Further evidence for a link between silica dust and esophageal cancer
Article first published online: 1 DEC 2004
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 114, Issue 3, pages 479–483, 10 April 2005
How to Cite
Yu, I. T. S., Tse, L. A., Wong, T. W., Leung, C. C., Tam, C. M. and Chan, A. C. (2005), Further evidence for a link between silica dust and esophageal cancer. Int. J. Cancer, 114: 479–483. doi: 10.1002/ijc.20764
- Issue published online: 8 FEB 2005
- Article first published online: 1 DEC 2004
- Manuscript Accepted: 28 SEP 2004
- Manuscript Received: 7 MAY 2004
- Research Grants Council of the Hong Kong Special Administrative Region, China. Grant Number: CUHK4328/99M
- esophageal cancer;
Our objective was to examine the relationship between silicosis and esophageal cancer in Hong Kong. The mortality of esophageal cancer was investigated among caisson and non-caisson workers in a cohort of 2,789 male silicotic workers in Hong Kong during the period 1981–99. The standardized mortality ratio (SMR) was calculated using the Hong Kong general population rates as reference. The indirect method proposed by Axelson was used to adjust for the confounding effects of cigarette smoking and alcohol drinking. The SMR of esophageal cancer in the entire cohort was 2.22 (95% CI 1.36–3.43, based on 20 deaths) and was 4.21 (95% CI 1.81–8.30, based on 8 deaths) in the subgroup of caisson workers who had a higher exposure to silica dust. The relative risk of esophageal cancer for caisson silicotics was reduced to 2.34 after adjusting for the effects of smoking and alcohol drinking. No more excess risk of esophageal cancer was observed among non-caisson silicotic workers after the adjustments. This historical cohort study revealed that there was an increased mortality risk of esophageal cancer among silicotics who had worked in underground caissons in Hong Kong after adjusting for cigarette smoking and alcohol drinking. We believe that the excess risk of esophageal cancer mortality among caisson workers with silicosis could best be explained by the very heavy exposure to free silica dust in their working environment. © 2004 Wiley-Liss, Inc.
The possible association between silica and esophageal cancer was first described in the English literature in 1968 among residents of Transkei in South Africa, where contamination of the diet by quartz or silica was thought to be partly responsible for the high incidence of esophageal cancer.1 Since then several other studies have reported that silica present in certain plants in fibrous form could be linked to the high incidence of esophageal cancer in northeast Iran and northern China by contamination of food.2, 3 Occupational exposures to silica dust are very common in many parts of the world, but not enough attention has been given to the risk of esophageal cancer among exposed workers. In the search for evidence on the role of silica dust exposure and silicosis on lung cancer, many studies among exposed cohorts or disease (silicosis) cohorts have been conducted over the past 2 decades. SMR for esophageal cancer was reported in a number of these studies along with other causes of death, but very few specifically discussed the implications of their findings on the possible role of silica on esophageal cancer. The main focus of research and debate was on lung cancer, and the International Agency for Research on Cancer (IARC) eventually classified crystalline silica as a Group 1 human carcinogen in 1997 for lung cancer.4
A study among Italian lead-zinc-silver miners first reported significant excess mortality from esophageal cancer, along with lung cancer and stomach cancer, among underground miners in 1989.5 The first study with a focus on the link between occupational exposure to silica dust and esophageal cancer did not appear until 10 years later.6 Evidence supporting an association between silica dust exposure and esophageal cancer remains sparse and in demand. We report here the mortality risk of esophageal cancer in a cohort of silicotic workers in Hong Kong and discuss the probable link to silica dust exposure.
Material and methods
The cohort and data collection
The cohort was comprised of all newly diagnosed male cases of silicosis seen at the Pneumoconiosis Clinic of the Tuberculosis and Chest Service of the Department of Health (the only referral center for all patients with pneumoconiosis in Hong Kong) from January 1, 1981 to December 31, 1998.
Each worker's demographic information, smoking habit, lifetime occupational history and medical history were abstracted using a data abstract form from the records kept by the Pneumoconiosis Clinic. A panel of doctors (2 chest physicians and an occupational physician) made the diagnosis after considering the medical history, the occupational history of silica dust exposure and the radiologic features on a standard chest X-ray. The radiographs were read and classified following the recommendations of the International Labor Organization (ILO), and the presence of silicosis was defined as profusion category 1/0 or higher.7 To avoid contamination, cases with a history of cancer or past occupational exposure to asbestos were excluded.
All cases were followed up until the end of 1999 to ascertain the vital status through various means: telephone interviewing the kin or next of kin and record linkage with the Pneumoconiosis Compensation Board, Cancer Registry, Chest Clinics, Hospital Authority and Registry of Death. The date of death, the death registry entry number and the causes of death as stated in the death certificate for each known death were reviewed. The relevant information of the deceased was sent to the Census and Statistics Department to retrieve the codes of the underlying causes of death. The 9th version of the International Classification of Disease (ICD-9th) was used throughout the study period.
The entry date of follow-up for each subject was defined as the date of diagnosis of silicosis. The exit date represented the date of death, the date of lost-to-follow-up or the study end date (December 31, 1999). The standardized mortality ratio (SMR) of esophageal cancer was calculated as the ratio of the observed number of deaths to the expected number of deaths. The 95% confidence interval (CI) for the SMR was derived assuming a Poisson distribution for the observed numbers.8 The expected number of deaths for esophageal cancer was calculated by multiplying the age-period-specific person-years at risk by the corresponding death rates of males in the total Hong Kong population obtained from the Census and Statistics Department. The SMR of liver cirrhosis was similarly calculated to estimate alcohol-related mortality.
Subjects of this cohort were classified into caisson workers, surface construction workers and workers in other dusty occupations for subgroup analysis. Caisson workers were those who had ever worked in underground caissons, where they were exposed to very high levels of silica dust. Surface construction workers worked only in surface construction or related jobs including stone cutting, pneumatic drilling, rock and sand blasting, stone crushing and quarrying, etc., and were exposed to lower levels of silica dust but not other known carcinogens. All other workers who had exposures to silica dust but not classifiable into the above 2 groups were regarded as workers in other dusty occupations.
As the prevalence of ever-smokers among silicotic workers (89.57%) was much higher than that among males of similar age (51.3%) in the general population,9, 10, 11 smoking could distort the association between silicosis and esophageal cancer. The indirect method proposed by Axelson was used to adjust for the potential confounding effect of tobacco smoking.12, 13 We applied the median odds ratio (1.81) of esophageal cancer for ever-smokers compared to nonsmokers obtained from a local study on Chinese subjects14 in the indirect adjustment. Ever-smokers included both current and ex-smokers. We used the same approach to adjust for the differences in the prevalence of alcohol drinking between the cohort and the general population by applying a median odds ratio of 3.40.14 Because we did not have good data on alcohol drinking in the cohort, we adopted a more conservative approach by assuming all workers in the cohort were drinkers. The prevalence of alcohol drinking in the Hong Kong male general population of similar age was 55.4%.15
We used the duration (years) of exposure and the radiologic severity of silicosis as surrogates to explore the dose-response relationship between esophageal cancer and silica dust exposure.
A total of 2,925 newly diagnosed cases of silicosis were seen at the Pneumoconiosis Clinic from January 1, 1981 to December 31, 1998; 21 cases were females, 53 had a history of cancer and 62 had a history of occupational exposure to asbestos and were all excluded, leaving 2,789 male subjects for follow-up. They contributed a total of 24,992.6 person-years of observations. The vital status was successfully ascertained in 2,707 workers (97.1%) of this cohort, and 853 known deaths (30.6% of the cohort) had occurred by the study end date of December 31, 1999. Death certificates or notifications were obtained for 809 (94.8%) deaths, and the causes of death were known for 804 deaths (94.26%). Details of the characteristics of the 3 occupational groups and the mortality experience from other major causes of deaths have been reported elsewhere (data submitted). Caisson workers were generally younger at entry into the cohort (diagnosis of silicosis), with a mean age of 49.2. The corresponding mean age was 57.6 and 57.0 for surface construction workers and workers in other dusty occupations, respectively. The mean age at death from esophageal cancer was 59.8, 60.5 and 72.7 for caisson, surface construction and other dusty occupations, respectively. The prevalence of ever smokers was 92%, 89% and 87% among caisson workers, surface construction workers and workers in other dusty occupations, respectively, and the corresponding mean pack-years were 21.8, 27.8 and 26.5, respectively.
The SMR of esophageal cancer in the entire cohort was 2.22 (95% CI 1.36–3.43), with 20 deaths. There were 9 deaths from liver cirrhosis (SMR 1.22, 95% CI 0.56–2.32). The highest risk of esophageal cancer deaths was observed among underground caisson silicotics (SMR 4.21, 95% CI 1.81–8.30). The risk was less elevated among the surface construction workers and workers in other dusty trades (Table I). Taking all non-caisson silicotics together, the SMR became 1.69 (95% CI 0.87–2.95, based on 12 deaths). The SMR of liver cirrhosis was slightly increased for non-caisson silicotic workers (SMR 1.48, 95% CI 0.64–2.91, based on 8 deaths), whereas no excess risk was found for caisson workers (Table I).
|Occupational groups||No. in cohort||Person-years of observation||Esophageal cancer SMR (95% CI) [observed no. of deaths]||Liver cirrhosis SMR (95% CI)|
|Caisson||1,027||8,500.52||4.21 (1.81–8.30) ||0.51 (0.01–2.83) |
|Surface construction||1,425||13,409.83||1.72 (0.82–3.16) ||1.36 (0.50–2.97) |
|Others||317||3,082.21||1.55 (0.17–5.58) ||2.00 (0.22–7.22) |
|All groups||2,789||24,992.56||2.22 (1.36–3.43) ||1.22 (0.56–2.32) |
The relative risks attributable to potential confounding by smoking and alcohol drinking and the indirectly adjusted SMRs are shown in Table II. The SMR of esophageal cancer for caisson silicotics was reduced from 4.21 to 2.34 after adjusting for cigarette smoking and alcohol drinking, with a lower 95% confidence limit of 1.01. The adjusted attributable risk fraction among caisson workers was 57.3%. No more excess mortality of esophageal cancer was found for non-caisson silicotic workers after the adjustments. A direct comparison of the risk of death from esophageal cancer between caisson workers and non-caisson workers using Cox's proportional hazards regression showed a risk ratio of 1.70 (95% CI 0.63–4.57) for caisson workers after adjusting for age and smoking pack-years. The risk ratio increased to 2.02 (95% CI 0.75–5.42) if we only count caisson workers who had worked in underground caissons for at least one year.
|Adjustment for risk factors||Caisson||Surface construction||Others||All groups|
|SMR before adjustment||4.21||1.72||1.55||2.22|
|RR attributable to smoking2||1.23||1.21||1.20||1.22|
|RR attributable to drinking3||1.46||1.46||1.46||1.46|
|Smoking- and alcohol-adjusted||2.34||0.97||0.88||1.25|
|SMR (95% CI)||(1.01–4.62)||(0.46–1.79)||(0.50–3.18)||(0.76–4.05)|
Table III shows the SMR of esophageal cancer by duration of exposure and severity of silicosis. A dose-response trend was seen among caisson workers for both the duration of exposure and the radiologic severity, although the trends were not statistically significant.
|Caisson workers||Non-caisson workers|
|O||E||SMR (95% CI)||O||E||SMR (95% CI)|
|Duration of exposure (years)|
|16–30||4||1.25||3.20 (0.86–8.19)||7||3.19||2.19 (0.88–4.52)|
|>30||4||0.46||8.77 (2.36–22.46)||2||3.10||0.64 (0.07–2.33)|
|Total||8||1.90||4.21 (1.81–8.30)||12||7.11||1.69 (0.87–2.95)|
|Test for trend||NS||NS|
|1||1||0.53||1.89 (0.02–10.54)||4||1.65||2.42 (0.65–6.19)|
|2||4||0.73||5.47 (1.47–14.01)||4||3.41||1.17 (0.32–3.00)|
|3||1||0.22||4.61 (0.06–25.64)||2||0.77||2.59 (0.29–9.37)|
|Subtotal||6||1.48||4.07 (1.48–8.85)||10||5.83||1.71 (0.82–3.15)|
|Test for trend||NS||NS|
|A||1||0.29||3.47 (0.05–19.32)||2||0.92||2.19 (0.25–7.89)|
|Subtotal||2||0.42||4.72 (0.53–17.03)||2||1.27||1.57 (0.18–5.67)|
|Test for trend||NS||NS|
By following up a cohort of 2,789 workers diagnosed to have silicosis in Hong Kong, we found that their mortality risk from esophageal cancer was high with an SMR of 2.22. The risk was particularly high in the subgroup that had worked in underground caissons where a very high silica dust level existed.16 The risk among this group of silicotic workers was still significantly raised, with an SMR of 2.34, after indirectly adjusting for cigarette smoking and alcohol drinking using very conservative estimates with possible over-adjustment. A dose-response relationship also existed with the duration of exposure and the radiographic severity of silicosis in this subgroup, although the trends were not statistically significant due to the small numbers.
Results from previous studies gave inconsistent and conflicting results on the association between silica dust exposure and esophageal cancer. A number of studies on workers exposed to silica dust or cohorts of silicotic workers reported positive associations,5, 17–24 but most of the results were not statistically significant due to the small numbers of esophageal cancer cases.18, 19, 20, 21, 22, 23 Other industry-based studies found negative associations.25, 26, 27 None of the above studies attempted to control for the potential confounding effects of cigarette smoking and alcohol drinking, as the major focus of these studies was on the association between silica dust exposure or silicosis and lung cancer.
As a follow-up to the earlier proportional mortality ratio (PMR) study that showed an increased standardized PMR of 2.2 among refractory brick makers in China,17 Pan et al. conducted a nested case-control study to specifically look into the relationship between esophageal cancer deaths and occupational exposure to silica dust,6 using deaths from other causes as controls. After adjusting for confounding factors (including alcohol drinking, cigarette smoking and meat and fruit consumption), occupational exposure to silica dust had an odds ratio of 2.8, and a clear dose-response relationship by length of exposed years was present. The authors postulated that ingestion of silica particles after lung clearance increased the risk of esophageal cancer among workers exposed to silica. A population-based case-control study in Japan also found a 2-fold increased risk of esophageal cancer among silicosis patients after adjusting for age and smoking but with a wide confidence interval (95% CI 0.87–6.23).28 However, using other cancer deaths as controls might have reduced the credibility of the findings. In the United States, a total population PMR study conducted using computerized U.S. vital statistics from the National Center for Health Statistics found that mortality from esophageal squamous cell carcinoma was high among occupations potentially associated with exposure to silica dust and chemical solvents and detergents.29 However, no adjustment was made to the possible confounding effects of cigarette smoking and alcohol drinking, and the authors claimed that their findings on the associations between occupational exposures and esophageal cancer could be due to confounding by smoking and drinking. On the other hand, a population-based case-control study among men in the Montreal area of Canada did not find any association between workplace crystalline silica exposure and esophageal cancer after adjusting for smoking and alcohol drinking.30 However, this study had only 99 cases of cancer of the esophagus to study occupational exposures to more than 30 substances. Furthermore, any exposures to crystalline silica were counted, and the average exposure level/dose would likely be much lower than cohorts of workers occupationally exposed to silica dust.
Our study was the first cohort study of silicotic workers to show an increased risk of death from esophageal cancer after taking into consideration 2 major nonoccupational risk factors, i.e., cigarette smoking and alcohol drinking.14, 31 Although our study subjects were based on a compensation register, we believe that selection bias was minimal, and we should have included most if not all workers with silicosis in Hong Kong. Compensation for silicosis in Hong Kong would be paid out for all confirmed cases even if they had normal lung function and were asymptomatic. Furthermore, the frequent publicity program on silicosis and the compensation system would have encouraged workers with mild silicosis to register for assessment also and helped minimize the selection bias of our cohort members. Misclassification of the underlying cause of death was possible but should not be serious. About 80% of all deaths in Hong Kong occurred in local hospitals, and 99.9% of death certificates were signed by medical practitioners.32 The misclassification, if present, would be expected to apply to both the cohort and the general population and would have been random, as we utilized the same source of information, i.e., the cause of death recorded by the Census and Statistics Department, for both groups and did not attempt to recode the death certificates ourselves. The random misclassification could have diluted the association. We believe that the RRs of esophageal cancer attributable to smoking and alcohol used in the indirect adjustment14 were appropriate and conservative, although higher RRs had been reported in other studies.33 The study by Cheng et al.14 was conducted among Hong Kong Chinese and would best represent the local situation. RRs of specific diseases for various risk factors can vary in different localities due to the interactions with other local risk factors as well as genetic differences. By using the highest RR of esophageal cancer for smoking (4.45) and alcohol (4.03) reported in the literature,33 the adjusted SMR among caisson workers became 1.86 (95% CI 0.80–3.66). The risk was still increased, though not statistically significant. Moreover, hospital-based case-control studies14, 33 that excluded patients with diseases related to tobacco and/or alcohol from the control groups might lead to overestimations of the relative risks attributable to smoking and alcohol.
Caisson workers in Hong Kong were exposed to very high levels of free silica dust in the form of quartz, which were frequently several hundred times above the threshold limit value.16 The increased SMR of esophageal cancer after adjusting for both cigarette smoking and alcohol drinking in this subgroup, but not the other two subgroups, would suggest that a very high exposure to silica dust could be responsible. This finding was similar to the original Italian study on lead-zinc miners that first showed excess mortality from esophageal cancer among underground miners but not among surface workers.5 Underground caisson workers were exposed to high levels of radon daughters in Hong Kong,34 but there has been no report of an association between radon exposure and esophageal cancer before. The esophagus is potentially exposed to inhaled dust particles because dust deposited in the respiratory tract is ultimately cleared by the mucociliary apparatus and is swallowed. The ingested sharp silica fragments may have the ability to cause local repeated injuries during their passage down the esophagus and to be buried in the mucosa, stimulating proliferation by providing anchorage.3
Although we have taken a conservative approach by assuming all silicotic workers were drinkers, residual confounding by alcohol drinking was still possible after the indirect adjustment if caisson workers drank more heavily than other workers. However, the SMR of liver cirrhosis by occupational groups (Table I) did not support this postulation. Data on smoking history in the 3 subgroups showed that caisson workers smoked fewer pack-years on average than the other 2 groups, and residual confounding by heavy smoking was unlikely. We do not have good data on the risk of esophageal cancer among ex-smokers and current smokers separately, and hence the indirect adjustment for smoking was made for all ever-smokers. If we assume the risk among ex-smokers to be half that of current smokers, the relative risk of esophageal cancer attributable to smoking would be less than 1.23 (Table II) and the adjusted SMR would be higher than 2.34 in the caisson workers. Data on occupational history did not show that our caisson workers were particularly engaged in or exposed to occupational settings that had been reported to be associated with an increased risk of esophageal cancer.6, 29, 30, 31, 35–37 Silicotic workers generally had lower socioeconomic status, which, as a surrogate for certain dietary habits and proximal risk factors, could have put them at a higher risk of esophageal cancer death. However, this cannot explain the difference in the risk of esophageal cancer between caisson workers and other workers as they all belonged to the same socioeconomic group.
In conclusion, this historical cohort study revealed that there was an increased mortality risk of esophageal cancer among silicotics who had worked in underground caissons in Hong Kong after adjusting for cigarette smoking and alcohol drinking. We believe that the excess risk of esophageal cancer mortality among caisson workers with silicosis could best be explained by the very heavy exposure to free silica dust in their working environment.
- 4International Agency for Research on Cancer. IARC monographs on the evaluation of carcinogenic risks to humans: silica, some silicates, coal dust, and para-aramid fibrils, vol. 68. Lyon: IARC Publications, 1997.
- 7ILO international classification of radiographs of pneumoconiosis. In: StellmanJM, ed. Encyclopaedia of occupational health and safety, 4th ed. Geneva: International Labour Office, 1998; 10: 32–7..
- 8Statistical methods in cancer research, vol. II. The design and analysis of cohort studies, vol. 2. No.82. Lyon: IARC Scientific Publications, 1987. 48–79., .
- 9Census, Statistics Department, Hong Kong Government. General Household Survey Special Topics Reports. No.7. 1991.
- 10Census, Statistics Department, Hong Kong Government. General Household Survey Special Topics Reports. No.11. 1994.
- 11Census, Statistics Department, Hong Kong Government. General Household Survey Special Topics Reports. No.15. 1997.
- 13Aspects of confounding in occupational health epidemiology. Scand J Work, Environ & Health 1978; 84–9..
- 19Lung cancer risk among pneumoconiosis patients in Japan, with special reference to silicotics. In: Occupational exposure to silica and cancer risk. Lyon: IARC, 1990. 95–104., , , .
- 26Further probe into the correlation between silica, silicosis and lung cancer in tungsten mines (in Chinese). Indus Health Occup Diseases 1999; 25: 209–13., , , , , , , .