhibition, MIC-2) and a slight reduction of growth (approximately 25% growth inhibition, MIC-3), relative to the drug-free control growth, were determined (Fig. 3). Each isolate was tested by both methods, using a single fungal suspension that was adjusted to the required concentration. The colony-forming unit (CFU) counts were affirmed each time via spread plate counts on Sabouraud dextrose agar plates, whereas the recommended C. krusei ATCC 6258 and C. parapsilosis ATCC 22019 strains were used as quality control strains for both methods. The CLSI and SYO MICs were evaluated via a visual inspection of the plates with the aid of a magnifying mirror by two blinded observers. In order to assess the potential variation in AMB MIC determination by SYO, a proportion of isolates (10 out of 65) were retested on different days and in different laboratories so as to determine the interday and the interlaboratory reproducibility, respectively.
Analysis. A head-to-head comparison of the generated MIC data sets, using the CLSI BMD as the reference methodology, was performed. For the quantitative analysis, the results of the two methods were analyzed using a paired Student's t test or a Wilcoxon matched-pairs signed rank test, depending on the validity of the normality assumption. The levels of CLSI-SYO EA within 61 and 62 twofold dilutions were calculated. For the qualitative analysis, the CA was determined, following the CDC's tentative AMB resistant breakpoint for C. auris ($2 mg/L) (10). A discrepancy was considered to be a MaE when the CLSI classified an isolate as susceptible (MIC , 2 mg/L) and the SYO classified the same isolate as resistant (MIC $ 2 mg/L) (false resistance). A very major error (i.e., false susceptibility, classified as resistant by CLSI and as susceptible by SYO) could not be determined, as there were no CLSI AMB resistant isolates. The SYO WT-UL, which was defined as the upper MIC value at which the WT distribution ended, was determined using the ECOFFinder program (31). The WT-UL was verified in a larger collection of 205 Greek C. auris isolates that clustered in clade I and were recovered from individual patients who were hospitalized in tertiary care hospitals that were located in densely populated metropolitan areas across the country from June of 2019 to May of 2022.

ACKNOWLEDGMENTS
No funding support was received for this study.
J.M. and S.P. contributed to the conception and design of the study, the data analysis, and supervision of the work. M.S., I.P., and M.-I.B. contributed to the identification and susceptibility testing of the isolates. T.D.G., B.S., and J.F.M. contributed to the sequencing and genotyping of the isolates. G.V. and A.T. provided the data for the large collection of isolates. M.S. wrote and prepared the original draft as well as participated in the data analysis. All authors reviewed and edited the writing. All authors have read and agreed to the published version of the manuscript.
The authors do not declare any conflict of interest related to the present study.