Antifungal Minimal Inhibitory Concentrations of Mold Isolates from Patients with Cancer; Single-Center Experience, 2018–2023

TL;DR

This study examines antifungal resistance in mold isolates from cancer patients, finding rising resistance to commonly used drugs like isavuconazole and voriconazole.

Contribution

The study provides new data on antifungal susceptibility trends in cancer patients, highlighting resistance patterns in Aspergillus species.

Findings

Isavuconazole showed high non-wild-type MIC rates in multiple Aspergillus species.

A. fumigatus resistance to voriconazole remained low but stable over time.

Itraconazole and posaconazole had the lowest MIC distributions against Aspergillus spp.

Abstract

The increasing emergence of antifungal resistance poses potential clinical challenges, particularly among immunocompromised patients with cancer at risk of invasive mold infections, but data on antifungal susceptibility trends specific to this population are few. We evaluated distributions of minimal inhibitory concentrations (MIC), including minimal effective concentrations (MEC) for echinocandins, of 11 antifungal agents for 523 mold isolates (395 Aspergillus spp.) from cancer patients. Based on published Clinical and Laboratory Standards Institute guidelines, isavuconazole had notably high rates of non-wild-type MICs for A. fumigatus (19.6%), A. flavus/oryzae (34.8%), A. niger complex (26.1%), and A. terreus complex (8.33%). Persistent low baseline resistance of A. fumigatus to voriconazole was observed across multiple years (2.4–11.5% per year, average 8.41%) without significant trends in MIC change over time. Itraconazole and posaconazole demonstrated the lowest MIC distributions (MIC50 ≤ 0.06–0.5 µg/mL) of the azoles against Aspergillus spp. Amongst the A. niger complex, 29.4% (27/92) demonstrated non-wild-type MICs to itraconazole. While the A. nidulans group was less frequent (n = 24), bimodal peaks in MIC/MEC were noted for caspofungin (≤0.06 and 1 µg/mL). Non-Aspergillus molds of significance (Zygomycetes, Fusarium spp., Scedosporium spp., and Lomentospora prolificans) demonstrated variable but increased MICs to antifungal agents as previously described. Our results highlight increased rates of non-wild type MICs for Aspergillus spp. to isavuconazole and voriconazole, which are commonly used antifungal agents in cancer patients. Such AST trends should be closely monitored in populations with frequent antifungal use and encourage increased antifungal stewardship efforts.

Figures

Full text

1. Introduction

Invasive mold infections (IMIs) are increasingly recognized in patients with immune compromise or severe underlying conditions [1]. Out of an estimated 6.5 million cases of invasive fungal infections globally each year, approximately 2.1 million individuals develop invasive aspergillosis (IA) among patients with different underlying conditions including cancer [2]. An estimated 14,820 hospitalizations occurred due to Aspergillosis infection in 2014 in the USA [3]. IA is a serious opportunistic infection encountered following intensive chemotherapy for hematologic malignancies or hematopoietic stem cell transplantation [4]. Despite advances in antifungal therapies, IA in intensive care unit patients remains associated with high mortality rates, with crude mortality exceeding 80% and attributable mortality ranging between 18 and 48% [5]. The genetic adaptability of molds, limited antifungal agents available, breakthrough infection on anti-mold prophylaxis, and incomplete antimicrobial susceptibility testing (AST) standards and clinical breakpoint developments pose challenges in defining the optimal treatment for IMIs [6,7].

When clinical breakpoints are limited, epidemiological cutoff values (ECOFFs/ECVs) may help in monitoring local susceptibility trends to inform empirical treatment decisions [8]. The ECV values, defined as the upper limit of the wild-type population (wt-UL), marks the highest minimal inhibitory concentration (MIC) of the wild-type population, ensuring that the isolates used for its determination lack resistance mechanisms [9,10].

Isavuconazole, a triazole antifungal, is widely used for the treatment of invasive aspergillosis and other clinically important fungal infection caused by molds and yeasts [11,12]. Voriconazole is another potent antifungal compound used as a first line defense against invasive aspergillosis [13]. Posaconazole and amphotericin or liposomal amphotericin B are also used to treat infections caused by Aspergillus spp. and Mucorales [14,15]. Echinocandins (e.g., caspofungin, anidulafungin, and micafungin), despite their limited efficacy against Aspergillus spp., serve as an alternative option for treating invasive aspergillosis due to their favorable tolerability and ability to enhance the effects of other antifungal agents and utility in salvage therapy [14,16]. To date, there are few studies examining the trends in MIC and non-wild-type (NWT) populations in cancer patient populations receiving antifungal prophylaxis, empiric therapy, or directed therapy. The purpose of this study is to present the MIC distributions of commonly used antifungal agents for clinically relevant molds isolated from an immunocompromised cancer patient population who often receive triazole antifungal prophylaxis. We compare our data to published MIC distributions and ECVs to understand antifungal MIC trends in a high-risk patient population.

2. Materials and Methods

2.1. Mold Isolates and Analysis of Minimal Inhibitory Concentrations

This was a retrospective study of mold isolates obtained from fungal cultures performed as the standard of care, 2018–2023 (City of Hope Comprehensive Cancer Center, Duarte, CA, USA). Isolates were routinely identified by mass spectrometry (Vitek MS, Biomerieux, Durham, NC, USA), with morphologic assessment and rDNA gene sequencing used as secondary methods as needed. First mold isolated per patient source per encounter were routinely submitted as the standard of care for AST (ARUP Reference Laboratories, Salt Lake City, UT, USA) following the CLSI-recommended broth microdilution method using RPMI 1640 medium [17]. Per CLSI standard, AST tests were performed once per isolate following validated methods under strict quality control. Reported MIC and MEC (minimum effective concentration) data were collected and de-identified prior to analysis. The standard AST panel included flucytosine (5-FC), amphotericin B, fluconazole, isavuconazole, itraconazole, posaconazole, voriconazole, ketoconazole, micafungin, anidulafungin, and caspofungin. The concentration ranges tested were as follows: flucytosine (0.06–64 µg/mL), amphotericin B (0.12–8 µg/mL), anidulafungin, caspofungin, and micafungin (0.06–8 µg/mL), and isavuconazole, itraconazole, ketoconazole, posaconazole, and voriconazole (0.06–16 µg/mL).

MIC_50_, MIC_90_, and modal MICs were calculated for each antifungal–organism pair. MIC_50_ and MIC_90_ refer to the MIC values representing the 50th and 90th percentile MIC for each antifungal–organism agent combination, respectively. Additionally, MEC data were similarly analyzed for echinocandins and refer to the lowest concentration of antifungal agent that results in small, rounded compact hyphal forms. MEC_50_, MEC_90_, and modal MECs were calculated for each echinocandins antifungal–organism pair. Statistical calculations were performed in R (version 4.4.2) with the dplyr package (version 1.1.4). For the clinical context, population level-cancer diagnoses were extracted over the study time period and sorted by mold group, but not at the individual patient or isolate level due to the de-identification of the AST dataset. The study protocol was determined by the Institutional Review Board at City of Hope not to be human subjects research.

2.2. Determining ECVs and Wild-Type Upper Limit (wt-UL)

For the purposes of comparisons to published ECVs, we determined the ECVs for our dataset using the ECOFFinder algorithm following standard methods [18]. Briefly, cumulative numbers of MIC distribution data collected from organism–agent combinations were modeled using non-linear regression to fit a cumulative log-normal distribution [18]. ECOFFinder is best-suited for unimodal, normally distributed data that closely fit the distribution curve; otherwise, it may fail to determine ECVs, as we observed in many of the tested combinations. The wt-UL was defined as two MIC steps higher than the modal MIC [19]. Even though both ECVs and wt-UL define the upper limit of the MIC value that differentiates wild type isolates from NWT isolates, we designated the latter as wt-UL to distinguish it from ECOFFinder-generated ECVs.

3. Results

3.1. Distribution of Isolated Mold Pathogens

Overall, 1025 clinical fungal isolates from 784 cancer patients were submitted for fungal AST (Table S1, Figure S1). The most prevalent molds were Aspergillus fumigatus (n = 215), A. niger complex (n = 96), Zygomycetes (n = 49), Fusarium spp. (n = 27), Scedosporium apiospermum/S. boydii (n = 22), and Lomentospora prolificans (n = 17). According to Poisson regression analysis, there was a statistically significant upward trend in the observed counts of Aspergillus spp. over time from 2018 to 2023 (p < 0.001).

Most Aspergillus isolates were from the lower respiratory tract (376/412, 91.3%), as were Zygomycetes (40 respiratory, 1 blood, 8 tissue) and S. apiospermum/S. boydii (18 respiratory, 3 blood, 1 tissue). By contrast, Fusarium spp. were most-isolated from blood (13 blood, 8 tissue, 5 respiratory tract), as was L. prolificans (10 blood, 4 respiratory tract, 3 tissue). Additionally, the overall distribution of specimen sources of the mold isolates is presented in Figure S2. Briefly, respiratory samples contributed the highest proportion (81.16%) of total mold isolates (n = 605), followed by tissue (7.77%), blood (5.12%), wound (1.65%), body fluid (1.49%), bone, urine, bone marrow, cerebrospinal fluid, ear, skin, and other sources (0.17–0.83%).

Cancer diagnoses associated with Aspergillus spp. recovered in cultures were more balanced between solid organ malignancies (45.9%) and hematologic malignancies (51.7%), compared to Fusarium spp., which were only isolated from patients with hematologic malignancies (20 leukemia, 2 myelodysplastic syndrome, 4 lymphoma). Scedosporium/Lomentospora spp. and Mucorales were likewise predominantly associated with hematologic malignancies (87.5% and 85.2%, respectively) rather than solid organ malignancies (14.2% and 11.1%, respectively), with a small number of combined hematologic and solid organ malignancy cases seen for Mucorales (3.7%).

3.2. Distribution and Trends in MICs

The MIC/MEC_50_, MIC/MEC_90_, mode, range, and ECVs for each mold organism–drug combination were tabulated (Table 1, Table 2 and Table 3). For Aspergillus spp., A. terreus complex predictably demonstrated elevated MIC_50_ (2 µg/mL) and range (0.5 to ≥8 µg/mL) for amphotericin B. The other Aspergillus species had overall non-wild-type (NWT) MIC rates for amphotericin B of <1.0% (Figure 1). MIC/MEC_50_ values for echinocandins were uniformly low (≤0.06 µg/mL) except for caspofungin with A. nidulans (MIC/MEC_50_: 0.5 µg/mL), which demonstrated an unexpected bimodal pattern, with one mode of ≤0.06 µg/mL and a second mode of 1 µg/mL (Table 3). Among mold active azoles, itraconazole and posaconazole demonstrated the lowest MIC distributions (MIC_50_ ≤ 0.06–0.5 µg/mL) compared to isavuconazole and voriconazole (MIC_50_ 0.25–4 µg/mL) for Aspergillus spp. Specifically, A. nidulans and A. terreus complex showed the lowest MIC_50_ (≤0.06 µg/mL), whereas A. niger complex demonstrated the highest MIC_50_ (0.5 µg/mL). For isavuconazole, A. niger complex and A. lentulus had higher MIC_50_ (4 µg/mL) than other Aspergillus species (0.25–1 µg/mL). A. lentulus, while consisting of only nine isolates, demonstrated relatively high MICs to amphotericin B (MIC_50_ of 4 µg/mL) and azole agents except for posaconazole (range ≤ 0.06–0.5 µg/mL).

Changes in antifungal MICs for Aspergillus species were evaluated over time (Figure 2A,B). Overall, we visually observed increasing amphotericin B MICs over time for A. fumigatus, A. niger, and A. terreus (Figure 2A). Anidulafungin, caspofungin, and micafungin showed rising MIC trends against A. nidulans (Figure 2A). Interestingly, ketoconazole exhibited decreasing MIC trends across all Aspergillus species (Figure 2B). Some antifungal MICs remained stable throughout the study period of 2018–2023. Although these trends were visually suggested (Figure 2A,B), no statistically significant changes were identified (p > 0.05).

For non-Aspergillus mold species, MIC distributions largely confirmed the expected patterns by group (Table 2), though no CLSI ECVs are available for comparison [20,21]. Antifungal agents showed little overall activity against Fusarium spp., though some variation was observed in the MIC range for voriconazole, posaconazole, and amphotericin B. Scedosporium spp., which generally showed elevated MICs to most agents including amphotericin B (MIC_50_ ≥ 8 µg/mL) with more variability for voriconazole and posaconazole. For Zygomycetes, posaconazole had the most effective MICs among the azole agents, followed by itraconazole and isavuconazole. Due to small numbers, individual species of Zygomycetes and Fusarium were not analyzed separately. Zygomycetes MICs for isavuconazole significantly correlated with posaconazole (Spearman’s ρ = 0.857, p < 0.001) and itraconazole (Spearman’s ρ = 0.842, p < 0.001) MIC values. Against Amphotericin B, 93.8% (46/49) of Zygomycetes MICs were ≤2 µg/mL.

3.3. Emergence of Non-Wild-Type Mold Isolates

Compared to published CLSI ECVs for species with >90 isolates (A. fumigatus and A. niger complex), most ECVs were similar but several key differences emerged. For both species, our calculated ECVs for isavuconazole and voriconazole were two-fold higher. Accordingly, isavuconazole NWT rates based on CLSI criteria were notably increased across Aspergillus spp.: A. flavus/oryzae (34.78%), A. niger complex (26.09%), A. fumigatus (19.63%), and A. terreus complex (8.33%). Against voriconazole, 8.41% of A. fumigatus were resistant, though rates varied by year (2.4–11.5%) (Figure 1A,B). The Voriconazole NWT rate was similar for A. niger complex (8.33%). Finally, our ECVs for itraconazole were two- to four-fold lower than the published CLSI ECVs for A. fumigatus and A. niger complex. However, the MIC distribution for A. niger complex was not unimodal, with 27 of 92 isolates (29.3%) demonstrating NWT MICs of ≥8 µg/mL.

4. Discussion

Our study identifies Aspergillus as the most frequently isolated fungal pathogen from our cancer center. The dominance of Aspergillus spp. aligns with the existing literature, as this genus is commonly associated with IMIs in immunocompromised cancer patients, exceeding the frequency of invasive candidiasis [22,23]. IA remains associated with high mortality rates in critically ill patients, with crude mortality exceeding 80% and attributable mortality ranging between 18 and 48% [5].

The antifungal AST data presented here largely reflect other published studies based on the CLSI broth microdilution method, except for notable MIC elevations to isavuconazole for Aspergillus spp. (8.3–34.8% NWT) and decreased in vitro activity for voriconazole against Aspergillus compared to other triazoles [20,24]. The SENTRY surveillance program found A. fumigatus NWT rates for isavuconazole of 3.8% (2015–16) to 6.0–7.3% (2017–21) [20,25], while our NWT rate of 20.0% was appreciably higher. Among studies focused on cancer patient populations, one retrospective cohort analysis identified A. fumigatus in one or multiple specimens from 13 cancer patients, with isolates from three of those patients exhibiting resistance to triazole antifungals [26]. Another study conducted at a tertiary care cancer center found that approximately 13% of Aspergillus spp. isolates collected from 1999 to 2015 were classified as NWT for triazoles [27]. Additional studies in patients with cancer have utilized other AST methods such as gradient diffusion strips and may not be directly comparable, but nonetheless suggest a wide range of resistance rates in Aspergillus species to azole agents, from <1% to nearly 30% [28,29].

While we identified a significant correlation between isavuconazole and voriconazole MIC values for A. fumigatus (Spearman’s ρ = 0.529, p < 0.001), this correlation was weaker than others have previously reported (Spearman’s ρ = 0.885–0.887) [30,31]. The reasons for this difference are unclear, but inter-laboratory technical variations in MIC determinations may contribute. In addition, adaptive responses in mold isolates that differentially affect MICs for voriconazole and isavuconazole may be possible given that the practice at our center favors the usage of isavuconazole for either antifungal prophylaxis or treatment, with very low voriconazole utilization due to drug–drug interactions and less predictable pharmacokinetic profiles. Our study also revealed a notably higher percentage (29.4%) of A. niger NWT isolates for itraconazole. This finding aligns with broader increased itraconazole resistance trends observed in the A. niger complex organisms. One large study using the gradient diffusion method and EUCAST interpretations found that over half (52%) of A. niger complex isolates were resistant to itraconazole, with rates varying across cryptic species from 33 to 100% [32]. Cross-resistance to other azoles may be observed but is not a consistent finding, and the mechanism of resistance in A. niger has not been fully elucidated, although it does not appear to be due to typical cyp51 mutations or upregulation [32,33].

There is general concern over increasing azole resistance in IA pathogens associated with cyp51 mutations [11,34], but data vary over time and between studies [24,25]. We saw no significant temporal increases in MIC values but did find variability between years. Analytical test factors, institution-specific clinical practices, and environmental variables may affect mold biology and MIC values [11,35]. Our increased baseline NWT rates for isavuconazole coincide with its use as prophylaxis and empiric or targeted therapy at our institution, yet MICs were not appreciably increased for posaconazole, which is also routinely utilized at our site for prophylaxis and therapy. An investigation into whether the MIC trends correlate with clinical exposure to antifungal agents can provide further insights for antifungal stewardship and interpretation of MIC data.

Echinocandins are known to cause morphological changes and growth arrest in Aspergillus spp.; therefore, determining MECs is recommended in order to assess their activity more accurately [36] MICs and MECs were identical for all of the echinocandins–organism combinations tested. In a transplant-associated surveillance network study, most Aspergillus spp. isolates collected between 2001 and 2006 showed MEC values not exceeding the ECVs defined for echinocandins [37]. Although the echinocandin agents generally had low MICs below ECVs for Aspergillus spp. in this study, we also found a selective increase in MICs/MECs to caspofungin among echinocandins in roughly half of the isolates identified as A. nidulans complex. This AST pattern has been described in other Aspergillus section Nidulantes species, particularly A. spinulosporus, but not with A. nidulans sensu stricto [38,39]. A. spinulosporus is a rare cause of invasive fungal infection in immunocompromised hosts and it is unknown if commercial MALDI-ToF systems would identify this closely related species as A. nidulans. We also visually observed an increasing trend in the MIC/MEC values of the echinocandins against some of the Aspergillus spp. (Figure 2). These observations were not statistically significant, which may be due to the multiplicative nature of MIC/MEC data and the limited number of collection years.

Non-Aspergillus mold isolates pose challenges in immunocompromised patients and in demonstrating elevated MICs to multiple antifungal drug classes. There are no CLSI interpretative standards or ECVs for these molds, though ECVs using the CLSI-based broth microdilution method have been described by other study groups [40,41]. For Fusarium, amphotericin B ECVs ranged from 4 to 8 µg/mL across species, similar to our data, but for other antifungal agents, ECVs varied much more greatly by species; therefore, we are unable to make meaningful comparisons [41]. For Scedosporium/Lomentospora spp., amphotericin B MICs are commonly increased, though in our dataset the MIC distribution (MIC_50_ = 8 µg/mL) is skewed higher than reported by others [42]. Elevated voriconazole MICs were also common in our isolates compared to prior studies, raising concerns of effective therapy options since voriconazole is a primary therapeutic agent for scedosporiosis [42,43]. For Zygomycetes, others have found variable but elevated MICs to isavuconazole simultaneously with favorable MICs to posaconazole [21,44], like the findings in the present study, and we additionally describe somewhat less-favorable MICs to itraconazole compared to posaconazole. Our findings for amphotericin B MIC distributions (ECV 97.5% of 2 µg/mL) were similar to previously reported ECVs of 2–4 µg/mL [40]. The clinical utility of AST performance for these non-Aspergillus molds remains to be proven, as clinical factors and other therapeutic modalities may be correlated more strongly with outcomes than MIC results from AST [45]. Additionally, it is hoped that newer antifungal agents in clinical trials, including olorofim and fosmanogepix, will soon become available for the treatment of IMIs caused by molds that are difficult to treat and that appear highly resistant in vitro [46,47].

Our collection of index mold isolates for antifungal susceptibility was tested at the time of care, minimizing selection bias and the effects of freeze–thaw and culture passages on MIC values. Our MIC testing and analysis was based on CLSI standards, recognizing that while there are also EUCAST standards for antifungal testing, several key differences exist between the two published AST methods [48]. In the absence of clinical data, we did not determine if all isolates were associated with IMIs and could not determine if elevated MIC values for certain agents were associated with patient parameters such as the cancer diagnosis or the prior use of antifungal agents.

This study offers epidemiological data to support the understanding of antifungal MICs across pathogenic molds in a high-risk cancer patient population commonly receiving antifungal therapy. Among the antifungal agents tested, posaconazole along with other echinocandins demonstrated consistently lower MICs across Aspergillus spp. Posaconazole and had a more favorable MIC distribution than isavuconazole and itraconazole against Zygomycetes. We further make note of elevated MICs for certain organism–antifungal agent combinations that should be monitored, including isavuconazole and Aspergillus and Zygomycetes, caspofungin and A. nidulans, and itraconazole and A. niger. While we describe these MIC trends, there are no clinical breakpoints to help interpret the applicability of these MIC values. Regardless, antifungal stewardship efforts are needed for the surveillance of unfavorable MICs and to help prevent the additional emergence of NWT strains. Although fungal antigen and molecular testing are widespread and important diagnostic tools, the microbiologic sampling and antifungal AST of recovered isolates remain vital to understand MIC patterns and emerging trends in vulnerable populations. Genotyping for cyp51 and other alterations in Aspergillus spp. as well as clinical correlation studies can help further elucidate the significance of the elevated MICs and NWT populations that were observed.