Document N59KNZ5R8eDgObNMrgX9KdZD

^ Fiber Burden and Patterns of Asbestos-related Disease W ln Workers with Heavy Mixed Amosite and Chrysotile Exposure ANDREW CHURG and SVERRE VEDAL Departments of Pathology and Medicine, University of British Columbia, Vancouver, British Columbia, Canada To attempt to determine the mineralogic factors that relate to the appearance of specific types of asbestosrelated disease in workers with heavy mixed exposure to amphiboles and chrysotile, we analyzed the pul monary,asbestos fiber burden in a series of 144 shipyard workers and insulators from the Pacific North west. Amosite was found in all lungs, and tremolite and chrysotile in most lungs, but the vast majority of fibers were amosite. Tremolite and chrysotile concentrations were significantly correlated, indicating that the tremolite originated from chrysotile products, but no correlation was found between tremolite orchrysotile concentration and amosite concentration. Time since last exposure was correlated with decreasing amosite concentration and the calculated clearance half time was about 20 yr, in a multiple regression analysis that accounted for the presence of more than one disease in many subjects, a high concentration of amosite fibers was correlated with the presence of airway fibrosis and asbesiosis, whereas subjects with mesothelioma, lung cancer, pleural plaques, or no asbestos-related disease had about the same, much lower, amosite concentration. No relationship was found between the concentration of chrysotile or tremo lite and any disease. Analysis of fiber size measures (length, width, aspect ratio, surface, mass) showed that pleural plaques were strongly associated with high aspect ratio amosite fibers and suggested that mesotheliomas were associated with low aspect ratio amosite fibers. We conclude that, in this population, the major residual fiber is amosite, and only amosite concentrations correlate with the presence of specific diseases, raising questions about the role of chrysotile in disease induction. There are distinct differences in the relationship of fiber burden and disease comparing workers with heavy amosite exposure to chryso tile miners and millers; in particular, mesothelioma appears at much lower amosite burdens than does asbestosis, in contrast to the situation previously reported for chrysotile-induced mesothelioma. Amosite clearance from the lung is extremely slow, requiring decades. Except for pleural plaques, the association of fiber size and disease remains uncertain. Churg A, Vedal S. Fiber burden and patterns of asbestosrelated disease in workers with heavy mixed amosite and chrysotile exposure. Am J Reapir Crit Care Med 1994;150:663-9. That heavy occupational exposure to asbestos produces a vari ety of pleuropulmonary diseases is well established, but much less information is available about the relationship of pulmonary asbestos fiber burden, taken in its broadest sense of fiber con centration, fiber type, and fiber size, and the appearance of spe cific diseases. A variety of studies have investigated this issue (1--6), but most have considered only the issue of fiber concentraiion, and chrysotile has not always been separated from amphibole. Furthermore, many such studies compare exposed work ers with nonexposed controls, and this approach may yield misleading information, particularly in respect to fiber size differ ences (see Discussion). In this study we examine a large group of shipyard and insula- (Received m original form September 27. 199} ond in rented form February 2S, 1994) Supported by Grants MA6907 and MA7820 from the Medical Research Coun cil ol Canada and grants from the National Cancer Institute of Canada. Correspondence and requests for reprints should be addressed to Andrew Churg, M.D.. Department of Pathology, University of British Columbia. 2211 Wesbrook Mall, Vancouver, 8C V6T 285, Canada. Am f Respir Crit Care Med Vol 550. pp 663-669, 1994 tton workers and workers in related trades with heavy mixed amphibole and chrysotile exposure to ask the question: Within a group of heavily exposed workers, what are the mineralogic mea sures that correlate with the appearance of specific diseases? We also compare our results with data from a previous, similarly de signed, study of chrysotile miners and millers, and with various fiber size-related predictions from the experimental and human literature. METHODS Casas for this study were selected from a series ol approximately 160 autopsy lungs referred to our laboratory tor mineral analysis (and, in many cases, for pathologic diagnosis as well). These cases constituted all the cases received and analyzed by our laboratory from 1981 through 1991 for which the following criteria were met. All of the cases were from the Pacific Northwest (Oregon, Washington, British Columbia) and all of the patients had been at some time in their careers shipyard workers, asbes tos insulators, or in related trades such as pipefitters. Approximately hall the cases were medical-legal referrals and half were initially referred by pathologists, although after completion of diagnoses and mineral analy sis. the majority of cases referred by pathologists became medical-legal cases. We were thus able to obtain detailed occupational and smoking k HWBUI0010306 664 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VQL ISO . variable TABLE i DEMOGRAPHIC AND DISEASE OATA N Mean SO Pang Al! subjects. N * 144 Sex 134M, 10F Ago Smoking, pack-years Exposure, yr Latency, yr Years since lest exposure 136 126 129 114 113 66 30 20 43 25 Subjects without asbestos-induced disease. N - 8 Ssx 8M Ago 8 72 Smoking, pack-years 6 37 Exposure, yr 6 33 Latency, yr 5 47 Years since last exposure 5 14 Disease No. Subiects 9.8 33 15 78 15 37-86 0-150 1-50 20-64 1-47 8.8 60-84 10 20-50 8.7 19-43 9,4 38--63 63 5-20 No. Subjects Having Only This Disease Asbestosis Airway fibrosis Mesothelioma Lung cancer Pleural plaques No asbestos disease 23 16 83 32 103 S 3 0 13 12 9 Plaques Pius Asbestosis Airway fibrosis Mesothelioma Lung cancer Subjects with multiple diseases No. Lung Cancer Plus No. Mesothelioma Plus 17 Asbestosis 14 Airwayfibrosis 69 16 8 Asbestos's 5 Airway fibrosis No. 5 9 histories from legal records in about 75% of the cases. Historic data tor the remaining cases were obtained from referring pathologists or clini cians. All of the patients were either known lo have been exposed to mix tures of amphibole (amosite) and chrysolite, or presumed lo be so ex posed because of general historic information about the types and levels of asbestos exposure encountered in these specific occupations. The lungs for each case were reviewed by standard gross and micro scopic techniques to determine the presence of each of the asbestos- related diseases listed in Table 1. In most cases four standard histologic sections (two upper lobe, two lower lobe) were prepared. A number of cases were excluded because we could not determine whether pleural plaques were present from the pathologic material and neither autopsy nor radiologic data could be obtained to answer this question, leaving 144 cases for the study. Asbestos-induced airway fibrosis as used here denotes a specific pat tern of asbestos-induced fibrotic thickening of the walls of membranous and respiratory bronchioles {see ref. 7 for a complete description and illus trations of this lesion), and a case was considered to have airway fibrosis if any airways fitting this pattern were observed. Asbestosis is used to refer to diffuse interstitial fibrosis affecting alveolar walls along with the presence of at least one asbestos body in the tissue sections; airway dis ease by itself was not considered asbestosis. Only one case showed both of these lesions. For each case at least one 5-g sample of formalin-fixed iung tissue was dissolved in bleach, and the mineral fibers collected for analytical electron microscopy. In general if only one sample was analyzed it was taken from the peripheral upper lobe, and it two were analyzer}, then both peripheral and central upper lobe were used. An additional piece ol lung was dried to constant weight to allow expression of results m terms of fibers per gram dry lung. For each sample, a minimum of SO randomly selected grid squares were examined and all asbestos fibers longer than 0.S iim found were counted, measured, and identified using a combina tion of fiber morphology, and fiber chemistry as determined by energydispersive x-ray spectroscopy. Fiber concentrations were calculated using an algorithm relating weight of lung tissue used and number ol grid squares examined. Crocidolite fibers were found in only a very lew cases, usually in quite small numbers, and have been excluded from all analyses. The following fiber characteristics were included in lha analyses: fi ber type, fiber concentration, and mean fiber size (length, width, aspect ratio, surface area. mass). Initial inspection showed that fiber concentra tion was fog-normally distributed, and statistical analyses were performed using log-transformed values. Geometric means were calculated for pre sentation of the data on concentration. The fiber size data were best nor malized using a square root transformation and the data are presented using means derived from the square roots. Pearson correlation coeffi cients were used to assess ihe association between two continuous vari ables. For each fiber type, concentration or size values for subjects with each specific disease were initially compared with subjects without any disease using t tests. However, because most cases had more than one disease, multiple regression models in which the asbestos measure of interest was regressed on indicator variables for each of the diseases plus any relevant covariates were constructed. All statistical calculations were performed using SYSTAT (8). RESULTS Demographic data and the number of subjects in the sample with specific asbestos-related disease are shown in Table 1, along with data on the subjects with no asbestos-related disease. Exposure to asbestos was generally long (mean 20 yr, range less than 1 to 50 yr); no actual exposure measurements were available, but historic data indicate that exposure levels in many instances were high. Most subjects had more than one asbestos-related disease, as listed in Table 1. Table 2 shows correlations of asbestos exposure with log fiber concentrations. Amosite concentration decreased with time since last exposure, but was not associated with length of exposure or exposure latency, or with age. The amosite tissue half-life, esti mated from a regression of concentration on time since last expo sure, was 20 yr. Neither tremolite nor chrysolite concentration was associated with any measure of asbestos exposure. Table 3 lists correlations between the concentrations of the various fiber types; A moderately strong correlation (R = 0.37, p < 0.001) was seen between tremolite and chrysotile concentrations, but there was no correlation between tremolite or chrysotile concentration and amosite concentration. Amosite fibers were found in ail 144 cases, chrysotile fibers in 102 cases, and tremolite fibers in 125 cases. Mean fiber con centrations in subjects with the various asbestos-related diseases were compared with those with no asbestos-related disease for amosite (Table 4), chrysotile (Table 5), and tremolite (Table 6). Where no chrysotile or tremolite was detected, a concentration I TABLE 2 CORRELATIONS (p VALUE) OF ASBESTOS EXPOSURE AND FIBER CONCENTRATION Age Exposure Latency Time Since Last Exposure Log amostte Log iremeiite Log chrysotile -0.12(0.15) - 0.02 (0.85) 0.03 (0.71) 0.12 (0.18) 0.09 (0.31) 0.07 (0.47) -0.15 (0.12) 0.14 (0.13) -0.02 (0.83) -0.28 (0.002) -0.05 (0.59) -0.11 (0.24) HWBUI0010307 Churg jnd VecJdi; Amosite Asbestos and Disease 'ABLE 3 correlations of concentrations of different fibers Tremoete `chrysotile Amosite'cftrysotile Amosiie'Iremolite R . 0 37 p < OOOI R - o so p - o.Ji R = 0.02 p = 0 82 ol zero was assumed. Subjects with asbestesis had significantly higher amosite concentrations than subjects with no asbestosrelated disease, and this also appeared to be true for subjects with airway fibrosis. Multiple linear regression models, in which fiber concentrations were regressed on indicator variables for each asbestos-related disease, were generated in an attempt to deter mine associations between concentration and disease that were independent of the presence of other asbestos-related diseases. These models confirmed the association of asbestosls and air- _ 66a way fibrosis with high amosite fiber burdens {Tables 4 and 7). No correlations were found between chrysolite or tremolite concen tration and the presence of any disease; the few values which appeal significant in Table 5 actually reflect higher concentrations in the subjects without disease compared with those with disease. Table 7 shows the regression coefficienls and standard errors lor these models. We attempted to examine the effects of long versus short fibers of amosite using an arbitrary definition of "long" as greater than 8 pm (see Discussion), and regressing the concentration of long or short fibers on the same disease indicator variables used above. Because of the very high correlation (r = 0.99) between concen trations of long and short fibers, this procedure produced results essentially identical to those found using total fiber concentrations. The association of amosite fiber length and aspect ratio with the asbestos-related diseases is shown in Tables 8 and 9. Fiber TABLE 4 GEOMETRIC MEAN AMOSITE CONCENTRATIONS IN SUBJECTS WITH ASSESTOS-RELATEQ DISEASE COMPARED WITH SUBJECTS WITHOUT ANY OISEASE' Disease Asbsstosis Airway fibrosis Mesothelioma Lung cancar Pleural plaques Cases With Disease H Geometric Mean SD 23 10 6.6 18 4.3 12 83 0.S6 7.6 32 1.1 8.3 103 1.4 7.8 Cases Without Disease N Geometric Mean SD 8 0.67 5.1 8 0.67 5.1 8 0.67 5.1 8 0.67 5,1 8 0.67 5.1 l Teslt P> < o.oot 0.072 0.74 0.53 0.31 Multiple Regression Model t (PI < 0.001 0.001 0.21 0.20 0.08 Concentration values in millions ol Stems dry lung. t Comparison ol cases win specified disease to tries* with no disease by r test. t Multiple /egression mojet accounting tor toe presence of more Irian one disease in most subjects. TABLE S GEOMETRIC MEAN CHRYSOTILE CONCENTRATIONS IN SUBJECTS WITH ASBESTOS-RELATEO DISEASE COMPARED WITH SUBJECTS WITHOUT ANY DISEASE' Cases With Disease Cases Without Disease Disease N Geometric Mean SD N Geometric Mean SD Asbestosls Airway fibrosis Mesothelioma Lung cancel Pleural plaques 23 18 63 32 103 0.005 0 0.004 0.031 0.004 300 1006 2S0 150 250 8 8 8 8 a 0.035 0.035 0.035 0.036 0.035 81 SI 81 31 01 * COAcentrsiten valinM in miWon* f fibera/g dry lung. t Comparison ol easea with apeetfted titan*** to ttfosut with no dbaas* by t tost. * MuJUpte regression model accounting for !h# presence ol more than one disease m most subjects. f Test* (P) 0.36 0.006 0.29 0.95 0.29 Regression! Mode! (P) 0.38 0.06 0.67 0.30 0.30 TABLE 6 GEOMETRIC MEAN TREMOLITE CONCENTRATIONS IN SUBJECTS WITH ASBESTOS-RELATED DISEASE COMPARED WITH SUBJECTS WITHOUT ANY DISEASE' (Disease Asbesiosis Airway fibrosis Mesothelioma Lung cancer Pleural pisQues Cases With Disease N Geometric Mean SO 23 0.034 16 0.092 83 0.051 32 0.21 103 0.049 125 73 63 13 65 Cases Without Disease H Geometric Mean so 3 0.079 ss 3 0.079 ss 3 0.079 as 6 0.079 85 a 0.073 05 * Coneenuelion values in motions of fibers/? dry lung. t Comparison or cases witn specified disease to those no disease by t lest, t Multiple regression mode* accounting *or me presence of mote man one disease id most subjects, r (p> 0.67 0.93 0.78 0.43 0.76 Multipis Mode)* (p> 0.27 0.82 0.81 0.31 0.20 HWBUI0010308 666 AMERICAN IOURNAL OF RESPIRATORY AD CRITICAL CARE MEDICINE VOL I 50 1994 TABLE 7 LINEAR REGRESSION MODELS OF ASBESTOS FIBER CONCENTRATION ON ASBESTOS-RELATED DISEASES* Fiber Amosite Chrysolite Tremofitd Intercept 13.33 10.14 It.94 Asbestosis 2.59 (0.46) t -1.15 (1.31) -1.07 (0.96) Airway Fibrosw 1.70 (0.51 )f -2.77 (1.45)4 0.2S (1.06) Mesothelioma -0.55 (0.44) - 0.54 (1.25) -0 22(0.91) Lung Cancer -0.63 (0.49) 1.47 (1.39) 1.03(1.02) Pleural Plaques 0.69 (0.37)* -1.12 (1.07) -1.00 (0-78) r* 0.29 0.06 0 54 ' Conc8WaE><Hi data as 'bra/g dry rung. Gnrriet ir rag/essaon ecefficteffts 9t log fibf counts, with standard wrens at the coatfcwnts m pvsnuitisa, on eacft ot Ur independent vznadiee included in * f < 0.01. mocf$. * p < 0.10, TABLE 0 MEAN AMOSITE FIBER LENGTH IN SUBJECTS WITH ASBESTOS-RELATED DISEASE Oisease Asbestosis Airway fibrosis Mesothelioma Lung cancer Pleural plaques Cases With Disease Mean* H (pm) 23 6.4 16 6.0 83 5.5 32 5.7 103 5.9 Cases Without Dissass Mean* N pm) 8 4.6 8 4.6 8 4.6 8 4.6 8 4.8 l Testt (P) 0.18 0.29 0.48 0.39 0.32 Multiple Regression Modef$ (P) 0.33 0.64 0.32 0.73 0.16 ' Mean darned from square root transformed values. * Comparison of cases with specified disease to those with no disease by t tees using square root transformed values. * Muttfpte regretwo model accounting for the presence of more than one disease in moat subpets and using square root transformed values. TABLE 9 MEAN AMOSITE ASPECT RATIO IN SUBJECTS WITH ASBESTOS-RELATED DISEASE COMPARED WITH SUBJECTS WITHOUT ANY DISEASE Disease Asbestos Airway fibrosis Mesolhslioma Lung cancer Pleural plaques Cases With Disease N Mean* 23 45 10 37 83 36 32 39 103 40 Cases Without Disease N Mean* 8 36 3 36 8 36 8 36 8 36 f Test (P) 0.48 0.92 0.99 0.80 0.73 Multiple Regression Model* (P) 0.40 0.72 0.077 0.50 0.004 * Mean derived from square root transformed values. t Comparison et oases with spedfted disease to those wHh no disease by t test using square roe* transformed vahtaa. t Multiple regression model accounting for the presence of mere lhan one disease in mast subjects and using square root transformed vaiuas. length was not associated with disease. On initial comparison of the aspect ratio data (i-a.. analyzed by simple r test), no differ ences were seen because of confounding due to the presence of multiple diseases. The multiple regression analysis, which reflects the associations of each disease with aspect ratio inde pendent of associated diseases, indicated that the subjects with plaques had significantly higher aspect ratios than those without disease, and that the subjects with mesothelioma had near sig nificantly lower ratios. This latter conclusion is not apparent in Ta ble 8, but if one compares the mean aspect ratios in subjects with both mesothelioma and plaque (37, n = 69), subjects with plaque but not mesothelioma (47. n = 34), subjects with mesothelioma but not plaque (30, n 14), and subjects without disease (36, n ^ 8), then it becomes clear that the aspect ratios associated with plaque alone are higher and the aspect ratios associated with mesothelioma alone are lower than the aspect ratios in subjects with no disease. No associations between amosite fiber width, surface area. or mass and specific diseases were observed. Because of the low concentrations of chrysolite and tremolite fibers no attempt was made to evaluate associations between chrysolite or tremo lite size measures and disease. DISCUSSION In this study we have examined the relationship of a variety of fiber-related measures and specific asbestos-related diseases to attempt to determine which of these measures are responsible for the appearance of given diseases within a heavily exposed workforce. As noted in the introduction, this question has been studied by several groups, with somewhat inconsistent results. For both amphiboles and chrysotile there is general agreement that patients with asbestosishave the highest lung burdens (1-6), but no agreement about what size of fiber is related to asbestosis. Experimentally long fibers appear to be much more fibrogenic than short fibers (3-12). but in previous studies of both amosite- HWBUI0010309 , Cf'urg and Vedal: Amosite Asbestos and Disease and cbrysotile-inctaced asbestosis in humans (13, 14), we found that the concentration of short fibers correlated better with fibro sis grade than did the concentration of long fibers. Whether these findings reflect local interference with clearance mechanisms (15, 18), an effect at modifying agents such as cigarette smoke which particularly appears to increase short fiber retention (17), or whether the experimental data simply do not apply to humans, remains unresolved. There is also general agreement that, tor amosite- or crocidoiiteinduced disease, mesotheliomas and plaques appear at much lower burdens than does asbestosis (2-5). How cases of amphibote-induced mesothelioma and amphibole-induced plaques re late to each other in terms of fiber burden has been less clear (2-4). as, indeed, has the question of what factors discriminate between these two pleural diseases. The effect of fiber size on the induction of mesothelioma has been an area of considerable interest. Stanton and colleagues (18) originally proposed that, within a specific set at fiber lengths, relatively longer and thinner (ie., higher-aspect-ratio) fibers have a greater propensity to induce mesothelioma than do shorter and thicker (lower-aspect-ratio) fibers; experimental studies by Davis and colleagues (11,12) have suggested that fibers shorter than about 5 pm have little propensity to induce mesothelioma. Again, the question of how well such predictions apply to hu mans is unresolved. In a general sense, the reported human data on the ability of tremolita to induce mesothelioma are consistent with the Stanton hypothesis, because populations exposed to long high-aspect-ratio tremolite fibers (for example, those at the Libby vermiculite mine [19] and those in the Metsovo area of Greece [20D have much higher incidences of mesothelioma than do Que bec chrysotile miners and millers whose exposure is to shorter fibers with lower aspect ratios (23). McDonald and coworkers (21) and Rogers and coworkers (22) performed case control studies of patients with mesothelioma compared with control subjects; the former group found that the concentration of amphiboles (amo site, crocidolite, tremolite) longer than 8 pm was the best predic tor of mesothelioma, with no contribution by shorter fibers, whereas the latter group found that shorter fibers, including fibers of chryso tile, also appeared to be important. It should be noted that both these studies used a general pop ulation control group, and this may have influenced the results, because persons with occupational asbestos exposure generally have much longer fibers in their lungs than do persons in the gen eral population inhaling asbestos from ambient air (23). This obser vation also raises a related issue which returns to the basic ques tion raised in this study; namely, that while it is clear that persons with asbestos-induced pleuropulmonary disease have consider ably higher fiber burdens than members of the general popula tion, it is not clear whether fiber concentration differences, fiber dimension differences (24,25), or some other factors explain the occurrence of disease among members of an exposed workforce. In this report we have studied a workforce with heavy, but none theless quite variable, exposure to both amphiboles and chryso tile. The limitations of our approach must be stressed. The group is not ideal in many respects. No details of exposure levels are available tor these workers, and we were unable to correlate expo sure in years with measured fiber burdens. This is not surprising, because many members of the group had exposure for only a few years to the very high liber levels that prevailed inside ships during construction and ripout during World War II. A second potential problem is the selected nature of the sub jects for whom we obtained lung tissue, in our previous study of chrysotile miners and millers (7) we used sequential autopsy cases, and in the present study, sequential cases referred for mineral 667 analysis (and sometimes diagnosis). Both sets of cases tend to include those who have a disease which either is clearly com pensable, or will be if a certain diagnosis or a certain mineral bur den can be established. Thus neither group can be regarded as providing any useful data about the relative frequency of the vari ous asbestos-related diseases. However, we are interested in examining the relationship of disease to fiber concentration and size measures, and the latter data could not be known by those referring the cases. Thus selection per se should not affect our results. Selection does, however, limit the statistical power of the study. Most of the cases referred to our laboratory are referred because they have disease. Lungs from workers with this type of exposure and no disease are hard to obtain and thus the reference (no dis ease) group is quite small. Power calculations for amosite show that we would be unable to statistically detect anything less than a tenfold increase in fiber counts in the disease groups relative to the group with no disease. But the multiple linear regression analysis, by estimating differences in fiber counts between each disease and the no disease group independent of the associa tions of each disease with the other diseases, effectively enhanced the ability lo detect differences due to airway fibrosis and to pleu ral plaque by removing the effect of the diseases (in this case mesothelioma and lung cancer) that did not have higher fiber counts than the no disease group. It is clear that many of the dif ferences that are not statistically significant (for example, the amo site concentrations in subjects with mesotheliomas and lung can cers--see Table 4) are very small and are unlikely to be biologically important. Moreover, we are dealing with a population in which each sub ject tends to have more than one asbestos-related disease. This problem can be accounted for to some extent by multiple regres sion analyses, but might in theory be better approached using groups of subjects having only one disease. However, since expo sure sufficient to produce disease appears to produce multiple abnormalities in many patients, it is probably unrealistic to expect to find such a set of subjects. Indeed, although we and others have proposed a general schema which relates fiber concentra tion and disease pattern in amphibole-exposed workers (2-5), the failure to account for the effects of multiple diseases turns out to have created some subtle errors, as discussed subsequently. Lastly, we have deliberately chosen not to include exposure in our models. In part we have made this choice because there is, for reasons mentioned previously, no correlation between expo sure as measured in years and fiber concentration in this group of workers. More important, however, we have excluded years of exposure because we are specifically Interested in the effects of fiber burden, irrespective of nominal exposure, and disease. In this sense fiber burden serves as our measure of exposure, and including both fiber concentration and years of exposure appears to us to be analytically incorrect. Similar comments apply to years since last exposure. Our results show clearly that, despite known historic exposure to amosite and chrysotile, amosite is by far the predominant resid ual fiber, and there are correlations between amosite measures and disease. Chrysotile was present inconstantly and in relatively small amounts, and no correlations were found between chryso tile measures and disease. Similar observations have been made by others (5) and there is general agreement that chrysotile does not persist in lung tissue (1). For this reason analyses of this type by definition grossly underestimate chrysotile exposure, and there fore considerable caution must be exercised in drawing conclu sions about the effects of chrysolite. We have proposed elsewhere that tremolite, which is a natural HWBUI0010310 V 668 AMERICAN IOURNAI OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL ISO 1994 minoT constituent of chrysotile ore, might serve as a substitute marker for chrysotile. and that tremolite rather than chrysotile may actuaffy be the agent responsible for `chrysolite-induced" mesothe lioma and perhaps other diseases as well (6). In a previous study of fiber burden in chrysotile miners and millers (6), tremolite con centrations and sizes correlated well with the presence of spe cific diseases, whereas chrysotile measures did not provide cor relations after accounting for the presence of tremolite. The fact that there is a correlation between chrysotile and tremolite con centration in the present study confirms the idea that the tremo lite seen in these lungs is derived from the chrysotile, but we were unable to show any correlations between tremolite measures and disease. It is possible that this failure simply reflects very low con tamination of processed chrysotile products with tremolite, and hence very low lung burdens of tremolite. But the failure to find correlations of disease and either chrysotile or tremolite burden again raises questions about the role of chrysotile in workers with this type of mixed exposure. Our results emphasize the importance of high fiber concen tration in the induction of airway fibrosis and asbestosis by amo site; this conclusion is similar to that reached in our study of chryso tile miners and millers (6). However, there is a marked difference in the relationship of mesothelioma and fiber burden between these two studies. For amosite our data confirm in a formal fash ion that mesothelioma typically appears at much lower burdens than are seen with asbestosis or airway fibrosis, whereas our pre vious study showed that chrysctile-induced mesothelioma ap peared at very high fiber burdens of tremolite and chrysotile, bur dens comparable to those that produce asbestosis or airway fibrosis. These findings support th9 idea that, even where a role for chrysotile ore (chrysotile plus tremolite) is unequivocal, there are major differences in the relationship of fiber concentration and disease between these fibers and amosite or crocidolite. A new finding from the present work is that, in contrast to pre vious conclusions that we and others have reached that mesothe liomas appear at greater fiber burdens than do plaques (2-4), the current data suggest that fiber concentration does not discrimi nate among heavily exposed subjects with mesothelioma, lung cancer, or no asbestos-related disease, and the association of pleu ral plaques with increased amosite concentration is marginal. The differences between this study and previous ones most likely arise from the use of regression techniques that account for the pres ence of multiple diseases in some subjects. This observation also emphasizes the necessity to examine measures other than fiber concentration to attempt to determine which mineralogic measures are associated with different diseases. Measures of fiber size do provide some information in this re gard. In particular, there is an association of high aspect ratio with the presence of pleural plaques, perhaps implying that highaspect-ratio fibers specifically play a role in the genesis of plaques. Exactly the same association of plaques and high-aspect-ratio fibers was found for tremolite in our study of chrysotile miners and millers (6). suggesting the importance of aspect ratio in the production of pleural plaques with all types of asbestos. It is also clear that, in our study, mesotheliomas are not asso ciated with long fibers and in fact are probably associated with lower-aspect-ratio fibers than are found in subjects without asbestos-related disease. This is a surprising finding that contra dicts the experimental data cited previously and also the studies of McDonald and coworkers (21) and Rogers and coworkers (22) that concluded that mesotheliomas are associated with long fibers. The latter may possibly be explained by examining Tables 8 and 9. because, overall, subjects with asbestos-related disease tend. as a group, to have longer and higher-aspect-ratio fibers than those with no asbestos-related disease, and even our subjects with expo sure but no disease have longer fibers in their lungs than mem bers of the genera! population (23). the group used as control sub jects by McDonald and coworkers (21) and Rogers and coworkers (22). Nonetheless, we cannot explain the anomalous finding of low-aspect-ratio fibers in subjects with mesothelioma compared with exposed subjects with no disease, and again must ask whether the experimental data really apply to humans, and what specific role long, high-aspect-raiio fibers play in the genesis of mesothelioma. ft has also been proposed from experimental studies (26) and reviews of human data (15, 24, 25) that fiber surface area might be the most important predictor of some diseases. This idea makes intuitive sense if available surface reactants (for example, iron to catalyze the generation of active oxygen species) are important in producing disease. However, in this study we were unable to find an association of fiber width, surface area, or mass with disease. An additional point that emerges from our data is the slow clear ance of amosite from the lung. In a recent review (1), we noted that, despite marked interfaboratory variations in fiber counting methods and despite quite discrepant interlaboratory results for any given specimen (27), the available data suggest that clear ance half times for amosite and crocidolite are on the order of years to decades. The present calculation of a clearance half time of 20 yr for amosite fits into this pattern. In contrast, in most stud ies, including this one, there is no correlation between chrysotile concentration and time since last exposure when times are mea sured in years. This finding implies that the bulk of chrysotile clear ance must be completed within weeks to months of exposure. in one sense our results are disappointing because the mod els examining correlations with fiber concentration or with fiber size measures (and even models combining concentration and size measures) fail to provide fiber-related discriminating factors tor some diseases. In particular, it remains unclear what factors favor the development of amphibole-induced mesothelioma, a question of considerable importance. This issue needs further examination. References 1. Churg A, Wright J. Persistence of natural libers in lung tissue. Environ Hlth Perspectives (In Press). 2. Churg A. Analysis of lung asbestos content. Br J Ind Medicine 1991:48:649-52. 3. Roggli VL Human disease consequences of fiber exposures--a review of human lung pathology and fiber burden data. Environ Hlth Perspec tives 1990:58:295-303. 4. Roggli VL. Pratt PC, Brody AR. Asbestos content of lung tissue in asbes tos associated disease: a study of 110 cases. Br J Ind Medicine 1986;43:18-29. 5. Wagner JC, Newhouse ML, Corrin 8. Rosslter CER, Griffiths DM. Corre lation between fibre content of the lung and disease in East London asbestos factory workers. Br J ind Medicine 1988:45:305-8. 6. Churg A. Wright J. VedaJ S. Fiber burden and patterns of asbestos-related disease in chrysotile miners and millers. Am Rev Respir Ois 1993:148:25-31. 7. Wright JL, Cagle P. Churg A. Colby TV, Myers J. State of the art: disease of the small airways. Am Rev Respir Dis 1992;146:240-62. 8. Wilkinson L. SYSTAT: the System for Statistics. Evanston. IL: SYSTAT. Inc. 1991. . 9. Adamson IYR, Bowden DH. Response of mouse lung to crocidoliteasoes- tos. 1. Minimal fibrotie reaction to short fibers. J Pathol 1987:152:99- w. 10. Adamson iYR. Bowden DH. Response of mouse lung to crocidolite asbes tos. 2. Pulmonary librosis after long fibers. J Pathol 1987;152:109-17. 11. Davis JMG, Addison J. Bolton RE. Donaldson K. Jones AD. Smith T. t he pathogenicity of long vs short fibre samples of amosile asbestos a ro istered to rats by inhalation and intraperitoneal injection. BrJ Exp Path 1986:67: 415-30. HWBUI0010311 Churg and Veda): Aftiosite Asbestos and Disease 2. Davts JMG. Jones AD. Comparisons of the pathogenicity ol long and short fibre chrysolite asbestos m rats. Br J Exp Path 1906;69:717-37. 13. Churg A. Wright JL, OePaoli U Wiggs 8. Minerafogic correlates of fibro sis in chrysotile miners and millers. Am Rev Respir Ois 1989; 139:891-6. u. Churg A. Wright J. Wiggs 8. DePaoli L. Minerafogic parameters related lo amosite asbestos induced fibrosis in man. Am Rev Respir Dis 1990:142:1331-6. 15. Timbrell V. Ashcroft T, Goldstein 8. Hayworth F. Meurman L. Rendall A EC. Reynolds JA. Shilkin KB. Whitaker D. Relationships between re tained amphibole libras and fibrosis in human iung tissue specimens. Ann Oocup Hyg 1988:32{Suppl l):323-40. 16. Coin PG. Roggli VL. Brody AR. Deposition, clearance, and translocation of chrysotile asbestos from peripheral and central regions of the lung. Environ Res 1992:58:97-1(6. 17. Churg A. Wright JL. Hobson J, Stevens 8. Effects of cigarette smoke on the clearance of long and short asbestos fibers from ihe iung. IntJExp Pathol 1992;73:287-97. 18. Stanton MF. Layard M, Tegeris A, Miller E. May M. Morgan E, Smith A. Relation of particle dimension lo carcinogenicity in amphibole asbes toses and other fibrous minerals. JNCI 1981:67:965-75. 19. McDonald JC, McDonald AD, Armsirong B. Sesbastian P. Cohort study of mortality of vsmiculita miners exposed to Iremolita. Br J Ind Medi 669 cine 1986;43:436-44. 20. Langer AM. Nolan RP. Constantopoulos SH. Moulsoooulos HM. Associ ation ot Metsovo lung and pleural mesothelioma with exposure lo irsmolite-contaming whitewash. Lancet 1987;1:965-7. 21. McDonald JC. Armstrong B. Case B. Ooell D. McCaughey WTE. McDon ald AD, Sebastion P. Mesothelioma and asbestos fiber type. Evidence from lung tissue analysis. Cancer 1989;63:1544-7. 22. Rogers AJ. Leigh J, Barry G. Ferugson DA, Mulder H8, Ackad M. Rela tionship between lung asbestos liber type and concentration and rela tive risk of mesothelioma. Cancer 1991;67:1912-20. 23. Churg A, Wiggs B. Fiber size and number in users of processed chryso tile ore. chrysotile miners, and members of the general population. Am J Ind Med 1986:9:143-52. 24. Lippmann M. Effects of fiber characteristics on lung deposition, reten tion, and disease. Environ Hlth Perspectives l990:88:3tl-7. 25. Lippmann M. Asbestos exposure indices. Environ Res >98fl;46:8S-i06. 26. Goodglick LA. Kane AB. Cytotoxicity ol long and short crocidolite asbes tos libers in vitro and in vivo. Cancer Res 1990:50:5153-63. 27. Gylseth B. Churg A. Davis JMG, Johnson N. Morgan A. MoweG. Rogers A, Roggli V. Analysis of asbestos fibers and asbestos bodies in human lung tissue samples. An international interlaboratory trial. Scand J Work Environ Hlth 1985;11:107-10. HWBUI0010312