In Vivo Efficacy of Liposomal Amphotericin B against Wild-Type and Azole-Resistant Aspergillus fumigatus Isolates in Two Different Immunosuppression Models of Invasive Aspergillosis

ABSTRACT

Using an immunocompetent murine model of invasive aspergillosis (IA), we previously reported that the efficacy of liposomal amphotericin B (L-AmB) (Ambisome) is not hampered by the presence of azole resistance mutations in Aspergillus fumigatus (S. Seyedmousavi, W. J. G. Melchers, J. W. Mouton, and P. E. Verweij, Antimicrob Agents Chemother 57:1866–1871, 2013, https://doi.org/10.1128/AAC.02226-12). We here investigated the role of immune suppression, i.e., neutropenia and steroid treatment, in L-AmB efficacy in mice infected with wild-type (WT) A. fumigatus and with azole-resistant A. fumigatus harboring a TR₃₄/L98H mutation in the cyp-51A gene. Survival of treated animals at day 14 in both immunosuppressed models was significantly better than that of nontreated controls. A dose-response relationship was observed that was independent of the azole-resistant mechanism and the immunosuppression method used. In the neutropenic model, 100% survival was reached at an L-AmB dose of 16 mg/kg of body weight for the WT strain and the TR₃₄/L98H isolate. In the steroid-treated group, 90.9% survival and 100% survival were achieved for the WT isolate and the TR₃₄/L98H isolate with an L-AmB dose of 16 mg/kg, respectively. The 50% effective dose (ED₅₀) was 1.40 mg/kg (95% confidence interval [CI], 0.66 to 3.00 mg/kg) for the WT isolate and 1.92 mg/kg (95% CI, 0.60 to 6.17 mg/kg) for the TR₃₄/L98H isolate in the neutropenic model and was 2.40 mg/kg (95% CI, 1.93 to 2.97 mg/kg) for the WT isolate and 2.56 mg/kg (95% CI, 1.43 to 4.56 mg/kg) for the TR₃₄/L98H isolate in the steroid-treated group. Overall, there were no significant differences between the two different immunosuppressed conditions in the efficacy of L-AmB against the wild-type and azole-resistant isolates (P > 0.9). However, the required L-AmB exposure was significantly higher than that seen in the immunocompetent model.

INTRODUCTION

Invasive aspergillosis (IA) caused by Aspergillus fumigatus is an important opportunistic fungal infection, particularly in immunocompromised patients (1–4). Azole antifungals, such as voriconazole (VRC), isavuconazole (ISA), and posaconazole (POS), are drugs recommended to manage Aspergillus diseases (5). VRC and ISA are currently recommended as first-choice treatments for IA, and POS is indicated for prophylaxis and salvage therapy (6, 7). However, the management of IA has become more complicated due to the emergence of azole resistance in A. fumigatus (8–18). Surveillance studies indicate that azole resistance is increasing in Europe, the Middle East, Asia, Africa, and Australia and, most recently, in North and South America as well (19–30). Azole-resistant IA has been shown to be associated with a very high mortality rate; therefore, alternative treatment regimens need to be investigated to improve the outcome of patients (18, 31, 32).

We have previously shown that liposomal amphotericin B (L-AmB) (Ambisome) is effective against azole-resistant IA using a nonneutropenic murine model of disseminated infection (31). A dose-response relationship was observed that was independent of the presence of an azole resistance mechanism. However, the clinical response to antifungal therapy in patients with IA is due to multiple factors, including host factors (2, 7, 32). A nonneutropenic model might therefore overestimate the efficacy of L-AmB, as many patients with IA might be neutropenic or might receive corticosteroids (3, 33).

We investigated the efficacy of L-AmB in models of two common immunocompromised conditions, namely, neutropenia and corticosteroid therapy. We determined the pharmacodynamics and dose-response relationships of L-AmB against a wild-type (WT) A. fumigatus isolate and an azole-resistant A. fumigatus isolate in two different setting of immunosuppression, utilizing our murine model of disseminated aspergillosis. The levels of efficacy of L-AmB seen under the two immunocompromised conditions and with our previously conducted nonneutropenic model were then compared. In all models, the same A. fumigatus isolates, mouse strains, and experimental designs were used.

(Parts of these results were presented at the 26th European Congress of Clinical Microbiology and Infectious Diseases 2016, 9 to 12 April 2016, Amsterdam, the Netherlands [34].)

RESULTS

Survival curves.

Figure 1 shows the survival curves of L-AmB-treated mice by dose. The survival curves for all control groups receiving intravascular 5% glucose showed a mortality rate of 90% or 100%. Mouse survival at day 14 in both immunosuppressed models was significantly better than that of the controls. A dose-response relationship was observed independently of the immunosuppression used. In the neutropenic mice, the maximum effect (100% survival) was reached at an L-AmB dose of 16 mg/kg of body weight for both the WT isolate and the TR₃₄/L98H isolate. In the steroid immunosuppression groups, 90.9% survival and 100% survival were achieved with the highest dose of L-AmB (16 mg/kg) for the WT isolate and the TR₃₄/L98H isolate, respectively.

FIG 1.

Efficacy of L-AmB against wild-type control and azole-resistant A. fumigatus isolates in two different immunosuppression models. Survival curves are depicted by strain. Animals were treated intravenously at days 1, 2, and 5 postchallenge with increasing 4-fold doses of L-AmB ranging from 0.004 to 16 mg/kg/day and were observed for 14 days. Placebo groups received 5% glucose. For all groups, n = 11. In the neutropenic groups, mice were rendered neutropenic by injection of 150 mg/kg cyclophosphamide intraperitoneally on days −4 and −1 and day +4. For steroid immunosuppression, cortisone acetate (100 mg/kg/day) was administered subcutaneously 3 days before and 5 days after infection.

Dose-response analysis.

The dose-response curves for dosing regimens and control groups of L-AmB monotherapy in two different immunosuppression settings are shown in Fig. 2. In both the neutropenic and the cortisone immunosuppression groups, L-AmB treatment improved the survival of the mice in a dose-dependent manner. A dose-response relationship was observed that depended on the L-AmB dose level but was independent of the presence of an azole resistance mechanism and of the immunosuppression used. The Hill-type model with a variable slope fitted the relationship between the dose and the 14-day survival rate well, with R² = 0.98 (neutropenic mice) and R² = 0.99 (steroid-treated mice) against the WT isolate and R² = 0.93 (neutropenic mice) and R² = 0.97 (steroid-treated mice) against the TR₃₄/L98H isolate, respectively.

FIG 2.

Fourteen-day survival as a function of L-AmB dose in two different immunosuppression models. Data are shown for the wild-type strain (AZN81-96; VRC MIC, 0.25 mg/liter) and the TR₃₄/L98H isolate (V52-35; VRC MIC, 4 mg/liter). L-AmB treatment improved the survival of the infected mice in a dose-dependent manner for both isolates independently of the azole-resistance mechanism and immunosuppression method used. The curves indicate fits with the Hill equation for each isolate.

The 50% effective doses (ED₅₀) were 1.40 mg/kg (95% confidence interval [CI], 0.66 to 3.00 mg/kg) for the WT isolate and 1.92 mg/kg (95% CI, 0.60 to 6.17 mg/kg) for the TR₃₄/L98H isolate in the neutropenic model and were 2.40 mg/kg (95% CI, 1.93 to 2.97 mg/kg) for the WT isolate and 2.56 mg/kg (95% CI, 1.43 to 4.56 mg/kg) for the TR₃₄/L98H isolate in the steroid-treated groups. In addition, the levels of efficacy of L-AmB were not different between the two immunosuppression models (P > 0.05).

Comparative efficacies of L-AmB in two different immunosuppressed models.

In order to compare the efficacies of L-AmB in treating infection caused by the WT and azole-resistant A. fumigatus isolates in two different settings of immunosuppression, the best-fit value parameters for the dose-response curves (i.e., the ED₅₀ values of L-AmB) were defined and the data were compared (Table 1).

TABLE 1.

Comparison of efficacies of L-AmB against wild-type control and azole-resistant A. fumigatus isolates in two different immunosuppressed models based on the log ED₅₀

Efficacy of L-AmB	Values for indicated isolates
	Cyclophosphamide setting		Cortisone setting
	Wild type	TR34/L98H mutant	Wild type	TR34/L98H mutant
ED50 (95% confidence interval)a	1.40 (0.66 to 3.00)	1.92 (0.60 to 6.17)	2.40 (1.93 to 2.97)	2.56 (1.43 to 4.56)
Hill slope (95% confidence interval)	1.00 (0.37 to 1.65)	0.91 (0.05 to 1.77)	1.01 (0.82 to 1.21)	1.07 (0.31 to 3.13)

^aED₅₀, 50% effective dose. Data are not statistically significant as determined by the F-test and one-way analysis of variance (ANOVA).

One-way analysis of variance (ANOVA) followed by Tukey's multiple-comparison test showed that there were no significant differences between the survival rates and the dose-response curves of the isolates with an azole resistance mechanism and WT controls (P > 0.9), and no differences in the levels of efficacy were found in the comparisons of the different immunosuppression models (P value = 0.96). In addition, the null hypothesis was not rejected in an F-test (P = 0.53; F = 0.81; DFn [degrees of freedom numerator] = 4; DFd [degrees of freedom denominator] = 31), indicating that the curve fits did not significantly differ among the four groups.

Comparative levels of efficacy of L-AmB in the immunocompetent and immunocompromised models.

We previously investigated the efficacy of L-AmB monotherapy in an immunocompetent murine model of IA. In that study, the ED₅₀ values were 0.29 mg/kg (95% CI, 0.05 to 1.65 mg/kg) for the WT isolate and 0.59 mg/kg (95% CI, 0.02 to 14.80 mg/kg) for the TR₃₄/L98H isolate (31). There were also no significant differences in L-AmB efficacy against the WT and azole-resistant isolates in immunocompetent mice (P value = 0.83).

To determine the effect of immunosuppression, the pooled data from the immunocompetent model were compared to the pooled data from the immunosuppressed model. As shown in Fig. 3, the dose-response curves are different for the two models. Subsequent analysis showed that the ED₅₀ values determined for these two groups, 0.4 mg/kg (95% CI, 0.10 to 1.6 mg/kg) and 2 mg/kg (95% CI, 1.6 to 2.6 mg/kg), respectively, were significantly different (two-tailed P value < 0.0001; t = 29; df = 37).

FIG 3.

Fourteen-day survival as a function of L-AmB dose in the immunocompetent model versus the immunosuppressed model. The pooled data of the immunocompetent model were compared to the pooled data of the immunosuppressed model in order to obtain a larger number of subjects and increase the power of analysis. There were significant differences in best-fit parameters obtained from analysis of the dose-response relationships of L-AmB in the immunocompetent group versus the immunosuppressed group (two-tailed P value < 0.0001). The data for the immunocompetent model were obtained from a study published previously (31).

DISCUSSION

The results of our experiments indicated that L-AmB treatment was able to prolong survival in vivo and that the dose-response relationships of L-AmB treatment were independent of the presence of an azole-resistance mutation. Moreover, there were no significant differences in the levels of efficacy of L-AmB against wild-type and azole-resistant isolates under the two different immunosuppressed conditions. Analysis of the differences between the dose-response curves indicated that higher L-AmB exposure was required under immunocompromised conditions than in the immunocompetent model in order to achieve similar outcomes.

As expected, the presence or absence of azole resistance did not have an impact on outcomes, as there was no difference in the response curves for the azole-susceptible and azole-resistant strains. The mechanism of action of L-AmB is different from that of the azoles, and the amphotericin B MICs for the wild-type strain and the azole-resistant strain were similar, confirming that both were part of the WT distribution with respect to amphotericin B susceptibility (35). Thus, L-AmB could likely be used for therapeutic purposes irrespective of the presence of an azole resistance mechanism.

One could argue that the effects observed might be inoculum size dependent. Of note, a lower inoculum was needed to reproduce a successful infection (90% mortality) in immunosuppression models (Table 2). There could therefore be an under- or overestimation of the L-AmB dose response and exposure response required (36).

TABLE 2.

The inoculum size used for the immunocompetent animal model, cortisone treatment model, and cyclophosphamide immunosuppression model^a

Infection setting	No. of conidia/mouse for:
	Aspergillus fumigatus wild type (AZN8196)	Aspergillus fumigatus TR34/L98H mutant (V52-35)
Immunocompetent modelb	2.4 × 107	2.5 × 107
Cortisone model	1.75 × 106	1.75 × 106
Cyclophosphamide model	1.16 × 105	1.16 × 105

^aAnimals were infected with an inoculum corresponding to the 90% lethal dose (LD₉₀) of each isolate into the lateral tail vein.

^bThe data for the immunocompetent model were obtained from a study published previously (31).

Generally, in animal models of aspergillosis, the capacity to control different variables is important; among them, inoculum size and immunosuppressive regimen are elements whose control is essential both to evaluate of the efficacy of monotherapy, combinatorial therapy, and immunotherapy and to investigate vaccines (37). This is also of significant importance, since A. fumigatus is an opportunist fungus with low virulence that is mainly pathogenic for severely immunocompromised human patients. Therefore, in immunocompetent humans and animals, lower fungal inoculums commonly fail to establish a lethal infection and a larger inoculum is probably required. However, when the host is immunosuppressed with cytotoxic agents or with corticosteroids, much lower levels of fungal spores can result in a reproducible lethal infection (38). Therefore, fatal infection could be established with a lower inoculum than that seen in the nonneutropenic model.

The host susceptibility and underlying disorders that chronically (e.g., cancers and hematologic malignancies) and acutely (e.g., major surgery and multiple trauma) impair host responses to infection contribute to the severity of the infections, thereby resulting in a worse outcome (39). The clinical signs and symptoms of an invasive fungal infection (IFI) can be absent until the infection is at an advanced stage due to the reduction or absence of a systemic inflammatory response in patients with profound neutropenia or receiving steroids (40, 41). Therefore, it is generally believed that in addition to antifungal therapy, control of underlying medical or immunosuppressive conditions is mandatory for successful management of patient with IFIs (32, 42). Previously, Hope et al. showed that underlying factors which are extraneous to the inherent activity of antifungal agents may be equally important in determining the ultimate therapeutic outcome in the treatment of experimental disseminated candidiasis (43). In another study, Wiederhold et al. also compared the rates and extents of activity of antifungals in neutropenic and nonneutropenic mice with Candida albicans invasive infection (44). In both the group of immunocompetent animals and the group of immunosuppressed animals, anidulafungin significantly improved survival versus the rates seen with control mice, and significant reductions in (1-3)-β-d-glucan levels and fungal burdens were observed. However, in neutropenic animals, higher doses of drug were required to improve survival and reduce fungal burden. The researchers further concluded that the extent of anidulafungin in vivo efficacy may be dependent on host immune status.

As a limitation of the current study, we used only two A. fumigatus isolates. One could surmise that inclusion of a larger set of isolates with various MIC phenotypes (susceptible, intermediate, and resistant) and also of isolates with an isogenic background might provide significantly more insights from which to draw a robust conclusion on the kinetics of infection and immune status. Moreover, as a comparison, one might consider investigating the efficacy of an azole antifungal (i.e., voriconazole) in immunosuppressed versus immunocompetent mice. It could well be that the differences found here for L-AmB, being a fungicidal agent, would be greater for azoles, these being more fungistatic or at least showing far lower kill rates over time (45).

In the present study, however, our results showed that targeting pharmacological factors such as the choice of the effective antifungal against fungal species and prediction of the expected drug exposure may play a more important role in optimizing the management of invasive Aspergillus infections to aim at improved outcome. The data also confirm our previous findings in experimental murine models of IA, in which equal levels of exposure-response relationships were required for treating neutropenic and nonneutropenic mice with azole antifungals (46–48). Of note, this effect might be related to the fungicidal or fungistatic mode of action of the antifungals studied (49).

In conclusion, our results show that targeting the invading pathogen with an agent with a different mechanism of action is an effective approach for treatment of infections caused by azole-resistant A. fumigatus. In addition, the host immune status has a major impact on the size of inoculum needed to establish A. fumigatus infection in mice as well as on the efficacy of the antifungals.

MATERIALS AND METHODS

Fungal strains.

The following two clinical A. fumigatus isolates used for our in vivo experiments (Table 3) were obtained from patients with proven IA (classified according to European Organization for the Research and Treatment of Cancer/Mycoses Study Group [EORTC/MSG] consensus definitions) (50): a WT isolate without mutations in the cyp51A gene (AZN8196) and an azole-resistant isolate harboring the TR₃₄/L98H mutation (V52-35).

TABLE 3.

Disease classification, history of previous azole exposure, underlying azole resistance mechanisms, and in vitro antifungal susceptibilities of A. fumigatus isolates used in the in vivo study^a

ID no.	Cyp-51A substitution	MIC (mg/liter)b
		AMB	ITC	VRC	POS
Aspergillus fumigatus (AZN8196)	None	0.5	0.125	0.25	0.031
Aspergillus fumigatus (V52-35)	TR34/L98H	0.5	>16	4	0.5

^aAll isolates were classified as having proven IA according to EORTC/MSG consensus definitions. None of the isolates were from mice with prior exposure to azoles. ID, identifier.

^bIn vitro antifungal susceptibility testing was performed by using the EUCAST (European Committee for Antimicrobial Susceptibility Testing) broth microdilution (two-dimensional [8 by 12]). AMB, amphotericin B; ITC, itraconazole.

Strain identifications and the cyp51A gene substitutions were confirmed by sequence-based analysis, as described previously (10). In addition, microsatellite genotyping of the isolates was performed to confirm that they were genetically distinct (22). The isolates were stored in 10% glycerol broth at −80°C and were cultured on Sabouraud dextrose agar (SDA) supplemented with 0.02% chloramphenicol for 5 to 7 days at 35 to 37°C. All isolates were cultured again on SDA for 5 to 7 days at 35 to 37°C before preparation of the inoculum.

In vitro antifungal susceptibility testing was performed by using the EUCAST (European Committee for Antimicrobial Susceptibility Testing) (51) broth microdilution (two-dimensional, 8 by 12) guidelines. The final concentrations of the antifungal agents ranged from 0.016 to 16 mg/liter for amphotericin B (AmB), itraconazole (ITC), VRC, and POS. Aliquots of 100 μl of each drug at a concentration 2 times the targeted final concentration were dispensed in the wells of flat-bottom 96-well microtiter plates (Costar, Corning, NY). Trays were maintained for a period of less than 1 month at −70°C until the day of testing. After the microtitration trays were defrosted, 100 μl of the inoculum, corresponding to a final concentration of 2 × 10⁵ to 5 × 10⁵ CFU/ml from each isolate, was added to each well. The microtiter plates were incubated at 35 to 37°C for 48 h. Growth inhibition was quantified visually by using an inverted mirror. Paecilomyces variotii (ATCC 22319), Candida parapsilosis (ATCC 22019), and C. krusei (ATCC 6258) were used for quality control in all experiments. All experiments were performed in three independent replicates and on three different days. The MIC was defined as the lowest concentration that completely inhibited growth in comparison to that seen in the drug-free well (control) as assessed by visual inspection (51). The EUCAST breakpoints were used to classify azole-susceptible and azole-resistant isolates (52).

Antifungal agents.

The commercial formulation of L-AmB was obtained from the manufacturer (Gilead Sciences, Inc. Amsterdam, the Netherlands). Drug solutions were prepared on the day of study following the instructions of the manufacturer and were diluted with a standard 5% glucose solution to obtain the desired concentration.

Infection model.

The efficacies and exposure-response relationships of L-AmB monotherapy were determined in two different immunosuppressed murine mouse models of disseminated aspergillosis using intravascular administration of the A. fumigatus inoculums. In the neutropenic groups, mice were rendered immunosuppressed by injection of 150 mg of cyclophosphamide/kg of body weight intraperitoneally on days −4 and −1 and day +4. For steroid immunosuppression, cortisone acetate (100 mg/kg/day) was administered subcutaneously 3 days before and 5 days after infection as described previously (46, 53).

A total of 528 outbred CD-1 (Charles River, the Netherlands) female mice (4 to 5 weeks old, weighing 20 to 25 g) were randomized into groups of 11 mice for L-AmB monotherapy. Animals were infected via injection of 0.1 ml of the conidial suspension (corresponding to the 90% lethal dose [LD₉₀] for each isolate) into the lateral tail of the mouse, using a procedure described before (31).

Before the experiment was performed, the isolates were cultured once on SDA (Sabouraud dextrose agar) for 5 days at 35 to 37°C and subcultured twice on 15-cm-diameter Takashio slants for 5 days at 35 to 37°C. The conidia were harvested in 20 ml of sterile phosphate-buffered saline (PBS)–0.1% Tween 80 (Boom B.V., Meppel, the Netherlands). The conidial suspension was filtered through sterile gauze folded four times to remove any hyphae, and the number of conidia was counted in a hemocytometer. After the inoculum was adjusted to the required concentration, the conidial suspension was stored overnight at 4°C.

For each of the immunosuppression models, the LD₉₀ was determined separately for each isolate; groups of 11 mice were infected intravenously (i.v.) with four different inoculum sizes of each isolate ranging between 1 × 10⁵ to 1 × 10⁸ conidia, and survival was monitored for 14 days. The LD₉₀ was 1 × 10⁵ conidia/mouse in the neutropenic model and 1 × 10⁶ conidia/mouse in the steroid immunosuppression groups for both the WT control isolate and TR₃₄/L98H isolate. Confirmatory postinfection viability counts of the injected inocula were determined to ensure that the correct inoculum had been injected.

For efficacy studies, groups of 11 mice were infected via injection of an inoculum corresponding to the LD₉₀ of the isolate into the lateral tail vein. Mice then were treated intravenously at days 1, 2, and 5 postchallenge with increasing 4-fold doses of L-AmB ranging from 0.004 to 16 mg/kg once daily and observed for 14 days. Control mice were infected but received only 5% glucose.

The animals were housed under standard conditions, with drink and feed supplied ad libitum. The animal studies were conducted in accordance with the recommendations of the European Community (Directive 2010/63/EU revising Directive 86/609/EEC on the protection of animals used for scientific purposes adopted on 22 September 2010), and all animal procedures were approved by the Animal Welfare Committee of Radboud University (RU-DEC 2013-021).

In all survival studies, experienced individuals blind to the animal treatment performed the monitoring. The infected mice were examined at least three times daily. Clinical inspections focused on dehydration, torticollis, staggering, high weight loss (a decrease of 15% within 48 h or 20% within 24 h), and body temperature (drop to below 33°C). Mice demonstrating these clinical signs were humanely euthanized following strict protocols. On day 15 postinfection, all surviving mice were humanely euthanized under conditions of inhalational isoflurane anesthesia. The survival value (measured in days postinfection) was recorded for each mouse in each group and was considered an outcome effect measure to assess the therapeutic efficacy of L-AmB monotherapy (31).

Statistical analysis.

All data analyses were performed using GraphPad Prism, version 5.3, for Windows (GraphPad Software, San Diego, CA). Mortality data were analyzed by the log rank test. The relationship between in vivo efficacy (survival) and dose was determined by the use of nonlinear regression analysis and the Hill equation with a variable slope fitted to the data, with the maximum effect (maximum survival) constrained at ≤100%. The goodness of fit was checked by determination of the R² value and by visual inspection. Statistical significance was defined as a P value of ≤0.05 (two-tailed).

Student's t test, one-way analysis of variance (ANOVA) followed by Tukey's multiple-comparison test, and the F-test were used to define whether there were significant differences between the dose-response curves. In addition, multiple-comparison analysis based on the best-fit value parameters of each curve, the standard error, and the degrees of freedom for that fit was performed to test for statistical significance differences between the immunocompetent and immunosuppressed groups and within each of the data sets. The data for the immunocompetent model were obtained from a study published previously (31).