High Rate of Exophiala dermatitidis Recovery in the Airways of Patients with Cystic Fibrosis Is Associated with Pancreatic Insufficiency

Abstract

The black-pigmented fungus Exophiala dermatitidis is considered to be a harmless colonizer of the airways of cystic fibrosis (CF) patients. The aim of this study was to establish the recovery rate of E. dermatitidis in respiratory specimens from CF patients, transplant recipients, and subjects with other respiratory disorders in Sweden. Second, we wished to determine if particular clinical traits were associated with E. dermatitidis colonization of the airways and the antifungal susceptibility profiles of Exophiala strains. Sputum and bronchoalveolar lavage samples (n = 492) derived from 275 patients were investigated. E. dermatitidis was isolated in respiratory specimens from 19% (18/97) of the CF patients but in none of the other patient categories. All isolates were recovered after 6 to 25 days of incubation on erythritol-chloramphenicol agar (ECA) medium. Morphological and genetic analyses confirmed species identity. Pancreatic insufficiency was positively associated with the presence of E. dermatitidis in sputum samples (P = 0.0198). Antifungal susceptibility tests demonstrated that voriconazole and posaconazole had the lowest MICs against E. dermatitidis. In conclusion, E. dermatitidis is a frequent colonizer of the respiratory tract in CF patients in Sweden and appears to be associated with more advanced disease. Whether E. dermatitidis is pathogenic remains to be elucidated.

INTRODUCTION

Cystic fibrosis (CF) is the most common genetically inherited disease among Caucasians, estimated to affect over 70,000 individuals worldwide (6, 9). The disease is a result of mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, which is located on chromosome 7 and encodes a chloride channel expressed by the plasma membrane of epithelial cells in the respiratory and gastrointestinal tracts (25). Defective expression of the CFTR leads to the production of thick and sticky bronchial mucus that facilitates the entrapment of airborne bacterial and fungal cells, providing an environment suitable for these microorganisms.

Staphylococcus aureus, Pseudomonas aeruginosa, Burkholderia cepacia, and Stenotrophomonas maltophilia are typical opportunistic bacteria recovered from the sputa of CF patients (19). Candida albicans and Aspergillus fumigatus are the fungi most frequently isolated from respiratory secretions in CF. In addition, fungi such as Exophiala (Wangiella), non-albicans Candida spp., Penicillium spp., and Scedosporium spp. have been reported to colonize the respiratory tracts of these patients (1, 11).

Exophiala dermatitidis is a dark-pigmented dimorphic fungus with a worldwide distribution. It is known to cause local infections of the skin, which may spread and cause disseminated disease and fungemia; the latter is seen particularly in immunocompromised patients (1, 4, 18). Only 2 decades ago, E. dermatitidis was revealed to be able to colonize patients with cystic fibrosis (12). The fungus is traditionally viewed as having low virulence and hence is considered to be a harmless inhabitant of the CF airways (20). Reports on the pulmonary prevalence of E. dermatitidis in CF range from 5% in Germany to 15% in Belgium (20). The differences may be due to genetic heterogeneity or differences in the life styles of the host populations, as well as a lack of standardization in the mycological examination of respiratory specimens (20). No reports exist of the recovery of E. dermatitidis among CF patients in Scandinavia.

The aim of the present study was to assess the isolation rate of E. dermatitidis in respiratory specimens from patients with CF; transplant recipients attending the transplant program at Sahlgrenska University Hospital in Gothenburg, Sweden; and patients with other respiratory disorders residing in the western part of Sweden. Second, we wished to determine if particular clinical traits or cocolonizing microorganisms were associated with E. dermatitidis colonization of the airways. Finally, we assessed the susceptibilities of E. dermatitidis isolates to antifungal agents.

MATERIALS AND METHODS

Respiratory specimens.

During the 6-month period from July to December 2008, 491 sputum and bronchoalveolar lavage (BAL) fluid specimens from CF patients, transplant recipients, and patients with other respiratory disorders were examined at the Mycology Unit of the Department of Clinical Bacteriology, Sahlgrenska University Hospital, for the presence of E. dermatitidis (Table 1). The frequency of sampling varied between twice a month and once a year for CF patients. Respiratory specimens were processed prior to culture: sputum was liquefied by the addition of pancreatin (10 mg/ml) in a volume ratio of 1/1, vortexed, and incubated at room temperature (RT) for 5 min. BAL fluid was centrifuged at 1,000 × g for 10 min, and the resulting pellet was cultured. Processed sputum and BAL fluid specimens (20 μl) were plated onto horse blood agar, streptococcus agar, haemophilus agar, Sabouraud agar, and erythritol chloramphenicol agar (ECA) (13). The last two media are selective for the isolation of E. dermatitidis, while the others were used to determine bacterial cocolonization. The cultures were incubated at 37°C and examined for black fungal growth for up to 25 days (or 2 days for bacteria). Black-pigmented colonies were subcultured onto ECA plates for confirmation of purity and maintenance of the strains. Black fungal colonies were identified to the species level using morphological criteria and growth at 40°C and on mycobiotic agar plates supplemented with cycloheximide.

Table 1.

Patient data and respiratory tract specimens

Patient type	No. of BAL/sputum specimens	No. of patients			Median age (yr) (25th–75th percentile)
		Male	Female	Total
CF	0/214	52	45	97	22 (15–33)
Pancreatic insufficiencya		38	39	77
Diabetes mellitusb		11	6	17
Transplant recipientsc	91/110	56	52	108	59 (49–63)
Other respiratory disorders	0/75	46	24	70	60 (48–74)

^aEnzyme replacement required.

^bInsulin required.

^cSolid organ or bone marrow transplant recipients.

Clinical data.

The diagnosis of CF was based on clinical presentation, together with abnormal values on a sweat chloride test. Data regarding age, gender, weight, height, body mass index (BMI), CFTR genotype, and occurrence of pancreatic insufficiency and diabetes mellitus were collected. The most common disease-causing mutation among CF patients was ΔF508, which occurred in 81% (79/97), of which 43/97 were compound homozygotes for ΔF508. Pancreatic insufficiency was defined by the requirement for enzyme replacement and was present in all ΔF508 homozygotes (n = 43) and in 58% (21/36) of the ΔF508 heterozygotes.

Direct microscopic examination of sputum and BAL samples.

Drops of pretreated sputum and BAL fluid specimens were placed on microscope slides and allowed to air dry, stained with the fluorochrome dye blankophore, and examined for the presence of fungal structures using a Zeiss fluorescence microscope (Carl Zeiss Inc., NY).

Genotypic identification of black yeasts by DNA sequencing.

Genomic DNA from fungal strains was extracted using the Amplicon Respiratory Specimen Preparation kit (Roche, Mannheim, Germany), according to the manufacturer's recommendations, and stored at 4°C until it was used. For PCR and DNA sequence analysis, the conserved primers UNI1 (forward primer; 5′-GTC AAA CTT GGT CAT TTA-3′) and UNI2 (reverse primer; 5′-TTC TTT TCC TCC GCT TAT TGA-3′) were used to amplify the region encompassing the rRNA internal transcribed spacer regions 1 (ITS1) and 2 (ITS2), including the 5.8S rRNA (3, 26) moiety. Each PCR was performed in a total volume of 50 μl, as described previously in detail (23). The PCR amplicons were purified using the QIAquick PCR Purification Kit (Qiagen, Düsseldorf, Germany), following the manufacturer's instructions. The purified PCR products were then subjected to automated fluorescent sequencing using the same primers that were used for amplification and the ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction kit, version 3.1 (Applied Biosystems, Inc., CA).

The DNA sequence quality was checked manually, and sequences were subjected to similarity searches using the FASTA program of the European Molecular Biology Laboratory (EMBL) Nucleotide Sequence Database for Fungi (http://www.ebi.ac.uk/Tools/fasta33/nucleotide.html). The sequences were edited to remove adjacent 18S, 5.8S, and 28S rRNA gene sequences and aligned using the BioNumerics v. 4.61 Sequence Types module (Applied Maths, Sint-Martens-Latem, Belgium). ITS1 (199 nucleotide positions) and ITS2 (191 nucleotide positions) sequence dissimilarities were determined, and cluster analyses were done using the unweighted-pair group method using average linkages (UPGMA) algorithm. ITS1-ITS2 sequences from reference strains of related fungi were obtained from the EMBL Nucleotide Sequence Database and included in comparative analyses (5).

Antifungal susceptibility tests.

Antifungal susceptibilities were determined by Etest (AB Biodisk, Solna, Sweden). The concentrations of the antifungal agents on Etest strips ranged from 0.002 to 256 μg/ml for fluconazole and from 0.002 to 32 μg/ml for itraconazole, voriconazole, posaconazole, caspofungin, amphotericin B, and flucytosine. Fungal colonies were suspended in 0.85% saline solution, and the turbidity was adjusted to 0.5 on the McFarland scale (1 × 10⁶ to 5 × 10⁶ CFU/ml). The suspensions were further diluted 1:5 in 0.85% saline and flooded on agar plates consisting of RPMI 1640 supplemented with 2% glucose and buffered with morpholinopropanesulfonic acid. Excess fluid was removed, and the plates were allowed to dry at RT before the Etest strips were applied and incubated for 72 h at 37°C. The MICs of antifungal agents were read as the lowest concentration at which the border of the elliptical inhibition zone intercepted the scale of the strips. The MIC₅₀ and MIC₉₀ results are the concentrations of each antifungal agent necessary to inhibit 50% and 90% of the isolates, respectively.

Statistical analysis.

Microbiological data were analyzed by the multivariate technique of orthogonal projection to latent structures (OPLS) (21), using the SIMCA-P software (Umetrics AB, Umeå, Sweden). OPLS is a regression extension of principal-component analysis (PCA) in which variables termed “score vectors” that covary with a selected Y variable are identified, for example, presence of E. dermatitidis in the airways. Fisher's exact test was performed using GraphPad Prism version 4.00 (GraphPad Software, San Diego, CA). A P value of <0.05 was considered to be statistically significant in univariate analyses.

RESULTS

Isolation of black fungi from patients with respiratory disorders.

Black fungi were recovered only from patients with cystic fibrosis, not from the other studied patient categories. Black fungi were isolated from 23% of the sputum samples (50 isolates/214 specimens) derived from 18 out of 97 CF patients (19%). These 97 patients constituted 69% of the 140 patients registered at the Gothenburg Cystic Fibrosis Center during the study period. Many E. dermatitidis strains were recovered from one individual patient, indicating longstanding colonization as opposed to transient appearance. Fourteen out of 18 CF patients in the cohort had at least 2 positive cultures during the 6-month-long study period. The majority of black fungi were isolated from culture within 2 weeks (n = 46; 90%) but required a minimum of 6 days of incubation (Fig. 1). Only one black fungal specimen was recovered after 3 weeks (day 22) of incubation (Fig. 1). The yield of black fungi was highest when ECA medium was used. With this medium, 50 isolates were recovered, compared to 41 when Sabouraud agar was used (Fig. 1). The numbers of black fungal specimens and other pathogenic microorganisms isolated from the sputa of CF patients are listed in Table 2. S. aureus and P. aeruginosa were the bacterial species most commonly isolated from the sputa of CF patients, and black fungi were more frequently recovered from the sputa of CF patients than A. fumigatus (Table 2). No patients received antifungal prophylaxis. However, two E. dermatitidis culture-positive patients (one patient with 1 positive culture and one patient with 3 positive cultures) were given intermittent antifungal treatment because of concomitant infection with A. fumigatus.

Fig. 1.

Recovery of E. dermatitidis from sputa of CF patients cultivated on ECA and Sabouraud agar (SAB), respectively, following various incubation times.

Table 2.

Pathogenic microorganisms isolated from sputum samples from CF patients

Isolate	CF patients				P valuea
	E. dermatitidis positive (n = 18)		E. dermatitidis negative (n = 79)
	n	%	n	%
A. fumigatus	4	22	6	8	0.085
P. aeruginosa	4	22	38	48	0.064
S. aureus	5	27	35	44	0.289
Stenotrophomonas spp.	2	11	7	9	0.671
Burkholderia spp.	2	11	1	1	0.095
Haemophilus influenzae	1	6	8	10	1

^aFisher's exact test; significance level, P < 0.05.

Identification of black fungi as E. dermatitidis.

Sputum specimens that were culture positive for black fungi were reexamined by fluorescence microscopy for the presence of fungal structures. All of the remaining available sputum samples (n = 42) contained fungal elements (Fig. 2). Morphological characterization of the colonies of the fungal isolates revealed them to be brown to olive-black in color, moist, shiny, and yeast-like. Microscopic analysis of young cultures showed oval to round budding cells that sprouted dark septate hyphae (7). Further, the fungal cells tolerated cycloheximide and temperatures as high as 42°C, which are characteristic traits of E. dermatitidis.

Fig. 2.

Direct microscopy examination of CF patient sputum samples that were culture positive for E. dermatitidis. Sputum was stained with the fluorescent dye blankophore and examined by fluorescence microscopy (magnification, ×400). Chains of yeast-like structures (a) and/or fungal hyphae (b) are seen.

The strains were genotypically characterized to ensure their species identity. The 5.8S rRNA gene sequences of the 50 isolates were identical to each other and to that of the type strain of the species, E. dermatitidis strain CBS 207.35^T (EMBL accession number AF050269). Moreover, the ITS1 and ITS2 sequences of almost all strains were identical to each other and to those of the reference strain E. dermatitidis CBS 207.35^T (Fig. 3). A single remaining strain, denominated strain 374, exhibited a slight variation compared to the other strains. This strain had one nucleotide insertion in the ITS1 sequence and differed with respect to a single nucleotide in the ITS2 sequence compared to the other identical isolates. The ITS1 and ITS2 sequences of strain 374 were identical to the sequences of E. dermatitidis reference strain CBS 149.90 (EMBL accession number AF050268) (Fig. 3). Hence, genetic analyses confirmed that the isolated black fungi were indeed E. dermatitidis.

Fig. 3.

Dendrogram of sequence similarities between the rRNA operon combined ITS1 and ITS2 regions of strains of E. dermatitidis derived from an unrooted UPGMA cluster analysis. Strains isolated from samples from patients at Sahlgrenska University Hospital, Gothenburg, Sweden, are numbered 274 to 3305. Sequences from reference strains of E. dermatitidis, representing each known sequence type, are included in the analysis. Strain designations: BMU, Research Center for Medical Microbiology, Beijing University, Beijing, China; CBS, Centralbureau voor Schimmelcultures, Utrecht, Netherlands; EM, Medical Microbiology, Athens Medical School, Athens, Greece; IFM, Research Institute for Pathogenic Fungi, Chiba, Japan; UTHSC, University of Texas Health Science Center, San Antonio, TX. The scale bar indicates sequence dissimilarities.

Clinical and microbiological features associated with E. dermatitidis colonization.

The clinical features associated with E. dermatitidis colonization in CF patients are summarized in Table 3. Fisher's exact test showed that the recovery of E. dermatitidis was significantly correlated with pancreatic insufficiency alone (P = 0.0198). Multivariate analysis of microbial agents isolated from the CF sputum samples revealed that E. dermatitidis was positively associated with colonization by Aspergillus and negatively associated with P. aeruginosa (Fig. 4). Univariate analyses confirmed that E. dermatitidis was positively associated with colonization by Aspergillus and negatively associated with P. aeruginosa, although this did not quite achieve statistical significance (Table 2).

Table 3.

Clinical features associated with E. dermatitidis colonization in patients with cystic fibrosis

Clinical feature	Value for E. dermatitidis in CF patients (n = 97)				P valuea
	Positive (n = 18)		Negative (n = 79)
	Median (25th–75th percentile)	Incidence (%)	Median (25th–75th percentile)	Incidence (%)
BMI	20 (19–22)		21 (18–23)
Age (yr)	21 (13–29)		22 (15–34)
Pancreatic insufficiency		18/18 (100)		59/79 (75)	0.0198
Diabetes mellitus		5/18 (28)		12/79 (15)	0.299b
ΔF508 genotype		18/18 (100)		69/79 (87)	0.200b

^aFisher's exact test.

^bNot significant (P < 0.05).

Fig. 4.

Multivariate analysis of OPLS was used to relate score vectors (microbial agents isolated from the CF sputum samples, e.g., S. aureus, Haemophilus influenzae, P. aeruginosa, Stenotrophomonas spp., A. fumigatus, and Burkholderia spp.) to the chosen Y variable (E. dermatitidis isolated from CF sputum samples). Loading plots with jackknifed confidence intervals are indicated by boxes. Positive plots indicate that variables are positively associated with E. dermatitidis. Negative plots are variables that are associated negatively with E. dermatitidis.

Antifungal MIC profiles of E. dermatitidis isolates.

The isolated E. dermatitidis strains grew well on RPMI agar plates, resulting in clear and detectable Etest endpoints. The MIC ranges, including the MIC₅₀ and MIC₉₀ values obtained, are shown in Fig. 5. Voriconazole and posaconazole exhibited the lowest MICs, followed by itraconazole and amphotericin B. Caspofungin had a very broad MIC interval, suggesting it may be active against certain Exophiala strains. Finally, fluconazole and flucytosine appeared to lack antifungal activity against Exophiala.

Fig. 5.

MICs of fluconazole, itraconazole, voriconazole, posaconazole, caspofungin, amphotericin B, and flucytosine according to Etests performed on 51 E. dermatitidis strains isolated from sputa of CF patients. The concentration of each antifungal agent necessary to inhibit 50% and 90% of the isolates, respectively, are indicated as MIC₅₀ and MIC₉₀. GM, geometric mean (μg/ml).

DISCUSSION

In this study, we documented a high recovery rate (23%) of a black fungus, identified as E. dermatitidis, in sputum specimens from CF patients but from none of the other investigated patient groups. One or more positive samples were recovered from 18 CF patients, comprising 19% of the study population. To date, this is the highest reported detection rate of Exophiala in a CF population. Figures from other countries range from 5% in Germany to 15% in Belgium (11, 20). Part of the variation in detection rates may simply be a consequence of microbiological techniques for the isolation of these fastidious fungi. We confirmed previously published findings demonstrating that the selective culture medium ECA, which inhibits the growth of bacteria, in combination with prolonged incubation times of up to 3 weeks, significantly boosted the detection rate of E. dermatitidis (13).

The high recovery rate of E. dermatitidis in sputum cultures from patients with CF, after prolonged incubation on specific media, was an unexpected finding of unclear clinical significance. At the time of the study, no patient had received specific treatment for E. dermatitidis. However, two patients received intermittent courses of voriconazole and itraconazole, respectively, for concomitant chronic colonization with A. fumigatus, suspected to have contributed to exacerbation of infection and bronchial inflammation.

We found a positive correlation between pancreatic insufficiency and the presence of E. dermatitidis in sputum. One possible explanation is that most CF patients with pancreatic insufficiency have more advanced lung disease and frequently require treatment with intravenous antibiotics directed against bacteria, which might promote fungal colonization. Although not statistically significant, a positive association was seen between the presence of A. fumigatus and E. dermatitidis in sputum while a negative association was seen for P. aeruginosa and E. dermatitidis. Another possibility is that growth of Exophiala may have been influenced by other pathogenic microbes harbored by the patient. E. dermatitidis is a slow-growing microorganism, and routine isolation techniques are often insufficient to detect the fungus in clinical specimens. This is particularly true when fast-growing bacteria, such as P. aeruginosa, are present in the sample. Additionally, P. aeruginosa produces pyocyanin and I-hydroxyphenazine, both of which have been reported to inhibit fungal growth (2, 16).

Retrospective analysis of E. dermatitidis culture-positive sputum specimens by fluorescence microscopy revealed fungal structures in almost all examined specimens. Unfortunately, direct microscopic examination cannot discriminate between fungal structures that belong to E. dermatitidis and those of other filamentous fungi. Nevertheless, demonstration of fungal structures by microscopy might still be useful in prompting the ordering of a fungus culture if one has not been ordered.

The antifungal susceptibility testing demonstrated that voriconazole and posaconazole were the most active drugs against the E. dermatitidis strains associated with CF patients, which is in agreement with previous studies by Johnson et al. using the broth microdilution method (14). Etests revealed that amphotericin B and itraconazole had good antifungal activity in vitro, confirming the results of Szekely et al. (24), who studied the antifungal activities of amphotericin B and itraconazole against 10 E. dermatitidis isolates. Amphotericin B, alone or combined with other drugs, has been used to treat invasive Exophiala infections (8, 15). One study reported clinical improvement when invasive E. dermatitidis infection was treated with a combination of amphotericin B and flucytosine, although definitive eradication of E. dermatitidis was achieved only after treatment with itraconazole (17). We found that caspofungin had a very broad MIC range. The activity of caspofungin in vivo may be unsatisfactory, as suggested by a study in which no reduction in fungal burden was noted after treatment of disseminated Exophiala infection in mice (22). The bimodal distribution of caspofungin MICs suggests that certain isolates with low MICs may be susceptible to the antifungal in humans despite data to the contrary in animal studies (22). It should be emphasized that there is poor agreement between Etest results and minimum effective concentration by the reference broth dilution method for some fungal species (10).

In conclusion, the recovery rate of E. dermatitidis was higher in respiratory specimens from patients with cystic fibrosis in western Sweden than in those reported previously from other countries. The isolates in this study exhibited reduced susceptibilities to several antifungal drugs. The clinical importance of E. dermatitidis in CF patients is still uncertain. To determine the impact on lung function decline in CF, prospective studies in which lung function improvement is assessed after E. dermatitidis eradication are needed.