Routine processing procedures for isolating filamentous fungi from respiratory sputum samples may underestimate fungal prevalence

Colonization of the airways by filamentous fungi can occur in asthma, chronic obstructive pulmonary disease (COPD) and cystic fibrosis. A recent study found IgE sensitization to Aspergillus fumigatus to be associated with reduced lung function. Significantly higher rates of A. fumigatus were detected in sputum from asthmatics sensitized to this fungus compared to non-sensitized asthmatics. The rate of positive cultures was far higher than equivalent historical samples analysed by the local clinical laboratory following protocols recommended by the UK Health Protection Agency (HPA). This study compares the HPA procedure with our sputum processing method, whereby sputum plugs are separated from saliva and aliquots of approximately 150 mg are inoculated directly onto potato dextrose agar. A total of 55 sputum samples from 41 patients with COPD were analyzed, comparing fungal recovery of five dilutions of sputa on two media. Isolation of A. fumigatus in culture was significantly higher using the research approach compared to the HPA standard method for mycological investigations (P < 0.001). There was also a significant difference in the recovery rate of A. fumigatus (P < 0.05) between media. This highlights the need for a standardized approach to fungal detection which is more sensitive than the method recommended by the HPA


Introduction
Airway diseases such as asthma, chronic obstructive pulmonary disease (COPD) and cystic fi brosis (CF) are common, important causes of disease and ill health. Colonization of the airways by fi lamentous fungi can occur in all three disease groups, although the clinical relevance is unclear. Allergic bronchopulmonary aspergillosis (ABPA) is well recognized as a severe complication of airway colonization associated with a fl orid hypersensitivity reaction to Aspergillus fumigatus reported in up to 8% of asthmatics [1] and 13% of CF patients [2]. Fungal colonization may offers the advantage of being non-invasive to obtain, and is therefore more readily available and suitable for repeated measurements. The use of an induction protocol can result in samples being obtained from over three quarters of normal and asthmatic subjects who cannot produce sputum spontaneously [5].
While mycology laboratory accreditation programs are common, most countries including the USA, Canada and Australia, have no national standard guidelines for processing respiratory samples. In the UK most National Health Service (NHS) clinical microbiology laboratories follow the national standard method set out by the Health Protection Agency (HPA) in BSOP57 [6].
A recent study by our group found IgE sensitization to A. fumigatus to be associated with reduced lung function in asthma. In addition, signifi cantly higher rates of A. fumigatus were detected in sputum from A. fumigatus-IgEsensitized asthmatics (63%) compared to non-sensitized asthmatics (31%) and healthy subjects (7%) [7]. One striking fi nding was the high rate of recovery of fungi in culture compared to the routine NHS clinical laboratory (using the HPA method); where on historical samples Ͻ 10% of patients had a positive culture. There were a number of differences in technique that could explain the disparity in culture prevalence, including quantity of inoculating material and media used for culture. The research and clinical samples were not taken at the same time so the results could not be directly compared.
The aim of this study was to directly compare our methodology with that recommended by the HPA, in particular comparing the effect of dilution of sputum on incidence of A. fumigatus positive cultures, and to determine the infl uence different media had on culture rates of A. fumigatus and yeast.

Patients
Sputum samples were obtained from patients with COPD as they represent a cohort which readily produces the large volumes of to adequately compare the techniques. Samples were obtained from patients recruited from the general respiratory clinic at Glenfi eld Hospital (Leicester, UK) who had a physician ' s diagnosis of COPD according to the global initiative for chronic obstructive lung disease (GOLD) criteria [8] and were able to produce Ͼ 2 ml of sputum. Subjects were recruited as part of a study investigating the use of biomarkers to target therapy during exacerbations of COPD [9]. The study was approved by the Leicestershire, Northamptonshire and Rutland Ethics Committee, and all subjects gave informed written consent.

Sputum induction, processing and identifi cation of fungal isolates
Some patients produced sputum spontaneously, for others sputum induction was performed as described previously [5,10]. Expectorated samples were stored on ice, and processed within 2 h within a class II hood. Quality of sputum was evaluated in accordance with previous work [11]. The sputum sample was divided into two parts, ensuring an approximately equal quantity of plug and saliva was distributed between aliquots. One part was used to obtain a homogenized sample, the other to obtain sputum plugs.
The HPA BSOP57 guidelines are designed for the identifi cation of both bacteria and fungi from a single sample. For routine microbiological investigations that are not specifi cally designed to detect fungi, the sputum sample is homogenized, diluted, inoculated onto Sabouraud dextrose agar (SDA) containing 50 μ g/ml chloramphenicol (SC) and incubated at 37 ° C for two days. For specifi c mycological investigations the samples were undiluted homogenized sputum and incubation period was increased to fi ve days. The whole expectorate of sputum plug and saliva is used. In contrast, our routine approach sputum plugs are carefully removed from saliva prior to their inoculation onto potato dextrose agar (PDA) containing 16 μ g/ml chloramphenicol, 4 μ g/ml gentamicin and 5 μ g/ml fl uconazole (PGCF). In studies of airway infl ammation, careful selection of sputum plugs contributes to less salivary squamous cell contamination [5], leading to fewer bacterial species recovered in culture, indicating less oropharyngeal contamination [10]. The use of undiluted sputum was used previously to investigate the prevalence of A. fumigatus in patients with cystic fi brosis [12].
The sputum for homogenization was mixed with an equal volume of 0.1% DL-dithiothreitol (DTT) and incubated at 37 ° C for 15 min. After incubation, 10 μ l of homogenized sputum was diluted 1/500 in sterile water. Aliquots (10 μ l and 100 μ l) of both homogenized and diluted-homogenized sputum were inoculated in parallel onto both PGCF and SC plates. For the second part, plugs were separated from saliva, and approximately 150 mg (100 -250 mg) inoculated onto PGCF and SC plates. As an additional control, 100 μ l of 0.1% DTT was inoculated onto both media. All plates were sealed with nescofi lm and then transferred to a separate laboratory where they were incubated for seven days at 37 ° C. Plates were inspected, without opening, three times (after 40 -48 h, between days 4 and 6, and on day 7) and the number of visible colonies recorded at each time point. After seven days, fi lamentous colonies were examined and A. fumigatus identifi ed based on macroscopic and microscopic features [13]. when a neat sputum plug was inoculated. One sample was culture negative with a sputum plug, but culture positive with 100 μ l homogenized sputum.
The proportion of plates that were culture positive for either yeast or A. fumigatus was either the same on the two media or higher on PGCF at all dilutions. Based on multiple logistic regression analyses, media (B ϭ Ϫ 0.528, SE ϭ 0.204, P ϭ 0.010) dilution (B ϭ 1.403, SE ϭ 0.108, P ϭ 0.000) and organism (B ϭ Ϫ 5.333, SE ϭ 0.331, P ϭ 0.000) all independently had a signifi cant effect, after correction for each other, on whether positive cultures were obtained.
In the majority of cases (11 of 16) fi lamentous fungal growth was observed after two days incubation at 37 ° C, with fungi detected in three samples after 4 -6 days of incubation and growth was not detected with two until the 7th day.

Discussion
Both the choice of media and quantity of sputum inoculated onto the tested media have an effect on the detection of fungi from respiratory samples. The primary aim of this study was to directly compare our research focused approach for the recovery of fungi from sputum specimens to the HPA recommended approach. If the samples obtained in this study had been referred for microbiological investigation without specifi cally requesting fungal culture, all

Statistical analysis
Matched data was analyzed by McNemar test or Cochran ' s Q test. A multiple logistic regression was performed to explore the factors relating to fungal culture. Explanatory variables included media (SC or PGCF), organism (yeast or A. fumigatus ), and dilution of sputum. McNemar test was calculated using GraphPad ' s QuickCalcs web page at www.graphpad.com, Cochran ' s Q test and multiple logistic regression were performed using SPSS version 18. All P -values were two-tailed. P -values Ͻ 0.05 were considered as statistically signifi cant.

Results
A total of 55 sputum samples from 41 patients were analyzed. Each sample was obtained on a separate visit and was independently analyzed. A total of 10 culture plates were inoculated with each sample. Yeasts were isolated from Ն 1 plate from all but one of the samples (98%) that were inoculated on Ն 1 plate, and 29% of samples (16 of 55) yielded A. fumigatus .
Detection of A. fumigatus was highly dependent on quantity of sputum inoculated onto the culture plate (PGCF, Q ϭ 33.667, df ϭ 4, P Ͻ 0.000; SC, Q ϭ 13.867, df ϭ 4, P Ͻ 0.008). Of the 16 samples that were A. fumigatuspositive on Ͼ 1 plate, growth of the fungus was not detected with diluted-homogenized sputum samples (Table 2). Not considering differences due to the media, 19% and 44% of A. fumigatus -positive samples were detected using 10 μ l and 100 μ l of homogenized sputum, respectively, and 94% only 12.5% of the ' ever A. fumigatus -positive ' samples would have been detected. A higher starting quantity of sputum resulted in a clear increase in number of A. fumigatus colonies retrieved from three samples. The presence of fungi in the remaining samples was often detected on a single plate, and 50% resulted in the isolation of a single colony. It is unclear whether there is a clinically relevant difference between recovery of multiple colonies compared to a single colony. However, multiple colonies could be hypothesized as refl ecting a longer term or more severe colonization event. Unfortunately counting colonies is not always feasible as A. fumigatus is fast growing, often appearing within a day or two of incubation and some colonies rapidly spread over the entire plate, particularly with PGCF media, preventing growth of additional colonies.
In our original asthma study [7], we assumed that the quantity of inoculating material was the prime factor accounting for differences between approaches, and stated that the other factor, i.e., choice of media, was less likely to have affected recovery rates. In this investigation we found isolation rates for yeasts and A. fumigatus were much higher on PGCF than SC, which was unexpected. This surprising result was particularly true for A. fumigatus . Both PDA and SDA are commonly used generalpurpose mycological media. We selected PDA over SDA based on a comparison of conidial formation in pure cultures of seven allergenic species representing different fungal genera, i.e., Alternaria , Aspergillus , Botrytis , would have been negative for A. fumigatus . This suggests that unless a clinician specifi cally suspects fungi as a cause of a respiratory problem, fungal colonization is likely to be missed. If a mycological investigation had been requested,  Cladosporium , Epicoccum , Leptosphaeria , and Penicillium (unpublished data). In all cases the fungi readily grew on both media, but a higher yield of conidia was obtained on PDA. We added antibiotics to the PDA at optimum concentrations for isolation of pathogenic fungi [14], and fl uconazole to enhance recovery of A. fumigatus through suppression of Candida [15]. It is unclear as to whether A. fumigatus grows preferentially on PDA compared to SDA or the higher concentration of chloramphenicol in SC plates inhibits A. fumigatus development or the fl uconazole enhances A. fumigatus recovery. Further testing would be required to address these issues, but this study highlights the fact that choice of media does make a difference and that SC, used routinely in many clinical mycological laboratories, may result in the underestimation of the prevalence of A. fumigatus colonization.
The HPA guidelines recommend two-day incubation of specimens for standard microbiology studies and fi ve days for mycological analysis. In this study the fungus in the majority of A. fumigatus positive samples was observed within 2 -5 days, although seven days of incubation was needed for fungal isolation with two samples. While A. fumigatus and yeasts are fast growing organisms some clinically relevant fungi grow more slowly. A. fumigatus is the most prevalent fi lamentous fungi isolated from respiratory samples in cases of CF [3,4,12,16], and asthma [J. Agbetile, unpublished results]. Once established A. fumigatus can spread to cover the entire plate preventing growth of other fungi. In the absence of A. fumigatus , samples may be incubated for longer periods. The HPA recommendation is that if Paracoccidioides brasiliensis infection is clinically indicated it is best to incubate samples for up to six weeks [6]. In these instance bijoux bottles are recommended instead of culture plates to prevent the media from drying out.
The clinical relevance of the isolation of fungi from sputum samples is still a matter of controversy, particularly with regard to more infrequently detected species. In CF, persistent fungal colonization of the airways is thought to exacerbate lung damage [16]. Studies looking at the treatment of CF-ABPA patients with antifungal agents have shown an increase in FEV 1 as one of the outcomes [17], suggesting elimination of colonizing fungi leads to better lung function. In asthma, IgE-sensitization to fungi is common in individuals with severe disease, being reported in up to 66% of people with severe asthma in one study [18], and treatment with antifungal drugs has been shown to improve quality of life [19]. Furthermore, we have found lung function to be worse in A. fumigatus-IgE-sensitized asthmatics compared to non-sensitized asthmatics [7]. There is a growing body of evidence suggesting fungi may be having a deleterious effect in a far higher number of individuals than once suspected. To fully understand the link between fungi and ill health, and to determine the effi cacy of treatments for eliminating fungi from the airways, more sensitive means of identifying fungal colonization are required.
This study aimed to directly compare our research method for culturing fungi from sputum samples to the HPA protocol followed by NHS clinical laboratories. We do not claim that our approach is the ideal way to isolate fi lamentous fungi from respiratory specimens. What this study has shown is that the approach most commonly used may be very insensitive. A recent multi-centre study looking at prevalence of fungi isolated from sputum of CF patients using different culture protocols has highlighted the need for a standardized approach to be adopted, and the pressing need for an optimal method for analysis of the fungal component of CF microbiology [16]. We echo that sentiment, but believe a standardized approach is required for all studies of fungal colonization of the respiratory tract, not just CF, and that a more sensitive approach is needed to truly understand the health impact of fungi.