Evaluation of MIC Strip Isavuconazole Test for Susceptibility Testing of Wild-Type and Non-Wild-Type Aspergillus fumigatus Isolates

ABSTRACT We evaluated the MIC Strip Isavuconazole test against EUCAST E.Def 9.3 by using 40 wild-type and 39 CYP51A mutant Aspergillus fumigatus strains. The strip full inhibition endpoint (FIE) and 80% growth inhibition endpoint were determined by two independent readers, reader 1 (R1) and R2. The essential (within ±0, ±1, and ±2 twofold dilutions) and categorical agreements were best with the FIE (for R1/R2, 42%/41%, 75%/73%, and 90%/89% for essential agreement, and 91.1%/92.4% categorical agreement, with 6.3/8.9% very major errors and 0/1.3% major errors, respectively). The MIC Strip Isavuconazole test with the FIE appears to be useful.

supplemental material) (7). The percent essential agreement between the tests was calculated. Isolates for which the MICs were above scale by both methods (EUCAST, Ͼ16 mg/liter; strip test, Ͼ32 mg/liter) were considered in agreement within Ϯ0 twofold dilution. The categorical agreement between the methods was calculated as the percentage of isolates classified equally by both methods. Very major errors (VMEs) were defined as isolate categorization as resistant (R) by EUCAST but susceptible (S) by the strip test, and major errors (MEs) were defined as isolate categorization as S by the EUCAST method but R by the strip test.
Most isavuconazole strip MICs were above the recommended ranges for the two control Candida strains (Table S1). In contrast, the strip MICs for A. fumigatus ATCC 204305 and A. flavus CM1813 were within Ϯ1 twofold dilution of the EUCAST MICs, suggesting better agreement for the Aspergillus strains and best when using the FIE for Aspergillus.
Nine isolates (11.4%) failed to grow sufficiently well to allow strip MIC reading on day 1, when, in general, zones were fuzzy and difficult to read. Day 2 MICs were lower with the 80% IE than with the FIE (Fig. 1). This was particularly evident for isolates harboring TR 34 /L98H alterations, for which the modal 80% IE MICs were 2 and 4 mg/liter, respectively but Ͼ32 mg/liter for both readers with the FIE. The essential agreement between the strip MICs from the two readers was highest, 97% at Ϯ1 twofold dilution and 100% at Ϯ2 twofold dilutions, when using the FIE (Table 1).
Isavuconazole MICs for isolates with wild-type CYP51A or single alterations at the G54 codon were all below the EUCAST ECOFF for the strip test with the FIE, as well as for EUCAST (Table 2). Likewise, the MICs for isolates harboring M220I alterations or TR 34 /L98H or TR 46 /Y121F T289A were all above the clinical breakpoint for both methods when the FIE was used for the strip test. However, the MICs for TR 34 /L98H isolates were higher when determined by the strip test (MIC range, Ͼ32 mg/liter) than when determined by EUCAST (MIC 50 of 8 mg/liter; range, 4 to Ͼ16 mg/liter) ( Table 2). The overall essential agreement between strip MICs and EUCAST MICs within Ϯ0, Ϯ1, and Ϯ2 twofold dilutions was best when using the FIE (R1/R2: 42/41, 75/73, and 90/89%) than when using the 80% IE (R1/R2: 25/30, 66/70, and 91/91%). At least 95% essential agreement between the strip test and EUCAST within Ϯ2 twofold dilutions was seen for all CYP51A genotypes except those harboring the TR 34 /L98H mechanism or the M220I alteration. Similarly, the categorical agreement was better for the FIE reading of the strip test (91.1 to 92.4% with 6.3 to 8.9% VMEs and 0 to 1.3% MEs) than for the 80% IE (89.9% with 10.1% VMEs and 0% MEs for both readers). VMEs included four isolates with the wild-type CYP51A genotype and one to four isolates harboring M220V, M220I, or G54R N248K alterations, respectively.
The MIC Strip Isavuconazole test manufacturer recommends an 80% IE reading, but in this study, higher interreader essential agreement, better separation between wildtype and resistant strains, and greater essential and categorical agreement compared to EUCAST results were achieved with the FIE. Thus, the FIE criterion was found to be superior although the MICs for the recommended Candida control strains were out of range (7). When using the FIE, the essential agreements with EUCAST within Ϯ1 and Ϯ2 twofold dilutions were 73 to 75% and 89 to 90% and thus better than previously found for the isavuconazole Etest versus the CLSI method, even though challenged here with a strain collection including a significant number of non-wild-type isolates (16). The categorical agreement was Ͼ91% when interpreting the MICs according to EUCAST breakpoints, and notably, among the 6 to 9% VMEs, half were isolates with a wild-type CYP51A target gene that either may be harboring other resistance mechanisms or may be isolates that are truly susceptible but misclassified as R by the EUCAST reference method because of the conservative EUCAST susceptibility breakpoint (7). Finally, the separation between wild-type and TR 34 /L98H and TR 46 /Y121F T289A mutant isolates   was greater for the MIC strip test, rendering it a potentially promising routine lab tool for detecting R environmental mutants, provided the FIE is used (2,5,(19)(20)(21). The CYP51A amino acid alterations have been associated with a codon-specific susceptibility pattern (4,13). Here, both the strip and EUCAST isavuconazole MICs indeed straddled the clinical breakpoint for isolates harboring M220 alterations and for the G54R N24K double mutant, which will inevitably lead to the random classification of such isolates as S or R in routine testing. Hence, as long as clinical outcome data are unavailable for such mutants, other measures such itraconazole MIC testing or CYP51A sequencing should be undertaken to detect these genotypes.
This study has limitations. We investigated strip test reader-to-reader agreement but no other factors associated with variation, such as variation across different lots and brands of RPMI agar plates, inoculum preparation, etc. Therefore, the promising performance reported here needs confirmation in a multicenter study.

ACKNOWLEDGMENTS
Isavuconazole MIC strips and the pure substance were kindly provided at no cost by Basilea. We thank Birgit Brandt and Désiré Mageme Nahimana for excellent technical assistance.
Maiken Cavling Arendrup has received research grants or speaker honoraria from Astellas, Basilea, Gilead, MSD, Novartis, Pfizer, and T2Biosystems. She is the current chairman of the EUCAST-AFST and has previously served on advisory boards for MSD (until 2014) and Pfizer (until 2012). Paul Verweij has received research grants from Astellas, Basilea, F2G, Gilead Sciences, Merck, and Pfizer; has been a consultant to Basilea, F2G, Gilead Sciences, Merck, and Pfizer; and has received speaker's fees from Basilea, Gilead Sciences, and Merck. Henrik Vedel Nielsen has no conflicts to declare.