Comparison of MIC Test Strip and Sensititre YeastOne with the CLSI and EUCAST Broth Microdilution Reference Methods for In Vitro Antifungal Susceptibility Testing of Cryptococcus neoformans

We compared MIC test strip (MTS) and Sensititre YeastOne (SYO) methods with EUCAST and CLSI methods for amphotericin B, 5-fluocytosine, fluconazole, voriconazole, and isavuconazole against 106 Cryptococcus neoformans isolates. The overall essential agreement between the EUCAST and CLSI methods was >72% and >94% at ±1 and ±2 dilutions, respectively. The essential agreements between SYO and EUCAST/CLSI for amphotericin B, 5-flucytosine, fluconazole, and voriconazole were >89/>93% and between MTS and EUCAST/CLSI were >57/>75%.

ABSTRACT

We compared MIC test strip (MTS) and Sensititre YeastOne (SYO) methods with EUCAST and CLSI methods for amphotericin B, 5-fluocytosine, fluconazole, voriconazole, and isavuconazole against 106 Cryptococcus neoformans isolates. The overall essential agreement between the EUCAST and CLSI methods was >72% and >94% at ±1 and ±2 dilutions, respectively. The essential agreements between SYO and EUCAST/CLSI for amphotericin B, 5-flucytosine, fluconazole, and voriconazole were >89/>93% and between MTS and EUCAST/CLSI were >57/>75%. Very major error rates were low for amphotericin B and fluconazole (<3%) and a bit higher for the other drugs (<8%).

INTRODUCTION

Cryptococcus species have been reported to cause various infections in both immunocompetent and immunocompromised patients, such as those with HIV/AIDS, especially in resource-limited settings (1); whereas in resource-rich countries, infection has been observed in solid-organ transplant recipients. Cryptococcal meningoencephalitis (CM) caused by Cryptococcus neoformans isolates is the most frequently encountered manifestation of cryptococcosis and particularly affects patients with AIDS (2). The global incidence of CM has been estimated at >200,000 infections per year, with 73% of cases occurring in sub-Saharan Africa (3). C. neoformans most commonly enters its host through the airways via inhalation of infectious cells that cause primarily pulmonary infection, which may lead to a disseminated infection (4). The treatment of cryptococcosis consists of amphotericin B (AMB) with or without other antifungals, such as 5-flucytosine (5FC) or fluconazole (FLC) (5). However, studies showed that these treatment regimens have high failure rates (6). An increasing number of reports have been published on Cryptococcus isolates with high MICs particularly to azoles, although the impact of high MICs on clinical outcome remains controversial (7, 8). Clinical breakpoints have not been established for Cryptococcus species, but epidemiological cutoff values have been proposed for some antifungal drugs to differentiate wild-type (WT) from non-wild-type (non-WT) isolates (7, 8). Some clinical studies indicated that high MICs may be associated with treatment failure (9, 10).

Discrepancies between MIC values and clinical outcomes may be partly explained by the antifungal susceptibility test used because MICs may differ between various methods (11). Currently, several techniques for MIC testing of Cryptococcus species are available, such as the broth microdilution (BMD) reference methods recommended by the Clinical and Laboratory Standard Institute (CLSI) and European Committee or Antimicrobial Susceptibility Testing (EUCAST), commercial techniques that use colorimetric endpoints (e.g., Sensititre YeastOne [SYO] assay), agar-based methods that use concentration gradients of antifungals that diffuse into the growth media (e.g., Etest and Liofilchem MIC test strip [MTS]), and automated systems (e.g., Vitek 2). These commercial methods are standardized, effective, simple, and easy to implement in the clinical microbiology laboratory compared with the gold-standard reference techniques, which are tedious, time-consuming, and limited to few centers. It is therefore essential to evaluate these commercial tests and to determine their ability to give MIC values that agree with those from the reference methods. Although many studies have investigated the concordance between various techniques, the results are not conclusive, showing discrepancies in MIC results. Furthermore, most studies were limited to AMB, 5FC, and FLC; and only few studies investigated alternative azole compounds, such as voriconazole (VRC) and the new triazole isavuconazole (ISA), which, despite no indication yet in the treatment of cryptococcosis, may be a promising alternative.

Here, we evaluate the performance and concordance between the four techniques, i.e., EUCAST, CLSI, MTS, and SYO, in determining the in vitro activity of AMB, 5FC, FLC, VRC, and ISA against C. neoformans.

A total of 106 C. neoformans isolates from 60 patients were collected between December 2010 and March 2018 and stored at −80°C in 20% glycerol (Radboudumc, Nijmegen, The Netherlands) until testing. All isolates had been identified to the species level using the Bruker Biotyper matrix-assisted laser desorption ionization–time of flight mass spectrometry. Susceptibility testing was carried out using the following methods: CLSI M38-A2 (12), EUCAST E.Def 7.3 (13), MTS (Liofilchem, Roseto, degli Abruzzi, Italy), and SYO (Thermo Scientific, Waltham, MA). For BMD reference methods, stock concentrations were prepared in water for 5FC or in dimethyl sulfoxide (Sigma-Aldrich, St. Louis, MO) for the other drugs. The final drug concentration ranges were 0.008 to 8 μg/ml for AMB (Sigma-Aldrich), 0.063 to 64 μg/ml for 5FC (Sigma-Aldrich) and FLC (Pfizer), and 0.016 to 16 μg/ml for VRC (Sigma-Aldrich) and ISA (Sigma-Aldrich). All antifungal agents were dissolved in RPMI 1640 buffered to pH 7 with MOPS (morpholinepropanesulfonic acid) supplemented with glucose to a final concentration of 2% (for EUCAST) according to published protocols (12, 13). The 96-well flat-bottomed microplates were stored at –70°C until used, for no more than 6 months. Isolates were cultured on Sabouraud chloramphenicol agar (Oxoid, UK) for 48 h at 37°C, after which a suspension was prepared in sterile saline solution, adjusted to 0.5 McFarland, and diluted to reach the final inoculum of 1 to 5 × 10⁵ CFU/ml for EUCAST and 0.5 to 2.5 × 10³ CFU/ml for CLSI. Microplates were incubated at 37°C for 48 h for EUCAST and 72 h for CLSI. Turbidity was assessed spectrophotometrically at 405 nm (Anthos Labtec Instruments GmbH, Salzburg, Austria), and the MICs were defined as the lowest drug concentrations that caused ≥90% (for AMB) or ≥50% (for 5FC, FLC, VRC, and ISA) reduction in growth compared with that of a drug-free growth control.

For MTS and SYO susceptibility testing, reading and interpretation of the results were done according to the manufacturers’ instructions. MTS was performed in RPMI 1640 containing 1.5% agar supplemented with 2% glucose and buffered with MOPS to pH 7. For 5FC, an additional plate with RPMI 1640 containing 1.5% agar supplemented with 2% glucose and buffered with citric acid to pH 5 was used. After 72 h of incubation, MTS MICs were read at the point of intersection between the ellipse of growth inhibition and the strip (at 100% inhibition for AMB and 90% for 5FC and azole drugs), whereas SYO MICs corresponded to the first blue well for AMB or the first purple/blue well for the other drugs. Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were included in each antifungal susceptibility assay as quality control strains.

For method comparison, MTS MICs were converted to the next-highest 2-fold dilution value that matched the CLSI and EUCAST 2-fold dilution scheme. For all methods, high off-scale MICs were converted to the next highest 2-fold concentration, whereas low off-scale MICs for EUCAST and CLSI were left unchanged. The MTS and SYO MICs were truncated to the ranges used by EUCAST and CLSI. Thus, for VRC and ISA, MIC results of ≤0.016 mg/liter were converted to 0.016 mg/liter. The MIC distributions together with mode, median, and geometric mean (GM) MIC and MIC₉₀ values were presented for each drug and method. The median (range) 2-fold differences and agreement within 0, 1, and 2 dilutions between the methods were calculated. To calculate the categorical agreement between the reference and commercial methods, a local epidemiological cutoff value (L-ECOFF, which is defined as the upper limit of WT for the test strains included in this study) was determined for each drug and method using the derivatization method based on the first peak of the second derivatives of the MICs after the modal MIC (14) and ECOFFinder software based on the 95% subset of MICs. An isolate was classified as WT or non-WT with each method when its MIC was less than or equal to the L-ECOFF and greater than the L-ECOFF, respectively. Very major and major error rates were calculated as the percentage of isolates classified as non-WT with a reference method and WT with a commercial method and vice versa, respectively.

Table 1 summarizes the in vitro susceptibilities of the 106 C. neoformans isolates to AMB, 5FC, FLC, VRC, and ISA. The data are presented as MICs, GM MICs, MIC ranges, modal MICs, median MICs, and MIC₉₀s. All of the isolates tested grew well in both round-bottomed (CLSI) and flat-bottomed (EUCAST) microtiter plates, giving detectable endpoints within 48 h. All the isolates grew well with SYO and MTS as well. MIC results of AMB spanned a narrow range of 0.016 to 0.25 μg/ml with the CLSI and EUCAST methods, 0.125 to 0.25 μg/ml with SYO, and 0.016 to 1 μg/ml with MTS. GM MICs and MIC₉₀s of ISA, VRC, and AMB were usually lower than those of 5FC when tested with the CLSI, EUCAST, MTS, and SYO methods. Accordingly, large interstrain variations were found for VRC and ISA using MTS, with MICs ranging from 0.002 to 0.25 μg/ml. Overall, AMB, VRC, and ISA showed the lowest MICs with all four methods, with VRC and ISA MICs of <0.125 μg/ml against all isolates. The mode and median MICs of AMB, 5FC, VRC, FLC, and ISA showed no differences among any of the methods. 5FC exhibited higher MICs (0.063 to >64 μg/ml) with CLSI, with respective median MICs, MIC₉₀s, and GM MICs of 2, 4, and 2.3 μg/ml with the CLSI method; 4, 8, and 3.2 μg/ml with the EUCAST method; 4, 8, and 2.9 μg/ml with SYO; and 2, 32, and 1.6 μg/ml with MTS. In addition, FLC showed high MICs against C. neoformans isolates, with respective median (range) and MIC₉₀ values of 4 (0.063 to 16) and 8 μg/ml with the CLSI method, 4 (0.125 to 32) and 8 μg/ml with the EUCAST method, 2 (0.25 to 16) and 4 μg/ml with SYO, and 4 (0.125 to 32) and 8 μg/ml with MTS. The MIC ranges for the serial isolates from the same patient were all within 2 dilutions except for two isolates from one patient, in which we observed an elevated MIC for 5FC from 2 to >32 μg/ml by all four methods.

TABLE 1.

In vitro susceptibilities of 106 Cryptococcus neoformans to five antifungal agents as determined by CLSI, EUCAST, Sensititre, and MIC test strip

Drug and method	MIC (μg/ml)a																	L-ECOFFb	Non-WTb (%)
	0.002	0.004	0.008	0.016	0.03	0.06	0.125	0.25	0.5	1	2	4	8	16	32	>32	GM
Amphotericin B
CLSI				1	12	46	43	4									0.081	0.25	0
EUCAST				1	6	29	49	21									0.108	0.25	0
SYO							91	15									0.133	0.25	0
MTS				7	22	28	21	20	7	1							0.088	0.5	1
5-Flucytosine
CLSI						1		1	4	12	61	18	3	3		3	2.325	4	8
EUCAST							2		6	10	25	38	17	5		3	3.266	8/16	8/3
SYO									6	17	28	38	11	3		3	2.922	8	6
MTS						2	8	7	20	20	24	5	2	1	7	10	1.613	4	19
MTS pH 5.0	74c	17	8	4											1	2	0.004	0.004	14
Fluconazole
CLSI						1		1	8	16	16	43	18	3			2.738	8/16	3/0
EUCAST							1	4	6	14	10	30	29	11	1		3.441	16/32	1/0
SYO								6	9	17	34	30	8	2			1.987	8	2
MTS							3	1	9	21	17	32	18	4	1		2.433	8/16	5/1
Voriconazole
CLSI				36	54	10	6										0.029	0.06	6
EUCAST				25	54	19	8										0.034	0.06	8
SYO			26	42	27	10	1										0.019	0.06	1
MTS	3	3	3	18	33	31	14	1									0.036	0.125	1
Isavuconazole
CLSI				86c	14	6											0.019	0.03	6
EUCAST				49c	42	13	2										0.026	0.06	2
MTS	14	12	24	29	16	9	2										0.012	0.03/0.06	10/2

^aItalicized numbers indicate modal MIC, underlined numbers indicate median MIC, and boldfaced numbers indicate MIC₉₀.

^bL-ECOFFs determined with the derivatization method and the ECOFFinder software based on 95% subset of MICs. The second number represents the L-ECOFF determined with the ECOFFinder whenever there were discrepancies with the derivatization method. For those cases, 2% of non-WT isolates are presented corresponding to each L-ECOFF.

^cTruncated.

Table 2 shows the absolute and essential agreements (within 1 and 2 dilutions) between the CLSI, EUCAST, SYO, and MTS MICs for each drug tested. The essential agreement within two 2-fold dilutions between methods for all drugs was >90%, except between CLSI and SYO for 5FC (75%), EUCAST and SYO for FLC (89%), and EUCAST and MTS for ISA (80%) and 5FC (57%). Overall, EUCAST and CLSI gave similar L-ECOFFs for all drug methods with some exceptions, i.e., where ECOFFinder gave L-ECOFFs 1 dilution higher than the derivatization method (Table 1). Because of truncated MIC distribution, no L-ECOFFs were determined with ECOFFinder for EUCAST and CLSI for ISA. The percentage of non-WT isolates was low (<10%) for all methods except for 5FC with MTS at pH 7 (19%). Very major error rates between the methods were low (<8%) for all drugs except for ISA between EUCAST and MTS (13%). Major errors between methods were low (<11%) for all drugs except for 5FC with MTS and EUCAST (14%/16%) (Table 2).

TABLE 2.

Comparison of different methods for antifungal susceptibility testing of Cryptococcus neoformans

Drug	Method comparison		Median (range) 2-fold dilution difference	Agreement within 2-fold dilution (%) of:			Error ratea (%)
				0	1	2	ME	VME
Amphotericin B	CLSI vs:	EUCAST	0 (−2 to 3)	42	82	99	0	0
		SYO	−1 (−3 to 2)	31	86	99	0	0
		MTS	0 (−4 to 3)	23	67	94	1	0
	EUCAST vs:	SYO	0 (−4 to 3)	53	92	98	0	0
		MTS	0 (−3 to 3)	25	65	90	1	0
5-Flucytosine	CLSI vs:	EUCAST	−1 (−3 to 4)	24	72	94	3/0	4/6
		SYO	–0.5 (−5 to 3)	27	82	96	1	4
		MTS	1 (−10 to 5)	19	50	75	12	2
	EUCAST vs:	SYO	0 (−4 to 3)	35	77	92	1/3	3/0
		MTS	2 (−9 to 6)	11	34	57	14/16	3/0
Fluconazole	CLSI vs:	EUCAST	0 (−3 to 5)	38	80	94	1/0	3/0
		SYO	0 (−5 to 4)	37	77	93	2/2	3/0
		MTS	0 (−6 to 5)	33	77	95	5/1	3/0
	EUCAST vs:	SYO	1 (−4 to 4)	28	70	89	2/2	1/0
		MTS	0 (−4 to 5)	29	69	90	5/1	1/0
Voriconazole	CLSI vs:	EUCAST	0 (−2 to 3)	52	88	99	5	3
		SYO	1 (−2 to 4)	68	91	100	1	6
		MTS	−0.5 (−2 to 3)	69	92	100	1	6
	EUCAST vs:	SYO	1 (−2 to 4)	50	91	99	1	8
		MTS	0 (−3 to 4)	50	92	100	0	7
Isavuconazole	CLSI vs:	EUCAST	0 (−2 to 2)	52	88	100	2/11	6/3
		SYO	ND	NDb	ND	ND	ND	ND
		MTS	1 (−2 to 4)	69	92	100	8/1	4/5
	EUCAST vs:	SYO	ND	ND	ND	ND	ND	ND
		MTS	1 (−2 to 4)	50	92	100	9/1	1

^aA major error (ME) and a very major error (VME) were considered when an isolate was classified as WT with the reference method and non-WT with the alternative method and vice versa. Whenever two numbers are shown, two L-ECOFFs were determined (see Table 1).

^bND, not determined.

We studied the agreement between SYO, MTS, CLSI, and EUCAST methods in determining the in vitro MICs of C. neoformans isolates originating from clinical samples within the same laboratory. For AMB, the essential agreement within one and two 2-fold dilutions was excellent among the four techniques: 82% and 99% between the BMD methods, 92% and 98% between SYO and EUCAST, 86% and 99% between SYO and CLSI, 65% and 90% between MTS and EUCAST, and 67% and 94% between MTS and CLSI. Our findings are in accordance with previous studies that used the other agar-based method, the Etest. Maxwell et al. (15) reported an agreement of 99% between the Etest and CLSI methods within the 2-fold dilutions, and Ochiuzzi et al. (16) showed that the agreement between the Etest and EUCAST methods within the 2-fold dilutions was 100%. However, in other studies, the agreement between the Etest and the BMD methods for AMB was very low (9, 17).

In our study, 17 (16%) of 106 isolates had MICs of >32 to 5FC by MTS at pH 7.0, whereas only 3% to 8% were non-WT with the other methods. Previous studies also reported high MICs to 5FC with the Etest performed at pH 7.0. Viviani et al. (17) concluded that yeast nitrogen base at pH 7.0 was not a suitable medium for testing the susceptibility to 5FC. On the other hand, Dannaoui et al. (18) reported that 87% of the tested strains had MICs of >32 μg/ml against 5FC. Mahabeer et al. (19) even discontinued use of the Etest method, using RMPI 1640 at pH 7.0 after having high MICs for the first 10 isolates. Despite these reports, RMPI 1640 at pH 7.0 is still recommended by the manufacturer. The concordance within 2 dilutions between the reference methods CLSI and EUCAST and between CLSI, EUCAST, and SYO was good at >92%; however, the agreements between MTS at pH 7.0 and EUCAST and between MTS at pH 7.0 and CLSI were low (57% and 75%, respectively), indicating that the 5FC Etest at pH 7 is not suitable for antifungal susceptibility testing of Cryptococcus, may be explained by the fact that 5FC activity is pH dependent and is decreased at pH 7.0 (20).

For FLC, we obtained good concordance of 94% within 2 dilutions between the EUCAST and CLSI methods. The agreements between SYO and CLSI and between SYO and EUCAST were 93% and 89%, respectively. In a recent study Vena et al. (11) found very low categorical agreement between CLSI and Sensititre (55.5%) and between EUCAST and Sensititre (49.2%). For MTS, the agreement was 95% with the CLSI method and 90% with the EUCAST method. This was concordant with the findings presented in previous reports; Mahabeer et al. (19) and Ochiuzzi et al. (16) reported high essential agreement (within 2 dilutions) between Etest and CLSI (95.1% to 98.5%). However, lower rates of 73%, 76.7%, and 81.1% were found by Aller et al. (9), Vena et al. (11), and Dannaoui et al. (18), respectively. Dias et al. (21) reported high levels of essential agreement between EUCAST and Etest (93%), whereas Vena et al. (11) presented lower levels (74.5%).

For VCZ, the essential agreement within one and two 2-fold dilutions (88% and 99%, respectively) was excellent between the reference methods. In addition, the essential agreements within one and two 2-fold dilutions between SYO and CLSI were 91% and 100%, respectively, and between SYO and EUCAST were 91% and 100%, respectively. The one and two 2-fold dilution agreements between MTS and EUCAST and between MTS and CLSI were 92% and 100%, respectively. These results agreed with previous reports, in which the essential agreement within 2-fold dilutions was >86.4% (15, 18, 21). The rate of non-WT isolates was low (<8%) with all four methods. For ISA, the essential agreement was perfect (100%) among the three methods; the same results were obtained by Thompson et al. (22), who found an essential agreement of 97.8% between CLSI and MTS within 2-fold dilutions. Furthermore, ISA has potent activity in vitro against Cryptococcus isolates, as demonstrated by low MICs in our study (MIC₅₀ and MIC₉₀ values against Cryptococcus neoformans infection of ≤0.03 and 0.06 μg/ml, respectively, for all four methods), with promising in vivo efficacy in the treatment of cryptococcal meningitis (23, 24).

In conclusion, the reference BMD methods CLSI and EUCAST have good correlation even though the MIC values were slightly higher with the EUCAST method and the incubation time was shorter with the EUCAST (48 h) than with the CLSI (72 h) method. In our study, the MTS and SYO methods correlated well with the reference methods for all antifungals (>90%) except for 5FC MTS, where the levels of agreement were lower (57% to 75%). The percentage of non-WT isolates was low (<10%) for all drugs and methods except for 5FC with MTS (14%) independent of the medium’s pH. The very major error rates were low for AMB and FLC (<3%) and a bit higher for the other drugs (<8%). Further investigations are needed that include more non-WT isolates to evaluate the performance of each test to detect antifungal resistance. Understanding the molecular mechanisms of resistance to antifungal drugs in Cryptococcus isolates is now mandatory to identify molecular markers for resistance, and this may overcome the problems of phenotypic antifungal susceptibility testing.