Document wq7mzqQX9mXRx4XYd4yzMmbm4

Eur Respir J, 1996, 9, 1000-1005 DOI: 10.1183/09031936.96.09051000 Printed in UK - all rights reserved Copyright ERS Journals Ltd 1996 European Respiratory Journal ISSN 0903 - 1936 Asbestos bodies in bronchoalveolar lavage in relation to asbestos bodies and asbestos fibres in lung parenchyma A. Karjalainen*, R. Piipari*, T. Mantyla**, M. Monkkonen*, M. Nurminen*, P. Tukiainen+, E. Vanhala*, S. Anttila* Asbestos bodies in bronchoalveolar lavage in relation to asbestos bodies and asbestos fibres in lung parenchyma. A. Karjalainen, R. Piipari, T. Mantyla, M. Monkkonen, M. Nurminen, P. Tukiainen, E. Vanhala, S. Anttila. ERS Journals Ltd 1996. ABSTRACT: In Finland, unlike other countries, anthophyllite asbestos has been widely used due to its domestic production in 1918-1975. In this particular con text, the aim of the present study was to analyse the relationship between asbestos bodies (ABs) in bronchoalveolar lavage (BAL) fluid and the concentration of ABs and the different amphibole asbestos fibres in lung tissue. Sixty five BAL lung tissue sample pairs from patients with pulmonary disease were analysed. The concentration of ABs in BAL fluid and lung tissue was deter mined with optical microscopy, and the concentration, type and dimensions of asbestos fibres in lung tissue with scanning electron microscopy. There was a significant correlation between the concentrations of ABs in BAL fluid and in lung tissue (r=0.72; p<0.001), between the concentrations of ABs and amphibole asbestos fibres in lung tissue (r=0.73; p<0.001), and between the con centration of ABs in BAL fluid and the concentration of amphibole asbestos fibres in lung tissue (r=0.64; p<0.001). In patients who had been exposed mainly to com mercial anthophyllite, significantly higher concentrations of ABs were observed per total pulmonary amphibole fibre burden, as compared to patients whose main expo sure was to crocidolite/amosite. The anthophyllite fibres in lung tissue were longer than the crocidolite/amosite fibres. The relationship between asbestos body counts in lung tissue and in bronchoalveolar lavage fluid was similar to previous international observations. When using the asbestos body count to predict the underlying total pulmonary amphibole asbestos burden in Finnish patients, however, it should be borne in mind that the relation ship between the two parameters seems to be different with anthophyllite as com pared to crocidolite/amosite fibres. Eur Respir J., 1996, 9, 1000-1005. *Finnish Institute of Occupational Health, Helsinki, Finland. **Laakso Hospital, Helsinki, Finland. +Dept of Pulmonary Medicine, Helsinki University Central Hos pital, Helsinki, Finland. Correspondence: A. Karjalainen Finnish Institute of Occupational Health Topeliuksenkatu 41 a A FIN-00250 Helsinki Finland Keywords: Anthophyllite asbestos bronchoalveolar lavage crocidolite lung tissue Received: April 23 1995 Accepted after revision January 19 1996 Occupational exposure to asbestos has been common in most industrialized countries during the past decades. In Finland, the use of asbestos peaked in the early 1970s [1]. Due to the long latent period of asbestos-related dis eases, their incidence will continue to increase in most countries during the next 10-20 years. According to a recent analysis, the maximum incidence may be higher than previously predicted [2]. The lung cancer and meso thelioma of asbestos-exposed workers, thus, constitute a growing medicolegal problem. Accurate exposure assess ment is crucial in these cases. The risk estimates of lung cancer and mesothelioma associated with increased pulmo nary concentrations of asbestos fibres as well as refere nce fibre concentrations observed in the general population, based on electron microscopy, have recently been pub lished in Finland [3-5]. In clinical routine, the intensity of asbestos exposure is often characterized by counting asbestos bodies (ABs) with optical microscopy in bronchoalveolar lavage (BAL) fluid samples. ABs are asbestos fibres that have been coated with ferroprotein by macrophages in the lung tissue [6, 7]. Their occurrence in the lungs is closely associated with the concentration of amphibole asbes tos fibres [8], and their concentration in BAL fluid reflects their concentration in lung parenchyma [9, 10]. In Finland, the use of amphibole asbestos has been com mon due to the domestic production of anthophyllite asbestos. Approximately 40% of all asbestos used in Finland consists of anthophyllite [1]. ABs are formed preferably on long fibres [6]. As the distribution of fibre lengths differs between anthophyllite and other am phibole fibres [11, 12], it is probable that there is also a difference in the coating rate between these fibre types. The aim of this study was to analyse the relationship between the concentration of ABs in BAL and lung parenchyma, the relationship between the concentrations of ABs and different amphibole asbestos fibres in lung parenchyma, and the value of BAL AB counting in the estimation of the pulmonary asbestos burden. ASBESTOS BODIES IN BAL AND LUNG PARENCHYMA 1001 Materials and methods Subjects All individuals with both a BAL sample and a lung tissue sample analysed in 1992-1994 in the Laboratory of Pathology of the Finnish Institute of Occupational Health were included in the study. Altogether, 65 sam ple pairs were identified when the BAL and lung tissue analysis files of the laboratory were compared. The mean age of the subjects was 65 (range 44-85) yrs. There were 59 patients with lung cancer, one with mesothelioma, one with lymphoma, one with pneumonia, and three with be nign pleural disease. All the subjects had been inter viewed personally for their complete work history and asbestos exposure during an ongoing BAL study. The occupational histories were classified into four exposure categories [13]: 1) Heavy exposure - at least 1 yr of ex posure in insulation work or in the manufacture of asbes tos products, or at least 10 yrs of exposure in shipyard or construction work with definite or probable intermit tent exposure to asbestos; 2) Moderate exposure - less than one year of exposure in insulation work or in the manufacture of asbestos products, or 1-10 yrs of expo sure in shipyard or construction work with definite or probable intermittent exposure to asbestos, or more than 10 yrs of exposure in garage work involving brake re pair; 3) Possible exposure - persons employed in other industrial occupations; and 4) Unlikely exposure - per sons with lifelong employment in occupations with no known exposure to asbestos, e.g. office work and farm ing. Forty five (69%) of the subjects studied were classi fied as heavily or moderately exposed, 15 as possibly exposed, and five as unlikely exposed. The high per centage of exposed individuals follows from the fact that most of the lung tissue samples (39 samples) were collected at necropsies performed due to suspicion of an asbestos-related death. The rest of the samples were collected from a series of surgically treated lung cancer patients without any selection as to exposure history. BAL sampling and AB counting in BAL samples BAL fluid samples were collected from patients who underwent diagnostic bronchoscopy for a suspected pul monary malignancy at the Department of Pulmonary Medicine of the Helsinki University Central Hospital or at Laakso Pulmonary Hospital, Helsinki. The study pro tocol was approved by the Local Ethics Committee, and informed consent was obtained from the patients. BAL was performed with a fibreoptic bronchofibreoscope under local anaesthesia. A segment of the right middle lobe of the lung was washed with 10x 20 mL aliquots of saline solution as described in detail previously [14]. In case of an intrabronchial tumour in the right middle lobe, the lingula was washed. The fresh lavage fluid was placed on ice in a glass container. No filtration was performed to trap mucinous contamination. For the determination of ABs, 40 mL of each original well-mixed sample was fixed in 20 mL of 96% ethanol. A sample containing 10 mL of BAL fluid was filtered through a 5 pm Millipore filter, using slight negative pressure at the end of the filtration. The filter was stained with an iron-staining Prussian blue dye, mou nted on a glass slide and covered with a coverslip. ABs were identified according to previously described crite ria [6], using an optical microscope and 200 fold mag nification. The number of ABs was expressed per mL of BAL fluid. The variation in AB concentration between consecutive 10 mL aliquots of each BAL fluid sample was insignificant. Lung tissue sampling Lung tissue samples were collected during surgery (26 samples) or necropsy (39 samples). The delay between BAL fluid and lung tissue sampling ranged from 7 days to 34 months (mean 4.5 months). The tissue pieces for AB counting and electron microscopic fibre analysis were taken from the peripheral part of the lung, not including pleural or tumour tissue. In case of surgical bilobectomy or pulmectomy, the sample was taken from the lobe which appeared to be closest to normal. The necropsy samples were taken, if possible, from the left upper lobe. AB counting in lung tissue samples About 1 g of formalin-fixed lung tissue was dried at 60C overnight, weighed, and digested with 30 mL of 5% potassium hydroxide overnight at 70C in 50 mL disposable polypropylene tubes. The sample was then centrifuged at 1,800xg for 20 min. The sediment was washed three times (distilled water, 0.5 N HCl, and dis tilled water). The sample was then resuspended in 10 mL of distilled water in a mild ultrasound bath (1 min). AB concentration was determined according to a modi fied cytocentrifuge method of Ehrlich and Suzuki [15]. Altogether 600-900 pL of the suspension were placed in a cytocentrifuge (Cytospin 3) and centrifuged on three glass slides (200-300 pL each) at 1,040 rpm (202xg) for 10 min. The glass slides were air-dried and covered with a coverslip. ABs were counted by phase contrast micro scopy at 400 fold magnification. All three slides per sam ple were thoroughly examined. The results are presented as number of ABs per gram of dry weight. Electron microscopic fibre analysis of lung tissue samples A tissue piece of about 100 mg wet weight was taken for the fibre analysis. Organic tissue was removed by low temperature ashing. Fibres were detected with a JEOL 100 CX-ASID4D electron microscope in scanning elec tron microscope (SEM) mode at a 5,000 fold magnifi cation [4]. A length to width ratio greater than three and roughly parallel sides was used as a fibre criterion. Fibres longer than 1 pm could be detected. A minimum of 400 viewing fields were evaluated to find at least 4-30 fibres per sample, depending on the density. An analytical sen sitivity (one fibre per sample) of about 0.1 million fibres per gram (f-g-1) of dry tissue could be achieved. An energy dispersive X-ray microanalyser (Tracor TN 5500) was used to determine the fibre type by compar ing peak ratios to standard spectra. Amosite and crocidolite have very similar X-ray spectra and are poorly 1002 A. KARJALAINEN ET AL. distinguishable. They are, therefore, not presented sepa rately. In a previous Finnish study, crocidolite fibres accounted for the great majority of amosite/crocidolite fibres identified with transmission electron microscope (TEM) [16]. Chrysotile fibres are poorly detected with SEM and, consequently, the results represent the con centration of amphibole fibres. Chrysotile fibres were observed in four samples and tremolite fibres in four samples. Fibre dimensions were measured directly on the screen in, altogether, 100 anthophyllite and 100 crocidolite/amosite fibres identified in 12 samples. Magnifications of up to x 100,000 were used. Statistical methods The distributions of concentrations of ABs in lung parenchyma and BAL and the distribution of concentra tions of asbestos fibres in lung parenchyma were approxi mately log normal. Log transformed concentrations were, therefore, used in the linear regression analyses, and the correlation between the log transformed concentrations was calculated. In nine of the BAL samples and one of the lung tissue samples no ABs were found, and in one of the lung tissue samples no amphibole asbestos fibres were found. For data analysis, concentrations of 0.05 AB-mL-1, 20 AB-g-1, and 0.05 million f-g-1, respectively, were used for these data points. Results ABs in BAL and lung tissue ABs were detected in 86% of the BAL samples and 98% of the lung tissue samples. Their concentrations var ied from <0.1 to 800 AB-mL-1 in BAL and from <40 to 890,000 AB-g-1 in lung tissue. The correlation between the concentrations of ABs in BAL and lung tissue is illustrated in figure 1. The simple regression equation log(AB-g-1 dry lung) = 3.39 + 0.590-log(AB-mL-1 in BAL) (r=0.72; p<0.001) was found between these two con centrations. Asbestos fibres in lung tissue Amphibole asbestos fibres were detected in all but one of the 65 lung tissue samples. Their concentrations ranged from <0.1 to 740 million f-g-1. Most of the lung tissue samples contained both anthophyllite and crocidolite/amosite fibres. The concentrations of anthophyllite fibres ran ged from <0.1 to 230 million f-g-1, and the concentrations of crocidolite/amosite fibres from <0.1 to 740 million f-g-1. The concentration of amphibole asbestos fibres correlated with the concentration of ABs in lung tissue (fig. 2) and BAL (fig. 3). When the concentrations of anthophyllite and crocidolite/amosite fibres were included in the same statistical model (multiple linear regression), both were found to correlate closely with the parenchy mal concentration of ABs (p<0.001 for both fibre types). The coefficient of anthophyllite fibres in the regression equation was, however, higher than the coefficient of cro cidolite/amosite fibres: log(AB) = 3.91 + 0.79-log(antho) 1000000i 100000 cn CD < 10000 10001 CO CD < 100 10 0.01 0.1 p<0.001 1 10 100 1000 Amphibole fibres in lung tissue million f-g-1 Fig. 2. - Relationship between concentration of asbestos bodies (ABs) and amphibole asbestos fibres in lung tissue samples of 65 Finnish patients. 1000 1 D5 O= 100 : C3D CO 10 1 ) '--............. I ......................I 0.01 0.1 1 ..................... I ...................... I .................... I 10 100 1000 ABs in BAL AB-mL'1 Fig. 1. - Relationship between asbestos body (AB) counts in bronchoalveolar lavage (BAL) fluid and lung tissue in 65 Finnish patients. _cu o 0.1 -Q 1Qc_ E 0.01 < 0.01 0.1 p<0.001 1 10 100' 1000 ABs in BAL AB-mL'1 Fig. 3. - Relationship between concentration of asbestos bodies (ABs) in bronchoalveolar lavage (BAL) fluid and concentration of amphibole asbestos fibres in lung tissue in 65 Finnish patients. ASBESTOS BODIES IN BAL AND LUNG PARENCHYMA 1003 Table 1. - Concentration of asbestos bodies (ABs) and amphibole asbestos fibres in lung tissue samples with pre dominant exposure (more than 80% of all amphibole fibres) to anthophyllite or crocidolite/amosite fibres Case ABs in No. lung tissue AB-g-1 Amphibole fibres in lung tissue million f-g-1 Proportion of ABs of amphibole fibres % Proportion of the predominant fibre type of all amphibole fibres % Main type of exposure Predominant exposure to anthophyllite 1 890000 253 2 780000 3 170000 4 105000 146 19.0 12.9 5 25000 6 13000 7 8800 8 8400 34.8 3.5 2.8 4.1 Predominant exposure to crocidolite/amosite 1 120000 2 87000 3 50000 739 100 13.1 4 20000 5 20000 6 15000 7 3100 335 35.0 10.8 5.6 8 2600 9 730 2.1 102 0.35 0.53 0.89 0.81 0.072 0.37 0.31 0.20 0.016 0.087 0.38 0.006 0.057 0.14 0.055 0.12 0.0007 91 >97 85 90 86 93 87 85 >97 >97 84 >97 89 92 86 89 >97 Asbestos product plant Asbestos product plant Asbestos product plant Construction Asbestos product plant Construction Construction Stevedore Asbestos sprayer Shipyard Construction, shipyard Shipyard Shipyard Construction Construction Power plant Shipyard + 0.43-log(cro/amo), where AB refers to pulmonary con centration of asbestos bodies (AB-g-1) and antho and cro/ amo to pulmonary concentrations of anthophyllite and crocidolite/amosite fibres (million f-g-1), respectively. Table 1 presents the concentrations of ABs and amphibole asbestos fibres in the lung tissue samples of eight cases with anthophyllite as the main fibre type and 9 cases with crocidolite/amosite as the main fibre type. All the samples that contained at least 2 million f-g-1 and in which the predominant fibre type accounted for at least 80% of all fibres are included in table 1. In samples that contained mainly anthophyllite fibres, the number of ABs in lung tissue divided by the number of amphibole asbestos fibres in lung tissue was significantly higher (p=0.0043; t-test). In the samples that contained mainly anthophyllite fibres, the concentration of ABs in lung tissue (opti cal microscopy) was on the average 0.44% (range 0.07-0.89, sd=0.29) of the concentration of amphibole asbestos fibres in lung tissue (electron microscopy). In the samples that contained mainly crocidolite/amosite fibres the respec tive figure was 0.096% (range 0.007-0.38, sd=0.12), and in the four cases that contained only crocidolite/amosite fibres, even lower. The subjects with main exposure to an thophyllite, had been exposed either in an asbestos prod uct plant processing the anthophyllite asbestos from Paakkila mine or in handling anthophyllite-containing pipe and boiler insulations in construction work or in stevedore work. The subjects exposed mainly to crocidolite/amosite fibres, had been exposed as an asbestos sprayer or as a bystander during asbestos spraying in shipyards, construction work, or power plants. Anthophyllite fibres detected in the lung tissue sam ples were longer (median length 7.5 pm, range 1.6-91 pm) than crocidolite amosite fibres in the same samples (median length 3.2 pm, range 1.1-40 pm). The pro portion of fibres longer than 10 pm was 45% in anthophyllite and 7.5% in crocidolite/amosite. Distribution ofparenchymal concentrations according to AB concentration in BAL fluid Tables 2 and 3 present the distribution of AB and amphibole asbestos fibre concentrations in lung tissue by concentration of ABs in BAL fluid. Of the cases with less than 1.0 AB-mL-1 in BAL, 92% showed less than 10,000 AB-g-1 and 96% less than 10 million amphibole fibres per gram dry lung tissue. Of the cases with >10 AB-mL-1 in BAL, all had >1,000 AB-g-1 and 90% >1.0 million amphibole fibres-g-1 in lung tissue. Table 2. - Comparison of AB in BAL fluid and lung tissue AB-mL-1 in BAL fluid n AB-g-1 dry lung tissue <1000 1000-9999 >10000 <0.10 9 6 (67) 3 (33) 0 (0) 0.10-0.99 17 8 (47) 7 (41) 2 (12) >1.0 39 3 (8) 18 (46) 18 (46) >5.0 27 1 (4) 11 (41) 15 (56) >10.0 19 0 (0) 6 (32) 13 (68) AB: asbestos body; BAL: bronchoalveolar lavage. Percentage values are presented in parentheses. Table 3. - Comparison of AB concentrations in BAL fluid and amphibole asbestos concentrations in lung tissue AB-mL-1 in BAL fluid Amphibole asbestos fibres million fibres-g-1 dry lung tissue n <1.0 1.0-9.9 >10.0 <0.10 0.10-0.99 >1.0 >5.0 >10.0 9 6 (67) 17 13 (76) 39 12 (31) 27 5 (19) 19 2 (11) 3 (33) 3 (18) 16 (41) 12 (44) 9 (47) 0 (0) 1 (6) 11 (28) 10 (37) 8 (42) AB: asbestos body; BAL: bronchoalveolar lavage. Percentage values are presented in parenthesis. 1004 A. KARJALAINEN ET AL. Discussion The concentrations of ABs in BAL fluid and lung parenchyma were found to correlate with each other and with the concentration of amphibole asbestos fibres in lung parenchyma. Most of our patients were exposed both to anthophyllite and crocidolite/amosite fibres, and the concentration of ABs in lung parenchyma correlat ed with both the concentration of anthophyllite and crocidolite/amosite fibres. According to multiple linear regression analysis, however, higher AB concentrations would be expected for a given pulmonary anthophyllite content as compared to an identical pulmonary concen tration of crocidolite/amosite fibres. A similar statisti cally significant difference was found when only cases with predominant exposure to one of the fibre types were considered. The general objective of asbestos fibre measurement in biological samples is to provide clinicians or epi demiologists with better information on past asbestos exposure. When AB counting in BAL samples is used as an indicator of past cumulative asbestos exposure, the validity of the following two general assumptions is cru cial: 1) the number of ABs in BAL fluid correlates with the number of ABs in lung parenchyma; and 2) the num ber of ABs in lung parenchyma correlates with the past cumulative asbestos exposure. In three previous studies, the correlation between parenchymal and BAL AB concentrations has been analysed with optical microscopy [9, 10, 17]. Table 4 summarizes the results of two of these studies and our results. The regression equations predicting the under lying parenchymal concentration from the concentration of ABs in BAL fluid are very similar. The equations predict nearly identical parenchymal concentrations for concentrations ranging 1-10 AB-mL-1 in BAL. Despite the numerous methodological differences in sample pre paration and AB counting, the similarity of the results is striking, and strongly supports the view that there is a well-established structural relationship between the con centrations of ABs in BAL and lung parenchyma. How ever, all three series contained only few cases with more than 100 AB-mL-1 in BAL or more than 100,000 AB-g-1 in lung parenchyma, and the consistency of the results in such high concentrations is not equal to that in mode rate concentrations. In each of these studies, there has also been a wide variation in the ratio of BAL and parenchymal concentration readings in individual cases, and the correlation between the two concentrations is lower if only restricted concentration areas are analysed [17]. Part of this variation may arise from the regional variation of AB concentrations in the different anatom ical sites of the lungs [9, 18]. Chrysotile fibres and short amphibole fibres are rarely coated [6, 7], and the concentration of ABs in BAL or lung parenchyma therefore reflects mainly the concen tration of long amphibole fibres in lung parenchyma. Our results indicate that this relationship is different in persons exposed to anthophyllite as compared to persons exposed to crocidolite/amosite. The average number of ABs divided by the total amphibole asbestos concentra tion was about five times higher among those exposed mainly to anthophyllite as compared to those exposed mainly to crocidolite/amosite. This is to be borne in mind when the number of ABs in lung parenchyma or BAL is used as an estimate of the cumulative asbestos expo sure in Finland. The observed difference in the coating of anthophyllite and crocidolite/amosite fibres is most probably due to the difference in fibre length between these fibre types. In lung parenchyma, fibres longer than 10 pm were about six times more frequent among anthophyllite than among crocidolite/amosite fibres. In this respect, our results apply only to the anthophyllite fibres detected in the lungs of Finnish patients who have been exposed to commercially exploited and industrially pro cessed anthophyllite fibres. Such fibres are probably on average longer than anthophyllite fibres from environ mental exposure or contamination of other minerals that sometimes occur in lung tissue samples of patients from countries where anthophyllite has not been in commer cial use. When the relationship between AB concentration and total fibre burden is interpreted, it must be emphasized that wide interlaboratory variation occurs in the deter mination of pulmonary fibre concentrations with elec tron microscopy [19]. Our results are consistent with previous reports in that a low or negative count of ABs in BAL does not rule out heavy exposure [9, 10, 17, 20]. This is best illus trated in two samples with a very high concentration of crocidolite/amosite fibres and only a relatively low con centration of ABs in lung tissue and BAL; 100 million f-g-1 of crocidolite/amosite, 730 AB-g-1, and 6 AB-mL-1 were observed in the samples of a shipyard worker with mesothelioma, and concentrations of 330 million f-g-1 cro cidolite/amosite, 20,000 AB-g-1, and 23 AB-mL-1 in the samples of a shipyard worker with pulmonary adeno carcinoma and asbestosis. In these cases, the regression equations of figures 2 and 3 would have predicted a much lower amphibole fibre concentration in lung tissue. Fig ure 3 also shows a number of cases with an elevated parenchymal amphibole fibre concentration but with a BAL AB content at or below the detection limit. All but one of the six cases with >1 million amphibole f-g-1 in lung parenchyma and <0.1 AB-mL-1 in BAL, had more than 1,000 AB-g-1 in lung parenchyma. This probably Table 4. - Correlation between parenchymal and BAL asbestos body (AB) concentrations (optical microscopy) in three international studies Regression equation Predicted parenchymal AB concentration (AB-g-1) Cases for 1, 10 and 100 AB-mL-1 in BAL______________ 1 AB-mL-1 10 AB-mL-1 100 AB-mL-1 n [Ref] log(AB-g-1) = 3.25 + 0.77 log(AB-mL-1) (r=0.74) log(AB-g-1) = 3.34 + 0.684 log(AB-mL-1) (r=0.73) log(AB-g-1) = 3.39 + 0.590 log(AB-mL-1) (r=0.72) 1800 2200 2500 10500 10600 9500 62000 51000 37000 69 100 65 [10] [9] Present study AB: asbestos body; BAL: bronchoalveolar lavage . ASBESTOS BODIES IN BAL AND LUNG PARENCHYMA 1005 demonstrates that, in addition to cases with an unex pectedly low fraction of coated fibres, this cluster may contain cases with a poorly representative BAL sample. Most of the ABs extracted from the human lungs are formed around amphibole fibres [6], and the pulmonary concentration of ABs correlates with the pulmonary amphibole concentration but not with the concentration of chrysotile fibres [8]. Elevated concentrations of ABs formed on chrysotile fibres can, however, be found in the lungs of workers with heavy exposure to chrysotile [21], but the concentration of chrysotile fibres in lung tissue or BAL is a poor indicator of cumulative chrysotile exposure [8, 20]. In Canadian chrysotile miners, tremolite fibres constitute only a small percentage of the air borne fibres in the mines, but have accounted for up to one half or more of the pulmonary asbestos fibres in the miners [22, 23]. We used scanning electron microscopy, which is known to underestimate the concentration of chrysotile fibres. Consequently, only one of the 65 lung tissue samples in our series contained chrysotile fibres as the main fibre type. This was from an Estonian immi grant who had worked as a pipe insulator in the former Soviet Union from 1970 to 1990. A concentration of 0.1 AB-mL-1 was detected in BAL in 1993 and 70 AB-g-1 and 2.6 million fg-1 in lung tissue in 1994. Due to the above-mentioned restrictions of bronchoalveolar lavage asbestos body counting, a complete chro nological work history based on personal interview is often the best way to evaluate past exposure to asbestos. In cases with inconclusive or lacking work history, bronchoalveolar lavage asbestos body counting offers, how ever, a reasonably reliable estimate of the past exposure. Acknowledgements: The authors thank T. Kaustia for linguistic revision of this manuscript. References 1. Ministry of Labour. Asbestikomitean mietinto (Report of the Asbestos Committee). Report series of the Gov ernmental Committees No. 66, Helsinki 1989. (In Finnish with English summary). 2. Peto J, Hodgson JT, Matthews FE, Jones JR. Continuing increase in mesothelioma mortality in Britain. Lancet 1995; 345: 535-539. 3. Tuomi T, Huuskonen M, Virtamo M, et al. Relative risk of mesothelioma associated with different levels of expo sure to asbestos. Scand J Work Environ Health 1991; 17: 404-408. 4. Karjalainen A, Vanhala E, Karhunen P, Lalu K, Penttila A, Tossavainen A. Asbestos exposure and pulmonary fiber concentration of 300 Finnish urban men. Scand J Work Environ Health 1994; 20: 34-41. 5. Karjalainen A, Anttila S, Vanhala E, Vainio H. Asbestos exposure and the risk of lung cancer in a general urban population. Scand J Work Environ Health 1994; 20: 243-250. 6. Churg AM, Warnock ML. Asbestos and other ferrugi nous bodies: their formation and clinical significance. Am J Pathol 1981; 102: 447-456. 7. Pooley FD. Asbestos bodies, their formation, composi tion and characters. Environ Res 1972; 5: 363-379. 8. Albin M, Johansson L, Pooley FD, Jakobsson K, Attewell R, Mitha R. Mineral fibres, fibrosis, and asbestos bod ies in lung tissue from deceased asbestos cement work ers. Br J Ind Med 1990; 47: 767-774. 9. De Vuyst P, Dumortier P, Moulin E, et al. Asbestos bod ies in bronchoalveolar lavage reflect lung asbestos body concentration. Eur Respir J 1988; 1: 362-367. 10. Sebastien P, Armstrong B, Monchaux G, Bignon J. Asbestos bodies in bronchoalveolar lavage fluid and in lung parenchyma. Am Rev Respir Dis 1988; 137: 75-78. 11. Tuomi T. Fibrous minerals in the lungs of mesothelioma patients: comparison between data on SEM, TEM and per sonal interview information. Am J Ind Med 1992; 21: 155-162. 12. Karjalainen A, Taikina-aho O, Anttila S, Heikkila L, Tossavainen A, Vainio H. Asbestos exposure among Finnish lung cancer patients: comparison of scanning and transmission electron microscopy in the analysis of lung burden. Ann Occup Hyg 1994; 38 (Suppl. 1): 657-663. 13. Karjalainen A, Anttila S, Mantyla T, Taskinen E, Kyyronen P, Tukiainen P. Asbestos bodies in bronchoalveolar lavage fluid in relation to occupational history. Am J Ind Med 1994; 26: 645-654. 14. Taskinen E, Tukiainen P, Renkonen R. Bronchoalveolar lavage: influence of cytologic methods on the cellular picture. Acta Cytologica 1992; 36: 680-686. 15. Ehrlich A, Suzuki Y. A rapid and simple method of extracting asbestos bodies from lung tissue by cytocentrifugation. Am J Ind Med 1987; 11: 109-116. 16. Anttila S, Karjalainen A, Taikina-aho O, Kyyronen P, Vainio H. Lung cancer in the lower lobe is associated with pulmonary asbestos fiber count and fiber size. Environ Health Perspect 1993; 101: 166-170. 17. Teschler H, Hoheisel G, Fischer M, Muller KM, Konietzko N, Costabel U. Asbestkorperchengehalt der bronchoalveolaren Lavageflussigkeit als Parameter einer erhohten Asbestbelastung der Lunge. Dtsch Med Wochenschr 1993; 118: 1749-1754. 18. Teschler H, Konietzko N, Schoenfeld B, Ramin C, Schraps T, Costabel U. Distribution of asbestos bodies in the human lung as determined by bronchoalveolar lavage. Am Rev Respir Dis 1993; 147: 1211-1215. 19. Gylseth B, Churg A, Davis JGM, et al. Analysis of as bestos fibers and asbestos bodies in tissue samples from human lung. Scand J Work Environ Health 1985; 11: 107-110. 20. Teschler H, Friedrichs KH, Hoheisel B, et al. Asbestos fibers in bronchoalveolar lavage and lung tissue of for mer asbestos workers. Am J Respir Crit Care Med 1994; 149: 641-645. 21. Dumortier P, De Vuyst P, Strauss P, Yernault JC. Asbestos bodies in bronchoalveolar lavage fluids of brake lining and asbestos cement workers. Br J Ind Med 1990; 47: 91-98. 22. Churg A, Wiggs B. Fiber size and number in workers exposed to processed chrysotile asbestos, chrysotile min ers, and the general population. Am J Ind Med 1986; 9: 143-152. 23. Pooley FD. An examination of the fibrous mineral con tent of asbestos lung tissue from the Canadian chrysotile mining industry. Environ Res 1976; 12: 281-298.