Five University Medical Centers (UMCs) participated in the surveillance network, including Leiden UMC (Leiden), Erasmus Medical Center (Rotterdam), Amsterdam UMC, VU Medical Center (Amsterdam), UMC Groningen (Groningen), and Radboud UMC (Nijmegen). The geographic regions include the west of the Netherlands (Rotterdam, Leiden, and Amsterdam), which is the most heavily populated region; the north (Groningen); and the east (Nijmegen). The centers were asked to screen A. fumigatus isolates cultured from clinical specimens by using an agar-based screening test (VIPcheck; MediaProducts, https://www.mediaproductsbv.nl). VIPcheck contains 3 agar wells supplemented with itraconazole, voriconazole, and posaconazole, and a growth control well (11). A. fumigatus colonies from the primary culture were inoculated on the 4-wells plate, incubated for up to 48 hours, and inspected for growth. If an isolate grew on any of the azole-containing wells, the isolate was sent anonymously to Radboud UMC for MIC testing and genotypic characterization. MIC testing was performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) broth microdilution reference method (12–14) for amphotericin B (AmB), itraconazole, voriconazole, posaconazole, and isavuconazole (added in 2015 after the drug was clinically licensed). Azole resistance was defined as resistance to >1 azole drug, according to EUCAST clinical breakpoints (itraconazole, >2 mg/L; voriconazole, >2 mg/L; posaconazole, >0.25 mg/L, and isavuconazole, >1 mg/L). EUCAST broth microdilution plates were made at Radboud UMC in batches of 250 96-well plates and complied with the recommended quality control standards (11–13). For A. fumigatus isolates with a confirmed azole-resistant phenotype, the full Cyp51A gene was analyzed by PCR amplification and sequencing (7). The Cyp51A sequence (GenBank accession no. AF338659) was used for mutation analysis. A spore suspension of all isolates was stored at −80°C in 10% glycerol.

Results of phenotypic testing were sent to the surveillance laboratories as soon as these were available. Analysis of the resistance genotypes was batched, and once a year each center received a list of isolates with resistance genotype and phenotype. The list of isolates was checked by the centers, who also provided the number of A. fumigatus culture-positive patients who had been screened for azole resistance during the year and the number of patients who harbored an azole-resistant isolate. Clinical information regarding underlying disease and classification of Aspergillus disease was not collected. Data on A. fumigatus resistance epidemiology are reported and published annually (15).

We calculated mean MICs with 95% CIs with GraphPad Prism 5.03 (https://www.graphpad.com). For calculations, we recoded MICs >16 mg/L as 32 mg/L. We calculated statistical tests on differences in MIC distributions by using Kruskal-Wallis test and tests on differences in classification according to clinical breakpoints by using the Fisher exact test.

Our resistance surveillance showed an increasing frequency of azole resistance in clinical A. fumigatus isolates during 2013–2018 in the Netherlands. Although the resistance frequency varied between the 5 UMCs, the increasing trend was observed in all centers. In 2017 and 2018, the azole-resistance frequency exceeded 10% in all centers, a threshold above which experts recommend reconsidering the use of first-line voriconazole monotherapy (17,18). Alternative regimens that cover resistance include voriconazole/isavuconazole in combination with liposomal-AmB or an echinocandin, or monotherapy with liposomal-AmB. As such, the Netherlands national guideline was revised and now recommends combination therapy in patients with suspected invasive aspergillosis, in particular for critically ill patients or when azole resistance cannot be excluded (19).

How the 10% resistance threshold should be determined remains unclear (17,18). Our study indicated that despite the increasing azole-resistance frequency, the resistance frequency of voriconazole remained below this threshold (8.8% in 2018). This observation warrants the question of which threshold (i.e., azole resistance or voriconazole resistance) should be used to guide decisions regarding primary treatment choices. The trend toward lower voriconazole MICs was observed mainly in TR₃₄/L98H isolates, whereas most nonresistant isolates exhibited a voriconazole MIC of 2 mg/L (intermediate). The optimal management of voriconazole-intermediate invasive aspergillosis remains unclear, but an increased failure rate is anticipated in patients treated with voriconazole monotherapy at the standard dose. An increased voriconazole trough level target of 2–6 mg/L is recommended or combining voriconazole with an echinocandin or liposomal-AmB monotherapy (18). Furthermore, there is currently insufficient evidence that infection caused by azole-susceptible or intermediate susceptible A. fumigatus isolates that harbor a resistance mutation can safely be treated with azole monotherapy.

Resistance was highly dominated by environmental resistance mutations, such as TR₃₄/L98H and TR₄₆/Y121F/T289A. Indeed, a recent study identified sites at high risk for resistance selection in the environment, referred to as hotspots (20). Particularly, stockpiling of decaying plant waste containing azole fungicide residues was found to harbor high numbers of resistant A. fumigatus, and isolates with identical resistance mutations to those recovered from clinical specimens were cultured from these hotspots (20,21). The observations support a strong link between environmental resistance selection and azole-resistant disease in humans. The increasing trend in azole-resistance frequency and the emergence of new resistance mutations indicate that the current use of azole fungicides is not sustainable and over time the medical use of azoles will be further threatened. Given the limited alternative treatment options for Aspergillus diseases, further research aimed at reducing the burden of azole resistance in the environment is urgently needed.

An important question is how the changing voriconazole phenotype in TR₃₄/L98H isolates can be explained. One possibility is that TR₃₄/L98H isolates harbor additional mechanisms or compensatory mutations that affect the overall azole-resistance phenotype. Additional changes in peptide sequence of the 14-α-sterol demethylase enzyme have been reported in TR₃₄/L98H isolates, such as F495I, which confers resistance to imidazole (16). Recombination experiments showed that recombinants with S297T/F495I in the TR₃₄/L98H background conferred high resistance to imidazole but also produced lower voriconazole MICs compared with the TR₃₄/L98H parent strain (16). Because TR₃₄ isolates originate in the environment through exposure to azole fungicides, changes in azole fungicide exposures could prompt changes in resistance phenotypes. However, only 9 isolates with an additional F495I mutation in the Cyp51A gene were found, of which 8 were voriconazole-susceptible, indicating that other or multiple factors might have contributed to the observed phenotype change.

Another possibility is the presence of other resistance mechanisms in TR₃₄/L98H isolates. Recent studies analyzing transcriptional control mechanisms of Cyp51A, have identified transcription factors, such as ABC transporter regulator, which was found to regulate many different processes involved in drug resistance, metabolism, and virulence (22). Furthermore, other steps in the ergosterol biosynthesis might be affected, for example, through mutations in the 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase-encoding gene (hmg1) (23). Mutations in the hmg1 gene resulting in peptide sequence changes were found to be frequent in TR₃₄/L98H isolates, possibly affecting the azole phenotype (23). These recent insights indicate that the resistance phenotype is likely to be multifactorial and that more research is needed to characterize possible mechanisms that explain the observed variation in TR₃₄/L98H phenotypes.

The main clinical implication of the observed phenotypic variation in TR₃₄/L98H relates to direct detection of resistance mutations through PCR tests. Several commercial PCR tests are available that enable detection of TR₃₄/L98H and TR₄₆/Y121F/T289A directly in clinical specimens (24,25), which are used to guide selection of antifungal drugs. However, this approach can only be used if the resistance genotype predicts the azole phenotype. Although TR₄₆/Y121F/T289A mutation detection is uniformly associated with resistance to voriconazole, the wide spectrum of voriconazole MICs in TR₃₄/L98H isolates hampers the use of direct detection of this mutation because voriconazole therapy might be withheld in cases of voriconazole-susceptible infection. However, as stated previously, the efficacy of voriconazole in voriconazole-susceptible and voriconazole-intermediate TR₃₄/L98H infection would need to be investigated before any treatment recommendations involving azole monotherapy can be implemented.

Our resistance surveillance had several limitations. All A. fumigatus isolates cultured from clinical specimens were screened for azole resistance, involving diverse patient groups and isolates not regarded clinically relevant. Including all isolates has the advantage of collecting a meaningful number of isolates, but the resistance frequency might not be representative for specific patient groups such as those with invasive aspergillosis. Furthermore, within hospitals the resistance frequency might vary between years (26), thus further complicating establishing a meaningful (local) resistance frequency. Another limitation was the lack of clinical information and Aspergillus disease classification. Such information would be more suitable to guide treatment decisions and local antifungal guidelines, but gathering the data is laborious, given that all aspergillosis cases need to be identified and classified. Various cohort studies have been performed in the Netherlands showing differences in resistance rates between patient groups. A resistance frequency of 26% (10 of 38 patients) was reported in A. fumigatus culture-positive intensive-care unit patients in a single hospital (27) and 29% (4 of 14 patients) in a national study of intensive-care unit patients with influenza-associated aspergillosis (2). In contrast, a low resistance frequency was reported in a 5 year single-center cohort of patients with hematologic malignancy (28), but the culture-positivity rate was only 6% and thus, in most patients, the presence of resistance remained unknown. To date, only few studies have determined the frequency of resistance mutations in culture-negative patients. In 1 study, the rate was found to be similar to that observed in culture-positive patients (29), although higher resistance rates were reported in culture-negative patients in comparison with the rates found by culture in another study on patients with chronic rather than acute invasive aspergillosis (30). Differences could be explained in part by coincidence because the number of cases in these single-center studies is small. Another explanation might be the number of A. fumigatus colonies that were tested for resistance, given that clinical cultures might contain azole-susceptible and azole-resistant colonies in cases of mixed infection (31,32).

In conclusion, azole resistance in A. fumigatus has been reported worldwide and provides major challenges regarding management of invasive aspergillosis and other Aspergillus diseases (9,32). Nevertheless, A. fumigatus is not included in global action plans to combat antimicrobial resistance (33), and no international surveillance programs monitor resistance in A. fumigatus. As a consequence, the presence and frequency of azole resistance remains unknown in most countries (34,35). Despite the challenges we face in performing resistance surveillance in A. fumigatus, our national surveillance has proved important to guide the national treatment guideline and provided insights in trends in resistance genotypes and phenotypes. Furthermore, continued surveillance will help to monitor effects of interventions aimed at reducing the resistance burden in the environment. We believe that global A. fumigatus resistance surveillance programs are warranted and should be implemented in initiatives to combat antimicrobial resistance.