Rapid Detection of Fungal Keratitis with DNA-Stabilizing FTA Filter Paper

Abstract

Purpose.: Polymerase chain reaction (PCR) is increasingly important for the rapid detection of fungal keratitis. However, techniques of specimen collection and DNA extraction before PCR may interfere with test sensitivity. The purpose of this study was to investigate the use of DNA-stabilizing FTA filter paper (Indicating FTA filter paper; Whatman International, Ltd., Maidstone, UK) for specimen collection without DNA extraction in a single-step, nonnested PCR for fungal keratitis.

Methods.: Specimens were collected from ocular surfaces with FTA filter discs, which automatically lyse collected cells and stabilize nucleic acids. Filter discs were directly used in single-step PCR reactions to detect fungal DNA. Test sensitivity was evaluated with serial dilutions of Candida albicans, Fusarium oxysporum, and Aspergillus fumigatus cultures. Test specificity was analyzed by comparing 196 and 155 healthy individuals from Switzerland and Egypt, respectively, with 15 patients with a diagnosis of microbial keratitis.

Results.: PCR with filter discs detected 3 C. albicans, 25 F. oxysporum, and 125 A. fumigatus organisms. In healthy volunteers, fungal PCR was positive in 1.0% and 8.4% of eyes from Switzerland and Egypt, respectively. Fungal PCR remained negative in 10 cases of culture-proven bacterial keratitis, became positive in 4 cases of fungal keratitis, but missed 1 case of culture-proven A. fumigatus keratitis.

Conclusions.: FTA filter paper for specimen collection together with direct PCR is a promising method of detecting fungal keratitis. The analytical sensitivity is high without the need for a semi-nested or nested second PCR, the clinical specificity is 91.7% to 99.0%, and the method is rapid and inexpensive.

Infectious keratitis, especially fungal keratitis, is a major cause worldwide of visual loss. ^1,2 The incidence of fungal keratitis keeps rising, partially in association with the extended wear of contact lenses, as recently emphasized by a global outbreak of Fusarium keratitis. ^3–6 Because of the slow growth of fungi in culture medium, the identification of the causative species is often delayed and may even fail in 20% to 60% of clinically evident cases. ^7–9 On the other hand, the clinical diagnosis of fungal keratitis correctly predicts fungi in only 38% of microbiologically proven cases. ¹⁰ The delay in diagnosis and adequate treatment may lead to endophthalmitis or the need for emergency keratoplasty. ^11–13  

Recently, several polymerase chain reaction (PCR)–based assays have been developed as rapid diagnostic tools for detecting fungal keratitis. ^7,14–17 Although these tests have the potential to improve and accelerate the diagnosis of keratomycoses, several technical problems remain. First, the area of infection is usually small, and it is therefore difficult to recover a sufficient amount of DNA with cotton swabs or metal spatulas. Second, calcium-containing swabs used to collect corneal samples can inhibit polymerase activity. The third problem arises in the laboratory, where DNA has to be extracted from the samples and purified as a starting template for the subsequent PCR. This step carries the risk that some of the DNA retrieved from the cornea will be lost. Many DNA extraction methods require sample volumes of 200 μL or more, meaning that corneal specimens have to be diluted further. Consequently, the risk of false-negative results with a single-step amplification from corneal samples is high. Therefore, most assays use a two-step nested PCR, to increase test sensitivity. However, boosting test sensitivity with a nested PCR step at the end of the diagnostic chain is associated with the risk of lower specificity because minimal amounts of inevitably present airborne fungi may lead to false-positive results. Hence, an ideal diagnostic test for fungal keratitis features a sampling technique that has a high retrieval rate for cellular and liquid material from the cornea, does not interfere with PCR enzymes, and avoids the need for nested PCR. 

DNA-stabilizing FTA filter paper (Indicating FTA filter paper; Whatman International, Ltd., Maidstone, UK) is commonly used in various medical fields, such as forensic medicine, ^18–20 and for the diagnosis of infectious and genetic diseases, ^21–24 because it fulfills the requirements outlined in the prior paragraph. It allows easy and reliable sampling of cellular tissue and body fluids. On direct contact of the specimen with the filter membrane, chelators and denaturants within the membrane lyse and inactivate microorganisms and cells. The released DNA becomes entrapped by forming a weblike structure around the FTA matrix, whereas cellular debris is rapidly washed off the card with a washing buffer. ²⁵ The extracted and amplifiable DNA remains stable, even if stored and transported at room temperature for several months. Once in the laboratory, the filter membrane is added directly to the PCR tube containing the usual enzymes and reagents. 

The purpose of this study was to investigate the potential of FTA filter paper for sampling conjunctival and corneal specimens for subsequent PCR-based diagnosis of fungal keratitis. In addition, we assessed the rate of clinically false-positive results in healthy volunteers, from an industrial population and a farming population, with FTA filter paper in combination with PCR. The PCR targeted the internal transcribed spacer (ITS) region, allowing separation of even closely related fungal species. ^26–28  

Materials and Methods

The study was approved by the ethics committee of the Canton of Lucerne and adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from each patient and participant in the control groups. 

Participants

Healthy Volunteers.

To evaluate overall specificity, we prospectively included two groups of healthy volunteers. The first group consisted of 196 volunteers (113 women and 83 men; mean age, 74 ± 12 years) from a predominantly urbanized area in Switzerland. These volunteers were recruited among patients scheduled for cataract surgery or intravitreal injections for age-dependent macular degeneration. The second group consisted of 155 volunteers (75 women and 80 men; mean age, 45 ± 16 years) from a rural area in Egypt. A detailed clinical history was taken from all volunteers, followed by slit lamp examination to exclude any ocular surface problems. Exclusion criteria for both groups were a history of any type of eye drops within 4 weeks before sample collection, the use of systemic or topical steroids or antibiotics within the past 3 months, any ocular trauma, the use of contact lenses of any type, preexisting ocular surface disorders, or systemic illnesses. Only one eye of each volunteer was used for sample collection. 

Patients with Infectious Keratitis.

To assess test sensitivity and specificity in infectious keratitis, we included in the study 15 patients referred because of suspected bacterial or fungal ulcers. After the eye was anesthetized with proxymetacaine (Alcaine; Alcon Pharmaceuticals Ltd., Hünenberg, Switzerland), conventional corneal Gram smears and scrapings were performed. Scrapings were plated on Sabouraud dextrose, Columbia sheep blood, and chocolate agar for microbiologic incubation for up to 14 days. After the conventional scraping, small FTA filter paper discs were applied directly to the corneal ulcer and analyzed as described in the following section. The correlation between microbiologic cultures and PCR results was analyzed. 

Specimen Collection and FTA Filter Paper Preparation

Indicating FTA filter discs (Whatman International Ltd.) were cut into 2 × 4-mm pieces with autoclaved scissors and forceps in a sterile air flow. Individual FTA filter paper strips were stored in 12-well plates under standard room conditions until used. Sterile forceps were used to apply the filter paper strips firmly for 1 second to the conjunctiva of healthy volunteers or to the corneal ulcer in patients with keratitis, with the same applanation force as otherwise used for obtaining impression cytology specimens (Fig. 1A). Before sample collection from healthy eyes, no topical anesthesia was applied to prevent any washout or dilution of the fungal load or any other factor that might interfere with fungal detection from the ocular surface. ²⁹ Successful sample capturing was indicated by a color change in the filter paper from pink to white (Fig. 1B). Filter paper strips were then transferred into dry 1-mL reaction tubes and stored at room temperature for 1 to 20 days, protected from light until processed by PCR. 

Figure 1.

 

Specimen sampling and processing: (A) The FTA filter paper is pressed onto the ocular surface for specimen collection. (B) The area of the collected specimen is indicated by a color change (white). A 2-mm disc is punched from the filter and transferred into the PCR tube.

A 2-mm-diameter disc was punched out of the color-indicated sample area of each filter paper with a sterile Harris Micro Punch (Whatman International Ltd.; Fig. 1B). After each disc was punched, the reusable punch was cleaned with 40% sodium hydroxide (Merck AG, Dietikon, Switzerland), 1 N (normal) hydrochloric acid, and 70% ethanol (Schweizerhall Chemie AG, Basel, Switzerland), then dried in a methane gas flame. The filter paper discs were transferred directly into PCR tubes and washed twice for 5 minutes with 200 μL FTA Purification Reagent (Whatman International, Ltd.), then washed twice in 200 μL TE buffer (10 mM Tris-HCl, 0.1 mM EDTA [pH 8.0]; Fluka BioChemika, Sigma-Aldrich Chemie, GmbH, Buchs, Switzerland) according to the manufacturer's guidelines. After the last washing step, the filter discs were left to dry for 5 to 10 minutes at room temperature. 

The paper discs that served as the control were handled the same as the test discs, but the negative control discs did not touch the ocular surface. Fungal suspensions instilled on the filter paper served as a positive control. 

Evaluation of PCR Sensitivity

Serial dilutions of quantified suspensions of Candida albicans (90028; ATTC, Manassas, VA), Fusarium oxysporum, and Aspergillus fumigatus (strain collections of the Institute of Medical Microbiology, University of Zurich, Zurich, Switzerland) were tested to evaluate the minimally detectable number of fungal organisms with the FTA filter paper and the following PCR protocol: A total of 5 μL of each dilution was applied on Indicating FTA filter discs, which were then examined by PCR. Simultaneously, 50 μL of the same dilution was cultured on Sabouraud dextrose agar for up to 14 days at 25°C. Colony counts in dilution series were compared with the PCR results. All experiments were performed in duplicate. As a positive control, the filter paper strips were rubbed directly onto culture plates with C. albicans, C. glabrata, F. oxysporum, F. solani, A. fumigatus, A. niger, A. terreus, or A. nidulans, and PCR was performed. 

DNA Amplification

PCR was performed in a commercial system (EpGradient Cycler; Eppendorf, Basel, Switzerland) in a 200-μL PCR tube containing a 2-mm diameter FTA filter disc, 10 μL of SYBR green master mix (Roche, Rotkreuz, Switzerland), 1 μL of each of the primers ITS-1 ³⁰ 5′-TCCGTAGGTGAACCTGCGG-3′ and ITS-4 ³⁰ 5′-TCCTCCGCTTATTGATATGC-3′ (5 μM each; Microsynth AG, Balgach, Switzerland), and 6 μL of H₂O, making a total PCR volume of 18 μL, plus the FTA filter disc. Depending on the fungal species, the amplified segment was expected to be 550 bp long for C. albicans, 500 to 600 bp for Aspergillus spp., and 550 bp for Fusarium spp. Primer ITS-2 ³⁰ (5′-GCTGCGTTCTTCATCGATGC-3′) was used in combination with ITS-1 for semi-nested PCR. 

Amplification parameters included initial denaturation for 10 minutes at 95°C followed by 35 cycles of denaturation for 15 seconds at 94°C, annealing for 30 seconds at 60°C, extension for 30 seconds at 72°C, and 1 cycle of final extension at 72°C for 5 minutes. 

The PCR product was mixed with 2 μL of gel-loading buffer, and 10 μL was loaded onto a 1.2% agarose gel containing ethidium bromide. The samples were subjected to electrophoresis at 100 V/cm for 30 minutes, with 0.25 TBE as the running buffer, and visualized under UV light. For sequencing (Microsynth AG), the remaining PCR amplification product was purified (MinElute PCR Purification Kit; Qiagen AG, Hombrechtikon, Switzerland). Sequences were compared with sequences available in GenBank (NCBI, Bethesda, MD). A PCR amplification was regarded as positive if a visible DNA band of the predicted segment length was present and confirmed by sequencing of the PCR product (DNA sequencing at Microsynth AG). 

Semi-nested PCR was performed for 20 negative samples from healthy volunteers without any signal after first-run amplification with primer ITS-1 and the reverse primer ITS-2. ²⁷ The amplification conditions were the same as just described. The amplification product was expected to be 215 bp long for C. albicans, 265 bp for A. fumigatus, and 235 bp for F. oxysporum. Furthermore, semi-nested PCR was performed when an initially positive PCR result could not be confirmed by sequencing. 

Results

Test Sensitivity

Direct PCR of the FTA filter discs detected as few as 3 C. albicans (Fig. 2), 25 F. oxysporum, and 125 A. fumigatus organisms. Nondissolved colonies sampled directly with the Indicating FTA filter discs from Sabouraud culture were easily detectable in cases of C. albicans, C. glabrata, F. oxysporum, and F. solani, but remained negative in cases of A. fumigatus, A. niger, A. terreus, and A. nidulans. 

Figure 2.

 

Ethidium-bromide–stained agarose gel of PCR reaction products from a C. albicans dilution series with 1 to 500 colonies per filter. Lanes 1 and 13: 100-bp ladder marker; lanes 2–12: serial dilution of C. albicans. Sequencing was positive for C. albicans dilution from lanes 3–12.

Test Specificity in Healthy Volunteers

Direct PCR amplifications of filter discs collected from the 196 healthy volunteers in Switzerland resulted in two weakly positive reactions (Fig. 3). However, sequencing of these PCR products was not possible in either case. Only an additional semi-nested PCR sequencing revealed the presence of C. albicans in one case. The two weakly positive tests resulted in a specificity of 99%. 

Figure 3.

 

Ethidium-bromide–stained agarose gel of PCR reaction products from healthy volunteers from Switzerland demonstrating two weak positive bands within all samples. Lanes 1 and 26: 100-bp ladder marker; lane 2: negative control; lane 3: positive control; lanes 4–25: conjunctival swabs from healthy volunteers including two positive PCR bands (lanes 9 and 18).

Among the 155 clinically healthy eyes from Egypt, 13 tests showed a weak positive band on the agarose gel (Fig. 4). In this farming population, the test specificity was 91.61%. Sequencing of the PCR product revealed C. albicans in one case. In the other 12 samples, the product of the first-run PCR was too minimal to be sequenced; therefore, a semi-nested PCR was performed for these 12 cases. Sequencing showed C. albicans in 10 samples, and 2 samples revealed sequences with no similarity to any fungal agent in GenBank; the nature of these sequences could not ultimately be identified. 

Figure 4.

 

Ethidium-bromide–stained agarose gel of PCR reaction products from healthy volunteers from Egypt demonstrating weak bands. Lane 1: 100-bp ladder marker; lane 2: negative control; lanes 10 and 18: positive controls; lanes 3–9, 11–17, and 19–26: conjunctival swabs from clinically healthy volunteers. Positive PCR bands were observed in lanes 9, 11, 12, 19, 20, and 24.

To further evaluate the effect of highly sensitive semi-nested PCR on specificity, we used semi-nested PCR to examine 20 additional punched discs from volunteers with a negative result in the standard PCR. In this subgroup of clinically healthy volunteers, all samples revealed weak but positive results. However, the products were too weak to be sequenced. 

Comparison of Microbiology and FTA Filter Paper/PCR-Based Diagnostics in Infectious Keratitis

Table 1 gives a summary of FTA filter paper/PCR results and the microbiologic culture results from conventional smears and scrapes from 15 patients referred with suspected bacterial or fungal corneal ulcers. 

Table 1.

 

Comparison of Microbiology Results with PCR Results in Cases of Keratitis

Case	Culture	Fungal PCR	Sequencing Result
1	C. albicans	Positive	C. albicans
2	C. albicans	Positive	C. albicans
3	P. lilacinus	Positive	P. lilacinus
4	No growth	Positive	P. cinerea
5	A. fumigatus (weak growth)	Negative	NP
Bacterial keratitis (n = 10)	Bacterial growth	Negative	NP

 NP, sequencing not performed because fungal PCR was negative.

Five ulcers were clinically suspected to be caused by fungi. PCR was strongly positive in three specimens (Figs. 5, 6). Sequencing of these PCR products identified C. albicans as the source of the DNA in two cases. This finding was confirmed by the culturing method, which revealed positive growth for C. albicans in both instances. In the third case (Fig. 6), sequencing of the PCR product from the corneal sample (Fig. 6, lane 3) was not successful. Cultures from this corneal ulcer revealed Paecilomyces lilacinus, and sequencing of the PCR product amplified from the filter of the culture plate confirmed P. lilacinus (Fig. 6, lane 4). Since the PCR products from the corneal filter and the culture plate filter (Fig. 6, lanes 3 and 4) were equal in size and run at a position different from the positive control C. albicans (Fig. 6, lane 10), we conclude that the product amplified from the corneal filter was derived from P. lilacinus as well. In the fourth case, which was clinically suspected to be a fungal keratitis, the FTA/PCR was also positive (not shown), and sequencing revealed Peniophora cinerea, which belongs to the Basidiomycetes family. Fungal and bacterial cultures from conventional scrapes in this case remained negative, but the keratitis improved after administration of antifungal therapy (voriconazole 200 mg/d twice daily). Therefore, we concluded that the infection had a fungal etiology. In the fifth case, fungal PCR and Gram-staining were negative, but after 14 days of incubation, the fungal cultures revealed slowly growing A. fumigatus. 

Figure 5.

 

PCR of clinical samples from corneal ulcers. Lane 1: 100-bp ladder marker; lane 2: negative control; lane 3: positive control; lanes 4 and 5: corneal ulcer with culture-proven C. albicans; lanes 6 and 7: corneal ulcers with positive bacterial but negative fungal cultures.

Figure 6.

 

PCR of clinical samples from corneal ulcers. Lane 1: 100-bp ladder marker; lanes 2 and 9: negative control; lane 3: filter from a corneal sample with culture-proven P. lilacinus; lane 4: filter from the culture plate with P. lilacinus; lanes 5–8: filter discs from culture-proven bacterial ulcers; lane 10: filter from a culture plate with C. albicans.

Ten ulcers had a culture-proven bacterial etiology: Pseudomonas aeruginosa (n = 4), Staphylococcus aureus (n = 3), coagulase-negative staphylococci (n = 2), and Streptococcus pneumoniae (n = 1). None of these 10 samples had a positive fungal PCR. 

Discussion

In this study, Indicating FTA filter paper was suitable for collecting and stabilizing fungal DNA from the ocular surface. The method allowed rapid and sensitive detection of fungal keratitis with a remarkably low risk of false-positive results in healthy volunteers and in culture-proven bacterial keratitis. The filter paper is an elegant alternative to conventional swabs for performing PCR, because it allows precise sampling of small affected areas with recovery of a sufficient amount of DNA, one of the main requirements for a successful amplification. An additional advantage of the Indicating FTA filter paper is that the color change it exhibits allows the clinician and the laboratory technician to clearly identify the area where the disc contacted the patient's eye. There is no need for DNA extraction before the DNA amplification step, which simplifies and accelerates the laboratory process and reduces the risk of losing any DNA during sample processing. Within less than 4 hours after taking the sample, a conclusion about whether an ocular infection is caused by fungi can be made. 

The rate of false-positive test results is another important aspect in the diagnosis of fungal infections with highly sensitive diagnostic tools such as PCR. Fungal cells are ubiquitous; thus, it is a delicate task for a diagnostic test to discriminate between healthy volunteers with a physiologically low copy number of fungal DNA on the ocular surface and patients with active fungal keratitis. The FTA filter paper used in this study is designed to capture DNA from wet surfaces or liquid media such as the cornea but not to collect dry airborne hyphae. This feature keeps the samples protected from contamination by airborne fungi in the examination room. Indeed, the rate of false-positive results in healthy individuals living in an urbanized area in central Europe was only 1%. Adding sequencing as a confirmation test further reduced the number of false positives to 0.5%. In subtropical Egypt's farming area, the environmental fungal load is much higher, and a rate of 8.3% false-positive results in healthy volunteers is not surprising. However, the high rate of false-positive results in healthy Egyptian farmers indicates that the low rate of false-positive results in Switzerland is not attributable to a lack of sensitivity but arises from real test specificity. This conclusion finds further confirmation in the fact that sequencing the positive results from the Egyptian farmers revealed fungi in only 1 of 13 cases. A false-positive rate of 1% to 8.3% found in our study compares favorably to the reported rate of 3% to 55% false-positive fungi results found in cultures from specimens obtained from healthy eyes in Europe and North America. ^31–34  

Performing a nested or semi-nested PCR (i.e., the subsequent amplification of a shorter segment within the area of the initial PCR product) would have the advantage of increasing the test's sensitivity because of the second amplification. This additional step lowers the threshold for successful detection down to 1 to 2 fungal organisms for C. albicans, ¹⁶ 10 organisms for F. solani, and 5 organisms for A. fumigatus. ⁷ However, a nested PCR decreases the specificity, as shown by the 20 samples of healthy participants in our study, in which the initial PCR was negative but became positive when boosted with a subsequent nested PCR. Unlike nested PCR, the single-step amplification can maintain a low rate of false-positive results, even in a warmer climate like that in Egypt. 

Several primers have been described for amplifying fungal DNA. Common primers target the 18S ribosomal RNA with a reported detection threshold in the range of 38 colonies with nested PCR. ⁷ However, the rate of false-positive results has not been reported with these primers. In a protocol reported very recently, Ghosh et al. ²⁶ used primers targeted to the ITS region. With semi-nested PCR, they achieved an extraordinarily low detection threshold of only one C. albicans cell, indicating that targeting the ITS region may be a more promising method. In our study, we also targeted the ITS region with a simple PCR protocol. This approach explains the somewhat higher detection threshold with serial dilutions for C. albicans (3 organisms) and F. oxysporum (25 organisms) but does not fully explain the much higher detection threshold of 125 organisms for A. fumigatus. However, from a clinical point of view, it should be stressed that such analyses of the lower threshold of detection are primarily important in comparing the sensitivity of various diagnostic techniques and microorganisms under standardized laboratory conditions. In cases of infectious keratitis, the number of microorganisms that is usually present on the ocular surface remains unknown. 

To investigate analytical sensitivity for molds in this study, we used a serially diluted spore suspension. The problem with the detection of Aspergillus may be explained by the hydrophobic property of Aspergillus spores that cannot perfectly adhere to the filter paper, or it may be due to the structure of the cell walls. The differences in cell wall constituents, chitin in particular, may affect DNA extraction efficiencies from fungi. Indeed, DNA release from spores of A. niger (filamentous fungi) was reported to be much more difficult than that from C. albicans (yeast). ³⁵  

The FTA filter paper with a single-step PCR failed to detect the single Aspergillus keratitis case in our series. However, it is not known whether Aspergillus hyphae of a corneal ulcer, similar to the spores, do not adhere or only weakly adhere to the filter, resulting in an insufficient DNA template for the subsequent single-step PCR assay used in this study. Hence, for the detection of Candida, Fusarium, and Paecilomyces species, which readily adhere to the FTA filter paper, we do not recommend performing a nested PCR. In selected cases when searching for Aspergillus species, an option might be to combine the FTA filter paper with a nested or semi-nested PCR to improve test sensitivity. Unfortunately, this test was not performed in our keratitis case in which the initial PCR was negative but culture results revealed A. fumigatus. 

The high specificity of the FTA filter paper combined with a simple PCR that was observed in the healthy volunteers was also demonstrated in the clinical cases. None of the 10 cases with culture-proven bacterial keratitis was positive for fungi, indicating an excellent predictive value of the test to exclude fungi at an early stage. Microbiology and PCR showed identical results in three of five cases of fungal keratitis, but the PCR results were available 6 days before the culture results. In clinical terms, PCR-based tests are faster diagnostic tools that are suitable for initiating early antimicrobial therapy. However, microbial cultures remain important because they are required for drug-sensitivity testing. 

In summary, the use of FTA filter paper represents an elegant, simple, and fast technique for sampling the ocular surface. It allows recovery of a sufficient amount of DNA without the need for subsequent DNA extraction. Because of the high retrieval rate of fungal DNA, the analytical sensitivity of the PCR assay was high and comparable to that of previously reported nested PCR protocols. Further studies with a larger case series of fungal keratitis are needed to determine the clinical sensitivity and to compare the FTA filter paper sampling technique with current sampling techniques (e.g., swab and corneal scrapings). The clinical specificity in our study was very high (91.7%–99.0%). FTA filter paper as a direct template in the PCR reaction tube in ITS-based PCR protocols can reliably detect Candida spp., Fusarium spp., and Paecilomyces spp., but it is somewhat limited in identifying Aspergillus spp. 

Footnotes

 Supported by a Departmental Research Grant, Cantonal Hospital Lucerne.

Footnotes

 Disclosure: N. Menassa, None; P.P. Bosshard, None; C. Kaufmann, None; C. Grimm, None; G.U. Auffarth, None; M.A. Thiel, None

The authors thank Mohamed El Kateb (Department of Ophthalmology, Alexandria Medical School) and Osama Ibrahim (Roayah Cornea Center, Alexandria, Egypt), for support in the collection of samples from Egypt, and Michaela Schulze (Institute of Medical Microbiology, University of Zurich), for cultivation of strains and preparation of the dilution series.