E TABLE 2.-Smoking characteristics of silicaexposed workers Number and type Study of population Smoking characteristics (percent) Comments Prow 240 gold miners. (1970) South Africa Brinkman et al. 301 automotive industry (1972) workers. aged 40-65 SluisCremer Men expmed to dust, (1972) Carletonville. South Africa Theriault et al. (3 papers) (1974) 792 granite workers. Vermont Armstrong et al. Coal and gold miners, (1979) Australia . Born et al. (1963) Trona miners. Wyoming Light (I-200)' Moderate (2Olm) Heavy (>6CO) Exposed workers Nonexpcmed workers SM 65.8 11 22.3 32.5 70 60.7 SM 60.4 SM EX' NS `Not smoked in 57 30 12 last 6 months Gold miners SM 66.3 42.6 NS/EX 34.2 ' Numerical rating of cigs/day x years smoked 30 39.3 EX NS 25.6 13.9 NSIEX 33.7 33.6 23.6 NOTJZ.: S.M=Smoker; EX=Ehsmoker; NS=Nonsmoker. Epidemiological Findings Early observers of occupational diseases, including Ramazzini in 1713 (1964), wrote about the respiratory problems of miners and stone cutters, and recognized silicosis among miners, stone cutters or hewers, and potters. Silicosis, and its previously described associated health effects, have been given a variety of names that reflect the several faces of silica exposur&ust consumption, ganister disease, grinders' asthma, grinders' consumption, grinders' rot, grit consump- tion, masons' disease, miners' asthma, miners' phthisis, potters' rot, rock tuberculosis, stonehewers' phthisis, and stonemasons' disease (Hunter 1955). Greenhow (18781, in his treatise on bronchitis, recognized that "irritants which act immediately upon the bronchial membrane may produce inflammation by means of either mechani- cal or chemical irritation. Fine coal and metal dust, stone and porcelain grit, and even the flue of cotton wool . inhaled into the lungs during various industrial processes are all of them mechanical irritants which become fruitful causes of bronchitis in certain classes of operatives" (p. 30). Mortality studies of silica-exposed cohorts have consistently shown increased mortality rates for tuberculosis and nonmalignant respira- tory disease, largely accounted for by silicosis (Guralnick 1962; Registrar General 1958, 1978; Davis et al. 1983; Armstrong et al. 1979; McDonald et al. 1978; Fox et al. 1981). Although none of these studies accounted for the effects of smoking, the consistency and magnitude of the increased rates suggest a causal relationship between silica exposure and these cause-specific mortality rates. Davis and colleagues (1983) demonstrated dose-response relation- ships between exposure category, tuberculosis, and silicosis, but found no excess mortality from bronchitis and pneumonia. Finkel- stein and colleagues (1982) investigated mortality among 1,190 Ontario miners receiving compensation awards for silicosis and found nonmalignant respiratory disease (excluding tuberculosis) to be the most frequent cause of death (standard mortality rate, 765). NIOSH recently assessed causes of disability among employees of the mining industry, based on the Social Security Disability Benefit Awards and Allowances to Workers for 1969-1973 and 1975-1976 (Osborne and Fischbach 1985). The observed proportional morbidity rate (PMR) for pneumoconiosis from silica and silicates (they were not distinguished) was found to be somewhat higher (4,894) than for other mining occupations. Workers employed in boring, drilling, and cutting jobs appeared to experience increased disability from respira- tory diseases, specifically pneumoconiosis including silicosis. These findings, based on somewhat more recent exposures than the previously cited mortality studies, confirm the major mortality findings, but suffer from the same methodological problems. Again, smoking data were not available or analyzed, and it is recognized 327 that those disabled in the mid-1970s very likely were exposed to silica three or four decades previously; therefore, their disabilities reflected previous dust exposures. Early morbidity studies of workers exposed to silica dust focused on rates of sickness, respiratory symptoms, and physical findings, supplemented in the 1920s with chest radiography. The U.S. Public Health Service (US PHS) conducted the first major U.S. silicosis study of the hard-rock mining industry in 1913-1915 (Higgins et al. 1917; Lanza and Higgins 1915). Their studies reported that 60.4 percent of the 720 miners examined suffered from pulmonary diseases attributable to mine rock-dust exposure. Dust samples collected with a Draeger liter bag-granulated sugar filter apparatus were reported to average from 30 to 50 mg/m3 (Higgins et al. 1917). Although these concentrations would appear to be quite high, they are difficult to interpret according to modern-day respirable dust sampling and analysis (x-ray diffraction for free silica content). Subsequent US PHS silica studies included Harrington and Lanza's (1921) 1916-1919 study of copper miners in Butte, Montana, in which 42.4 percent were judged to have some dust-induced lung injury and 25.5 percent to have advanced disease. Dreessen and colleagues (1942) studied 727 metal miners in 1939, and Flinn and colleagues (1963) studied 67 underground mines employing 20,500 miners from 1958 to 1961, but found varying silicosis prevalence from mine to mine and widely divergent exposures to free silica. The silicosis prevalences of 9.1 percent and 3.4 percent, respectively, were found to be associated with longer duration of exposure and especially with face work exposures (Dreessen et al. 1942; Flinn et al. 1963). Earlier, Flinn and colleagues (1939) had reported an impor- tant study (19361937) of West Virginia potteries that included 2,516 workers with an overall silicosis prevalence of 7.8 percent. Free silica content ranged from 1 to 39 percent, dust concentrations varied from 3 to 440 million particles per cubic foot (mppcfl, and mean particle diameters were judged to be 1.2 pm (but without data on the concentration of respirable dust). A strong dose-response relation- ship between dust concentration, duration of pottery exposure, and silicosis prevalence was documented. It was suggested that no new cases of silicosis would occur if dust concentrations in this industry were brought below 4 mppcf. Renes and colleagues (1950) studied 18 ferrous foundries in 1948-1949, and found 9.2 percent of 1,937 foundrymen to have pulmonary fibrosis. Free silica content averaged 30 percent, with a mean particle size of 3 pm, and 82 percent of the samples had levels below 6.9 mppcf. Mechanical shakeout operations were found to have the highest dust concentrations (10 to 75 mppcfl, and silicosis was noted to be more prevalent among foundrymen with 20 or more years of exposure. It was suggested that conditions had improved in foundries and that most of the pulmonary fibrosis was 328 due to previous exposures. Early studies of the refractory (silica) brick industry documented high percentages of free silica, often in the form of cristobalite and tridymite from burned bricks. Keatinge and Potter (1949) and Fulton and colleagues (1941) studied 1,035 exposed workers in this industry, finding 52 percent to have some stage of silicosis. A relationship with dust concentration and duration of exposure was again documented, as was an apparent increased risk among men exposed to burned brick dust (Keatinge and Potter 1949; Fulton et al. 1941). Epidemiological studies of workers in the Vermont granite indus- try have provided an important and interesting chronology of data on the natural history of silica-associated respiratory diseases. Early US PHS studies of this industry (Russell et al. 1929) documented high dust concentrations (37 to 59 mppcf) and a very high prevalence of silicosis. On the basis of dust with a free silica content of 35 percent, a presumptive "safe limit" of dustiness was suggested to lie between 9 and 20 mppcf. A subsequent US PHS study (Russell 1941) essentially confirmed the findings of the original study, noted an increased progression of silicosis among the highly exposed cutters, and concluded that a limit below 10 mppcf for this industry would be desirable. Subsequent followup studies in 1955 by the US PHS and the Vermont State Board of Health (Hosey et al. 1957) found that the prevalence of silicosis had decreased from 45 percent in 1937-1938 to 15 percent in 1956, that the silica content of the dust averaged a somewhat lower 22 to 25 percent free silica, and that nearly all workers with silicosis had been exposed prior to implementation of dust controls in 1937. This report was consistent with an earlier report by Ashe (19551, and was subsequently supported by a further followup study by Ashe and Bergstrom (19651, which reported no cases of silicosis among 1,478 granite workers employed after 1937, and a study by Davis and colleagues (1983) that reported only one case in the same population. All of these early studies of silica exposure concentrated on radiographic evidence of silicosis and tuberculosis and the associa- tion with silica content and concentration. These studies formed the basis for environmental control of silica exposures, demonstrated the effectiveness of dust control, and provided a widely held impression that silica exposures, and hence disease arising from silica expo- sures, were well controlled. Beginning in the 195Os, British epidemiologists introduced stand- ardized respiratory questionnaires, field spirometry, and sound epidemiological methodology to the study of bronchitis and chronic obstructive lung disease, and were the first to use these methods to assess respiratory effects among industrial workers. This allowed assessment of other risk factors, including cigarette smokmg, and quantitation of major risk factors compared with appropriate 329 157-5L.4 3 - 66 - 1,' reference populations. At the same time, on the basis of clinical case series, it was becoming clear that bronchitis and nonspecific airways obstruction were more common than pneumoconiosis among work- ers exposed to coal mine dust and silica dust. The significance of these health effects was not clear. Modern epidemiological studies began with the Higgins and colleagues (1959) investigation of Stavely, an English industrial town of 18,000 and home to a significant number of coal miners and foundry workers. This study and other cross-sectional studies of silica exposure that have assessed standardized respiratory symptoms, lung function, smoking, and occupation (only those with non-coal-mining silica exposures) are summarized in Table 3. Review of these studies has found them to be heterogeneous in regard to workforce composition, free silica content and dust concentration (if reported), and other associated occupational expo- sures that may contribute to respiratory symptoms and declines in lung function. In some instances associated occupational exposures other than silica dust appear to be as important or more important than silica dust (Higgins et al. 1959; Gamble et al. 1979; Manfreda et al. 1982; Graham et al. 1984). Two of these studies found lung function to be somewhat better among exposed workers than among reference subjects (Clark et al. 1980; Graham et al. 1984). However, in both of these studies, one of potash miners and one of taconite miners, it is very likely that the free silica exposure, although not documented, was low. One study of fluorspar miners (Parsons et al. 1964) and one of copper miners (Federspiel et al. 1980) suggest a significant dust effect on bronchitis prevalence and a somewhat lower lung function among exposed miners. Specific environmental data on free silica content or dust concentration were not provided in either study, although most likely some of the dust exposure in these mines was silica. Four of the studies reviewed in Table 3 have documented significant exposures to free silica with the relative absence of other exposures: the Welsh slate workers study (Glover et al. 19801, the Vermont granite workers study (Theriault, Burgess et al. 1974; Theriault, Peters, Fine 1974; Theriault, Peters, Johnson 1974) and two studies of South African gold miners (Sluis-Cremer et al. 1967; Wiles and Faure 1977). Silicosis was reported in all four study populations, ranging from 5 to 33 percent. Sluis-Cremer and colleagues (1967) surveyed the prevalence of chronic bronchitis in a mixed mining and nonmining population in Carletonville on the Witwatersrand, South Africa. Chronic bronchitis was more common among miners who smoked than among nonminers who smoked, but there were no significant differences in prevalence of bronchitis between miners and nonminers who did not smoke. The prevalence of bronchitis was substantially higher among smokers than among 330 Study, country Number and Age type of (mean or population range) Bronchitis (ratio) S/NS Exp/Not Lung function P"eum* conic& S NS A EXP Not A (percent) Comment Higgine et al. (1959), United Kingdom Current and ex- 55 to foundry workera; 64 105 exposed, 81 nondusty occupations Not available for foundry workera alone 1.2 Not available for foundry 82.1 90 -7.9 14.0 Foundry workera both "pure" workers alone with free silica exposure and (Indirect MBC based on "mixed" with chemical fumes FEVd (HCI. H,SQ, caustic soda, and benzol) and other dusts; increased respiratory symptoms and decreased lung function mainly in "mixed" foundry workers suggests other dusts and fumes likely more important than "pure" foundry work Parsons et al. (1964), Canada Fluorspar mining; 301 ev-4 5% co"trola ,383 (20 to 70) Not available 5.5 Generally higher for nonminers; decreased lung function in chronic bronchitic me" appeara more important than dust eategory @a.& on indlrect MBC, MMF, PFR) 1.93 No specific SiO, expasurea or dust measurements available; exposure determined from job category and tenure Sluis Cold mining; 35 and 1.8 1.22 Not studied 5 Free silica content range 5& Cremer et 582 expceed, 285 older 70%, but generally low dust al. (19871, community levels; smoking somewhat South controls more common among miners; Africa significant increase in chronic bronchitis prevalence among smoking and ex-smoking but not nonsmoking miners suggests possible interaction between smoking and E "underground aerial w pollution" TABLE 3.-Continued Number and Age Bronchiti 3 (ratio) Lung function Pneume Study, type or (mean or --- coniosie country population range) SINS l&p/Not S NS h EXP Not h (percent) Comment Higgins et 60 foundry 25 to Not Not Not reported separately 3.49 3.55 -.c6 23.1 No environmental al. w66), workers, 100 34 reported reported for foundrymen measurements, but typical United "nondusty" SiO, foundry exposures; 9- Kingdom workers year followup mortality higher among foundry 43 foundry 55 to 2.27 2.36 -39 workers than others, workers, 52 64' appreciably higher among "nondusty" (FEV7S%) foundry workers with silicwis workers Theriault 792 granite 44 Not Not 4.2 4.1 + IO 4.2 4.1 +.10 31 Single "A" radiograph reader; L al. (3 workers =pod repoti dow+response relationship PapeM (FVC) between FVC and granite (1974). 189 marble 41 duet and quartz dust; 2 mL United workers FVClduetyear decline, 9 mL States F'VC/smoking-year decline Wilea and 2,209 gold 4.5ta 2.3 5.3 2.53 3.77 -1.24 3.60 3.64 -.24 6.7 7@-250 particles/cm' dust Fare miners with 54 (1977). counts; 75% free BO,; 2 10 years' (low duet (MMEF) flow duet significant dust South service YS. exposure/chronic bronchitis Africa highYsbust high dust doee-reeponee relationship in C&gOries) C&gOrieS) all smoking categories; much stronger smoking effect on lung function, but no dase- response &lationehip TABLE 3.-Continued Number and Age Bronchitis (ratio) Lung function Pneume Study, type of (mean or - conk& country population range) SINS Exp/Not S NS A EXP Not A (percent) Comment Gamble et 121 talc miners 39.7 3.5 1.2 3.74 4.13 -.39 3.66 3.64 -.14 2.2 2040 w/m' free silica; 0.2% al. (1979), talc, 2.96 mg/m' respirable dust; United 1,077 potash 39.2 (FEV,) .I07 anthophyllite, tremolite, and States miners potash crysotile asbestos fibers found; (reference group) 17/24 jobs ;,2f/cc; decrea.. lung function and pleural thickening significantly aesaciated Clark et 240 iron ore 49.3 None 1.0 c478.5 61.4 -2.9 861.4 80.0 t1.4 c2 Taconite has iron. quartz. al. (198X, miners. 220 among and numerous silicates, esp. United yE2X-S nonsmoking mv,/Fvc) grunerite-cummingtonite; 25- States underground miner8 40% total dust quartz; repxted significant smoking effect, no 86 not exposed 50.1 dust effect on lung function - Federspiel 133 surface Not 98.4 None GM.5 92.5 -7.0 492.5 98.0 -5.5 Not No dust level or SiO,% data; et al. workers given among reported no SO, miner exposure. little (196OJ nonsmoking FEV,) or no surface worker SO, United 112 copper nonminers exposure; mining and States miners smoking additive effect on bronchitis; signiticantly reduced nonsmoking miner FEV, and FVC and smoking miner FVC i% TABLE 3.-Continued 6 Number and Age Bronchitis (ratio) Lung function Pneumw Study, type of (mean or conioaia country population range) SINS Exp/Not S NS A Exp Not A (percent) Comment Glover et al. (19801, United Kingdom 725 slate workers 530 nonexposed 1.8 4.4 3.06 3.23 mv,1 -.17 3.23 3.53 -30 33 13-325 reepirable quartz in reapirable dust; no smoking category/radiographic opacity association; no dust concentrations available, but thought "high"; respiratory symptoms dependent on pneumonconiceie category per multiple regression, except nonsmokers with previous TB (in 4&50% of slate workew age >w Manfreda et al. (1982). Canada 241 hardrock 25to 9.0 9.5 23 7 +16 0 7 -7 TLV; some 382 nonexposed percentile of nonsmoking general population) underground worker (95) NO, men (community exposure, some smelter sample) worker (107) SO, exposure (10% >TLV); low sulfm? worker (39) duet exposure; nonsmoking miner bronchitis significantly increased; significantly reduced lung function in smelter workers, not miners; probable selection processes noted NOTE: Bronchitin rat& and lung function comparisons are of nonexpwed smokers 6) and nonsmokers (NS) to 888e88 smoking effect and of nonamokiing exposed workers (Exp) and nonexpoeed worken, (Not) to BBBBBB exposure effezt. Combined smoking and exposure et&& are not shown, but are add- under omxnent. nonsmokers in both the dust-exposed and the nonexposed popula- tions. Evaluation of Vermont granite shed workers (Theriault, Burgess et al. 1974; Theriault, Peters, Fine 1974; Theriault, Peters, Johnson 1974) revealed that both smoking and cumulative dust exposure contributed to the differences in FVC and FEV, among these workers, but the effect of smoking was larger than the effect of dust exposure, using a multiple regression technique. A dose-re- sponse relationship between silica dust exposure and decreased lung function was demonstrated in both the Vermont granite shed workers and the South African gold miners (Wiles and Faure 1977). Glover and colleagues (1980) examined 725 workers and former workers from the slate mines and quarries of North Wales and 530 men from the same area who had never been exposed. The prevalence of chronic cough ranged from 5.2 percent in the nonsmok- ers not exposed to dust to 19.4 percent in the nonsmokers with dust exposure. Smokers with no exposure to dust had a prevalence of cough of 27.5 percent; the prevalence was 38.9 percent among the smokers with dust exposure. FEV, (standardized to a fixed height) was lower in the smokers than in the nonsmokers. The dust-exposed nonsmoking workers had a lower mean FEV,, but the values for the dust-exposed and the nonexposed smokers were similar. The regres- sion coefficients for FEV, with age were 20 mL per year in the nonexposed nonsmokers and 38 mL per year in the dust-exposed nonsmokers, but the coefficients for smokers were similar between the dust-exposed (40 ml/year) and nonexposed (46 ml/year) men. The absence of an effect of dust exposure among the smokers in some of these studies may be the result of the cessation of smoking by those workers with declining lung function, as suggested by the observation that the mean FEV, and regression coefficient for decline in FEV, with age among slate workers was worse in ex- smokers than in either current smokers or nonsmokers. In contrast, the values for ex-smokers in the general population were between those for smokers and those for nonsmokers. Only a few prospective studies of silica-exposed workers have been reported in the literature. Four of these studies are summarized in Table 4 (Higgins et al. 1968; Pham et al. 1979; Kauffmann et al. 1982; Manfreda et al. 1984). Two other prospective studies of silicaexposed workers are not included in this table because of methodological questions. Brinkman and colleagues (1972) followed a group of foundry workers with silica exposure and with silicosis over an ll- year period. Only a third of the men known not to have died in that interval were restudied, however, raising questions about the validity of the finding of no apparent difference between the silica- exposed workers and the unexposed workers in decline in lung function over time. The original cross-sectional study found poorer lung function among silica-exposed workers and silicotics. Musk and 335 colleagues (1977) conducted a 4-year followup study of Vermont granite workers and reported a substantially higher annual loss in lung function than predicted from previous cross-sectional studies of this population. However, reassessment of some of these data has raised questions about the adequacy of the pulmonary function testing (Graham et al. 1981). Reana!ysis of the population revealed that Vermont granite workers had an annual decline in FEV, of 44 mL per year and those who had left the industry had a decline of 72 mL per year (Eisen et al. 1983). The smoking habits of those who continued working (20 pack-years) and those who had left the industry (27 pack-years) were similar. There was no statistically significant relationship between lifetime dust exposure and decline in FEV, for either the workers who were still working or those who had left the industry. One of the four studies reviewed in Table 4 found no increased decline in lung function over time among silica-exposed workers (Higgins et al. 1968). On followup, however, the mortality rate among the foundrymen in the original study (Table 3) was appreci- ably higher, particularly among those with silicosis and among older workers. Smoking habits were recorded in this study, and the foundry workers who smoked had lower mean FEV, values than the nonsmoking foundry workers in both the 25 to 34 and the 55 to 64 age groups. Pham and colleagues (1979) found consistently increased declines with age in all measures of lung function studied (FEV,, FVC/FEV,, RV/TLC, and fractional uptake of CO) among silica- exposed steel workers compared with unexposed workers. Results for smoking and nonsmoking workers were not reported separately. Lung function of the exposed men in the original survey was somewhat higher than in the unexposed workers (although they had much more bronchitis), suggesting that selection processes (healthy worker effect) occurred in this study. Kauffmann and colleagues (1979, 1982) also found increased declines in smoking-adjusted lung function over time among workers exposed to mineral dust (especial- ly silica), and argued that the mineral dust and silica exposures were most likely to be causal. Their findings are consistent with this conclusion, but exposures were assessed by type of job, and informa- tion on silica dose or interval progression over the 12 years of study is lacking. Manfreda and colleagues (1984), in a 5-year followup study of hard-rock miners and smelter workers, reported significant declines in FEV,/FVC for both smoking and mining industry exposure. These effects were quantitatively similar, but may reflect more of a smelter effect than a mining (silica dust) effect, as that was the finding on their original cross-sectional study. The prospective study abstract does not address this question. 336 TABLE 4.-Prospective studies of workers occupationally exposed to silica Study, country Number and Age type of (mean or population range) Annual decline in lung function s NS A EXP Not A Comments Higgins et al. (19661. United Kingdom 60 foundry workers, 100 "nondusty" workers 43 foundry workers, 100 "nondusty" workers 25-34 5574 (Ages in 1957) 36' 21 -17 29 30 -1 Somewhat higher increased annual lung function decline in older men, not strongly associated with occupation, but strongly influenced by smoking 32' 54 -22 37 34 +3 ' Heavy smokers only (FEV 76) Pham et al. 119791, France 196 steel (foundry and roll sheet) workers 166 unexposed workers 49.5 49.6 (Ages at first exam) 7.4% 0.6% -6.6% At baseline, bronchitis prevalence significantly higher in steel vs. unexposed workers (37.8 vs. 17.51, lung function somewhat higher in steel workers; over Byear followup. somewhat more (%FEV increased steel worker bronchitis prevalence predicted) (45.3121.9); more steelworker than unexposed worker all-parameter lung function consistent Matched for age, decline; no silica content or dust concentration height, smoking environmental data status TABLE 4.-Continued Number and Age Annual decline in lung function Study, type of (mean or country population range) S NS A Exp Not A Comments Kauffmann 178 mineral dust 41 No smokingspecific lung 52 42 -10 la-year followup study of 11 factories, including et al. expmd workers (55 function decline given several mineral dust exposures, of which only ww, expoeed to silica) (41) o (57) (42) C-15) silica is separable; no silica monitoring data; France significant FEV, annual adjusted declines in 177 unexpceed 41 CFEV,) mineral dust exposed workers (esp. silica exposed) workers interpreted a~ work related and consistent with Annual declines silicaexposed worker original crw-aectional adjusted for decreased lung function assessment smoking statue and amount `Numbers in parentheses, of 55 silica-expceed workers only Manfreda 179 hard rock 2544 3.4% 2.0% -1.4% 3.1% 1.6% -1.5% (See Table 2 for crowsectional results) et al. miners, 254 Cough and phlegm prevalence greater among (19&o. unexpcwd (Percentage decrease in (Percentage decrease in miners at baseline, no change over time; adjusted Canada (community sample) FJW,/FVC, over 5 years, FEV,/FVC, over 5 years, FEV, and FJW,/FVC declines significant for adjusted for age and adjusted for age, height, smoking and FEV,/FVC decline significant for height) and smoking) mining expoeure; data suggest FEV,/FVC more sensitive indicator; data con.&.ent with mining and smelter exposure and smoking additive effect NCFlW S=Smoker; NS=Nonemoker; Exp=Expaeed; Not=Not exposed. Pathogenesis of Silica-Related Health Effects The characteristic pathology of the various forms of silicosis are well described in recent texts dealing with occupational respiratory diseases (Parkes 1982; Kleinerman and Merchant 1983). The mature lesion of silicosis is the hyalinized nodule that is spherical and typically varies in size from 3 to 12 mm. The nodules are more commonly found in upper lobes, but are found throughout the lung and are frequently subpleural. Microscopically, the nodules have a whorled appearance composed of lamina of acellular hyalin. The borders of the lesions are typically serpiginous and are composed of pigment (especially if associated with a coal exposure), chronic inflammatory cells (mainly lymphocytes and plasmacytes), and connective tissue extending into the surrounding lung parenchyma. With phase microscopy, doubly refractile silica particles 1 to 5 urn in size may be observed within the lesions and within macrophages in the surrounding infiltrate. Acute silicosis, or acute silicoproteinosis, differs from classical nodular silicosis in that the principal finding is alveolar proteinosis associated with a diffuse interstitial reaction. Scanning electron microscopy and x-ray microanalysis have demonstrated small bire- fringent silica and silicate particles (less than 1 urn in diameter) in these processes (Abraham 1978,1984). Progressive massive fibrosis may develop on a background of silicosis through the enlargement and sometimes the coalescence of the nodular lesions of silicosis into conglomerate silicosis. These lesions form most commonly in the apical or middle portion of the upper lobes and are frequently complicated by tuberculosis. Cavita- tion of these lesions may occur with or without tuberculous infection (Kleinerman and Merchant 1983). The mechanisms that produce silicosis, and particularly conglom- erate silicosis, are still not fully understood. The cellular events leading to lung injury appear to arise from the cytotoxicity of the respirable silica particle for a principal lung defender, the alveolar macrophage. Upon phagocytosis of the silica particle, cell death is caused by the release of proteolytic and hydrolytic enzymes into macrophage cytoplasm. The release of these cytoplasmic constitu- ents, including the still biologically active silica particle and fibroblast stimulating factor, may then lead to fibrosis (Allison et al. 1966,1977). As has been noted with other types of pneumoconiosis, silicosis (and particularly conglomerate silicosis) is associated with a high prevalence of circulating autoantibodies (ANA and RF) (Jones et al. 1976; Turner-Warwick et al. 1977). Although silica exposure and particularly silicosis may be associated with rheumatoid arthritis and several other collagen-vascular diseases, the role of these antibodies in the etiology, onset, and progression of silicosis is not 339 clear (Parkes 1982). Angiotensin,-converting enzyme (ACE) elevation has also been reported among silicotics (Gronhagen-Riska 1979; Nordman et al. 1984). Nordman and colleagues (1984), in a case- reference study of the Finnish Occupational Diseases Register from 1965 to 1977, reported an association between ACE activity and progression of silicosis. Smoking, age, and bronchitis .were not related to ACE activity, which was thought to reflect accumulation and increased degradation of macrophages. Histocompatibility anti- gens (HLA) have also been studied as possible genetic risk factors for silicosis, but with variable results. Koskinen and colleagues (1983) found that the prevalence of HLA-Awl9 was higher in their Finnish silicosis patients than in the silica-exposed referent population and that the highest risk of developing advanced silicosis was associated with the phenotypic combination Awl9 and B18. However, Sluis- Cremer and Maier (1984) reported only a decrease in HLA-B40 among 45 South African gold miners. These variable associations are statistically weak and may be related to the number of statistical tests performed on the multiple HLA antigens. The pathogenesis of airways obstruction in silica-exposed workers is less well understood. Although increased rates of bronchitis and decreased lung function have been observed in epidemiological studies comparing silica-exposed and unexposed workers, it appears that these findings are largely separate from the clinical and epidemiological picture of silicosis. In the largest and best controlled study of a reasonably pure silica-exposed sample of 1,973 white gold miners (Irwig and Rocks 1978), chronic bronchitis was found to be equally common among those with radiographic evidence of silicosis and those without. The smoking habits of miners with radiographic silicosis and of those without silicosis were not significantly different statistically. Silicotics reported only more days away from work, a finding the authors suggested may have been as related to their compensation status as to their disease. Comparison of lung function between silicotics and their silica-exposed referants revealed equiva- lent FVC, FEV,, and FEFz~:~~ among silicotics. Recent pathological evidence of a nonfibrous mineral dust small airway lesion has been provided by Churg and Wright (1983). This pathological process, which they labeled mineral dust airways disease (MDAD), is similar to that produced by tobacco smoke. This pathological process involves primarily respiratory bronchioles, with a lesser extension into alveolar ducts. This lesion generally involves more pigmentation and more thickening of the bronchiole walls than is typically found in cigarette smokers. In a recent study by Churg and colleagues (1985) of 13 cases of patients with MDAD, 7 had occupational histories consistent with a primary silica exposure. Only 1 of 121 cases without a clear history of dust exposure was found to have MDAD. Those. with MDAD, matched for age and 340 smoking habit with 13 cases without MDAD, were found to have significantly poorer lung function, including clinically relevant lower mean levels of FEVI, FEF25-75~1, and FVC and increased RV/TLC and AN, per liter as percentages of that predicted. It was also noted that significantly more membranous and respiratory bronchiole fibrosis occurred among subjects with MDAD. The similarity in location, morphology, and physiological impairment between the pathology induced by mineral dust and that observed with cigarette smoking suggests that the cellular events giving rise to them may be similar. Although a good deal is known about the pathogenesis of the process arising from cigarette smoke (US DHHS 1984), systematic experimental studies of mineral dust airways disease have not been reported. Silica Exposure and Cancer Initial concerns about the association between silica exposure and cancer arose during the 1930s among investigators in England, Canada, and South Africa. In the early research on this topic, the focus was on the proportion of lung cancers arising among autopsied cases of silicosis compared with that among nonsilicotics or members of the general public. All of the early research (Dible 1934: Anderson and Dible 1938; Kennaway and Kennaway 1947; Klotz 1939; Irvine 1939) was characterized by the lack of any data on smoking. Early assessments of the association between silicosis and lung cancer were summarized by Hueper (1966). More recently, Hepple- ston (1985) summarized the autopsy findings from South Africa (Becker and Chatgidakis 1960; Chatgidakis 19631, from Switzerland (Ruttner and Heer 19691, and from Germany (Otto and Hinuber 19721, but again no smoking data were presented. The reports from South Africa and Switzerland showed no differences in the ratio of lung cancers between silicotics and controls. However, Otto and Hinuber (1972) showed that porcelain workers with silicosis had more than twice the proportion of lung cancers as the noncases. Early studies suggested that silicotics have an increased lung cancer risk (Dible 1934; Klotz 1939; Mittmann 1959) or that silicotics with respiratory cancer have greater concentrations of silica in lung tissue (Anderson and Dible 1938). However, data from Bridge (19381, Heppleston (19851, Hueper (1966), and Irvine (1939) suggested that lung cancer risk among silicotics is less than or equal to that of men without silicosis, regardless of their occupation. In reviewing the evidence, Hueper (1966) observed that the data support the idea that lung cancer is a coincidental finding among silicotics and that there is no etiological relationship. None of these studies addressed the smoking status of the subjects, a crucial omission in any study of lung cancer. Furthermore, age was 341 not adjusted, nor were there any quantitative estimates of the silica exposure or assessments of the severity of the silicotic lesions. Epidemiologic Studies of Smoking, Silica Exposure, Silicosis, and Cancer Silica-Exposed Cohort Studies Occupational silica dust exposure is common in many industries; therefore this section is organized so that exposure studies in work settings that are similar can be examined together, i.e., metal ore mining, the steel industry, and workplaces where exposures are to silica only. Metal Ore Mining McDonald and colleagues (1978) conducted an enlarged followup study from 1937 to 1973 of the Homestake Veterans Association cohort that included 1,321 men with at least 21 years of employment at the mine. Standardized mortality ratios (SMRs) were calculated using South Dakota mortality rates as opposed to U.S. rates. The South Dakota lung cancer rates were lower than those for the United States as a whole. Using dust exposure data from company midget impinger samples, the authors examined the pneumoconiosis (mostly silicosis) and cancer risks in five categories of dustiness, collapsing them when indicated owing to small numbers. The data showed striking trends for pneumoconiosis and tuberculosis, but no gradi- ents emerged for respiratory cancer. Brown and colleagues (1985) also conducted an assessment of the Homestake gold miners. The cohort included 3,328 white male miners employed at least 1 year between 1940 and 1965 and followed until June 1, 1977. The authors calculated SMRs using person-years and contrasted mine mortality rates with rates for U.S. white men. An index of dust exposure by job location was assembled for the purpose of assessing dose-response gradients. The SMR for malig- nant neoplasms of the trachea, bronchus, and lung was 100, with no trends in latency or dust exposure by length of employment. Katsnelson and Mokronosova (1979) examined the mortality at a U.S.S.R. gold mine and at several brick plants from 1948 to 1974. Dust concentrations were not specifically stated for workers in the gold mine, and an approximation of the SMR (which the authors termed "relative risk") was calculated to compare the cancer risk among gold miners with the cancer risk of residents of a nearby town (excluding those who worked with chromate dusts and adding to the comparison group those who worked less than 3 years in the plants under study). The authors reported a relative risk (RR) of 7.9 (p < 0.001) for lung cancer among the male underground gold miners; without the silicotics, the RR was 3.1 (~~0.02). Surface workers had 342 a nonsignificant RR of 1.6. No lung cancer deaths occurred among women during these years. No smoking data were presented for gold mine workers, although data presented for workers at a silica firebrick plant and an aluminosilicate brick plant indicated that two- thirds to three-fourths of the men smoked, whereas only 0 to 15 percent of the women did. There appeared to be an inverse gradient of the proportion of lung cancers by stage of silicosis or silicotubercu- losis (although no standard classification such as that of Internation- al Labour Office (1980) is given). Armstrong and colleagues (1979) followed 1,974 Kalgoorlie gold miners from Western Australia (whose smoking habits were mea- sured between 1960 and 1962) for silicosis incidence and mortality through 1975. Expected death rates were obtained from the age- specific death rates of Western Australia during 1963-1967, 1968- 1972, and 1972-1976. There were significant (p