Rabbit Model of Candida albicans Biofilm Infection: Liposomal Amphotericin B Antifungal Lock Therapy

Catheter-related infections due to Candida albicans biofilmsare a leading cause of fungal nosocomial bloodstream infection.In this paper, we describe the development of a model of catheter-associatedinfection with C. albicans biofilms and show that antifungallock therapy with liposomal amphotericin B is an effective treatmentstrategy for these infections. Silicone catheters surgicallyplaced in New Zealand White rabbits were infected with C. albicans,and the rabbits were randomized into three groups: (i) untreatedcontrols, (ii) liposomal amphotericin B lock, and (iii) fluconazolelock. Upon completion of therapy, blood cultures were obtainedand the catheters were removed for quantitative culture andscanning electron microscopic analyses. Quantitative culturesrevealed that catheters treated with liposomal amphotericinB yielded 0 CFU, which was significant compared to the untreatedcontrols (P < 0.001) and the fluconazole-treated group (P= 0.0079). Although fluconazole treatment tended to have lowerCFU compared to untreated controls, there was no differencein mean colony counts between these two groups (1.128 ±0.764 and 1.841 ± 1.141 log₁₀ CFU/catheter segment, respectively;P = 0.297). Scanning electron microscopy revealed abundant biofilmin the control and fluconazole groups, while the liposomal amphotericinB group was virtually cleared. These findings suggest a possibletreatment strategy for the successful salvage of catheters infectedwith C. albicans biofilms and describe an animal model thatmay play an important role in the further study of C. albicansbiofilm pathogenesis and evaluation of potential antibiofilmagents.

INTRODUCTION

Top

Introduction

Nosocomial bloodstream infections are a significant cause ofmorbidity and mortality among hospitalized patients, with Candidaspecies representing the fourth most common cause of such infections(9). Central venous catheters (CVCs) are responsible for manyof these infections, and up to 40% of patients with Candidaisolated from their CVCs have underlying fungemia (1). Despitethe introduction of new antifungal therapies, the attributablemortality of patients with candidemia remains unacceptably high,ranging from 39 to 49% (10, 26).

Adherence of microorganisms and biofilm formation on cathetersurfaces is a common mechanism for catheter-associated bloodstreaminfection (20). These biofilms are characterized by the productionof a thick extracellular matrix and an altered resistance phenotype(8). Electron microscopic evaluation of catheters removed frompatients has demonstrated widespread biofilm colonization, confirmingthe importance of biofilms in catheter-associated infections(20). CVCs infected with Candida biofilms are especially problematicsince Candida species adhere avidly to catheter material andantifungal therapy alone is insufficient for cure (6). Thisexplains the rationale behind the recent national guidelinesrecommending the removal of catheters infected with Candidain order to eradicate a potential nidus of bloodstream infection(17, 19). However, catheter removal is not always feasible forpatients with coagulopathy or limited vascular access and isassociated with increased healthcare expenses as well as complicationsrelated to catheter replacement (18).

Data from several laboratories have demonstrated the near-totalresistance of in vitro Candida albicans biofilms to common antifungalagents (6, 7, 13, 15). More recently, our group and others haveobserved that in vitro C. albicans biofilms show susceptibilityto the lipid formulations of amphotericin B and the echinocandins(3, 16). Although these data suggest a possible strategy forthe salvage of CVCs infected with C. albicans biofilms, theclinical significance of these observations remains unknown.There has been no in vivo validation of these findings because(i) no animal model of catheter-related C. albicans biofilminfection is available and (ii) no prospective, randomized clinicaltrials have been performed. In this study, we describe the developmentof the first clinically relevant animal model of C. albicansbiofilm-associated catheter infection and evaluate the effectivenessof liposomal amphotericin B antifungal lock therapy.

MATERIALS AND METHODS

Top

Materials and Methods

Model development. (i) Catheter placement. All animal experimentation was performed in accordance withthe guidelines for animal health and welfare required by theInstitutional Animal Care and Use Committee at Case WesternReserve University School of Medicine, Cleveland, Ohio. Surgicalplacement of the CVCs was modified from previously describedmethods (11, 25). Female New Zealand White rabbits weighing2.5 to 3.0 kg (Covance Inc., Kalamazoo, Mich.) were used forall procedures. Silastic tubing, 0.04-in. internal diameterx 0.085-in. external diameter (Dow Corning, Midland, Mich.),was precut to 30 cm, and a polyethylene cuff (PE 240; BectonDickinson, Sparks, Md.) was slipped over the catheter 4 cm fromone end and superglued in position. The purpose of the cuffwas twofold: (i) it marked how far to insert the catheter, thuspreventing damage to the tricuspid valve and possible endocarditis,and (ii) it protected the catheter from collapsing when thedistal suture was tied. Polyurethane catheters (5 French) werepurchased commercially (Instech Solomon, Plymouth Meeting, Pa.).

The tubing was sterilized by ethylene oxide. Rabbits were givenintramuscular anesthesia with ketamine (70 mg/kg of body weight)and xylazine (7 mg/kg of body weight). The right cervical, shoulder,and scapular regions were shaved with electric clippers, andthe area was washed with betadine soap followed by isopropylalcohol and betadine solution. A 4-cm incision was made in theright anterolateral cervical region, exposing the external jugularvein. A segment of the external jugular vein, just distal tothe bifurcation of the internal and external maxillary veins,was freed from subcutaneous fat, and an 18-in. x 1/4-in. Penrosedrain (Medline Industries, Inc., Mundelein, Ill.) was placedunder the vein (Fig. 1A). Two segments of 3-0 vicryl suture(Ethicon Inc., Somerville, N.J.) were placed proximally anddistally to the Penrose drain. The catheter was flushed withsterile saline and clamped with a hemostat. Using a no. 11 scalpelblade, an incision in the isolated segment of vein was made,and bleeding was controlled with upward traction of the Penrosedrain. The catheter was inserted into the vein caudally 4 to5 cm, up to the cuff (Fig. 1B), placing the catheter tip inthe right anterior vena cava as demonstrated in the venogramin Fig. 1G. The proximal and distal ligatures were tied, andblood was withdrawn to test catheter patency (Fig. 1C). Thecatheter was then flushed with heparinized saline. Passing ahemostat cephalad through a 1.0-cm incision in the intrascapularregion to the external jugular vein incision site created asubcutaneous tunnel. The hemostat was used to pull the catheterthrough the subcutaneous tunnel, the excess catheter was cut,and an 18-gauge Luer stub adapter (Becton Dickinson) and a sterileheparin lock device (Medex, Hilliard, Ohio) were placed on theproximal end. Catheter patency was tested again by withdrawingblood and flushing the catheter with heparinized saline. Theneck incision was closed by using 3-0 vicryl sutures. The Luerhub was buried in the subcutaneous tract, and a purse stringsuture was used to attach the heparin lock device flush to theskin (Fig. 1D to F). For the first three postoperative days,the animals received buprenorphine for postsurgical pain controland penicillin G and gentamicin to prevent surgical site infection.

View larger version (43K):
[in this window]
[in a new window]
FIG. 1. Surgical placement of the intravenous catheter. (A to C) Catheter insertion into the external jugular vein; (D to F) attachment of the heparin lock device to skin; (G) postoperative venogram of catheter placement.

(ii) Inoculation of catheter. The C. albicans isolate (M61) used in this study was obtainedfrom a CVC tip of a patient with catheter-associated candidiasisas previously described (16). The mean inhibitory concentrationsof liposomal amphotericin B and fluconazole against this strainin the planktonic and biofilm form were 0.5 and 0.25 µg/ml(planktonic) and 1 and >256 µg/ml (biofilm), respectively.C. albicans M61 cells were grown, harvested, washed, and countedas previously described (6). A standard 300-µl inoculumconsisting of 10⁷ CFU of C. albicans, 100 U of heparin (AbbottLaboratories, North Chicago, Ill.), and sterile normal salinewas prepared. After catheter placement, the inoculum was "locked"in the internal lumen of the catheter and allowed to dwell for24 h. The inoculum was removed, and the catheter was flusheddaily with 300 µl of heparinized saline (100 U).

(iii) Catheter removal. Seven days postinfection, the rabbits were anesthetized as describedabove, and 5 ml of blood was obtained for blood culture (BDBactec, Becton Dickinson) through the catheter and via a cardiacpuncture. The animals were euthanized with an intracardiac injectionof pentobarbital. By using sterile technique, the catheter wasremoved and divided into proximal (subcutaneous tunnel) anddistal (intravenous) 4-cm segments. Each segment was dividedin half (2 cm) for (i) quantitative culture and (ii) examinationby scanning electron microscopy (SEM).

(iv) Quantitative catheter culture. Using sterile technique, catheter segments were cut in halflengthwise and placed in 10 ml of sterile saline. Specimenswere sonicated at 40,000 Hz (Bransonic 1510; Branson UltrasonicsCorp., Danbury, Conn.) for 12 min at 4-min intervals and thenvortexed for 15 s. Serial dilutions were performed prior toplating 1-ml aliquots onto Sabouraud dextrose agar (Difco Laboratories)supplemented with chloramphenicol and gentamicin. The plateswere incubated for 48 h at 37°C prior to counting CFU. Basedon prior clinical studies, catheters growing 10² CFU or moreof C. albicans were considered infected (17).

(v) SEM of catheter segments. Catheter segments were fixed in 2% glutaraldehyde and preparedfor SEM observation according to previously published protocol(7). The fixed and dehydrated catheter segments were sputtercoated with gold-paladium (60/40) and viewed under a PhillipsXL30 scanning electron microscope.

Antifungal lock experiments. (i) Antifungals. Liposomal amphotericin B was obtained from Fujisawa Healthcare,Inc., and Gilead Sciences (San Dimas, Calif.). Fluconazole wasprovided by Pfizer Pharmaceuticals Group (New York, N.Y.). Liposomalamphotericin B was reconstituted in sterile water and dilutedwith 5% sterile dextrose, while the fluconazole was reconstitutedand diluted with sterile water.

(ii) Catheter treatment. Each catheter was flushed with 300 µl of heparinized saline(100 U) on a daily basis for the first three days of the study.On day 3 postinfection, 5 ml of blood was drawn through eachcatheter for culture. To determine the efficacy of antifungallock therapy, the rabbits were randomized into three groups,each consisting of seven animals. Group I received daily catheterflushes with 300 µl of heparinized (100 U) saline. Catheterlumens of animals in group II were locked with a 300-µlsolution containing (i) 3 mg of liposomal amphotericin B, (ii)100 U of heparin, and (iii) 5% dextrose solution. Rabbits ingroup III had their catheter lumens locked with a 300-µlsolution containing (i) 3 mg of fluconazole, (ii) 100 U of heparin,and (iii) sterile normal saline. The concentration of liposomalamphotericin B used in this study was guided by dosages in priorcase reports (4, 14, 24) The fluconazole concentration was basedon one in vitro study (2) and was limited by solubility. Thefluconazole dose used in this study required a 10-min incubationin a hot water bath to completely dissolve. The antifungal solutionswere locked in the lumen of each catheter for 8 h per day for7 days. Upon completion of each daily treatment, the antifungallock solution was removed and the catheter was flushed with300 µl of heparinized saline. If the antifungal solutioncould not be withdrawn, it was flushed into the systemic circulation.After 7 days of antifungal lock therapy, the animals were anesthetized,blood was obtained through the catheter and via a cardiac puncture,and the animals were sacrificed with an intracardiac injectionof pentobarbital. The catheters were removed under sterile conditions,and quantitative catheter cultures were performed as describedabove. Catheters from three animals in each group were examinedby SEM.

(iii) Statistical evaluation. The mean CFU from quantitative catheter cultures were comparedby the Student's t test using StatView version 5.0.1 software(SAS Institute, Cary, N.C.). A P value of < 0.05 was consideredsignificant.

RESULTS

Top

Results

Model development. During the initial development of the animal model, experimentswere conducted with both silicone and polyurethane cathetersin order to determine if there was a difference in C. albicansbiofilm formation based on catheter substrate. Seven days postinfection,quantitative catheter cultures consistently yielded greaterthan 2 log₁₀ CFU/catheter segment, which is considered the thresholdfor catheter-related infections (17). There was no significantdifference between the numbers of CFU found on silicone catheters(3.537 ± 0.752 log₁₀ CFU/catheter segment) and polyurethanecatheters (3.286 ± 1.123 log₁₀ CFU/catheter segment;P = 0.45). Quantitative cultures of the proximal versus distalcatheter segments for both silicone (3.555 ± 0.323 versus3.519 ± 1.068 log₁₀ CFU/catheter segment; P = 0.9) andpolyurethane catheters (3.459 ± 0.639 versus 3.114 ±1.567 log₁₀ CFU/catheter segment; P = 0.7) revealed no significantdifference. SEM analyses of intraluminal catheter segments consistentlyrevealed the same biofilm architecture regardless of the cathetermaterial used (data not shown). Therefore, we selected siliconeelastomer catheters for use in this in vivo model, allowingus to maintain consistency with our in vitro model in whichC. albicans biofilms are grown on silicone elastomer disks.

Previous in vitro experiments by our group have shown that matureC. albicans biofilms form within 48 to 72 h (6). In order todetermine if in vivo biofilm formation might occur sooner than7 days, as initially described, we used quantitative catheterculture and SEM analyses to examine biofilm formation in threeanimals 72 h postinfection. Mean CFU counts revealed no significantdifference between biofilms grown on silicone catheters at 3versus 7 days for either the proximal (2.257 ± 0.125versus 3.555 ± 0.323 log₁₀ CFU/catheter segment; P =0.705) or distal segments (2.617 ± 0.221 versus 3.519± 1.068 log₁₀ CFU/catheter segment; P = 0.884). SEM analysesof infected catheter segments at 3 and 7 days revealed similararchitecture consisting of patchy regions of biofilm adherentto the catheter surface, consisting of yeast blastospores encasedin thick extracellular material (Fig. 2). Based on these data,the optimal time course of 3 days was selected for mature biofilmformation in subsequent experiments.

View larger version (143K):
[in this window]
[in a new window]
FIG. 2. Mature in vivo biofilm formation during model development. Scanning electron micrographs of C. albicans biofilms adherent to the intraluminal surface of catheters showing no difference in biofilm architecture at 7 days postinfection (magnification, x6,500) (A) and 3 days postinfection (magnification, x2,500) (B) are shown.

Antifungal lock experiments. To confirm the presence of catheter-associated infection priorto the introduction of antifungal lock therapy, blood cultureswere drawn through the catheters 3 days postinfection. C. albicanswas present in 67% of all cultures and 100% of cultures drawnfrom unclotted catheters (Table 1). Quantitative culture ofcatheters after 7 days of liposomal amphotericin B lock therapyyielded 0 CFU (Table 1). This finding was significant comparedto the untreated controls (P < 0.001) and the fluconazole-treatedgroup (P = 0.0079). Although fluconazole sterilized both proximaland distal catheter segments in two animals after 7 days oftreatment, overall there was no significant difference in meanCFU between the animals treated with fluconazole (1.128 ±0.764 log₁₀ CFU/catheter segment) and the untreated controls(1.841 ± 1.141 log₁₀ CFU/catheter segment; P = 0.2973).

View this table:
[in this window]
[in a new window]
TABLE 1. Culture results of antifungal lock studies^a

Catheter segments removed from the animals upon completion oftreatment were sectioned lengthwise and prepared for SEM examination.In general, the biofilm seen in the untreated controls was patchyin distribution but abundant, consisting of yeast cells encasedin a dense extracellular matrix (Fig. 3). Catheters treatedwith liposomal amphotericin B were completely clear except for1 to 2 small areas of biofilm per catheter. In these areas,the biofilms seemed to consist of yeast cells that were lackingthe thick matrix material evident in the untreated controls.Overall, the fluconazole group had less biofilm than the untreatedcontrols, but patches of yeast cells encased in thick extracellularmatrix were readily visible, and the biofilm structure appearedsimilar to that seen in the untreated controls (Fig. 3).

View larger version (85K):
[in this window]
[in a new window]
FIG. 3. Effectiveness of antifungal lock therapy. Scanning electron micrographs of intraluminal catheter surfaces following 7 days of therapy with heparinized saline (magnification, x5,000) (A), liposomal amphotericin B (magnification, x121) (B), and fluconazole (magnification, x3,500) (C) are shown.

DISCUSSION

Top

Discussion

In this paper, we describe the first clinically relevant animalmodel of catheter-related C. albicans biofilm infection. Usingthis model, we show that liposomal amphotericin B antifungallock therapy can sterilize infected catheters. Experiments aimedat optimizing this animal model established by both quantitativeculture and SEM analyses that mature biofilms formed equallyon polyurethane and silicone CVCs. Previously, Hawser and Douglasreported that C. albicans biofilm formation in vitro was significantlydecreased when grown on polyurethane compared to silicone elastomerdisks (12). Our in vivo studies, however, indicate that thereis no difference in C. albicans biofilm formation between cathetersmade of polyurethane and those made of silicone. One explanationfor this discrepancy between in vitro and in vivo data may bethe formation of conditioning films, or fibrin sheaths, on theintraluminal catheter surface after exposure to blood in vivo.Conditioning films, rich in fibronectin and fibrin, have beenshown to coat the surfaces of catheters exposed to blood withinhours after catheter insertion (20). Once formed, the conditioningfilm masks the properties of the underlying catheter materialand enhances the adherence of C. albicans (8, 21). This findingmay explain why there appeared to be no difference between invivo biofilm formation on silicone and polyurethane catheters.

During the development of this model, we were surprised to findthat only 26% of the untreated animals had positive peripheralblood cultures despite documented catheter infection by bothquantitative catheter culture and SEM. However, three of fiveanimals with positive peripheral blood cultures had clottedcatheters that had been flushed multiple times immediately beforethe cardiac puncture was performed. The injection of salinethrough the catheter lumen may have caused a detachment of yeastcells from the catheter-associated biofilms leading to systemicinfection. The continuous infusion of fluids through the infectedcatheters, as routinely occurs in the clinical setting, maylead to a greater percentage of positive peripheral blood cultures.However, in this model, catheters were flushed daily but wereotherwise closed to the systemic circulation.

To our knowledge, this study provides the first in vivo SEMimages of C. albicans biofilms adherent to CVC surfaces. Hawserand Douglas, as well as our group, have previously publishedin vitro SEM images of C. albicans biofilms (7, 12). The densenetwork of yeast cells, germ tubes, and hyphal forms that comprisedthe in vitro biofilms are difficult to distinguish in vivo.Instead, the in vivo biofilms were composed of a dense basallayer of yeast cells adherent to the catheter material encasedin a thick extracellular matrix. Similar findings have beenreported by our group using confocal scanning laser microscopyto examine C. albicans biofilms formed on silicone elastomerdisks in vitro (6). It is possible that the dense network ofyeast cells, germ tubes, and hyphal forms may exist in vivo,but they are not visible due to the limitation of SEM to visualizeonly the surface topography of the biofilm.

To document the clinical utility of this animal model, we comparedthe ability of fluconazole and liposomal amphotericin B antifungallock solutions to sterilize CVCs infected with C. albicans biofilms.We selected the antibiotic lock technique because it was developedas a means to overcome the high-grade antimicrobial resistanceobserved by bacterial biofilms infecting CVCs. By instillinghigh concentrations of antibiotics into the lumen of infectedcatheters, the antibiotic lock technique is able to (i) overcomethe increased resistance of microorganisms in the biofilm state,(ii) avoid systemic drug toxicity and eliminate the need tocheck drug levels in serum, and (iii) prevent distant spreadof microorganisms by closing the catheter during treatment (5).Several open trials of antibiotic lock therapy, with or withoutconcomitant systemic antibiotics, have reported successful cathetersalvage without relapse in 83% of those infected with bacteria;however, only 30% of catheters with fungal infections have beensuccessfully retained by using the antibiotic lock technique(17).

Two case reports have documented the successful salvage of cathetersinfected with C. albicans by using amphotericin B deoxycholateantifungal lock therapy. Johnson et al. (14) salvaged two cathetersby instilling 3 ml of a 2-mg/ml amphotericin B deoxycholatesolution and allowing it to dwell for 12 h per day for 10 to14 days. A report of the salvage of two more catheters infectedwith C. albicans after treatment with 3 ml of a 2.5-mg/ml amphotericinB deoxycholate solution locked into the catheter lumen for 14days, with 7 days of concomitant systemic antifungal therapy,was recently published by Viale et al. (24). These results suggestthat antifungal lock therapy may be an effective strategy forthe retention of catheters infected with C. albicans.

The chances of successful catheter salvage may be improved ifantifungal agents with enhanced antibiofilm activity are used.Recent in vitro studies performed by our group have identifiedlipid formulations of amphotericin B and echinocandins as havingimproved antibiofilm activity compared to other antifungal agents(16). In this study, we validated these findings with our newlydeveloped animal model. Catheters treated with a liposomal amphotericinB antifungal lock solution were completely sterilized, whilecatheters treated with a fluconazole antifungal lock solutionshowed no significant difference in C. albicans CFU comparedto untreated controls.

Suboptimal fluconazole concentrations in the lock solution mayexplain its inability to sterilize the infected catheters. Accordingto previously published in vitro data, the fluconazole and liposomalamphotericin B MICs for mature biofilms grown with C. albicansstrain M61 are >256 and 0.25 µg/ml, respectively (16).Therefore, the concentration of liposomal amphotericin B antifungallock solution used in this study was 12,000 times greater thanthe MIC for biofilm, while the concentration of fluconazoleantifungal lock solution was <11 times the MIC for biofilm.Increased concentrations of fluconazole (300 to 3,000 mg in300 µl) may improve its efficacy as an antifungal locksolution. However, dissolving the higher concentrations of fluconazoleinto the small volumes needed for the antifungal lock techniqueproved to be difficult in this study.

Blood cultures drawn through the infected catheters in the antifungallock experiments revealed some interesting observations. Thereseemed to be an exceptionally high rate of catheter thrombosisdespite allowing heparin to dwell in the catheter lumens atall times. Thirty-three percent of catheters 3 days postinoculationwere clotted. Of the cultures drawn at the completion of thetreatment period, 57% of catheters were clotted in the untreatedcontrol, while only 14% of catheters were clotted in both theliposomal amphotericin B- and fluconazole-treated groups. Therelationship between catheter thrombogenicity and catheter-relatedinfection has been well documented in several studies (21-23,27). However, it remains unclear whether catheter thrombosispromotes catheter-related infection or if the infection leadsto thrombosis. In this study, more episodes of catheter thrombosisoccurred in the untreated control group than the treated groups,suggesting that infection promotes catheter thrombosis.

Despite the complete sterilization of the catheters treatedwith liposomal amphotericin B by quantitative culture, 2 outof 7 blood cultures drawn from the catheters were positive foryeast. Although SEM of these catheters revealed almost completelyclean intraluminal surfaces, each catheter examined had 1 to2 small patches of C. albicans surrounded by minimal amountsof damaged matrix. There is a strong possibility that theseadherent cells were dead, but their viability could not be determinedby our methods. These findings suggest that perhaps extendingthe treatment period to 10 to 14 days, as in the successfulcase reports with amphotericin B deoxycholate, may result inthe complete eradication of fungal elements from catheters (14,24).

Even though fluconazole lock therapy was unable to sterilizethe intraluminal surface of catheters as documented by quantitativecatheter culture and SEM, only 1 out of 7 catheter-drawn bloodcultures grew yeast at the completion of the treatment period.The fluconazole lock solution may only kill free-floating planktonicC. albicans released from mature biofilm. The remaining yeastcells are encased in a thick extracellular matrix, as documentedby SEM, which may not be released with routine blood draws throughthe catheter. Sonication of the treated catheters causes a disruptionof the matrix with the release of yeast that is isolated byquantitative culture. Therefore, fluconazole lock therapy mayrepresent an effective suppressive therapy for preventing continuousseeding of the bloodstream despite not sterilizing the cathetersurface. A possible drawback to such a treatment strategy isthe return of infection once the suppressive agent is stopped.

In summary, we describe the development of the first animalmodel of catheter-related C. albicans biofilm infection andshow that this model has utility in preclinical evaluation ofantifungal agents and in the study of biofilm pathogenesis.The demonstration of liposomal amphotericin B lock therapy asan effective strategy for treating C. albicans catheter-associatedinfections may have significant clinical implications, and alarge prospective, randomized clinical trial should be performedin the near future.

ACKNOWLEDGMENTS

We thank A. Heuer and A. McIlwain at the Center for SurfaceAnalysis of Materials for assistance with the SEM, A. Morriseyand the microbiology staff at University Hospitals of Clevelandfor technical support with the blood cultures, T. J. Walsh foruseful discussions during the development of the animal model,R. Rodriquez for help developing the surgical technique, D.Kuhn for critical review of the manuscript, and K. Smith forsecretarial assistance.

Funding for this work was provided by the National Institutesof Health (Infectious Diseases/Geographic Medicine Traininggrant no. AI07024) to M.K.S. and (RO1 DE/AI 13932-01A1) to M.A.G.,the American Heart Association (grant no. 0335313N) to P.K.M.,the Dermatology Foundation to J.C., Fujisawa Healthcare to M.A.G.,and Bristol-Myers Squibb (Award for Distinguished Research inInfectious Diseases) to M.A.G.

FOOTNOTES

* Corresponding author. Mailing address: Center for Medical Mycology, Department of Dermatology, University Hospitals of Cleveland and Case Western Reserve University, 11100 Euclid Ave., Cleveland, OH 44106-5028. Phone: (216) 844-8580. Fax: (216) 844-1076. E-mail: mag3{at}cwru.edu.

REFERENCES

Top