ABSTRACT
Aspergillus spp. cause serious invasive lung infections, and Aspergillus fumigatus is the most commonly encountered clinically significant species. Voriconazole is considered to be the drug of choice for treating A. fumigatus infections; however, rising resistance rates have been reported. We evaluated a matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS)-based method for the differentiation between wild-type and non-wild-type isolates of 20 Aspergillus spp. (including 2 isolates of Aspergillus ustus and 1 of Aspergillus calidoustus that were used as controls due their intrinsic low azole susceptibility with respect to the in vitro response to voriconazole). At 30 and 48 h of incubation, there was complete agreement between Cyp51A sequence analysis, broth microdilution, and MALDI-TOF MS classification of isolates as wild type or non-wild type. In this proof-of-concept study, we demonstrated that MALDI-TOF MS can be used to accurately detect A. fumigatus strains with reduced voriconazole susceptibility. However, rather than proving to be a rapid and simple method for antifungal susceptibility testing, this particular MS-based method showed no benefit over conventional testing methods.
INTRODUCTION
Aspergillus spp. cause serious invasive lung infections, and Aspergillus fumigatus is the most commonly encountered clinically significant species (1). Voriconazole is considered to be the drug of choice for treating A. fumigatus infections (2); however, rising resistance rates have been reported (3, 4), and antifungal susceptibility testing (AFST) remains limited to a small number of specialized laboratories. Early initiation of appropriate therapy has been shown to lead to better clinical outcomes (5). Additionally, it has been demonstrated that patients infected with resistant strains have poorer outcomes than those infected with susceptible strains (6).
Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has been widely demonstrated to be an excellent tool for the rapid, accurate, and inexpensive identification of microorganisms (7, 8). Groups are now investigating this technology for potentially equally rapid, simple, and inexpensive antimicrobial susceptibility testing (9–12). One method that has been explored uses the composite correlation index (CCI), a statistical autocorrelation method that analyzes the relationships/similarity between spectra generated by microorganisms treated with different concentrations of drug (13). By comparing theses spectra against the spectra obtained at the null or the maximal concentration of drug, similarity of the spectra to the two extremes can be determined (14). This allows for establishment of the minimal profile change concentration (MPCC), which is the lowest drug concentration where the spectra is more similar to that of the maximum drug concentration than to that of the null drug concentration (14). The MPCC has been shown to approximate an MIC determined by broth microdilution (BMD) (14). Here, we sought to explore this method to determine if it could be used to determine voriconazole MICs for Aspergillus spp.
The objectives of this study were to (i) demonstrate whether a MALDI-TOF MS system could be used to determine MPCCs of voriconazole, which approximate MICs for Aspergillus spp., (ii) determine if these MPCCs could correctly classify A. fumigatus isolates as wild type (WT) (no acquired resistance) or non-WT (acquired resistance) with respect to voriconazole, and (iii) establish if this method had any advantages over traditional BMD.
RESULTS
BMD-based MICs, CYP51A genotype, and MALDI-TOF MS MPCC results are presented in Table 1, with A. fumigatus MICs and MPCCs interpreted as WT or non-WT using an epidemiological cutoff value (ECV) of 1 μg/ml. At 24 h incubation, all MPCC data allowed for the appropriate classification of A. fumigatus WT strains as WT. For the Cyp51A mutants, however, 2 of 4 mutant strains displaying non-WT BMD MICs were assigned WT MALDI-TOF MS MPCCs; these would be classified as very major errors (15). Slightly longer incubation times (30 and 48 h) allowed for accurate discrimination between WT and non-WT A. fumigatus strains. Barring the exceptions noted above at the 24-hour time point, the MALDI-TOF MS MPCCs for all Aspergillus strains were within ±1 dilution of the BMD MICs.
Results of antifungal susceptibility testing of Aspergillus spp. to voriconazole using genotyping, BMD MIC, and MALDI-TOF MS MPCC determination
In addition, we also tested the A. fumigatus strains by an abbreviated AFST method of categorization of the isolates as WT or non-WT. This abbreviated method, or “three-point” assay (16), compares the spectra obtained at only three drug concentrations, namely, no drug, maximum drug concentration (16 μg/ml voriconazole), and the voriconazole ECV (1 μg/ml) (see Table S2A and B in the supplemental material). For this method, an isolate was determined to be resistant (non-WT) if at a concentration of 1 μg/ml voriconazole the CCI score was more similar to what is seen with no voriconazole compared to the maximal concentration. It was found that by evaluating just these three drug concentrations, A. fumigatus isolates could be correctly classified as WT or non-WT after incubation periods of 30 and 48 h (Table 2).
Results of antifungal susceptibility testing of A. fumigatus to voriconazole using genotyping, BMD MIC determination, and an abbreviated MALDI-TOF MS method
DISCUSSION
In this proof-of-concept study, we demonstrate the use of MALDI-TOF MS technology for the generation of MPCCs and approximation of BMD MICs (±1 dilution) for filamentous fungi and the accurate classification of A. fumigatus strains as WT or non-WT in their response toward voriconazole based on the ECV. The establishment of ECVs has aided clinicians in determining whether A. fumigatus isolates behave more like WT strains toward a certain drug or more like non-WT isolates with known acquired mechanisms of resistance.
Variable rates of Aspergillus azole resistance, ranging from as high as 2 to 8% in some areas (3, 4, 17) to <1% in others (18), have been reported. Variability of resistance rates likely relates to differences in testing frequency between laboratories and the relative risk of different populations acquiring resistant fungal infections (4). Additionally, prevalence data may be biased by susceptibility testing being carried out in referral centers, where the samples received are likely from patients who are failing therapy, as opposed to testing a random cross section of samples (4). The demand for susceptibility testing is rising, particularly for isolates from patients receiving antifungal therapy, underscoring the need for a rapid, inexpensive method of susceptibility profiling (4).
Voriconazole resistance is most commonly mediated via modification of the binding site of the drug. The target site, 14 α-demethylase, is encoded by the cyp51A gene (19). There are two proposed sources of selective pressure for the fixation of Cyp51A mutations mediating resistance. The first source is the selective pressure provided during prolonged drug exposure. Consequently, resistance of Aspergillus to voriconazole is most commonly identified in patients with aspergillomas requiring prolonged courses of therapy (20). Second, humans may be infected by inhaling environmental molds that have developed resistance due to the selective pressure from azole fungicides. This proposed mechanism of resistance has been seen primarily in Europe. The majority of resistant isolates identified in the Netherlands all share a mutation, L98H, coupled with a tandem repeat in the promoter region, suggesting a common source of a resistant strain (20).
While the use of MALDI-TOF MS technology allowed for accurate identification of isolates as WT or non-WT, the current methodology did not save any time compared to traditional BMD assays. While the abbreviated MALDI-TOF MS method, which requires only 3 drug concentrations, may potentially reduce the amount of work required for setup, it still requires 30 to 48 h of incubation prior to analysis to allow time for sufficient growth for accurate differentiation between WT and non-WT strains (Table 1). Consequently, results are not obtained any faster with this method than with traditional BMD or antifungal disk diffusion susceptibility testing (21, 22). In addition, the MALDI-TOF MS analysis resulted in MPCCs, which are only surrogates for the well-established and more easily clinically interpreted MICs.
Currently, for determination of phenotypic susceptibility profiling, BMD and antifungal disk diffusion testing are still the preferred methods, and for accurate identification of Cyp51A mutations, gene sequencing remains the gold standard. Future development of alternative MALDI-TOF MS protocols and/or databases that can recognize the specific molecular changes associated with resistance rather than serving as an indicator of MIC would ultimately be of greater utility in the clinical mycology laboratory.
MATERIALS AND METHODS
Isolates.Twenty isolates of Aspergillus spp. were tested, 17 clinical isolates of A. fumigatus, 4 of which were previously demonstrated to be phenotypically and genotypically resistant to voriconazole (generously gifted by Susan Howard; 4), 2 clinical isolates of Aspergillus ustus, and 1 clinical isolate of Aspergillus calidoustus. A. ustus and A. calidoustus strains, which display intrinsic elevated MICs towards voriconzole, were included to challenge the assay against a broad range of voriconazole MICs. For each isolate, the MIC of voriconazole was determined using the Sensititre YeastOne microdilution panel (Thermo Fisher Scientific, Waltham, MA), which is an adaptation from the CLSI BMD reference method (21). The sequence of the cyp51A gene was verified for each non-wild-type A. fumigatus isolate using PCR and sequencing. Briefly, after DNA extraction, the promoter and full coding regions of the cyp51A gene were amplified and sequenced using primers described previously (23). The sequences were compared against the sequence of wild-type cyp51A (GenBank accession no. AF338659) to identify known resistance mutations.
Sample preparation.Stored isolates (−80°C) were subcultured on potato dextrose agar and incubated at 28°C until growth was sufficient (∼72 h). Samples were prepared according to previously published methods with a few modifications (16, 24). Fungal inocula were prepared by dislodging and suspending conidia in 0.01% Triton X-100 (108 conidia/ml). When necessary, suspensions were filtered through cotton wool to remove hyphal fragments. Wells of a 24-well microtiter plate were inoculated with RPMI broth, fungal suspension (final concentration, 107 conidia/ml), and voriconazole solution (drug concentration, 0.125 to 16 μg/ml) or no voriconazole (positive growth control), for a final volume of 2 ml. Samples were incubated for 24, 30, or 48 h at 37°C with gentle agitation (130 rotations/min). After incubation, the fungal material was collected and washed twice in water and once in 70% ethanol. Pellets were dried in a SpeedVac concentrator (Thermo Scientific Savant DNA120) for 10 min. Equal amounts of 70% formic acid (20 to 100 μl) and acetonitrile (20 to 100 μl) were added to each pellet in proportion to the biomass. Samples were vortexed and centrifuged again; the supernatant was used to spot on the target for analysis.
MALDI-TOF MS analysis.CCI and MPCC determinations were performed using the Bruker Microflex LT mass spectrometer system and Bruker MBT Compass Explorer software v1.4 as previously described (24). For each experimental condition, 12 to 24 spectra were acquired and used to generate the CCI scores used to compare all drug concentrations. MPCCs were determined by identifying the lowest drug concentration at which the spectra were more similar to that observed at the maximum drug concentration than that observed with no drug (see Table S1A and B in the supplemental material). For the A. fumigatus strains, a cutoff value of 1 μg/ml voriconazole was selected, based on the epidemiological cutoff value (ECV), to determine if MPCCs could be used to classify strains as WT or non-WT (25). ECVs combine phenotypic and genotypic data to establish whether an isolate responds as WT or non-WT to a drug (25). Published studies (19, 26) and the European Committee on Antimicrobial Susceptibility Testing (25) have determined the ECV for voriconazole against A. fumigatus to be 1 μg/ml; MICs of ≤1 μg/ml are interpreted as WT, whereas MICs of >1 μg/ml are considered non-WT and may have acquired mechanisms for resistance. For the 3 non-A. fumigatus strains, no interpretation criteria can be applied; they were tested only to assess how the MPCC compared to the MIC values.
ACKNOWLEDGMENTS
We thank Maria Witkowska and Nazareno Ocampo for their technical assistance with this project.
FOOTNOTES
- Received 18 February 2017.
- Returned for modification 13 March 2017.
- Accepted 5 April 2017.
- Accepted manuscript posted online 12 April 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/JCM.00254-17.
- © Crown copyright 2017.
The government of Australia, Canada, or the UK (“the Crown”) owns the copyright interests of authors who are government employees. The Crown Copyright is not transferable.
