Silicosis and risk of lung cancer or lung tuberculosis: A cohort study

ENVIRONMENTAL RESEARCH 41, 339-350 (1986)

Silicosis and Risk of Lung Cancer or Lung Tuberculosis: A Cohort Study P E T E R W E S T E R H O L M , * A X E L AHLMARK,'~ R E I N M A A S I N G , ~ AND INGA SEGELBERG§

*Swedish Confederation of Trade Unions, S-105 53, Stockholm, ?Department of Environmental Medicine, Ume~ University, S-901 87, Ume~, eKabi Vitrum AB, Stockholm, and §Group of Applied Statistics, Stockholm, Sweden Received April 16, 1984 This is a study of cancer mortality, cancer incidence, and incidence of lung tuberculosis among cases of silicosis reported to the National Swedish Pneumoconiosis Register during 1959-1977. Two occupational categories were extracted--"mining, tunneling, and quarrying" (n = 284) and "iron and steel foundries" (n = 428), respectively. Control groups were drawn from a national register of persons undergoing periodic health examinations with regard to silicosis risk. The controls were matched for occupationl age, and time of first exposure. The follow-up was performed through record-linkage operations to computerized information in Swedish Death Statistics, Swedish Cancer Register, and the Swedish Tuberculosis Index. End of follow-up was set at December 31, 1980. In cases drawn from mining, quarrying, and tunneling workers seven deaths in lung cancer were observed and two among the controls. Among iron and steel foundry workers the corresponding numbers were 10 and 6. The values for expected numbers, based on general population statistics, were 1.3 and 2.6, respectively, for these two occupational groups. When cancer incidence statistics were used, the case/control ratio for lung cancer was 2.1 for "mining, quarrying, and tunneling" and 0.6 for "iron and steel foundries." There were 29 cases of lung tuberculosis registered among the silicosis cases during the follow-up period. Only one tuberculosis case was observed among the controls. The results demonstrate that persons with silicosis contracted in the mining, quarrying, and tunneling occupations are subject to an increased risk of lung cancer. The risk is observed when both the general population and a closely matched control population from the same occupations are used for values of reference. The results also demonstrate the high risk of persons with silicosis to contract lung tuberculosis. © 1986AcademicPress, Inc.

INTRODUCTION The question of whether exposure to silica dust increases risk of lung cancer has recently been addressed in the scientific literature by several authors. Efforts have been made to solve this question by means of epidemiological studies and animal experiments. In the experimental field a carcinogenic effect--resulting from a promoting action of silica--has been suggested by Stenb~tck and Rowland (1979). A tumor-initiating effect--histiocytic lymphomas after intratracheal instillation of silica in rat lungs--has also been suggested by Wagner et al. (1980). At present, the position is far from clear as to what general inferences can be drawn from these findings. Epidemiological field studies demonstrating elevated lung cancer risk for workers exposed to quartz dust have been published. In 1981, Palmer and Scott reviewed the studies performed in iron and steel foundries where a number of 339 0013-9351/86 $3.00 Copyright© 1986by AcademicPress, Inc. All rightsof reproductionin any form reserved.

340

WESTERHOLM ET AL.

particular occupations and processes--including molding, casting, crane operation, and blacksmithing--have been identified to be associated with such risk. Other settings for occupational quartz exposure have also been reported, e.g., ceramic industry by Thomas (1982), sandblasting by Puntoni et al. (1979), and mining by Peterson and Milham (1980). Studies of persons who have contracted silicosis were performed by Westerholm (1980) on mining, quarrying, and tunneling workers and workers from iron and steel industries and by Finkelstein (1982) on miners only. In both studies an excess risk of lung cancer was observed. The indications in epidemiologic studies that exposure to quartz dust m a y m a t least in some occupational settings--be associated with lung cancer risk have been confirmed in many studies and can no longer be taken lightly. Goldsmith and associates (1982) have in a recent, extensive review of the issue argued in favor of the view that exposure to silica dust is causally related to lung cancer with silicosis as a possible intermediate step. Goldsmith discusses three plausible hypotheses: Silica directly induces lung cancer. Silica causes silicosis, which may be an intermediate pathologic state leading to lung cancer. Silica, linked with polyaromatic hydrocarbons either from smoking or from the ambient working environment, impairs lung clearance. Thus the effective dose and/or duration of exposure are increased, inducing neoplasia in the adjacent pulmonary tissue. It was decided to undertake a study of cancer mortality and cancer incidence among recently notified cases of silicosis derived from mining, quarrying, and tunneling workers and from workers employed in steel and iron industries. The cohort study design was chosen as basic method of investigation. Lung tuberculosis is a well-known complication of silicosis. It was decided to examine also the incidence of lung tuberculosis in the cohort chosen for study of lung cancer risk.

SOURCES OF INFORMATION The Swedish Pneumoconiosis Register was used as a primary source of information. The register consists of two separate parts. One part is a register of cases. All cases which are diagnosed and reported to the National Insurance Board and to regional public insurance agencies are registered in the pneumoconiosis register through an administrative routine. Silicosis is a notifiable occupational disease. The notification is compulsory by law. The registering procedures and register functions have beem described by Ahlmark et al. (1965) and by Westerholm (1980). The second part of the register is a register of exposure. In this register are included, through a reporting procedure, persons who are under legal requirement to undergo periodic health examinations for early diagnosis of silicosis. In many branches where silicosis risks exist, such periodic examinations are prescribed by the National Board of Occupational Safety and Health, which is the government agency supervising the law on the working environment. Ac-

SILICOSIS AND LUNG CANCER

34l

cording to this law, the employer is obliged to arrange for employees exposed to silica hazards to have periodic health examinations which include a chest X ray. The exposure register was set up as a manual register in 1962-1963 and transferred to a computer base in 1978-1979. In the two parts of the register the same criteria and nomenclature are used for classification of occupations. The National Swedish Cancer Register is a person-based register receiving notifications from the whole country of newly diagnosed cases of cancer in Sweden. Administratively, the register is operated by the National Board of Health and Welfare. The register was established in 1958. The notification is compulsory by a code of practice issued by the board. The registered information includes particulars of identity (person number and name), the cancer diagnosis classified according to ICD-7, and detailed information on histopathological findings and observations at autopsy if these examinations were performed. In addition, administrative information on the notifying physicians or institutions is registered. The functions of the Swedish Cancer Register and routine statistics are published in a series of annual publications from the National Board of Health and Welfare. The Swedish Register of Causes of Death is a computer-based register containing information on all deaths occurring in Sweden. The primary material for registration is provided by the death certificates. The register is administratively operated by the National Board of Statistics. The primary material is classified (ICD-8) and coded centrally at the board. All deaths are registered with personal identity data (person number and name), main and contributing causes of death, and some data of administrative nature. The Swedish Tuberculosis Index is a computer-based register based on notifications from physicians who diagnose a case of tuberculosis. The register has been operating since 1971. Reporting to this register is compulsory by code of practice; tuberculosis is classified as a contagious disease associated with a public health risk. A notification is to be made on the suspicion of tuberculosis. The register contains information on identity (person number and name), notifying physician or institution, and detailed information on the diagnosis including clinical observations and results of laboratory examinations. MATERIALS From the Silicosis Register (case records) were selected cases notified to the register from 1959 on and including 1977. Only male cases were used. The cases were drawn from the broad occupational categories of "mining, quarrying, and tunneling" (284 cases) and "steel and iron foundries" (428 cases). At least one control person for each case person was selected from the Silicosis Register (exposure records). Each control was selected to match its correspondent case with regard to occupation, age at first exposure to silica (___1 year), and calendar year of first exposure to silica ( _ 2 years). Within these two broad occupational categories the control person was selected from the same occupation as the case person. To qualify for selection as control person, exposure to silica during not less than 5 years was required. From the c~,tegory mining, quarrying, and tunneling 334 control persons were selected and ~rom the category iron and steel foundries 476 persons.

342

WESTERHOLM ET AL.

The mean age of first exposure was 26.6 years in the category mining, quarrying, and tunneling and 28.0 years in the category steel and iron foundries. The mean age at notification of silicosis was 51.5 years in mining, quarrying, and tunneling and 55.1 years steel and iron foundries. Age distributions at first exposure and at the time of notification for these study groups are shown in Figs. 1-4. FOLLOW-UP The cases and control persons were followed up with regard to the endpoints under s t u d y - - d e a t h and cancer d i a g n o s i s - - r e g i s t e r e d during the years 1961-1980. The follow-up was performed by means of record linkage operations using the magnetic tape containing information on the index and control populations and the computerized records of the Swedish Register of Causes of Death and the Swedish Cancer Register. The follow-up with regard to diagnosed lung tuberculosis was performed manually by means of the Pneumoconiosis Register and the Swedish Tuberculosis Index. The time period of the follow-up of tuberculosis was 1971-1980. STATISTICAL METHODS

Calculation of Expected Numbers For both the cases and the controls the expected number of deaths during a specific year x on a specific age y was calculated according to the formula x--1 y--1 i=a j=b

where q is the sex-specific death risk, p is the survival probability and a and b refer to the time of entry. The summation is made over all subjects. The product part of the formula expresses the cumulative survival probability and is thus an

50

%

40 30 20 10 ~ - - - - - = ' - - - - - ~ . , . Age 20- 25 30 35- 40- 45- 50- 55- 60- 65- 7019 24 29 34 39 44 49 54 59 64 69

FIG. 1. Mining, quarrying, and tunneling. Age at first exposure. Five-year intervals of age. Percentage distribution. (Reproduced, by permission of the publisher, from Goldsmith et al. (1985).)

343

SILICOSIS AND LUNG CANCER 50

To

40 30 20 10

~

. . . . . ; . . . . 20- 25- 30- 35- 40- 45- 50- 55- 60- 65- 70-19 24 29 34 39 44 49 54 59 64 69

Age

FIG. 2. Steel and iron industry. Age at first exposure. Five-year intervals of age. Percentage distribution. (Reproduced, by permission of the publisher, from Goldsmith et al. (1985).)

inverse m e a s u r e of the death risk. The expected n u m b e r of lung cancer cases was calculated according to a similar formula, x--1 y--1

Cxy = E qxy 1~ PO"sij, a

b

where q n o w is the risk of lung c a n c e r diagnosis, s is 1 - q, and p is the survival probability as before. Differences b e t w e e n the o b s e r v e d and expected n u m b e r of events were evaluated by the X2 test or, in case of small expected numbers (less than 5), b y the Poisson distribution.

Comparison of Events between the Cases and Controls Since a varying n u m b e r of controls were selected for each case, the following p r o c e d u r e was used to achieve a c o m p a r a b l e basis.

50

%

40 30 20 IO

F,',',',',',',',','~ ~ Age 20- 25- 30 35- 40- 45 50- 55- 60 65- 70

-19 24 29 34 39 44 49 54 59 64 69

FIG. 3. Mining, quarrying, and tunneling. Age at notification of silicosis. Five-year intervals of age. Percentage distribution. (Reproduced, by permission of the publisher, from Goldsmith et al. (1985).)

344

WESTERHOLM ET AL. 50 % 40. 30 20 10 . . . . ~ 20- 25- 30-35- 40- 45- 50- 55- 60- 65-19

24

29

34

39

44

49

54

59

64

~

Age

70-

69

FIG. 4. Steel and iron industry. Age at notification of silicosis. Five-year intervals of age. Percentage distribution. (Reproduced, by permission of the publisher, from Goldsmith et al. (1985).)

For the cases, the number of events--deaths or cancer diagnoses--was calculated as D=

~D i i=1

where D; = 0 if the case was alive at the end of follow-up and D i an event. For the controls the formula ~

E =

I

=

1 in case of

rni

--~

Eij

i=1 FF/i

was used. In the formula E corresponds to D above and mi is the number of controls selected for case i. For instance, consider the situation where four controls have been drawn for one of the cases and only one of the control persons contracts lung cancer. In such instances we have recorded the number of lung cancer among the controls as 1//4 = 0.25. With the controls taken as the standard population, a standardized ratio D/E was set up and the difference of this ratio from unity was evaluated using the Poisson distribution. This evaluation was, following the matched-pair design of study, limited to those case-control pairs where a within-pair difference was observed. RESULTS

Total Mortality Tables 1 and 2 present the observed and expected mortality from all causes in the index population (cases) and the control population. The columns give observed and expected numbers of deaths and ratios of observed/expected. The value "expected" is derived by application on the index population of the agespecific death rates of the general population for each year of follow-up. The data in Tables 1 and 2 do not indicate any increase in the overall mortality rate in either of the index groups under study in comparison with the general population level.

345

SILICOSIS AND LUNG CANCER TABLE 1 MINING, QUARRYING, AND TUNNELING GROUPS

Time period

Cases observed

Expected

Obs/Exp

Controls observed

Expected

Obs/Exp

1961-1965 1966-1970 1971 - 1975 1976-1980

1 5 14 18

0.8 3.6 12.2 19.5

1.2 1.4 1.1 0.9

0 0 16 25

0.9 3.9 13.6 24.0

--1.2 1.0

1961-1980

38

36.1

1.1

41

42.4

1.0

Lung Cancer Mortality In Table 3 is presented the number of observed lung cancer deaths among the case and control populations, respectively. In the table the expected number of deaths caused by lung cancer is included. The value for "expected" was calculated using the age-specific lung cancer death rates of the general population. There were no pairs where lung cancer deaths were observed both in case and in control persons. This means that numbers of observed lung cancer deaths in discordant pairs in Table 3 add up to all lung cancer deaths.

Lung Cancer Incidence In Table 4 the same procedure as that in Table 3 was used. This time the cancer register was used in search of incident cases of lung cancer. The number calculated by applying general population cancer incidence risks on the index population is given under the heading "expected number." As in Table 3, no pair was found to be concordant with regard to incident cancer in case and control persons. This means that the sums of observed lung cancer cases in discordant pairs equal the total number of such cases. Tables 3 and 4 also show the risks of lung cancer deaths or lung cancer diagnoses in the studied populations. Deaths in lung cancer are increased for both occupational categories. Incidence of lung cancer is increased in the category mining, quarrying, and tunneling workers. Among the controls an increased risk of dying from lung cancer can be

TABLE 2 IRON AND STEEL FOUNDRIES

Time period

Cases observed

Expected

Obs/Exp

Controls observed

Expected

Obs/Exp

1961-1965 1966-1970 1971-1975 1976-1980

2 8 25 42

1.7 8.1 29.7 38.8

1.2 1.0 0.8 1.1

1 1 28 45

1.8 8.6 27.2 40.8

0.6 0.1 1.0 1.1

1961-1980

77

78.3

1.0

75

78,4

1.0

346

WESTERHOLM ET AL. TABLE 3 LUNG CANCER DEATHS AMONG CASES AND CONTROLS. EXPECTED NUMBERS BASED ON GENERAL POPULATION DATAa

Occupation category

Cases obs

Exp

Obs/Exp

Controls observed

Observed case/control ratio

Mining, quarrying, and tunneling Iron and steel foundries

7 10

1.3 2.6

,5.4' 3.9*

2 6

3.5" 1.7

* P < 0.05.

Reproduced, by permission of the publisher, from Goldsmith

et

al. (1985).

observed in the category drawn from the iron and steel industries if the general population data were used as a basis of comparison. In the same way an increased incidence of lung cancer in control groups drawn from both the examined occupational categories can be seen. Tuberculosis Table 5 shows the number of incident cases of lung tuberculosis among the case and control populations, respectively. For a person to be labeled as a case, a positive finding of bacteria was required in addition to X-ray findings. This finding of tuberculosis must be verified by microscopy or alternatively by positive guinea pig or bacteriological culture tests. The numbers observed include both persons retrieved from the Pneumoconiosis Register (16/29) and persons identified in the Swedish Tuberculosis Index (13/29). The data in Table 5 show the strong association between silicosis and the risk of lung tuberculosis. DISCUSSION The important finding in this study is the elevated risk of dying from lung cancer in the occupational category mining, quarrying, and tunneling. The risk is elevated and achieves statistical significance when both observations on the control population and expected values for the general population are used as values of reference. For the same occupational category the increased risk is also observed in lung cancer incidence. In the occupational category of iron and steel industry, the risk for persons with TABLE 4 OBSERVED INCIDENCE OF LUNG CANCER AMONG CASES AND CONTROLS: EXPECTED NUMBER BASED ON GENERAL POPULATION DATAa

Occupation category

Cases obs

Exp

Obs/Exp

Controls obs (adjusted)

Observed case/control ratio

Mining, quarrying, and tunneling Iron and steel foundries

9 6

1.7 3.3

5.3" 1.8

4.3 9.5

2.1 0.6

* P < 0.05. a Reproduced, by permission of the publisher, from Goldsmith e t al. (1985).

347

SILICOSIS AND LUNG CANCER TABLE 5 NUMBER OF INCIDENT CASES OF LUNG TUBERCULOSISIN STUDY POPULATIONS 1971-1980 Occupational group

Cases observed

Controls observed

Mining, quarrying, and tunneling Iron and steel industry

16 13

-1

silicosis to die from lung cancer is seen only when the expected values derived from the general population are used as a basis of comparison. For lung cancer incidence no difference between case and control populations can be observed. Among the controls there is an increased number of lung cancer deaths and diagnosed lung cancer cases when the general population data have been used to calculate values of expectation. It is a well-known fact in occupational and lung medicine that persons with lung silicosis are subject to an increased risk of contracting lung tuberculosis. The literature on the association between silicosis and tuberculosis was reviewed by Gothe in 1970, by Bailey et al. in 1974, and by Worth and Stahlmann in 1976. See also comment by Parkes (1982). In 1980, Westerholm observed a six- to sevenfold increase in risk of lung tuberculosis among silicotic persons recorded during 1951-1969 using the general population as basis of comparison. The data of the present study serve only to support and confirm earlier observations on this wellknown complication. The difference between cases and controls in this regard is not likely to be explained with any observational or detectional bias, although the importance of these factors can not be denied. Cases of silicosis are likely to be observed very closely with regard to complications and disease progression. On the other hand, the control persons of the present study are also under compulsory schemes of periodic radiographic lung examinations. This should offset, at least to some extent, the impact of any existing detectional bias. The increased lung cancer risk in the mining, quarrying, and tunneling occupations is noteworthy. This observation, quantified as an observed/expected ratio or a case/control ratio for lung cancer mortality and incidence lying in the range of 2-5, fits well with the observations made earlier by Westerholm (1980). In considering this finding, the rationale and method of selecting the particular control population should be kept in mind. Persons working in the same occupation as the index cases were considered to be a better basis for comparison than were statistical values calculated for the general population. We are confident in saying that the index and control populations are drawn from the same socioeconomic subset of the general population. In our view these two populations should be very similar with regard to personal characteristics and life-style factors of importance for the health endpoints under study. It should also be mentioned that this method of selecting a control population results to some extent in control of the length of exposure in the occupations studied. One of the requirements to become a control person was that the length of employment in the same occupation as the correspondent case person was to be not less than 5 years. It should at the same time be observed that the sampling frame for the control groups was

348

W E S T E R H O L M ET AL.

established in the years 1962-1963. Also, the long time lapse between first exposure to silica dust and the time when silicosis can be diagnosed radiographically is to be considered. It means that the controls, matched for occupation and still active in that occupation in 1962-1963, had been exposed during many years, sometimes for longer times than the corresponding cases. In spite of these observations, precise matching for length of exposure or formal control of exposure duration could not be obtained with the sources of information used in this study. In any study of lung cancer the factor of smoking habits has to be acknowledged. We have in the present set of data no information on smoking habits in the index and the control populations. Therefore, the factor of smoking is not under formal control. We have, however, no reason to assume that the index and control populations should differ in this particular regard. Theoretically, if smoking causes an increased risk for contracting silicosis, it could be conceived that a population of silicotic persons contains a higher proportion of smokers than a population of persons who have not yet developed silicosis. In the absence of studies which demonstrate an association between smoking and silicosis we have chosen to regard the two populations as equivalent with regard to smoking habits. The factor of smoking therefore seems not to be the likely explanation for the observed differences in lung cancer risk between the index and control populations. It should also be remembered that exposure to other carcinogenic agents may have occurred in both the index population and the control population. The working environment of miners and of workers in iron and steel industries contains other factors which may be of importance to cancer risk. They may have effects in association with the exposure to quartz dust. Among Swedish miners, for instance, radiation of and exposure to radon and radon daughters should be taken into account as has been demonstrated by Axelsson and Sundell (1978). A recent comment on this issue has been made by Edling (1983). Also in this particular regard, just as for smoking, we have no reason to suspect that the comparison between cases and controls is biased. The silicotic cases are, just as are the control persons, drawn from the studied occupations in the whole country. There is in our view nothing to support the contention that the exposure to ionizing radiation should differ among the index and control groups. For foundry workers, exposure to polyaromatic hydrocarbons is considered to be of importance by many research workers--for instance, as described by Tola et al. (1979). Close similarity in the working environments of the case and control persons was therefore judged to be important. It would otherwise not be possible to separate the effects of the silicosis per se and the effects of other factors. Still, we must be aware of the fact that in the present study the length of exposure is not under full formal control. The case and control persons were not matched for length of exposure. We could only ascertain that the controls have at least 5 years of exposure in the same occupation as the cases. We do not, however, have reason to suspect that cases and controls differ with regard to length of exposure to account for the observed differences in lung cancer risk. On the other hand, admittedly, this possibility cannot be excluded with scientific certainty.

SILICOSIS AND LUNG CANCER

349

A methodological issue with possible implications on the interpretation of the findings in the present study arises from the fact that the index and control populations were derived from different sources. The cases were recorded chronologically in the Swedish Pneumoconiosis Register. The controls were selected from a sampling frame, defined in 1962-1963, when the exposure records were set up for working populations subject to periodic health examinations. Theoretically, the cross-sectional nature of this sampling frame could provide a source of error. It could result in a population with characteristics differing in some important regard from those of the control population. We have, however, no reason to assume this to be the case. In fact, the populations in the occupations studied are usually regarded as relatively stable with regard to health characteristics and turnover rates. We find it reasonable to infer that the results are consistent with the concept that quartz dust in mining, quarrying, and tunneling operations acts as a causal factor with regard to risk for lung cancer. This, of course, means that the causality cannot be conclusively proven with the data presently available--only suggested. In fact, the design and selection of study groups in the present study are not ideal in exploring the possible carcinogenic properties of silica dust per se. Both index groups and related control groups have exposure to quartz dust as a common feature. The difference lies in the fact that in the index groups silicotic disease has been found and reported, but not in the control groups. Naturally, on the basis of only the data of the present study there is no possibility to analyze the underlying mechanisms of the enhanced cancer risk. Nor can it be clarified how quartz could possibly operate in combination with other factors in the environments from which the cases were drawn. This can only be explored further in properly designed experimental studies. In cases derived from steel and iron industries no differences in risk for lung cancer deaths or lung cancer incidence among cases and controls could be observed. It should be remembered, however, that the period of observation for these study populations was relatively short. Therefore, it seems imprudent to accept the present findings as conclusively negative. It is our intention to seek confirmation of the present observations in an extended follow-up in the near future. When considering the nature of the association between silicosis and lung cancer risk wherever it is observed, it must constantly be borne in mind that it only means that silicosis may, at least in some occupational settings, act as a prerequisite or a predisposing factor for an increased lung cancer risk. From this it does not follow that silica dust per se has carcinogenic properties. An increased lung cancer risk may also result from a weakening of defenses or repair capacities in lung and bronchial tissues due to the silicotic disease process which leads to an increased susceptibility to other agents in the environment.

CONCLUSION The data of the present study show that persons with silicosis derived in mining, quarrying, and tunneling occupations are subject to an increased risk of dying from lung cancer and of getting the diagnosis of lung cancer. The increased

350

WESTERHOLM ET AL.

risk can be observed when both a control population, matched for age, calendar period, and occupation, and expected values calculated for the general population are used as bases for comparison. The data also demonstrate the enhanced risk of pulmonary tuberculosis in cases of silicosis derived from the occupations under study. REFERENCES Ahlmark, A., Bruce, T., and Nystrom, A. (1965). Det centrala pneumoconiosisregistret--dess tillkomst och anviindbarhet. Ldkartidningen 63, 198-202. Axelsson, O., and Sundell, L. (1978). Mining, lung cancer, and smoking. Scand. J. Work Environ. Health 4, 46-52. Baily, W., Brown, M., Buechner, H., Weill, H., Ichinose, H., and Ziskind, M. (1974). Silico-mycobacterial disease in sandblasters. Amer. J. Respir. Dis. 110, 115-125. Edling, C. (1983). "Lung Cancer and Radon Daughter Exposure in Mines and Dwellings." Link6ping University Medical Dissertations No. 157. Finkelstein, M., Kusiak, R., and Suranyi, G. (1982). Mortality among miners receiving workmen's compensation for silicosis in Ontario: 1940-1975. J. Occup. Med. 24, 663-667. Goldsmith, D., Guidotti, T., and Johnston, D. (1982). Does occupational exposure to silica cause lung cancer? Amer. J. Ind. Med. 3, 423-440. Goldsmith, D. E, Winn, D. M., and Shy, S. M. (Eds.) (1985). "Silica, Silicosis, and Cancer." Praeger, New York. G6the, C. J. (1970). Effect of intravenouslyinjected tubercle bacilli (BCG) on transport of quartz dust via the pulmonary lymphatics--An experimental study on rats. Scand. J. Respir. Dis. Suppl. 73, (1970). National Board of Health and Welfare. The Cancer Registry. "Cancer Incidence in Sweden." Annual Reports, 1959-1979. Palmer, W. G., and Scott W. D. (1981). Lung cancer in ferrous foundry workers: A review. Amer. Ind. Hyg. Assn. J. 42, 329-340. Parkes, W. R. (1982). "Occupational Lung Disorders," 2nd ed. Butterworths, London. Peterson, G. R., and Milham, S., Jr. (1980). "Occupational Mortality in the State of California 1959-61" DHEW Pub. No. (NIOSH) 80-104, NIH, NCI, Bethesda, Maryland and USDHEW, PHS, CDC, NIOSH, DSHEFS, Cincinnati, Ohio. Puntoni, R., Vercelli, M., and Merlo, E, et al. (1979). Mortality among shipyard workers in Genoa, Italy. Ann. N. Y. Acad. Sci. 330, 353-377. Stenbftck, E, and Rowland, J. (1979). Experimental respiratory carcinogenesis in hamsters: Environmental, physiochemical, and biological aspects. Oncology 36, 63-71. Thomas, T. L. (1982). A preliminary investigation of mortality among workers in the pottery industry. Int. J. Epidemiol. 11, 175-180, Tola, S., Koskela, R. S., Hernberg, S., and Jfirvinen, E. (1979). Lung cancer mortality among iron foundry workers. J. Occup. Med. 21, 753-760. Wagner, M. M. E, Wagner, J. C., Davies, R., and Griffiths, D. M. (1980). Silica-induced malignant histiocytic lymphoma: Incidence linked with strain of rat and type of silica. Brit. J. Cancer 41, 908-917. Westerholm, E (1980). Silicosis--Observation on a case register. Scand. J. Work Environ. Health (Suppl.) 2 6, 1-86. Worth, G., and Stahlmann, W. (1976). Silikose und Tuberkulose. In "Pneumoconiosen" (W. T. Ulmer and G. Reichel, Eds.), "Handbuch der inneren Medizin." Band 4: "Atmungsorgane," pp. 320-387. Springer-Verlag, Berlin.