Long-term sequelae of Farmer's lung disease in HRCT: a 14-year follow-up study of 88 patients and 83 matched control farmers
- First Online:
- Cite this article as:
- Malinen, A.P., Erkinjuntti-Pekkanen, R.A., Partanen, P.L.K. et al. Eur Radiol (2003) 13: 2212. doi:10.1007/s00330-003-1848-1
- 79 Views
The aim of this study was to compare high-resolution computed tomography (HRCT) findings of long-term farmer's lung (FL) patients and control farmers. We studied 88 FL patients and 83 matched control farmers with a mean follow-up time of 14 years. Emphysematous, fibrotic, and miliary changes were recorded by HRCT. The pattern of emphysema and location and distribution of other findings were evaluated in detail. Emphysema was found in 20 (23%) FL patients and in 6 (7%) controls (p=0.005). Recurrences of FL attacks increased (p=0.021) the risk of emphysema. Prevalence of fibrosis (17 vs 10%, p=0.16) and miliary changes (11 vs 4%, p=0.06) did not differ significantly in patients and controls. Among FL patients, emphysematous, fibrous, and miliary changes were more pronounced at the base than in the upper parts of the lung (p<0.02). In slice analysis, the pattern of emphysema was more polymorphous (p=0.001) and the distribution of fibrotic and miliary changes was more variable among FL patients than controls. Emphysema in HRCT is more common in FL patients than matched control farmers, and the occurrence is increased by recurrences of FL. Emphysematous, fibrous, and miliary changes in FL patients HRCT are multiform and predominate in the lower parts of the lung.
KeywordsComputed tomographyFarmer's lungSequelaeEmphysemaFibrosis
Farmer's lung disease (FL), a form of hypersensitivity pneumonitis (HP; also called allergic alveolitis) caused by repeated inhalation of mouldy material, is still an important occupational disease among farmers in some countries [1, 2]. Since it involves respiratory bronchioles and alveoli, the term "bronchioloalveolitis" has been suggested for it .
Clinically, FL is divided into three stages—acute, subacute, and chronic —each with different symptoms and radiological findings. On the basis of previous follow-up studies, impaired diffusing capacity (DLCO) and airway obstruction seem to be the most common clinical sequelae of FL [5, 6]. As these findings are commonly seen in emphysema, they suggest that emphysema may be an important long-term morphological and radiological sequelae of FL; however, long-term high-resolution computed tomography (HRCT) sequelae of FL have seldom been reported.
Previous studies of radiological findings in FL were based on chest X-rays [7, 8, 9], which showed miliary changes and nodular shadowing in the acute stage and mainly fibrosis in the chronic stage; however, conventional chest radiography is not very accurate in detecting minor lung parenchymal changes, especially emphysema [10, 11, 12]. Although the CT findings of HP of different causes and the distribution of findings have been described by several authors [6, 11, 13, 14, 15, 16, 17, 18, 19, 20], the study groups involved have been quite small. In addition, smoking habits have not always been registered, whereas disease stage, number of recurrences, and follow-up time are often unclear. Another limitation in previous studies is that control groups have seldom been used.
Several studies have shown that farming itself is a risk factor for respiratory disorders [21, 22, 23]: a study in Finnish farmers found a 40% excess mortality rate from respiratory causes, for example ; however, studies on the frequency of morphological changes in the lung parenchyma associated with farming itself and their HRCT characteristics are scanty. The effect of farming itself should, however, always be taken into account when evaluating the surplus changes caused by FL.
The present study was designed to systematically evaluate the long-term sequelae in FL patients' lung parenchyma by HRCT, comparing these farmers with matched control farmers.
Materials and methods
Basic characteristics of the study population. FL farmer's lung
Ex-smokers smoking years
Current smokers smoking years
At the time of the HRCT examination none of the FL patients had acute FL. One FL patient had had the last recurrence 8 months before the radiological examination. In all other patients the interval was more than 2 years. Of the FL patients, 66% had had only one FL attack, 22% had a history of one recurrence, 7% had two, and 2% had three recurrences.
This study was approved by the ethics committee of the university of our district and city university hospital.
The HRCT examination was carried out with a Siemens Somatom Plus-S scanner (Siemens, Erlangen, Germany). Imaging was performed with the subject in the supine position and at end-inspiratory state. One-millimeter-thick slices at 20-mm intervals (137 kVp, 275 mA, 1-s scanning time) were obtained from the diaphragm to the lung apex, seven to ten in total. A high-frequency reconstruction algorithm was used and the zoom factor was as large as possible. Scans were filmed in 12-in-1 format with a window setting of 1200/−500 Hounsfield units (HU).
The presence of low-attenuation areas was interpreted to represent emphysema according to criteria described by Sanders et al , in which emphysematous changes are decreased attenuation, few or no vessels, and bullae with a well- or ill-defined wall. The scale was: 0=normal; 1=emphysema in <25% of the slice; 2=emphysema in 25–50% of the slice; 3=emphysema in 50–75% of the slice; and 4=emphysema in >75% of the slice. The maximum possible score of ten slices was therefore 40. Emphysema related to fibrosis or scarring was not regarded as emphysema.
The fibrotic and miliary changes by HRCT were graded according to a modified Hapke's classification  in which 0=normal, 1=minimal changes, 2=definite changes, and 3=marked changes. For fibrotic changes score 1=reticular and diffuse, fine, linear or peripheral, ill-defined, nodular opacities of varying size; 2=coarse linear opacities, radiating from the hila, perhaps with slight contraction of one lobe; and 3=coarse linear opacities, small cyst formation, and deformity due to contractions of lung. For miliary changes, the score system was 1=abnormalities so slight that they would be passed as within normal limits by a clinician or radiologist nor alerted by the clinical history to the possibility of farmer's lung; 2=definite change—discrete, sharply defined, punctiform, or so fine as to be ground glass in appearance; and 3=changes similar to those listed in score 2 with additional poorly defined "soft" opacities which become confluent in places.
The HRCTs were analyzed by two experienced radiologists who were blinded to the clinical data. The final results were based on a consensus reading of these two radiologist (K.P. and H.R.). Concordance of interpretation has been published previously . A third radiologist (A.M.), who was also blinded to the clinical data and the results of the other findings, reread all HRCT examinations for emphysematous changes. Intra- and interobserver reproducibility proved to be good . In addition, after a 1-year interval the third radiologist (A.M.) classified the pattern of emphysema and distribution of fibrotic and miliary changes and recorded changes of micronodules and mosaic pattern of parenchyma to obtain more detailed information about the miliary changes. The results of a consensus reading were available. The micronodules were defined as discrete, small, or focal round opacity with a diameter of no more than 7 mm . The mosaic pattern was recorded if regional differences in lung attenuation in parenchyma were seen, and vessels in the lucent regions of lung appeared smaller than in denser lung regions . The pattern of emphysema was classified as centrilobular, paraseptal, or bullae according to Austin et al. . Emphysema was classified as multiform when two patterns of emphysema were seen in the same slice. The location of emphysema among the ten slices and distribution of all other findings within each slice were analyzed slice by slice. The distribution of fibrosis and miliary changes within each slice was classified as peripheral/central, and dorsal/ventral. Findings were considered widespread if they were found in all four distributions. The location of changes were defined as lower and upper part of the lung with respect to the level of carina.
For statistical analysis, the total emphysema, fibrosis, and miliary scores of each patient were characterized into the following two groups: a score of ≥5 represented pathological HRCT, and scores 0–4 represented normal HRCT. This is the classification used in producing the results herein, unless otherwise specified. Ten slices were obtained from the majority (134 of 171, 78%) of the subjects, but only nine slices were obtained from 27 subjects and eight slices from nine subjects and seven slices from one subject due to smaller thorax. Thirteen of these 37 subjects had definite changes; the rest had only minor changes, so cutoff point of five for definite changes was also used for these subjects. Fisher's Exact test or the chi-square test was used for group comparisons. In order to evaluate the associations between the severity of emphysematous, fibrotic, and miliary changes, the scores of each patient were divided into four groups: score 0 as normal; 1–4 as mild; 5–9 as moderate; and ≥10 as severe. The Spearman correlation coefficient was used to test concurrent occurrence of different findings. Normality of all continuous variables was tested with Kolmogorov-Smirnov one-sample test. Since the distribution of the variables clearly deviated from the normal distribution, Mann-Whitney test was used to analyze the effect of recurrent FL attacks with the severity of emphysematous, fibrous, and miliary changes. The non-smoking study subjects were also analyzed separately, and statistically significant results of the non-smoking subjects are reported separately.
Prevalence and severity of findings
Prevalence of high-resolution CT findings among FL patients and control farmers
No definite changes
Severity scores of the different findings in a slice of those FL patients and control farmers with definite (total score ≥5) emphysematous, fibrous, or miliary changes
Total no. of slices
Effect of recurrences of FL attacks on emphysematous, fibrous, and miliary changes. Groups were congruent according to age and smoking habits
One FL attack (n=58)
Two or more attacks (n=30)
Pattern and location of emphysema
Location and distribution of fibrosis and miliary changes
The majority of FL patients (63%) had no abnormal parenchymal findings at HRCT; however, as a long-term sequelae, significantly more FL patients than control farmers had mild but polymorphous emphysema spearing the upper parts of the lung. The recurrences of FL attacks significantly increased the risk of emphysema. The FL patients and control farmers did not significantly differ with regard to amount of fibrosis and miliary changes, but the location and distribution of the changes differed significantly. Among FL patients these changes were located at the lower parts of the lung and they had significantly more often fibrosis and miliary changes distributed centrally than control farmers. Severity of emphysema and fibrosis was correlated in FL patients even when cicatrix emphysematous changes were excluded. The reported long-term sequelae of FL patients varies, partly because of selection bias and partly because of the limitations of the imaging methods used [6, 11, 13, 16, 19, 20]. Only a few studies [13, 33] report open-lung or transbronchial biopsies of the chronic FL patients, but the results are not reported in detail. Pathological correlation is seldom available as deaths caused by FL are rare , and definite pathological diagnosis of emphysema, especially, would require the inflated whole lung or lobe specimen. Several studies have demonstrated fibrotic changes, as evaluated by plain chest radiographs, to be the major radiological long-term sequelae of FL [7, 35]. Detection of emphysema, especially if it is not very pronounced, is more accurate by HRCT than by chest radiograph [12, 28]. In our previous study [29, 36] emphysematous changes on HRCT correlated strongly with pulmonary function defects typical of emphysematous changes (impaired FEV1, FEV%, and DLCO). In that study also fibrosis on HRCT correlated with restrictive pulmonary function tests (pft) and impaired DLCO, but miliary changes on HRCT did not correlate significantly with pft. The HRCT has proved to be superior in showing micronodules and ground-glass opacification [6, 11, 14, 33] in HP patients.
Although some CT studies have reported chronic or long-term changes of HP [11, 13, 14, 16, 17, 33], the patients have been mainly other than FL patients. A thin-section CT study of bird breeder HP  suggests that emphysema must be considered as an integral part of bird breeder lung, varying from areas of focal air trapping to diffuse emphysematous changes. Lalancette et al.  found emphysema as the main abnormality in HRCT in 7 of 14 chronic and in 2 of 19 of acute FL patients. They concluded that emphysema develops late in the course of FL. A previous study  with chest X-rays suggests that repeated exposures may lead to interstitial fibrosis and emphysema. Recently, Cormier et al.  reported HRCT characteristics of 95 FL patients. They found that emphysema, as a long-term change, was more frequent than interstitial fibrosis. The results of the present study support those of Lalancette et al.  and Cormier et al.  that emphysema is the main abnormality to be found, and ground-glass abnormalities as a long-term sequelae of FL are seen in only a minority of the FL patients.
In a previous follow-up study of FL , symptomatic recurrences, especially five or more, correlated with impaired pulmonary functions and radiological abnormalities. Especially in chronic pigeon breeders lung fibrosis seems to increase even if exposition has ceased . In the present study, one or more recurrences were associated with the risk of emphysema but not with the risk of fibrosis.
Ground-glass opacities have been reported to be the most common finding among acute and subacute FL patients' thin-section CT, but these abnormalities seem to diminish among the FL patients who have avoided environmental contact with antigen and are in clinical remission [17, 19]. Remy-Jardin et al.  suggest that ground-glass attenuation might reflect reversible interstitial lung changes that characterize HP, and further that the concurrent presence of ground-glass attenuation and/or micronodules with honeycombing on CT scans strongly suggests subacute changes superimposed on chronic disease. Hapke et al.  previously noticed that micronodules persisting for more than 1 year should be looked upon as chronic type changes. Another study  suggests that the changes are fibrous, if ground-glass opacities are associated with other fibrous changes. It seems that ground-glass opacities seen at HRCT of FL patients might reflect subacute disease superimposed on the chronic changes if exposure is continuing, or fibrosis, if exposure has ceased, especially if the changes are permanent.
The distribution of changes of HP have been reported, but the results have varied, mainly as a result of the different stage (acute, subacute, chronic) of disease and/or paucity of patients [9, 11, 13, 14, 16, 17, 33]. The distribution of long-term sequelae, especially in FL disease, have seldom been reported. Both lower-zone predominance [11, 17] and middle-zone predominance  of fibrosis have been reported. Relatively spared lung bases and apex are regularly reported. A study of 45 subacute and chronic FL patients  reported linear attenuation limited to the lower-lung zones, whereas emphysematous changes were diffusely distributed. Another study of 95 acute and chronic FL patients  reported that ground-glass changes predominated in the lower lobes of the lungs, whereas other abnormalities had no preferential distribution. In the present study emphysematous, fibrous, and miliary changes as the long-term finding of the FL patients' HRCT predominated in the lower parts of the lungs. Even though the most typical distribution of fibrosis was purely peripheral, centrally distributed fibrosis was also seen in FL patients, but not in controls. Only 2 patients had advanced fibrosis at the lower parts of the lungs, but no honeycombing was seen as in idiopathic pulmonary fibrosis.
An important strength of the present study is that the control group was carefully and statistically relevantly selected. In addition, all FL patients of the present study were at the chronic stage of the disease, constituting a uniform population whose precise follow-up time is recorded. The limitations of the present study were associated with the HRCT classification methods. Firstly, Hapke's classification is established for chest radiographs, not for CT. The HRCT is known to reveal more detailed information, especially of nodules and ground-glass opacification [11, 14]; however, we evaluated the films also for mosaic perfusion and micronodules, even if Hapke's classification does not separate them. Another possible drawback may be the fact that the right and left lungs were not viewed separately; however, FL or HP diseases have been reported to be bilateral, symmetrical diseases [16, 33]. Some control farmers had more severe emphysema than FL patients. This might be explained by a selection bias; as the symptomatic control farmers were more willing to participate in research than the healthy farmers.
Bronchiolitis, which is seen in subacute FL, causes air trapping. In the future there is a need to distinguish air trapping from emphysema in thin-section CT, if it is possible. Repeated HRCT inspiratory and expiratory examinations, and pulmonary function tests, including plethysmography, may give further insight into this question.
In conclusion, in the majority of FL patients long-term HRCT findings remain normal. More FL patients than a matched control population have emphysema, and in addition, recurrences of FL significantly increase the risk of emphysema. The emphysematous, fibrous, and miliary changes in the cattle-tending FL patients are typically located at the lower parts of the lung, and the distribution of changes and the pattern of emphysema are different from those in matched control farmers.
Acknowledgements. This work was supported in part by grants given to A. Malinen from The Finnish Anti-Tuberculosis Association Foundation, Pehr Oscar Klingendahl Foundation, and EVO Funding grant.