Efficacies of Fluconazole, Caspofungin, and Amphotericin B in Candida glabrata-Infected p47^phox−/− Knockout Mice

Abstract

Candida glabrata is the second leading cause of adult candidemia, resulting in high mortality. Amphotericin B is considered the treatment of choice, while the efficacy of fluconazole is controversial and caspofungin efficacy is unknown. To ascertain drug efficacy in vivo, the utility of a murine model of C. glabrata infection was investigated. C. glabrata was found to cause progressive, lethal infection when injected intravenously into C57BL/6 mice with reduced oxidative microbicidal capacity due to knockout of the p47^phox gene. Spleen and kidney organ CFU counts were determined in groups of mice 2 days after the mice completed 6 days of daily intraperitoneal drug treatment, which began on the day of infection. Daily injections of fluconazole at 80 mg/kg did not reduce spleen or kidney CFU counts after infection with C. glabrata strains having in vitro fluconazole MICs of 2, 32, or 256 μg/ml compared to saline-treated controls. However, this fluconazole regimen reduced spleen CFU counts in mice infected with Candida albicans, an infection that is known to be responsive to fluconazole. Caspofungin at 5 mg/kg and amphotericin B at 5 mg/kg were both effective in reducing fungal burden in spleens and kidneys of C. glabrata-infected mice. Ten mice treated for 6 days with caspofungin at 1 mg/kg survived for 15 days, though all 10 saline-injected mice died or were so ill that they had to be sacrificed by 96 h postinfection. This murine model provided evidence of the efficacy of amphotericin B and caspofungin but not of fluconazole against C. glabrata infection.

Candida glabrata is second only to Candida albicans as a cause of candidemia in adults, accounting for ca. 16 to 24% of candidemia cases (4, 18). Septicemia, endocarditis, pyelonephritis, esophagitis, and other deep sites of infection may also occur (5, 14). Mortality in patients with C. glabrata sepsis is high, unadjusted for the underlying disease, and at least equivalent to mortality in patients with C. albicans sepsis (6). Amphotericin B is considered the drug of choice. Although only 5 to 6% of C. glabrata isolates are resistant (MIC ≥ 64 μg/ml) to fluconazole in vitro, the clinical efficacy of this drug against C. glabrata infections has been difficult to establish (18, 22). C. glabrata infection largely occurs in patients with severe underlying diseases, rendering drug efficacy more difficult to assess. Catheter-acquired infections are a major cause of candidemia and may respond to catheter removal alone (14, 16). A randomized trial comparing fluconazole with amphotericin B in candidemia of nonneutropenic patients, largely catheter acquired, found equivalent results in patients with C. glabrata sepsis who were treated with fluconazole or amphotericin B (20). Four of 11 patients failed to respond to fluconazole, including one with a mixed infection, whereas 2 of 13 failed to respond to amphotericin B (P > 0.05). The range of error with a sample size this small is such that fluconazole could have provided a proportion cured that was 50% less than amphotericin B and still the difference would have failed to be statistically significant (Fisher's exact test). Complicating any decision about the use of fluconazole against C. glabrata infection is the 30-fold-lower susceptibility of C. glabrata compared to C. albicans (18). Drug concentrations used to distinguish susceptible from resistant C. glabrata were obtained from experience with C. albicans and may not apply to C. glabrata (21).

An animal model for C. glabrata might provide some insight into the interpretation of fluconazole susceptibility testing, as has been reported for C. albicans (2). However, C. glabrata has failed to cause a progressive infection in normal mice or rats, necessitating the use of organ colony counts to assess drug efficacy. Intravenous inoculation of 10⁸ C. glabrata into normal rats (11) or mice (7) did not cause obvious illness, progressive increase in organ colony counts, or death despite the enormous inoculum. Fluconazole at doses of 20 or 40 mg/kg given once daily by gavage for 7 days to rats beginning 3 days after infection reduced kidney but not liver colony counts (11). Only three rats were used in the control group, raising questions as to the reproducibility of this test. The fluconazole MIC for the C. glabrata strain used was 100 μg/ml. There have been two reports of the use of fluconazole in C. glabrata-infected mice rendered neutropenic by fluorouracil or irradiation (3, 13). Mice were given a very large intravenous inoculum (1 × 10⁸ to 1.4 × 10⁸ CFU) of isolates with fluconazole MICs ranging from 2 to >64 μg/ml. Fluconazole at 100 mg/kg twice daily by gavage for 5 days caused a significant reduction in the kidney CFU count in one of two experiments with the most susceptible isolate but no reduction in the splenic CFU count (3). In four experiments with other isolates and the same fluconazole regimen, the drug had no effect on the CFU in the spleen or other organs (3). In another report, mice infected with an isolate having an MIC of 16 μg/ml had a fourfold reduction in kidney but not in the splenic CFU counts after 7 days of fluconazole at 10 mg/kg (13). The wide scatter of organ CFU and the marginal evidence of efficacy left the therapeutic effect of fluconazole unclear in this model. Other drugs were effective in this same model. Intraperitoneal amphotericin B at 3 mg/kg daily caused roughly a 10⁴-fold reduction in the colony counts of both the kidney and the spleen (3). Caspofungin, given intraperitoneally to normal (1) or neutropenic (13) mice inoculated with C. glabrata, reduced colony counts in the kidney (1, 13) but not in the spleen (13).

We have explored the utility of a knockout (KO) strain of mice to determine efficacy of antifungal drugs against C. glabrata in vivo. This model avoids the confounding effects of cytotoxic drugs and neutropenia. C57BL/6 mice with disruption of the p47^phox gene have deficient production of phagocyte oxidative products and are susceptible to infection with Aspergillus nidulans (8). Isolates of C. glabrata with a large range of fluconazole susceptibility were obtained from bone marrow transplant recipients or human immunodeficiency virus-infected patients and used to infect p47^phox−/− KO mice. The mice developed a progressively lethal infection and proved useful for evaluating the relationship between in vivo efficacy and in vitro susceptibility. Caspofungin proved to be effective, as did amphotericin B, but fluconazole was not effective at 80 mg/kg daily, irrespective of in vitro fluconazole susceptibility.

MATERIALS AND METHODS

Mice.

p47^phox−/− KO mice on a C57BL/6 background were supplied by Taconic Labs (Germantown, N.Y.). After the experiments were completed, it was discovered that the colony had accidentally been crossed into a C57BL/6 colony with disruption of gamma interferon (IFN-γ) gene. Tissue samples from 10 mice from the colony were tested by PCR for copies of the native and disrupted gene by using a published method (19), with the following exceptions. A DNA isolation kit (Gentra Systems, Minneapolis, Minn.) was used to purify the mouse DNA, and the PCRs were heated to 94°C for 3 min prior to the first cycle; run for 35 cycles at 94°C for 40 s, 60°C for 40 s, and 72°C for 1 min; and then kept at 72°C for 10 min after the last cycle. Two of the ten mice were found to have heterozygous disruption of the IFN-γ gene (+/−); no homozygotes (−/−) were found. The C57BL/6 wild-type (WT) colony of mice supplied by Taconic for our experiments had not been contaminated. Female mice (ca. 6 weeks old) were housed in autoclaved cages with hardwood bedding. Mice in the chemotherapy experiments all received trimethoprim (640 μg/ml) and sulfamethoxazole (128 μg/ml) in their drinking water to prevent bacterial infections from confusing the endpoints. All federal and institutional guidelines regarding animal use were followed. The combination of hunched backs, difficulty walking, and extreme sleepiness was used to determine whether mice had lethal infections and should be sacrificed.

Inoculum.

C. glabrata isolates were obtained from mouthwashes of people receiving fluconazole prophylactically during the course of hematopoietic stem cell transplantation. Single colonies were collected from initial culture media and stored frozen at −70°C in 10% glycerol. An additional isolate, 40a, obtained from our stock, was used for initial experiments to establish virulence of C. glabrata in the KO mouse model. Inocula were grown on either yeast extract peptone dextrose (YEPD) agar or Sabouraud dextrose agar (SAB; Difco, Sparks, Md.) overnight at 30°C. Cells were suspended in saline, and the cell concentration estimated by optical density or hemocytometer count. Actual CFU were determined by culture of serial dilutions. Mice were inoculated with a volume of 0.2 ml into their lateral tail vein.

Histopathology.

Organs were fixed in 10% buffered formalin (pH 7; sodium phosphate buffer), mounted in paraffin, and sectioned for staining with hematoxylin-eosin or methenamine silver stain.

Drug treatment.

Unless otherwise noted, mice received daily intraperitoneal injections of 80 mg/kg fluconazole (a kind gift of Pfizer) diluted in saline, 5 mg/kg of amphotericin B (Pharma-Tek, Huntington, N.Y.) diluted in water, 1 or 5 mg/kg caspofungin (Merck, Rahway, N.J.) diluted in water or an equivalent volume (400 μl) of saline for 6 days beginning 1 h after infection. Mice were either observed for death or morbidity as defined above or killed by cervical dislocation 2 days after the last drug dose to assess fungal burden in tissue. For organ colony counts, spleens, kidneys and, in some experiments, livers were weighed and homogenized in a 10× (volume/weight) of phosphate-buffered saline (8.1 mM phosphate [pH 7]) by using a Stomacher 80 (Seward, Westbury, N.Y.). Tenfold dilutions were duplicate plated in SAB containing chloramphenicol (Sigma, St. Louis, Mo.) at 100 μg/ml, and colonies were counted after 2 days of incubation at 30°C. The geometric mean CFU/gram values were calculated, counting a negative culture as 1 CFU/g; these values appear on the graphs of log₁₀ as a zero.

Susceptibility testing.

Drug susceptibility was measured by the NCCLS M27A microdilution assay (17) except for fluconazole susceptibility, which was measured by E-Strip Test (AB Biodisk, Solna, Sweden). The susceptibilities of the isolates are given in Table 1.

TABLE 1.

Drug susceptibility of Candida isolates used in this study

Strain	MIC (μg/ml) of:
	Fluconazole	Amphotericin B	Caspofungin
C. glabrata 40A	>250	1	4
C. glabrata 12175	32	0.5	2
C. glabrata 18434	256	1	4
C. glabrata 1219	2	2	4
C. albicans 5314	1.5	1	2

Statistical calculations.

SPSS for Windows version 10.1 (Chicago, Ill.) program was used to compare tissue colony counts by using the Mann-Whitney U-test.

RESULTS

Establishing virulence of C. glabrata in the KO mouse model. (i) Survival of C57BL/6 WT versus KO mice after infection with C. glabrata.

This initial experiment was conducted to determine whether C. glabrata could cause a lethal infection in KO mice. The isolate chosen was from our stock collection and not used for later chemotherapy experiments. Groups of five WT or KO mice were injected with either 5.7 × 10⁶ or 5.7 × 10⁷ viable C. glabrata and observed daily for 14 days (Fig. 1). WT mice given either inoculum remained healthy, whereas lethal infection occurred in the KO mice.

FIG. 1.

Survival of mice after inoculation with C. glabrata. Groups of five C57BL/6 (WT) or p47^phox−/− KO mice were inoculated intravenously with either 5.7 × 10⁶ (WT [⧫]; KO [▴]) or 5.7 × 10⁷ (WT [◊]; KO []) viable C. glabrata 40A.

Two additional experiments were performed to show that lethality of C. glabrata infection in p47^phox−/− mice did not require genetic contamination by heterozygous disruption of the IFN-γ gene. Eight 8-week-old p47^phox−/− female mice were obtained that were free of genetic contamination. These mice were obtained from a (129 × C57BL/6)F₁, backcrossed five times into C57BL/6 mice for five generations (N5) (15). Comparable N5 control mice were not available. The injection of C. glabrata 1219 at 2.5 × 10⁷ viable cells per mouse resulted in the death of all eight KO mice on days 5 and 6 postinjection. Additionally, two groups of 10 6-month-old female C57BL/6 mice were obtained from Jackson Laboratory (Bar Harbor, Maine). Ten were were WT, and ten mice had homozygous disruption of the p91^phox gene. This latter group of mice had a defective phagocyte oxidative burst, a phenotype similar to homozygous disruption of the p47^phox gene (9). Of 10 gp91^phox−/− mice, 7 died on days 6 to 9 postinfection with C. glabrata 1219 at 6.5 × 10⁷ cells per mouse, with the remaining moribund mice being sacrificed on day 10. None of the WT mice became ill.

(ii) Histopathology after infection of WT and KO mice.

Tissue response to C. glabrata infection was compared between WT and KO mice. Twenty WT and twenty KO mice were each injected with 1.03 × 10⁷ viable C. glabrata, again by using our stock isolate, and then observed daily. On days 5, 9, 12, 15, and 19 postinjection, spleens, kidneys, and livers from four living mice per group were harvested for histopathology. WT mice remained healthy, whereas KO mice began to exhibit signs of illness starting on the fifth day postinjection, and all were sick by the seventh day. One KO mice died on each of days 6, 9, 10, and 12, and three died on day 15. Only one KO mouse was available for sampling on day 19. Upon histopathology, the spleens of WT mice showed hypertrophy of germinal centers with normal tissue architecture and rare yeast in the spleen and none in the liver and kidneys, a finding consistent with what has been published elsewhere (7). KO mice on days 5, 9, 12, and 15 showed progressive replacement of the splenic architecture with macrophages. Neutrophils were scant, and necrosis was not present. Numerous yeast cells were found scattered throughout the spleen upon Gomori methenamine silver staining. The liver tissue showed progressive infiltration of macrophages into the portal triads and hepatic lobules between days 5 and 15. Organisms were seen in the same loci as the macrophages. Clusters of macrophages were seen in the renal cortex by day 15, but much less prominently than in the liver or spleen. The renal medulla appeared normal. Yeast were sparse in Gomori methenamine silver stains of kidney.

Chemotherapy trials comparing the KO and WT mouse models. (i) Rationale for the study design.

In view of the uncertainty surrounding the significance of fluconazole susceptibility tests, experiments were designed to test fluconazole therapy with infections caused by three isolates with a wide range of MICs: 2, 32, and 256 μg/ml. These isolates appeared to provide roughly equivalent splenic colony counts in KO mice at 1 week postinoculation, with an inoculum of 10⁵ giving ca. 10⁵ to 10⁶ CFU per g of spleen. Amphotericin B and caspofungin were included in some experiments as positive controls. Most experiments used organ colony counts rather than death as the endpoint in order to reduce the number of mice experiencing pain and suffering. WT mice of the same genetic background were included to determine whether KO mice, only available in small numbers, contributed additional information. Treatment was begun an hour postinfection to maximize chemotherapeutic effect, since we found no therapeutic effect of fluconazole in preliminary experiments. An 8-day study period was selected for organ colony counts in most experiments because KO mice started to die after this point and because C. glabrata colony counts in organs of normal mice had been reported to decline rapidly after 7 days (7).

(ii) Response of C. glabrata infections to fluconazole in WT and KO mice.

Experiments in WT and KO mice evaluating the efficacy of fluconazole in reducing the C. glabrata burden in organs of infected mice are summarized in Table 2, all with daily intraperitoneal injections of 80 mg/kg. Regardless of the inoculum, the wide range of fluconazole susceptibilities in KO versus WT mice, or the duration of therapy, no differences between fluconazole-treated animals and saline-injected controls could be detected based on organ colony counts in the kidney or spleen.

TABLE 2.

Organ fungal burden 2 days after 6 days of fluconazole treatment at 80 mg/kg^a

Fungal strain	FLC MIC (μg/ml)	Mouse strain	Inoculum (CFU) per mouse	Treatment	No. of mice	Geometric mean (log10 CFU/g) ± SE in:
						Spleen	Kidney
C. glabrata
1219	2	KO	1.30 × 105	FLCb	9	5.19 ± 0.243	NDc
1219	2	KO	1.30 × 105	Salineb	8	5.38 ± 0.320	ND
1219	2	KO	5.00 × 105	FLC	7	6.35 ± 0.122	4.05 ± 0.269
1219	2	KO	5.00 × 105	Saline	7	5.59 ± 0.425	2.88 ± 0.568
1219	2	KO	7.90 × 103	FLC	5	4.65 ± 0.092	1.52 ± 0.185
1219	2	KO	7.90 × 103	Saline	5	4.76 ± 0.061	1.68 ± 0.138
1219	2	KO	9.40 × 105	FLC	5	5.40 ± 0.230	2.61 ± 0.403
1219	2	KO	9.40 × 105	Saline	5	5.40 ± 0.587	2.35 ± 1.024
12175	32	KO	1.52 × 105	FLC	5	5.51 ± 0.030	2.89 ± 0.159
12175	32	KO	1.52 × 105	Saline	5	5.48 ± 0.084	2.80 ± 0.165
18434	256	KO	1.59 × 105	FLC	5	4.96 ± 0.259	3.25 ± 0.289
18434	256	KO	1.59 × 105	Saline	5	4.31 ± 0.632	3.51 ± 1.121
1219	2	WT	9.40 × 107	FLC	5	5.78 ± 0.085	4.00 ± 0.120
1219	2	WT	9.40 × 107	Saline	5	5.40 ± 0.241	4.22 ± 0.830
18434	256	WT	1.56 × 107	FLC	5	4.88 ± 0.370	3.45 ± 0.482
18434	256	WT	1.56 × 107	Saline	4	3.85 ± 0.617	2.01 ± 1.501
C. albicans
5314	2	KO	8.90 × 103	FLC	5	2.31 ± 0.087d	1.30 ± 0.181
5314	2	KO	8.90 × 103	Saline	5	4.74 ± 0.037	1.93 ± 0.304

^aFLC, fluconazole. Mouse strains: WT, wild-type C57BL/6 mice; KO, p47^phox−/− mice.

^bThese mice received fluconazole or saline for 13 days.

^cND, not done.

^dFluconazole versus saline: P = 0.008.

(iii) Response of C. albicans infection in KO mice to fluconazole.

In order to show that the regimen of fluconazole used in C. glabrata infections would have a therapeutic effect in mice infected with C. albicans, isolates of C. glabrata and C. albicans were selected that had nearly identical susceptibilities in vitro. Inocula were selected to be as identical as possible: 8.9 × 10³ C. albicans and 7.9 × 10³ C. glabrata (Table 2). Fluconazole was markedly effective in reducing splenic counts of C. albicans (P = 0.008)- but not of C. glabrata-infected KO mice. Kidney CFU counts were unchanged by fluconazole treatment in both C. albicans- and C. glabrata-infected mice, but the CFU counts were too low to be evaluable, even in undiluted organ homogenate. The difference in response of the two species to fluconazole cannot be due to the drug dose or MIC, which were the same.

(iv) Response of C. glabrata infections to amphotericin B or caspofungin.

Amphotericin B significantly reduced tissue CFU in spleens of KO and WT mice and kidneys of WT mice inoculated with C. glabrata 12175 (Fig. 2). The amphotericin B effect was more obvious in a repeat experiment (with C. glabrata 18434) that also employed caspofungin at the same dose as amphotericin B (5 mg/kg for 6 days) to groups of seven KO mice. Both drugs caused a highly significant reduction in spleen and kidney CFU counts (P ≤ 0.002) (Fig. 3). Cultures of organ homogenates from caspofungin-treated mice were grown in SAB-chloramphenicol agar with 0.25, 0.5, or 1 μg of caspofungin/ml. Geometric mean colony counts in caspofungin-treated mice were 7.42 × 10⁴ in the spleen and 5.20 × 10² in the kidney on SAB but 0 on all plates containing 1 or 0.5 μg of caspofungin/ml and 5 (a 98.7% reduction) at 0.25 μg of caspofungin/ml in undiluted tissue homogenate. Caspofungin resistance did not account for the tissue CFU 2 days after the last dose. Caspofungin at 1 mg/kg given for 6 days also protected all 10 mice from death over a 15-day observation period, whereas all controls were dead by the fifth day (Fig. 4).

FIG. 2.

Amphotericin B (AMB) treatment. Groups of five C57BL/6 (WT [solid symbols]) mice and five p47^phox−/− (KO [shaded symbols]) mice were injected intravenously with C. glabrata 12175 with inocula of 1.7 × 10⁷ for WT mice and 2.6 × 10⁵ for KO mice. Mice were given intraperitoneal AMB at 5 mg/kg or saline daily for 6 days and sacrificed 2 days later. Significant differences in organ colony counts between treated and control mice (P < 0.02) are indicated by an asterisk. The geometric means and the standard error are shown.

FIG. 3.

Treatment with caspofungin, amphotericin B (AMB), or saline. Groups of seven p47^phox−/− KO mice were injected intravenously with 5 × 10⁵ viable C. glabrata 18434 and intraperitoneally given caspofungin at 5 mg/kg, amphotericin B at 5 mg/kg, or saline daily for 6 days. The spleen (○) and kidney (□) tissue burdens were assessed 2 days later. Significant differences in organ colony counts between caspofungin- or amphotericin B-treated and control mice are indicated by an asterisk (P < 0.002). The geometric means and the standard errors are shown.

FIG. 4.

Survival of groups of 10 p47^phox−/− KO mice after intravenous inoculation with 5 × 10⁷ viable C. glabrata 18434 and treatment with caspofungin at 1 mg/kg () or saline () daily for 6 days.

DISCUSSION

The strain of KO mice used in this experiment provided the first animal model of progressive, lethal C. glabrata infection. Tissue colony counts were fairly consistent between experiments and reflected the appearance of illness in the mice. Although kidney colony counts tended to be more variable than spleen counts, the 1,000-fold differences in CFU reported among untreated neutropenic mice were not seen in most experiments with KO mice (3). The question remains whether presence of a heterozygous interruption of the IFN-γ gene in some of our KO mice influenced the results. Neutralization of endogenous IFN-γ by monoclonal antibody had no effect on replication of C. glabrata in the organs of immunocompetent Crl:CF-1 mice (7). That result plus the failure to discern two populations in the organ colony counts indicated to us that the admixture of mice with heterozygous interruption of the IFN-γ gene probably did not play a discernible role in our results. Additional confirmation was obtained from single experiments which found that lethal C. glabrata infection occurred in p47^phox−/− and p91^phox−/− mice without genetic contamination.

Fluconazole was not effective in lowering the burden of C. glabrata infection in the kidneys or livers of KO or WT mice with isolates having a fluconazole MIC of from 2 to 256 μg/ml. Extending the period of fluconazole treatment from 6 to 13 days also did not improve efficacy. The dose used here, i.e., 80 mg/kg by the intraperitoneal route daily, was effective against infection with a C. albicans isolate having the same MIC (∼2 μg/ml) as one of the C. glabrata strains used. The same or lower fluconazole doses have proven effective in murine infections due to Cryptococcus neoformans (10), Coccidioides immitis (12), and C. glabrata (13). It is possible that gavage doses of 100 mg/kg given twice daily would have provided the same occasional effect in the KO mouse model as in neutropenic mouse kidneys (3). The solubility of fluconazole limits mouse experiments to intraperitoneal doses of 80 mg/kg and individual gavage doses of 100 mg/kg. Fluconazole has a plasma half-life of 3.7 to 5.7 h in mice (10), and there has been no suggestion of either a dose response to fluconazole or differences in response when the infecting C. glabrata strain had an MIC of 2 or >64 μg/ml (3, 13). We could not document secondary fluconazole resistance in an isolate of strain 12175 obtained from a fluconazole-treated mouse (fluconazole MIC, 32 μg/ml [data not shown]).

Amphotericin B, considered the drug of choice in clinical C. glabrata infections, was effective in reducing the kidney and spleen colony counts of our KO mice, as has been reported in neutropenic mice (3).

Caspofungin (MK-0991, L-743,872) has had a significant chemotherapeutic effect at 0.375 to 0.5 mg/kg per day in normal mice and in neutropenic mice infected with C. glabrata (1, 13). We confirmed the efficacy of this compound in reducing kidney colony counts but also showed that the drug reduced spleen colony counts and prolonged survival. Mice appeared healthy 9 days after the last caspofungin dose, 10 days after all of the controls had died. These results suggest that caspofungin may prove to be effective in humans with deep C. glabrata infections.