INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Farmer's lung disease (FLD), the most common form of extrinsic allergic alveolitis, is a pulmonary disease with symptoms of dyspnea and cough resulting from repeated exposure to high concentrations or prolonged exposure to low concentrations of inhaled antigens from moldy hay or straw, both of which lead to sensitization and development of this disease. Its diagnosis has most often relied on an array of nonspecific clinical symptoms and signs developed in an appropriate setting, with the demonstration of interstitial marking on chest X-ray, serum precipitins against offending antigens, lymphocytic alveolitis on bronchoalveolar lavage (BAL), and/or a granulomatous reaction on lung biopsies.

Identifying the etiological agents is of primary importance for the immunological diagnosis of this disease. It is also a necessary step in the development of preventive methods. Saccharopolyspora rectivirgula and Thermoactinomyces vulgaris remain the main recognized etiologies of FLD (1, 2). However, other bacterial (3, 4) and fungal species (5) have been demonstrated or suspected to play a role in FLD. The diagnostic value of serological tests is much debated (1, 9). This may be partly due to the antigenic diversity of FLD. It would certainly seem worthwhile to consider adapting the panel of microorganisms used for serological tests to each region (10).

In Franche-Comté, a region in the east of France where FLD is common (11), a previous study based on a large number of farmers demonstrated the poor reliability of immunologic tests using S. rectivirgula despite the clinical probability of FLD being correlated with the presence of precipitins against total hay extracts (12). This is probably due to the infrequency of S. rectivirgula in this area (13). To identify the putative etiological agents of FLD in our region and, more generally, to propose a process for identifying the microorganisms responsible for the disease, we performed a three-step experiment:

1. Analysis of the occupational exposure to microorganisms in farmer's lung patients (cases) compared with that of a group of control healthy farmers matched for personal and occupational characteristics

2. Preparation of antigens from the most frequently identified microorganisms in both groups of farms

3. Comparison of case and control farmers for serous immunologic responses against these antigens

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

In this prospective case-control study, exposure and immunologic reactions were compared in FLD and control dairy farmers. Sera from healthy urban nonexposed subjects constituted the control sera. Informed written consent was obtained from each subject. The protocol was approved by the local review board for research involving human subjects.

Case subjects were all dairy farmers newly diagnosed with FLD (n = 11) from January 1998 to April 2000 in our department according to recently proposed criteria (1, 14). All subjects with FLD had chronic exposure to moldy material, respiratory symptoms suggestive of the diagnosis, inspiratory crackles, low CO-diffusing capacity, suggestive high-resolution thoracic computed tomography (CT) scan features, and BAL lymphocytic alveolitis (Table 1).

Control subjects were as follows:

1. Eleven healthy farmers with no respiratory or general symptoms, normal pulmonary auscultation, and no past history of pulmonary diseaseall attested by replies to a standardized medical questionnaire and subsequently verified by a pulmonologist. Control subjects were matched to the case subjects for age (± 5 yr), sex, smoking habits, geographical location of the farm, occupational characteristics of the farm (size of the farm and the cattle herd, structure of the barn and the cattle shed, mode of drying and storing fodder) and type and duration of daily farming activities. Each control subject was found by a subject with FLD residing in the same village or a neighboring one.

2. Twenty-two urban nonexposed control subjects matched to the case subjects for age (± 5 yr), sex, and living in the same region (towns up to 5,000 residents).

Farms

Microbiological analyses of air and fodder were performed in the farms of the 11 subjects with FLD and the 11 control subjects. All farms raised cattle, with herds varying from 50 to 100 head. Hay was stocked in round bales except for farms pair number 9, 10, and 11, where hay was stored in bulk. All farmers distributed hay or ensilage (farm pair no. 1) and flour, cleaned their straw bedding, and milked their cows twice a day. Workshifts were the same for each pair of farms and lasted between 25 min and 2 h.

Microbiological Analyses

Air sampling during a work shift (n = 22). Microbiological sampling was performed in FLD farms on the day after diagnosis. The same sampling was performed in the control farms within the next 2 wk. Hay and straw bales were handled as usual on the day of sampling, with no particular selection. Air sampling was done with a biocollector MAS 100 (Merck, Darmstadt, Germany). All measures were carried out with a set of four petri dishes (for mold, mesophilic and thermophilic actinomycetes, and thermotolerant nonfilamentous bacteria isolation). Each set of dishes was exposed to 3 aspirations28, 10, and 1 literresulting in 12 dishes per sampling. The number of samplings varied from 4 to 16 per farm.

Hay, straw, ensilage, and flour sample analysis. A sample was taken for each available batch of fodder. The number of samples per farm varied from 3 to 10. In all, 137 samples were collected (80 in FLD farms and 57 in control farms): hay (n = 87), straw (n = 29), ensilage (n = 5), and flour (n = 16). Each sample was frozen at 18° C overnight to kill the mites. Samples were weighed, rinsed with 20 ml of sterile distilled water, shaken vigorously for 1 min, and cultured on petri dishes.

Culture media included the following:

1. Dichloran-Glycerol 18 Oxoid medium (Unipath, Basingstoke, England) with 0.5% chloramphenicol (Merck) at 30° C for mesophilic mold and yeast isolation

2. Actinomycete isolation agar Bacto medium (Difco, Detroit, MI), at 30° C for mesophilic actinomycetes and at 52° C for thermophilic actinomycetes

3. Mueller-Hinton II medium (Becton Dickinson, Cockeysville, MD), at 37° C for thermotolerant bacteria

4. In addition, samples of fodder were cultured on 3% malt-agar Oxoid (Unipath) with 10% salt and 0.5% chloramphenicol, at 20° C, for osmophilic fungal species. The number of colony-forming units per plate was counted after 3 and 7 d of incubation.

Identification of microorganisms. Molds were identified by means of techniques commonly used in mycology: macroscopic and microscopic observation. The ID 32C system (Bio Mérieux, Marcy l'Etoile, France) was used to identify yeasts. Actinomycetes were classified by macro- and microscopic observation, hydrolysis of casein, culture in saline and V8 mediums, and temperature tests (15). They were then compared with ATCC reference strains. Thermotolerant bacteria were counted, but not identified.

Immunological Methods

Antigen extract. Seven antigen extracts were produced from the seven microorganisms most frequently isolated from air and fodder. Five somatic antigens were derived from fungi (Absidia corymbifera, Wallemia sebi, Eurotium amstelodami, Aspergillus fumigatus, and Aspergillus nidulans) and two, from actinomycetes (mesophilic and thermophilic Streptomyces sp.). Three other antigens were produced: two total antigens derived from total extracts of hay and ensilage, and one somatic antigen derived from the reference strain S. rectivirgula (ATCC 15347). The antigens were produced as follows (16): both fungal and bacterial strains were cultured in brain-heart infusion for 2 wk, crushed with an Ultraturax (IKA Labortechnik, Staufen, Germany), sonicated, extracted overnight in NH₄CO₃ at 4° C, centrifuged at 13,000 rpm, ultrafiltered with a Centricon 10 (Amicon Millipore, Saint Quentin en Yvelines, France), and dosed for standardization at a 100-mg/ml concentration of protein. Hay and ensilage extracts were produced by the classic procedure (2) and used at a 100-mg/ml concentration of freeze-dried product.

Precipitins. Serous precipitins were investigated by agar gel double diffusion (17) and electrosyneresis on cellulose acetate (Sartorius, Goettingen, Germany) (18). A preliminary analysis showed reproducibility to be 89% for double diffusion (n = 108) and 78% for electrosyneresis (n = 202), with results showing ±1 arc considered to be the same (our unpublished data). All results were read blindly by two people, and no difference between the two readings was noted. Moreover, to confirm the first results, a second reading was done after clearing by immersion of the band overnight in 10% acetic acid.

Statistical Analyses

Environmental and parasitological data are known not to show a normal distribution. In this case, classic linear models are usually more likely to generate both type I and II errors. Moreover, log transformation often fails to normalize these right-skewed distributions (19). Consequently, analyses were conducted in two ways. A nonparametric Wilcoxon signed ranked test was used for air samples and immunological tests when only one type of data per farmer or farm was collected. Samples of several fodders were collected from each farm. The samples within each farm are likely to be dependent, that is, correlated with one another (20). Multilevel models including negative binomial error were used to take this correlated structure of the data into account (21). The Wald method was used to test coefficients (22).

We used SYSTAT 8.0 (SPSS, Chicago, IL) for nonparametric tests and MLwiN 1.02 for multilevel modeling (23). p Values less than 0.05 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Air Sampling in Cowsheds during a Workshift

The maximum values for concentration of microorganisms per cubic meter of air varied from 10³ to 6.5 × 10⁵ CFU for molds and from 10³ to 6.4 × 10⁵ CFU for actinomycetes. There was no significant difference between the two groups of farms in terms of the concentration of either category of microorganisms (total fungi and total actinomycetes). Table 2 shows the results of peak air concentration for the main species, which represented more than 80% of the total microorganism count. In 10 of 11 pairs of farms, the concentration of A. corymbifera was equal or higher in FLD farms than in control farms (p < 0.05).

Repeated measurements at each farm showed that the peaks of air concentration in microorganisms occurred soon after the beginning of workshifts; the time necessary for the concentration of microorganisms to double was, on average, 5 min at FLD farms and 10 min at control farms. At the end of the workshifts, microorganism concentrations returned to initial values in 12 of 22 cases for fungi and in 9 cases for actinomycetes.

Hay, Straw, Ensilage, and Flour Sampling

Twenty-seven filamentous fungal species were identified; the main species both in terms of abundance and frequency were as follows: W. sebi, E. amstelodami, A. corymbifera, A. nidulans, A. fumigatus (Table 3). They represented 82% of all filamentous fungal species. The others, by decreasing order of frequency, were as follows: Alternaria spp., Penicillium spp., Aspergillus flavus, Aspergillus umbrosus, Mucor spp., Cladosporium spp., Aspergillus versicolor, Aspergillus niger, Aspergillus reptans, Aspergillus ochraceus, Absidia cylindrospora, Aspergillus herbarum, Epicoccum purpurascens, Aureobasidium pullulans, Acremonium spp., Aspergillus candidus, Paecilomyces variotii, Fusarium sp., Aspergillus terreus, Scopulariopsis brevicaulis, Blastobotrys nivea, Trichoderma sp., and Chaetomium sp.

Yeasts represented 17% of the fungi. Six different species were identified; of these, Candida lambica, Candida famata, and Saccharomyces cerevisiae were the three most frequent.

Thirteen actinomycetes species were isolated, 4 of which were mesophilic (mainly Streptomyces spp. with dry, white and gray colonies) and 9 were thermophilic (Streptomyces with dry, white colonies that grew only between 37 and 55° C). Thermoactinomyces spp. and Saccharomonospora viridis were more rarely identified; only a few colonies of S. rectivirgula were isolated in five farms.

Five fungal and two actinomycete species represented more than 80% of all microorganisms collected. Mean concentrations of these seven species were compared in both groups of farms (Table 3). The concentration of A. corymbifera was significantly higher in FLD farms (p < 0.01). There was no significant difference for any of the other microorganisms.

Precipitins

Precipitin results by the electrosyneresis method are detailed in Figure 1. The mean number of arcs was significantly higher in patients with FLD than in control subjects for A. corymbifera (p < 0.01), E. amstelodami (p < 0.01), and W. sebi (p < 0.05). A two-arc threshold with the A. corymbifera antigen sufficed to differentiate subjects with FLD from control subjects in 9 of 11 cases. The same two-arc threshold used with W. sebi antigen produced correct identification in only 6 of 11 cases (p = 0.02). A three-arc threshold for E. amstelodami antigen allowed differentiation of FLD and control subjects in 8 of 11 cases (p < 0.01). The results of precipitin tests for the other antigens showed no difference between FLD and exposed control subjects. Comparison of precipitin results between FLD and urban nonexposed subjects showed a significant difference between both groups for all but two antigens: mesophilic Streptomyces (p = 0.06) and A. nidulans (p = 0.15).

View larger version (23K): [in this window] [in a new window]   Figure 1.   Precipitin results for 10 antigens by electrosyneresis in farmer's lung and exposed control subjects.

Ouchterlony double diffusion gave comparable results (not detailed in this article), but the number of arcs of precipitation for the 10 antigens tested was lower and the differences between groups less significant as compared with electrosyneresis. Nevertheless, significant discrimination (p < 0.05) was also seen for A. corymbifera, W. sebi, and E. amstelodami.

No positive correlation was found between total cells, percentage of lymphocytes, or total lymphocyte number in the BAL and the precipitin test scores with electrosyneresis, either for each of the 10 antigens or for the sum of the 10 antigens.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, we found A. corymbifera to be the only microorganism able to discriminate subjects with FLD from control farmers both in terms of exposure and of sensitization. Wallemia sebi and E. amstelodami probably play a role as well, whereas S. rectivirgula and T. vulgaris, still usually described as the classic antigens of the disease (1, 2), do not appear to be involved in this area of France. These results were obtained by an original methodology, which could eventually be used in identifying the etiological agents of FLD in other farming regions or occupational settings.

One peculiarity of this study was the selection of control farmers by FLD patients themselves. They used their professional experience to find controls comparable not only with respect to the classic criteria (age, sex, smoking habits) but also in terms of professional activity (type of farm, agricultural techniques, working conditions, duration of exposition to antigens). A standardized questionnaire was used to check, and showed that none of the control subjects presented any general or respiratory symptoms. These data were verified by a pulmonologist, who also ensured that all control subjects had a normal auscultation. Moreover, instead of using commercially available tests and/or antigens produced from microorganisms identified in previous studies performed in other areas, farmers and control subjects were tested against antigens extracted from the microorganisms most frequently isolated from their environment.

In this prospective study, the 11 pairs of exposed control subjects from the same area were included over a 28-mo period. The 11 subjects with FLD represented all the new cases examined by our team during this period. All 11 clearly presented recently published diagnostic criteria (1, 14) and, in accordance with numerous recently reviewed studies (1), none of them were smokers (Table 1). This strengthens the diagnostic certainty of FLD. Exposed control subjects were not subjected to paraclinical investigations, given that the presence of respiratory symptoms and inspiratory crackles is known to be a sensitive diagnostic criterion (1). Hence, their absence probably excludes the diagnosis of FLD. Infraclinical forms of FLD in control subjects cannot be totally excluded, however, but their diagnosis would have required a BAL. This invasive investigation in control subjects was not in accordance with our medical ethics, nor would it have been accepted by the local review board for research involving human subjects.

The presence of A. corymbifera in air and in fodder was significantly associated with FLD farms. In 10 out of 11 cases, the maximal concentration of A. corymbifera in the air during a farmer's workshift was similar to or higher than the concentration of this fungus in the air of control farms. However, by questioning farmers, we were able to establish that the symptoms of FLD were not caused by all types of hay or straw, but only by certain batches. During the study of FLD farms, the choice of hay and straw samples used to study air contamination was left to the farmer. The samples used the day of the study were sometimes, but not always, those suspected by the farmer to be responsible for the symptoms. This could explain the partial discrepancy found between air sampling and analysis of hay and straw stored in the farms. The significantly higher sensitization of patients with FLD to A. corymbifera gives further reason to suspect this fungi to be one of the main etiological agents of FLD in our area. Precipitins have been considered by numerous authors to be good indicators of exposure (2, 3, 24). However, tests for precipitins have rarely been found to discriminate patients with FLD from exposed control subjects (10, 27). Consequently, numerous authors still doubt the reliability of diagnostic tests for precipitins (3, 25, 26, 30). In our study, tests for precipitins against A. corymbifera clearly distinguished farmers with FLD from control farmers: nine farmers with FLD presented with two or more arcs with electrosyneresis, whereas no control farmer had more than one. These results concur with those of a study performed in Finland in which the authors, using the enzyme-linked immunosorbent assay (ELISA) method, found a level of IgG against A. corymbifera three times higher in farmers with FLD than in exposed control farmers (8).

Although E. amstelodami antigen also discriminated well between patients with FLD and exposed control subjects, no significant difference was found between farms owned by patients with FLD and control farms for contamination of straw, hay, or air. This fungus is closely related to Eurotium umbrosum (31, 32), previously described as an etiology of FLD in Finland (5). Similarly, although W. sebi was not more frequent in FLD farms than in control farms, this fungus allowed discrimination of the sera of patients with FLD from those of exposed controls. Wallemia sebi have been suspected to be another etiological agent of FLD in Finland (7). Because W. sebi requires a special medium for culture, its role has long been underestimated. The reasons for the sensitization of FLD patients in our region against E. amstelodami and W. sebi remains unclear. These fungi could be seen either as etiological agents of the disease or only as agents acting as cofactors in patients previously sensitized with A. corymbifera. The possibility of cross-reactions has also been reported by some authors (33, 34).

In contrast, although mesophilic Streptomyces spp. were frequently isolated both at FLD farms and control farms, the sensitization of patients against their antigens was low. Thus, the fact that a microorganism contaminates a farmer's environmenteven considerablyis not an argument for its role in the occurrence of the disease.

In conclusion, it appears today that A. corymbifera, and perhaps E. amstelodami and W. sebi, can be considered as etiological agents of FLD in eastern France. Conversely, S. rectivirgula and T. vulgaris, which are rarely isolated in this area, do not play a substantial role in FLD. In our view, the constant evolution of agricultural techniques requires periodic evaluation of species involved in the specific ecosystem of a given geographic area.