Annals of Occupational Hygiene Advance Access originally published online on April 24, 2007
Annals of Occupational Hygiene 2007 51(5):495-500; doi:10.1093/annhyg/mem014
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Measurement of Asbestos Fibre Concentrations in Fluid of Repeated Bronchoalveolar Lavages of Exposed Workers
1 Section of Occupational Medicine and Toxicology, Department of Clinical Medicine and Immunological Sciences, University of Siena, Italy
2 Laboratory of Public Health, U.F. Occupational Hygiene and Toxicology AUSL n. 7 Siena, Italy
3 Department of Quantitative Methods University of Siena, Italy
* Author to whom correspondence should be addressed. Tel: +39 0577 586755; fax: +39 0577 586159; e-mail: sartorelli{at}unisi.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objectives: The aim of this study was to assess the reliability of asbestos fibre concentration in bronchoalveolar lavage fluid (BALF) by carrying out the mineralogical analysis of BALF at different times in the same patient and comparing the results.
Methods: Twenty two patients underwent diagnostic fibreoptic bronchoscopy twice: the first was to assess the past asbestos exposure and the second for different clinical reasons. Mineralogical analysis of BALF was carried out.
Results: In 16 patients (72.7%), a reduction of concentration in BALF of all asbestos fibres was observed. The concentrations of both chrysotile and amphiboles in the first bronchoalveolar lavage (BAL) were related to their concentrations in the second BAL and the observed differences were not statistically significant. A significant decrease in asbestos body concentration between the first and the second BAL was found (Wilcoxon test, P < 0.01).
Conclusions: The reliability of the fibre concentration in BALF as a marker of past asbestos exposure seems quite good. In most cases, it allows us to distinguish workers in different classes of exposure and gives useful information on the pattern of exposure. Uncertainties related in general to lung residues and in particular to mineralogical analysis of BALF (mainly due to the high coefficient of variation (CV) at low fibre concentrations and the results of the statistical analysis on total fibres) suggest that this biomarker is more likely suitable for a qualitative/categorical approach to exposure assessment than a quantitative one.
Keywords: asbestos • bronchoalveolar lavage • mineralogical analysis • transmission electron microscopy (TEM)
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
There is a lack of information regarding asbestos exposure (i.e. airborne fibre concentration and exposure pattern) in many working places. Exposure data on asbestos mines and manufacturing facilities producing of brake lining, clothes, ropes and asbestos-cement is available (Health and Safety Commission, 1979; Selles et al., 1984). There is little information on other working places such as the building industry and railway rolling stock production. Asbestos exposure has been intermittent for many workers such as welders, electricians and maintenance workers in shipbuilding yards, iron foundries and the steel industry. On the other hand, the existence of unusual and occult occupational exposures has been proved (Corhay et al., 1990; De Vuyst et al., 1997). Very often, the quality of quantitative exposure data is poor (Dupre et al., 1984). Consequently, the available environmental monitoring data is useful to quantitative risk assessment in epidemiological studies but very often is not sufficient to characterize individual exposure.
The measure of the fibre load of lung tissue using electron microscopy can only be performed after open lung biopsy, lung surgery or death. When this is not possible, mineralogical analysis of bronchoalveolar lavage fluid (BALF) by electron microscopy should represent the most suitable method for assessing asbestos exposure remembering that the presence of asbestos bodies (AB) does not reflect the exposure to chrysotile.
Fibre concentration in BALF has been the subject of considerable study as an asbestos fibre burden marker (Sartorelli et al., 2001), but it has been claimed that there is no standardized and validated approach to mineralogical analysis (Frank, 1995). In a previous study (Sartorelli et al., 2001), mineralogical analysis of BALF of 108 exposed workers and 57 non-professionally exposed patients showed a significant difference in the two populations, being positive in all exposed subjects. These results were confirmed in a larger population of 193 exposed workers and 84 non-professionally exposed subjects (Romeo et al., 2004). If fibre concentration in BALF allows us to distinguish professionally exposed populations from the non-professionally exposed, its suitability for exposure assessment of individual cases is not certain. In particular, it could suffer from a low repeatability, both for biological and analytical factors, with a variability of results in the same subject as a consequence.
A group of workers previously exposed to asbestos was studied on two different occasions by carrying out two bronchoscopies on the same patient. The aim was to assess the reliability of the asbestos fibre concentration in BALF as biomarker of past asbestos exposure in individual cases by carrying out the mineralogical analysis of BALF at different times on the same subject and comparing results: the absence of a significant difference between the first and the second measurement would demonstrate a good repeatability of the test.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Subjects
To assess the past asbestos exposure, mineralogical analysis of BALF was carried out in 193 occupationally exposed workers hospitalized in the last 12 years in the Section of Occupational Medicine and Toxicology of the University of Siena for suspected asbestos-related diseases. Some years after the first bronchoalveolar lavage (BAL), 22 patients underwent diagnostic fibreoptic bronchoscopy a second time for different clinical reasons. These subjects were generally patients suspected of having lung cancer or afflicted with fibrosis of the lung. A written informed consent to perform BAL was obtained from all patients. In this population, measurement of AB and uncoated fibre concentrations in BALF was repeated a second time. A detailed occupational history was obtained with a questionnaire. According to the occupational history, four main industrial activities were highlighted: (i) electric energy production (36.4%), (ii) railway rolling stock production (22.7%), (iii) shipbuilding yards (9.1%) and (iv) steel industry (9.1%). The others (22.7%) worked in a variety of fields.
When the 22 studied subjects (all males) underwent the first bronchoscopy, their exposure to asbestos had ceased. The mean age at the moment of the first BAL was 53 ± 7.93 years (median 56, range 41–63); 9.1% were smokers, 54.5% ex-smokers and 36.4% non-smokers. At the moment of the second BAL, the mean age was 57.22 ± 7.95 years (median 60.5, range 43–67) with the same percentage of smokers, ex-smokers and non-smokers. The mean lag time between the first and the second BAL was 4.0 ± 2.3 years (median 4, range 1–10).
On the basis of the radiological and clinical study, four categories of diagnosis were considered: (i) asbestosis (4%), (ii) pleural plaques (57%), (iii) asbestosis and pleural plaques (27%) and (iv) subjects with non asbestos-related diseases (12%) such as chronic bronchitis.
Bronchoscopy and lavage were performed in the right middle lobe or in the lingula by instilling normal saline solution in four 50 ml aliquots followed by immediate withdrawal (Begin et al., 1985; De Vuyst et al., 1987; Dodson et al., 1991).
AB counting by light microscopy and fibre counting and analysis by TEM
AB were counted with a phase contrast microscope at a magnification of x250 following the method described by De Vuyst et al. (1987). Ten ml of BALFwere used for asbestos fibre counting with transmission electron microscopy (TEM) at a magnification of x10 000. The organic material of BALF was dispersed by adding sodium hypochlorite and hydrogen peroxide for 24 h. The samples were filtered onto a 0.45 µm Millipore filter. All reagents (sodium hypochlorite, hydrogen peroxide, water) were pre-filtered onto a 0.22 µm Millipore filter to avoid asbestos contamination. The sample was treated using the Dodson method (Dodson et al., 1991). Only particles with an aspect ratio of 3:1 or more were counted, not using the lower limits of length and diameter. The detection limit was <102 fibres per ml of BALF.
The method used to analyse asbestos fibres in BALF was tested for its applicability to samples with low concentrations of fibres, such as those in subjects with little or no occupational exposure to asbestos (Scancarello et al., 1997). Analytical reproducibility was evaluated in three samples with high, medium and low concentrations and in 10 aliquots of BALF with medium-low concentrations of fibres. The coefficient of variation (CV) was 12.7% for samples with high concentrations of asbestos fibres and a satisfactory dispersion was found for samples with intermediate concentrations (CV = 31.3%). For samples with concentrations of total asbestos fibres, whether chrysotile or amphibole, which were very low, the coefficient of variation reached 65%. Although this is a high dispersion, it was possible to distinguish populations of exposed and non exposed subjects clearly at concentrations close to the detection limits.
Statistical analysis
Data is made up of observations corresponding to two couples of paired variables (i.e. the concentration of chrysotiles in the fluid of the first and the second BAL and the concentration of amphiboles in the fluid of BAL 1 and BAL 2). The paired data were reduced to differences as usual. These differences (corresponding to chrysotiles and amphiboles, respectively) constitute a set of bivariate data and the corresponding scatterplot is reported in Fig. 1
|
The biomarker turns out to be robust if the bivariate difference distribution is scattered about the axis origin (i.e. the measurements remain stable between BAL 1 and BAL 2). This statement implies that the marginal difference distributions for chrysotiles and amphiboles are scattered about the zero. In the classical inferential approach, the null hypothesis of a difference distribution scattered about the origin actually reduces to assess that the mean vector of the difference distribution is given by the null vector. However, the use of standard bivariate testing techniques commonly adopted with bivariate paired data (such as the paired T-Hotelling test) is precluded, since they assume the bivariate normality of the observations. A preliminary data screening (Fig. 1) shows that the bivariate normality for the differences barely holds, since the data are not scattered according to a "cloud" with the typical elliptical shape. Therefore, a sign-permutation procedure as suggested by Pesarin (2001) was suitable in this case. This permutation test aims to assess that the median vector of the difference distribution is given by the bivariate null vector. As usual in nonparametric statistics, the hypothesis deals with the median vector rather than the mean vector, since the mean vector may not exist for some distributions. In addition, this nonparametric technique avoids the bivariate-normality assumption and allows handling at the same time the marginal hypotheses (i.e. the two marginal difference distributions are marginally scattered about the zero) and the joint hypothesis (i.e. the joint difference distribution is scattered about the axis origin). Using this procedure, the hypothesis of a null median vector is decomposed into the intersection of two marginal hypotheses, each corresponding to the hypothesis of a null median for each marginal distribution. Accordingly, in a first stage, two sign-permutation tests are separately carried out for chrysotiles and amphiboles in order to assess that the differences distributions have marginal null medians. In a second stage, the P-values corresponding to each marginal test are combined in an overall test statistic. In turn, the significance of the overall test statistic is computed by means of the same permutations performed on the data. In this way, the dependence structure of the marginal statistics is non-parametrically captured by the permutation procedure. Hence, if the overall hypothesis is rejected, one can assess the marginal hypothesis on which the rejection depends.
AB concentrations in BAL 1 and BAL 2 were compared using nonparametric tests for paired data (Wilcoxon test, sign-permutation test). The analysis was performed on 17 subjects, since for 5 patients concentration either for AB in BAL 1 or AB in BAL 2 (or both) were missing.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The mineralogical data (AB, chrysotile and amphiboles concentrations) of repeated BAL is summarized in Table 1. In five cases, samples were not suitable for AB analysis and in three AB concentration was not revealed (<0.5 AB ml–1).
|
Fibre concentration detection limit was up in all the analyzed BALFs. The scatterplot for the measurements corresponding to the concentration of chrysotile in the fluid of first and second BAL is considered in Fig. 2. It is at once apparent that a linear dependence between the variables exists. The squared correlation coefficient is given by R2 = 0.63. Moreover, the coefficients of the regression line are given by
= 109.96 and 
= 0.63 (where and represent the estimated intercept and the slope of the regression line respectively), or equivalently the concentration of chrysotile in second BALF is likely to assume values which are about the 60% of the concentration in the first BALF.
|
As to the concentration of amphiboles in the fluid of first and second BAL, the scatterplot for the measurements corresponding to these variables is considered in Fig. 3. It follows that a linear dependence between the variables exists. In this case, the squared correlation coefficient is given by R2 = 0.64. As the coefficients of the regression line are given by
= 225.10 and = 0.55, the concentration of amphiboles in the second BALF is likely to assume values which are about the 60% of the concentration in the first BALF. The scatterplot for the different measurements is given in Fig. 1.
|
For the present data, the analysis shows that the hypothesis of a null median vector (i.e. the fibre concentration in BAL 1 and BAL 2 does not significantly differ) may be partially questionable since the global observed P-value was P = 0.05. However, as to the marginal analyses, the P-values were 0.19 for the chrysotiles and 0.10 for the amphiboles. These results are not conflicting and they may be explained by means of an exploratory analysis of Fig. 1. The ordinates axis approximately halves the data points into two sets and the abscissas axis approximately halves the data points into two sets. However, the data points are not evenly scattered in the four quadrants, since the first quadrant contains a slightly larger number of data points compared to the other quadrants. Hence, a positive difference for chrysotiles tends to arise in connection with a positive difference for amphiboles. Thus, the differences of chrysotiles and amphiboles have marginal distributions evenly scattered about the zero (i.e. the measurements marginally remain stable between BAL 1 and BAL 2) and the marginal P-values are not significant. However, the differences tend to be slightly scattered in the first quadrant when they are jointly considered and the presence of this mild dependence gives rise to a P-value, which is about 5%. Hence, even if the results are weakened by the joint analysis, there is not enough empirical evidence against the null hypothesis.
In 16 patients (72.7%) who underwent bronchoscopy twice, a reduction of concentration in BALF of all asbestos fibres was observed. Even if these measurements are influenced by the CV of the used methodology and the differences are not significant (P = 0.05), this trend could be referred to the natural clearance of the lung. In order to explore if the differences in fibre concentration depend on the lag time between BAL 1 and BAL 2, two regressions for chrysotiles and amphiboles were carried out. From an exploratory analysis of the scatter-plots (Figs 4 and 5), it is at once apparent that no dependence exists. Actually, the R-squared coefficients are approximately null both for the regressions of chrysotiles and amphiboles.
|
|
Finally, the one-sided Wilcoxon test demonstrated a significant decrease in AB concentration between the first and the second BAL (P < 0.01). The same result may be achieved by adopting a sign-permutation test for paired data, which even provides a smaller P-value.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The concentrations of both chrysotile and amphiboles in the first BAL are related to their concentrations in the second BAL. If one only considers the chrysotile fibres, 63.4% of the cases present a reduction in the concentration of the second BAL compared to the first, whereas for the amphiboles the reduction occurred in a percentage equal to 61.5% of cases. However, the observed differences were not statistically significant and cannot be statistically explained on the basis of the natural clearance of the lung because they are not related to the lag time between the first and the second BAL. In the specific case it seems that the lung clearance does not influence fibre concentrations in BALF while it could be responsible for the significant reduction of AB concentrations between BAL 1 and BAL 2. The lung clearance of AB seems to differ from fibres, their alveolar content decreasing in the long run. After a few months, the concentration of AB in lung specimens from autopsy decreases (Mollo et al., 2000) and this could also happen in vivo. It is possible that the mean lag time between the first and the second BAL (4 years) was too short to influence the fibre concentration while AB are cleared relatively rapidly. However, lung clearance and lung residues are very complicated issues (Howie, 2005; Rogers, 2005) and need a specific research work while the aim of the study was only to test the repeatability of mineralogical analysis of BALF. In any case, even if the decrease in AB concentration from the first to the second BAL is influenced by the lung clearance, considering that in two subjects AB were under the detection limit in the first BAL and positive in the second, the repeatability of the test is not as good as that of mineralogical analysis by electronic microscopy.
As far as the total number of fibres is concerned, a reduction of the concentrations can be observed in 72.7% of cases with a difference between the two BALF at the limit of the statistical significance (P = 0.05). Thus, there is no empirical evidence against the hypothesis that fibre concentration in BALF is a reliable biomarker, even if the results are weakened by the joint analysis. Hence, the statistical analysis simply suggests that the biomarker seems reliable, even if some caution should be adopted in its practical use.
For this reason, the effect of the total fibre concentration variation between BAL 1 and BAL 2 in terms of fibre burden assessment was evaluated in the individual cases. Considering the non-professionally exposed subjects (84 subjects, 65 males and 19 females) of all the studied population the identification and quantification of uncoated fibres with TEM shown a value of 650 ff ml–1 of BALF as the maximum concentration, while the higher confidence limit was 182 fibres ml–1 (Romeo et al., 2004). Following these results, three classes of exposure were defined on the basis of all asbestos fibre concentrations in BALF to compare exposed workers to the general population:
- >650 ff ml–1 (certain occupational exposure to asbestos),
- <650 ff ml–1 and >182 ff ml–1 (uncertain occupational exposure to asbestos),
- <182 ff ml–1 (non occupationally exposed subjects).
In 16 subjects, fibre concentration was >650 ff ml–1 both in the first and in the second BALF. In four patients, fibre concentration was <650 and >182 ff ml–1 both in the first and in the second BALF. In only two cases (9.1%), the class of exposure changed with the second BAL (from 991 to 546 and from 607 to 765, respectively). In no case were there concentrations inferior to 182 ff ml–1.
Fibre concentration in BALF seems a reliable marker of past asbestos exposure because its repeatability is good considering chrysotile and amphiboles separately and sufficient for total fibre concentration. In most cases, it allows one to distinguish workers in different classes of exposure giving useful information on the pattern of exposure.
Although specificity of AB is high, they are a less sensible marker of asbestos exposure: (i.e. the lack of AB does not necessarily exclude exposure). Moreover, there is a poor correlation between AB and exposure to chrysotile. So electron microscopy research on fibres can give more precise information on exposure.
Uncertainties related in general to lung residues and in particular to mineralogical analysis of BALF (mainly due to the high CV at low fibre concentrations and the results of the statistical analysis on all asbestos fibres) suggest that this biomarker is more likely suitable for a qualitative/categorical approach to exposure assessment than a true quantitative one (e.g. models based on such a data do not seem appropriate). Asbestos exposure assessment of individual cases should be based on the occupational history and the available information on exposure levels in the workplace, compared to AB and fibre concentration in the lung tissue or in BALF when available.
|
FOOTNOTES |
|---|
The originally published version of this paper in the Annals Occup. Hyg. (Vol.51 n. 5, pp. 495-500) was incorrect. The spelling of BRO-CHOALVEOLAR is wrong and should be BRONCHOALVEOLAR.
Received April 18, 2006; in final form February 7, 2007
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Begin R, Cantin A, Berthiaume Y, et al. Clinical features to stage alveolitis in asbestos workers. Am J Ind Med (1985) 8:521–536.[CrossRef][Web of Science][Medline]
Corhay JL, Delavignette JP, Bury T, et al. Ocult exposure to asbestos in steel workers revealed by bronchoalveolar lavage. Arch Environ Health (1990) 45:278–282.[Web of Science][Medline]
De Vuyst P, Dumortier P, Moulin E, et al. Diagnostic value of asbestos bodies in bronchoalveolar lavage fluid. Am Rev Respir Dis (1987) 136:1219–1224.[Web of Science][Medline]
De Vuyst P, Dumortier P, Gevenois PA. Analysis of asbestos bodies in BAL from subjects with particular axposures. Am J Ind Med (1997) 31:699–704.[CrossRef][Web of Science][Medline]
Dodson R, Garcia J, O'Sullivan M, et al. The usefulness of bronchoalveolar lavage in identifying past occupational exposure to asbestos: A light, electron microscopy study. Am J Ind Med (1991) 19:619–628.[Web of Science][Medline]
Dupre JS, Mustard JF, Uffen RJ. Report of the Royal Commission on matters of health and safety from the use of asbestos in Ontario. Vol. 1. (1984) Ontario: Ministry of Government Services Publications Services Branch.
Frank AL. Asbestos mineralogic analysis as indicator of carcinogenic risk. Med Lav (1995) 86:490–495.[Medline]
Health and Safety Commission. Asbestos Vol. 1: Final report of the advisory committee. (1979) London: Her Majesty's Stationary Office.
Howie R. Letter to the editor. Asbestos lung residue and asbestosis risk. Ann Occup Hyg (2005) 49(1):95–97.
Howie R. Letter to the editor. Asbestos lung residue and asbestosis risk. Ann Occup Hyg (2005) 49(4):364–365.
Mollo F, Cravello M, Andreozzi A, et al. Asbestos body burden in decomposed human lung. Am J Forensic Med and Path (2000) 21:148–150.[CrossRef]
Pesarin F. Multivariate permutation tests with applications in biostatistics. (2001) New York: Wiley.
Rogers A. Letter to the editor. Asbestos lung residue and asbestosis risk. Ann Occup Hyg (2005) 49(4):363–364.
Romeo R, Scancarello G, Cassano P, et al. Stima dell’esposizione ad asbesto mediante analisi mineralogica del liquido di lavaggio broncoalveolare. Med Lav (2004) 95:17–32.[Medline]
Sartorelli P, Scancarello G, Romeo R, et al. Asbestos exposure assessment by mineralogical analysis of bronchoalveolar lavage fluid. J Occup Environ Med (2001) 43:872–881.[Web of Science][Medline]
Scancarello G, Romeo R, Marcianò G. Variabilità analitica nella determinazione delle fibre di asbesto nel liquido di lavaggio broncoalveolare. Minoia C, Scansetti G, Piolatto G, Massola A, eds. (1997) L’amianto dall’ambiente di lavoro all’ambiente di vita . Nuovi indicatori per futuri effetti. Pavia Fondazione Salvatore Maugeri IRCCS; I Documenti – 12, 393–398.
Selles DJA, Isserow LW, Robok K. (1984) Results of sampling in mining and non mining areas. In: Proceedings of the 6th International Conference on Air Pollution 1984. Pretoria. 23–25 October, South Africa.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||