Detection and Identification of Fungi from Fungus Balls of the Maxillary Sinus by Molecular Techniques

Abstract

The aim of this study was to find a reliable method for the detection and identification of fungi in fungus balls of the maxillary sinus and to evaluate the spectrum of fungi in these samples. One hundred twelve samples were obtained from patients with histologically proven fungal infections; 81 samples were paraffin-embedded tissue sections of the maxillary sinus. In 31 cases, sinus contents without paraffin embedding were sent for investigation. PCR amplification with universal fungal primers for 28S ribosomal DNA and amplicon identification by hybridization with species-specific probes for Aspergillus fumigatus, Aspergillus flavus, Aspergillus niger, Aspergillus terreus, Aspergillus glaucus, Pseudallescheria boydii, Candida albicans, and Candida glabrata were performed for all samples. Furthermore, PCR products were sequenced. Fresh samples were also cultivated. Fungal DNA was detected in all of the fresh samples but only in 71 paraffin-embedded tissue samples (87.7%). Sequence analysis was the most sensitive technique, as results could be obtained for 28 (90.3%) fresh samples by this method in comparison to 24 (77.4%) samples by hybridization and 16 (51.6%) samples by culture. However, sequence analysis delivered a result for only 36 (50.7%) of the paraffin-embedded specimens. Hybridization showed reliable results for A. fumigatus, which proved to be the most common agent in fungus balls of the maxillary sinus. Other Aspergillus species and other genera were rarely found.

Aspergillus spp. is the most commonly reported cause of fungal sinusitis and fungus balls of the sinuses, followed by dematiaceous fungi (5). Since the viability of fungal elements in fungus balls is poor, fungi frequently fail to grow from hyphae-rich material obtained during surgery. Furthermore, other limitations, such as slow growth of many relevant fungi, delayed production, lack of characteristic fruiting bodies or macroconidia, special nutritional requirements of certain fungi, and similarities in macromorphology or micromorphology or both at the genus level, may prevent their detection and identification. Histopathology reveals matted, dense conglomerations of hyphae separated from, but adjacent to, the respiratory mucosa of the sinus. Distinct features such as fruiting bodies (e.g., Aspergillus heads) and characteristic conidia or spores are rarely produced in vivo; thus, exact identification at the species level by histopathology is rendered nearly impossible. This might be the reason why it is still unclear as to how many of these infections are actually caused by Aspergillus species and what species of other fungi are common causes of fungus balls of the maxillary sinus. To our knowledge, there exist no data concerning the epidemiology of fungus balls of the maxillary sinus in Europe.

PCR is a sensitive method which can be used to detect viable as well as nonviable fungal pathogens. Several fungus-specific primers have been described. A variety of PCR protocols for human samples have also been published, including panfungal PCR assays (12, 17, 26) and methods that detect a single species (22, 25), members of a fungal family (13), or several species (6, 14). Also, in situ hybridization was shown to allow for reliable and rapid identification of Candida spp., Aspergillus spp., and Zygomycetes in tissue sections (11). However, a single protocol that permits detection and identification at the species level would be advantageous.

The aim of this study was to find a reliable method for the detection and identification of fungi, primarily Aspergillus species, in the tissues of patients with fungus balls of the maxillary sinus diagnosed by histopathology. For this purpose, we used a PCR assay targeting universal fungal 28S ribosomal DNA (rDNA) gene sequences, followed by amplicon identification by hybridization with species-specific probes for Aspergillus fumigatus, Aspergillus flavus, Aspergillus niger, Aspergillus terreus, and Aspergillus glaucus. In addition, probes for Pseudallescheria boydii, Candida albicans, and Candida glabrata were used. PCR products were subsequently sequenced in order to confirm data obtained by hybridization and also to identify additional species that might be involved in the pathogenesis of fungus balls of the maxillary sinus.

MATERIALS AND METHODS

Patients' samples.

One hundred twelve samples were obtained from patients with histologically proven fungus balls. Eighty-one samples were paraffin-embedded tissue sections of the maxillary sinus. In 18 cases, fresh surgical specimens of the maxillary sinus (approximately 3 cm in diameter) were sent directly to our laboratory. Thirteen samples were frozen at −20°C and then sent to the laboratory. The 31 samples without fixation and paraffin embedding were also cultivated.

Fungi used in study.

For evaluation and control of PCR as well as specificity testing of probes, the following 24 isolates obtained from an external quality control program (Neqas, Central Public Health Laboratory, London, United Kingdom) were used: A. fumigatus, A. terreus, A. flavus, A. niger, Aspergillus nidulans, Aspergillus versicolor, A. glaucus, Aspergillus candidus, Aspergillus clavatus, Aspergillus niveus, Paecilomyces lilacinus, Fusarium solani, Scedosporium prolificans, Scedosporium apiospermum, Trichoderma sp., Blastoschizomyces capitatus, Saccharomyces cerevisiae (ATCC 9763), C. glabrata, and Candida lusitaniae. In addition, C. albicans (ATCC 90028), Candida tropicalis (ATCC 90874), Candida parapsilosis (ATCC 22019), and Candida krusei (ATCC 6258) were applied.

DNA extraction.

Paraffin tissue sections were cut with a microtome. Before the first cut and after each sample, the microtome and other instruments were cleaned with cleaning benzine followed by 2 M HCl and rinsed with sterile water. One tissue section (5 μm thick) or a loopful of fresh tissue or the control strains was suspended in 200 μl of sterile water, and DNA extraction was performed according to the procedure described by Sandhu et al. (23). The DNA pellet was dissolved in 25 μl of sterile bidistilled water. After every two samples, a negative control was included.

PCR.

Universal primers for the 28S rDNA originally described by Sandhu et al. (23) were used to amplify a DNA sequence 260 bp in length (primers U1 [5′-GTG AAA TTG TTG AAA GGG AA-3′] and U2 [5′-GAC TCC TTG GTC CGT GTT-3′]). Primers were synthesized by MWG-Biotech (Ebersdorf, Germany). PCR amplifications were carried out in 50-μl reaction volumes with a Primus thermocycler (MWG-Biotech). The reaction mixture was slightly modified by reducing the quantity of template DNA from 5.0 to 1.0 μl. Cycling conditions were as follows: initial denaturation at 95°C for 10 min followed by 49 cycles of denaturation at 95°C for 1 min, annealing at 50°C for 1 min, and extension at 72°C for 2 min followed by a final extension phase at 72°C for 10 min.

To minimize the risk of contamination, a laminar flow hood and aerosol-resistant micropipette tips were used and areas for the preparation of mastermix, extraction of DNA, template preparation, setting up of PCR, and post-PCR analysis were physically separated. Each PCR run included 10 samples, a positive control with purified DNA of A. fumigatus, and at least two negative controls with blank reagents.

Amplification products were separated by electrophoresis in a 1% agarose gel by using standard techniques (1), subsequently stained with ethidium bromide, and analyzed with a gel documentation system (MWG-Biotech). PCR products 260 bp in length were interpreted as evidence of successful target amplification.

The detection limit of PCR was determined by using 10-fold serial dilutions of DNA extracts of either A. fumigatus, A. flavus, A. niger, A. glaucus, C. albicans, or C. glabrata, ranging from 65 pg to 6.5 fg of DNA per reaction. DNA was quantified by using a spectrophotometer. Dilutions were subjected to PCR with primers U1 and U2.

Dot blot hybridization.

Eight DNA probes as described by Sandhu et al. (23) were used for dot blot hybridization. The sequences are listed in Table 1. Prehybridization and hybridization were performed by using the North2South Chemiluminescent Nucleic Acid Hybridization and Detection kit (Pierce, Rockford, Ill.) as prescribed by the manufacturer. Before starting, amplicons were diluted 1:1 with 20× SSC (3 M NaCl, 0.3 M sodium citrate, 0.1 M EDTA) (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) and then treated as prescribed. Briefly, 5 μl of each sample was fixed on a positively charged nylon membrane (Hybond-N+; Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom) and air dried for 3 h. After a denaturation step (1.5 M NaCl, 0.5 M NaOH) for 5 min and a neutralization step (1.5 M NaCl, 0.5 M Tris-HCl [pH 7.2], 0.001 M EDTA) for 1 min, DNA was cross-linked (125 mJ, GS Genelinker [Bio-Rad, Hercules, Calif.]) and prehybridization was performed. Subsequently, the biotinylated DNA probe was added (2.5 pmol/ml of buffer), and hybridization (at 40°C), washings, and detection were performed as prescribed by the manufacturer. The membrane was then exposed to X-ray film for 1 min.

TABLE 1.

Probes used for dot blot hybridization

Fungus	Probe sequence (5′-3′)
A. fumigatus	CTC GGA ATG TAT CA
A. terreus	GCT TCG GCC CGG TG
A. flavus	AGA CTC GCC TCC AG
A. niger	CCC TGG AAT GTA GT
A. glaucus	CTG TCA TGC GGC CA
P. boydii	GCG ATG GGA ATG TG
C. albicans	CCT CTG ACG ATG CT
C. glabrata	CTT GGG ACT CTC GC

Sequencing.

28S rDNA amplicons of all samples were purified by using Microcon-YM-100 (Amicon; Millipore Co., Bedford, Mass.) columns and subjected to further amplification with the Big Dye Terminator Cycle Sequencing Ready Reaction sequencing kit (Applied Biosystems, Foster City, Calif.). Cycle sequencing was performed in a volume of 20 μl consisting of the following components: 1 μl of primer U1 or U2 (10 pmol/μl), 2 μl of Big Dye Terminator Ready Reaction mix, 3 μl of 5× sequencing buffer (100 mM Tris-HCl [pH 9.0], 10 mM MgCl₂), 1 μl of sample containing 10 to 100 ng of purified PCR product, and 13 μl of sterile water. For cycle sequencing in the thermocycler Robocycler Gradient 96 (Stratagene, La Jolla, Calif.), the following parameters were selected: 30 cycles of 30 s at 95°C, 10 s at 50°C, and 4 min at 60°C. PCR products were purified with Centri-Sep columns (Princeton Separations Inc., Adelphia, N.J.), as suggested by the manufacturer.

Sequence analysis was performed with an automated sequence analyzer (ABI Prism 310 genetic analyzer; Applied Biosystems). Sequence similarities were assessed by a search for homology with GenBank sequences by using the BLAST search program. Sequence similarities of 98% or more over a range of at least 75% of the amplification product were considered significant.

As sequences submitted to GenBank are not peer reviewed, the most important strains of the mentioned test organisms (A. fumigatus, A. versicolor, S. apiospermum, C. albicans, and S. cerevisiae) were also sequenced in order to prove the accuracy of the results obtained by the BLAST search program.

Cultures.

Twenty-nine samples without fixation and paraffin embedding were cultivated on Sabouraud dextrose agar containing gentamicin (20 mg/liter) and chloramphenicol (50 mg/liter) at 37°C for 10 days and then on Sabouraud dextrose agar containing gentamicin (20 mg/liter) and chloramphenicol (50 mg/liter), inhibitory mold agar, and Sabouraud broth at 30°C for 4 weeks. All specimens were processed under a laminar flow hood to avoid contamination. Positive cultures were identified by macro- and micromorphology as described by Kwon-Chung and Bennett (16).

RESULTS

PCR.

The sensitivity of PCR was evaluated by performing PCR of serial dilutions of purified DNA (65 pg to 6.5 fg) from A. fumigatus, A. terreus, A. flavus, A. niger, A. glaucus, C. albicans, and C. glabrata with the primers U1 and U2. Signals derived from as little as 6.5 fg of DNA of all tested species could still be clearly detected. In comparison with histopathology, PCR had a sensitivity of 100% in fresh samples but only 87.7% (71 out of 81) in paraffin-embedded tissue sections. In order to evaluate specificity, 20 histologically negative samples were also studied. All of them tested negative on PCR.

The eight specific DNA probes were tested for their ability to identify the amplicon of the corresponding target organism; no cross-reactivity to the amplicons of any of the tested fungi was registered.

Fresh tissue specimens.

Dot blot hybridization and sequence analysis were studied in 31 tissue samples from patients with histologically proven fungus balls. Table 2 summarizes the data for all fresh tissue sections without paraffin embedding. Sixteen samples (51.6%) were culture positive, 25 samples (80.7%) showed a positive result on hybridization, and 28 samples (90.3%) were positive by sequencing. Eleven of 16 (68.8%) culture-positive samples were confirmed by both molecular methods. Two samples identified as Schizophyllum commune were confirmed by sequence analysis only, as a DNA probe for this organism was not included in the study. One sample was culture positive with a hitherto unidentified basidiomycete (result confirmed by the European reference center Centraalbureau voor Schimmelcultures, Utrecht, The Netherlands), but sequencing identified Aspergillus oryzae. In another sample C. glabrata was grown, but A. fumigatus was identified by both hybridization and sequencing. In a further sample, the cultivated A. fumigatus was detected neither by hybridization nor by sequencing.

TABLE 2.

Comparison of culture, hybridization, and sequencing results from fresh samples obtained from patients with suspected or histologically proven fungus balls (n = 31)

No. of samples	Histology result	Result determined by:			% Similarity with GenBank sequences
		Culture	Dot blot hybridization	Sequencing of 28S rDNA
10	Positive	A. fumigatus	A. fumigatus	A. fumigatus	99-100
1	Positive	C. albicans	C. albicans	C. albicans	100
2	Positive	S. commune	Negative	S. commune	99
1	Positive	Basidiomycete	Negative	A. oryzae	100
1	Positive	C. glabrata	A. fumigatus	A. fumigatus	100
1	Positive	A. fumigatus	Negative	2 sequences
11	Positive	No growth	A. fumigatus	A. fumigatus	99-100
1	Positive	No growth	Negative	A. spergillus phoenicis/A. niger	100
1	Positive	No growth	Negative	P. boydii	99
1	Positive	No growth	P. boydii	2 sequences
1	Positive	No growth	A. fumigatus	Insufficient detection	<98

Fifteen samples did not grow on the various culture media but could be identified by hybridization, sequencing, or both methods. Eleven samples revealed identical results with both molecular methods. One sample identified by hybridization as A. fumigatus delivered two overlapping electropherograms by sequence analysis. A further sample identified as P. boydii by hybridization showed sequence similarity with GenBank sequences below 98%.

Paraffin-embedded tissue specimens.

PCR detected fungal DNA in 71 of 81 samples with paraffin embedding (87.7%). Table 3 shows the results for paraffin-embedded tissue sections obtained with both detection methods. Results for 47 of 71 samples were obtained by molecular techniques. Fifteen samples (31.9%) revealed results with both detection methods, 11 samples (23.4%) revealed results with hybridization, and 21 samples revealed results (44.7%) with sequencing alone. In 12 samples, both methods showed an identical result, namely A. fumigatus. Twenty-four of 71 samples (33.8%) could not be identified by sequence analysis: they either showed two different electropherograms or the sequences were insufficiently similar when compared with GenBank sequences.

TABLE 3.

Comparison of results obtained with hybridization and sequencing (n = 47) from histopathologically positive paraffin-embedded tissues

No. of samples	Result determined by:		% Similarity with GenBank sequences
	Dot blot hybridization	Sequence analysis of 28S rDNA
12	A. fumigatus	A. fumigatus	99-100
2	A. fumigatus	Insufficient detectiona,b
1	A. fumigatus	A. versicolor	100
1	A. niger	Fennelia nivea	100
2	A. niger	Insufficient detectiona	<98
5	A. niger	Insufficient detectionb
1	P. boydii	Insufficient detectiona	<98
1	P. boydii	Insufficient detectionb
1	P. boydii	Malassezia restricta	99
1	Negative	A. versicolor	100
1	Negative	Eupenicillium crustaceum	99
1	Negative	Penicillium expansum	99
1	Negative	P. dimorphosum	98
1	Negative	I. perplexans	98
1	Negative	B. graminis	98
1	Negative	S. trullisatus	98
5	Negative	S. cerevisiae	99-100
2	Negative	Malassezia restricta	99
1	Negative	Malassezia globosa	98
3	Negative	Cryptococcus sp.	98-99
1	Negative	C. tropicalis	99
1	Negative	Candida oniarionensis	99
1	Negative	Trichosporon sp.	99

^aGaps not allowing identification (n = 1).

^bTwo overlapping sequences (n = 1).

DISCUSSION

The purpose of this study was to evaluate and establish a molecular diagnostic method for the detection and identification of fungi in tissue specimens obtained from the maxillary sinus. As cultures have poor sensitivity, a molecular diagnostic method was deemed useful. Several fungi, including a few rare varieties, have been reported to cause fungal sinusitis (2-5, 7, 20). To our knowledge, no study based on molecular techniques provides epidemiological data on fungi causing fungus balls of the maxillary sinus. PCR is a sensitive diagnostic tool for the detection of pathogens that are nonviable or dormant because of antifungal therapy or other unfavorable conditions. The 28S rRNA sequence was selected as a detection target because its large size reveals adequate species-specific differences to distinguish closely related organisms (23). In addition, 100 or more copies of ribosomal genes in fungi are present (18), which was shown to provide good detection sensitivity (23).

PCR amplified fungal DNA in all fresh and frozen samples, whereas the DNA of paraffin-embedded tissues could only be amplified in 71 out of 81 samples (87.7%). One reason might be the age of tissue sections. They were collected over a long period of time; some tissue sections were nearly 20 years old. Thus, degradation of DNA in numerous sections might have been the reason for unsuccessful target amplification. Another reason might be that DNA became cross-linked or was sheared into small pieces by the long-term formalin treatment.

Fresh samples were not only subjected to molecular methods but also cultured. The results demonstrate the superior sensitivity of PCR in comparison with culture. For several decades, culture was the only method that permitted reliable identification of fungi from fungus balls. According to our data, both hybridization and sequence analysis are effective when used in fresh tissue samples. In 22 samples, fungi were identified by both methods and showed concordant results for 100%. In comparison with hybridization and culture, sequence analysis was found to be the most sensitive method, as 90.3% of the microorganisms present in the samples were identified by sequencing, compared with 77.4% by hybridization and only 51.6% by culture.

Results of culture, hybridization, and sequence analysis of fresh samples were concordant for 73.3% of culture-positive cases. Ten of them were identified as A. fumigatus and one was identified as C. albicans, the latter being in agreement with histopathology, where the presence of structures resembling yeasts was reported. One sample that was culture positive with C. glabrata was identified as A. fumigatus by both molecular methods. Also, histopathology proved the presence of hyphae characteristic for Aspergillus spp. In this case, it seems justified to assume that A. fumigatus was no longer viable in the necrotic sinus contents and that C. glabrata is a contaminant rather than the infecting agent.

In one sample, a hitherto unidentified basidiomycete was grown, whereas sequencing detected A. oryzae. We cannot exclude the fact that A. oryzae may also have been present in this sample. However, in this case one would expect two overlapping sequences by sequence analysis. In two further samples, the filamentous basidiomycete S. commune was grown, which could be confirmed by sequencing. Filamentous basidiomycetes causing infections are rarely reported in the medical literature. Nevertheless, Buzina et al. found S. commune in patients with sinusitis more often than usually anticipated (3). Therefore, it is difficult to decide whether or not the basidiomycete is a mere indicator of exogenous contamination.

Comparing culture and sequence analysis, the high proportion of concordant results shows that sequence analysis can be reliably applied in fresh tissue samples. Thus, results can be obtained more rapidly and more efficiently than with culture. However, it has to be mentioned that sequences submitted to GenBank are not peer reviewed and, therefore, some of them may be erroneous. As a consequence, five test strains were also sequenced in order to minimize the possibility of erroneous results. All of these strains were identified correctly. Besides, the probe designed for A. fumigatus yielded reliable results. Only a limited number of fungi were identified by other probes, e.g., C. albicans, which was also confirmed by both culture and sequencing, whereas P. boydii in one sample could not be confirmed by sequence analysis. No other organisms that could have been detected by the utilized probes were present in fresh samples. Results for paraffin-embedded tissue sections obtained by both hybridization and sequence analysis were discordant in 20% (3 of 15) of the samples. This finding and alignments in GenBank showing that some of the applied DNA probes, though binding specifically on the amplicon, may still bind unspecifically on human DNA sequences (data not shown) demonstrate the limitations of this method for detecting fungi other than A. fumigatus. However, the number of PCR-negative results in paraffin-embedded tissue sections and the number of samples that are negative by hybridization, sequence analysis, or both techniques indicate a rather low quality of some of these samples.

From 21 samples of 47 paraffin-embedded tissue sections (Table 3), positive results could only be obtained by sequence analysis. Some of the detected organisms such as Penicillium spp. are reported to be typical causes of fungal sinusitis (2, 5). Consequently, they also may cause fungus balls of the maxillary sinus. Phialemonium dimorphosum, a hyaline filamentous fungus and a synonym of Phialemonium curvatum, is known to trigger invasive disease in immunocompromised patients (9, 10) and has also been found in localized infections (8, 15, 19, 24). In view of these facts, it appears reasonable to add Phialemonium to the list of potential causes of fungal sinusitis and fungus balls.

Fungi have also been reported to colonize nasal mucus and paranasal sinuses. Ponikau et al. (21) isolated up to four different fungal species belonging to the common environmental genera in the nasal lavage fluid of both volunteers and patients with chronic rhinosinusitis. Thus, these fungi may well exist in the sinus without causing sinusitis or a fungus ball. This is consistent with the findings of Vennewald et al. (27), who described a commensal colonization of the paranasal sinuses, mainly by Aspergillus spp. but also by other molds.

The finding of Itersonilia perplexans and Blumeria graminis suggests exogenous contamination of the specimens or inhalation of spores without infection, although histopathology showed hyphae. Both of the detected fungi showed sequence similarities of 98%, and other matches were far below 98%. Neither fungus has been implicated in human disease nor does current knowledge support human pathogenicity, although B. graminis has at least been described in a patient with a destructive facial bone infection, but without any evidence of its pathogenicity (12). Whether these fungi are incidental environmental organisms residing on the specimen surface or innocent bystanders of a pathological condition should be the subject of future research. The same appears to be true for Strobilurus trullisatus, a tiny white-capped fungus usually found on Douglas fir cones.

Yeasts such as Malassezia spp., S. cerevisiae, Candida spp., Trichosporon spp., and Cryptococcus spp., also shown in Table 3, seem very unlikely to be the infecting organisms in our samples, as histopathology showed hyphal elements indicating the presence of filamentous fungi. Thus, the most reasonable explanation seems that they are part of a transient flora or contaminants. As shown by Ponikau et al. (21), several yeasts also colonize the nasal secretions of healthy individuals. Therefore, it would be reasonable to assume that they colonize the maxillary sinus as well.

Our results obtained with fresh samples show that molecular techniques such as dot blot hybridization and sequence analysis may be useful tools in order to determine the distribution of fungi causing fungus balls in the maxillary sinus. The findings of the present study confirm that A. fumigatus is the most common species causing fungus balls of the maxillary sinus, as has been reported in previous studies which used microscopy and culture (5, 7). Other Aspergillus species and other genera are not as common but may also be present in fungus balls.