Objectives To examine mortality from digestive cancers in a Chinese miner cohort and to explore the exposure–response relationship between chrysotile mining dust and site-specific digestive cancers.
Methods A cohort of 1539 asbestos miners was followed for 26 years. Information on vital status and death causes was collected from personnel records and hospitals. Underlying causes of death from cancers were determined by combination of clinical manifestations and pathological confirmation. Individual cumulative dust exposures were estimated based on periodic dust measurements of different workshops, individuals’ job title and employment duration, and treated as a time-dependent variable. Standardised mortality ratios (SMR) were calculated according to Chinese national data and stratified by exposure (levels 1–3, from low to high). Cox proportional hazard models were constructed to estimate HRs in relation to cumulative exposure with adjustment of smoking.
Results Fifty-one deaths from digestive cancers were identified in the cohort, giving an SMR of 1.45 (95% CI 1.10 to 1.90). There was a clear exposure–response relationship between asbestos dust exposure and mortality from stomach cancer, with SMR of 2.39 (95% CI 1.02 to 5.60) and 6.49 (2.77 to 15.20) at exposure levels 2 and 3, respectively. The clear relationship remained in multivariate analysis, in which workers at the highest exposure level had HRs of 12.23 (95% CI 8.74 to 17.12). In addition, excess mortality from oesophageal and liver cancers was also observed at high exposure levels.
Conclusions This study provides additional evidence for the association between exposure to chrysotile mining dust and excess mortality from digestive cancers, particularly stomach cancer.
Statistics from Altmetric.com
What this paper adds
The association between asbestos exposure and digestive cancers remained uncertain.
This study found that workers with high cumulative exposure to chrysotile mining dust had significantly excessive mortality from cancer of stomach, oesophagus and liver. A clear exposure–response trend was seen in stomach cancer when either external or internal comparison was made.
The study provides additional evidence for increased mortality from digestive cancers, particularly stomach cancer, associated with the high level of exposure to chrysotile containing dust.
Although more studies on this topic are needed, the current result highlights the necessity of banning all forms of asbestos globally.
Asbestos exposure has been well recognised as a causal factor for lung cancer and mesothelioma. Since Selikoff et al1 first reported an increased risk of gastrointestinal cancers in insulation workers in 1964, a number of epidemiological studies were conducted to examine the association between asbestos exposure and digestive cancers. However, the association has remained uncertain due to conflicting results from different studies. For example, a study from Western Australia showed no association between exposure to crocidolite and mortality from stomach or colorectal cancer or other gastrointestinal cancers.2 Later on, a systematic review showed that none of the various methods used to estimate asbestos exposure yielded consistent exposure–response trends and the strength of the associations was consistently weak or non-existent for four forms of gastrointestinal cancers.3 However, recent studies conducted in a French cohort suggested a positive relationship with colorectal, small intestine and oesophageal cancer.4 ,5 An earlier literature review also showed that a high level of asbestos exposure, as indicated by a standardised mortality ratio (SMR) for lung cancer of at least 2.0, was associated with elevated SMR for gastrointestinal cancer.6 In 2011, a report from the International Agency for Research on Cancer (IARC) suggested positive associations between exposure to all forms of asbestos and cancer of pharynx, stomach and colorectum.7 Nevertheless, more evidence is needed to address the issue.
There have been a limited number of studies on the associations between different types of asbestos and digestive cancers. One study suggested that subjects exposed to mixed asbestos had a higher risk of digestive cancers than those exposed to chrysotile only.8 Data from a meta-analysis focusing on workers exposure only to chrysotile asbestos indicated that there was borderline significance of excessive mortality from stomach cancer (meta-SMR=1.24, 95% CI 0.95 to 1.62).9 Due to insufficient evidence, the committee in the US Institute of Medicine (IOM) did not draw conclusions on the difference of selected digestive cancers in relation to any specific type of asbestos.10 IARC also noted that there was insufficient information in the literature to discern whether any differences existed among asbestos fibre types in their ability to cause stomach cancer.7 The existence of discrepancies in mortality from these cancers among workers exposed to different asbestos types makes the relationship remain unclear.11
In the whole profile of cancers in China, digestive cancers have contributed to the largest proportion of all cancer deaths, with preponderance at the site of stomach, oesophagus and liver.12 Several studies suggested that occupational risk factors, including asbestos exposure, were related to oesophageal and stomach cancer.13–16 Although amphibole asbestos has been banned in China, chrysotile asbestos continues to be mined and used widely, with production of 440 000 tons in 201217 and consumption of 613 760 tons in 2010.18 In this study, we examined mortality from digestive cancers, in particular, cancer of stomach, oesophagus and liver, in an asbestos miner/miller cohort from the largest chrysotile asbestos mine in China and explored the exposure–response relationships between asbestos-containing dust exposure and the site-specific cancers.
Subjects and methods
The cohort members were recruited from a chrysotile asbestos mine located in Qinghai Province of China, with a relatively high altitude and dry condition. Being the largest in China, the mine has been in operation since 1958, and its annual production was about 12 000 tons. A total 1539 male workers who had been employed for at least 1 year on 1 January 1981 were enrolled. These workers did not have any forms of cancer and accounted for over 90% of workers in the mine at that time. Job titles mainly included mining, milling, transportation, maintenance, packaging and other services.
We followed the cohort from 1 January 1981 to 31 December 2006 for a total of 26 years. Although nearly half of the workers retired from the mine during the follow-up period, we were able to trace all workers in the cohort, irrespective of their working status and vital status, largely because the mine administration kept good records and contacts with the workers.19 No one was lost in the end, and the follow-up generated 34 736 person-years from the cohort.
Information on individuals’ occupational history and vital status was obtained from personnel records in the mine; the causes and dates of death were ascertained from the mine hospital and local hospitals. Digestive cancers were diagnosed by combining clinical manifestations and pathological evidence or autopsy reports. To avoid the misclassification of diagnosis,20 all of the cancer outcomes included in the study were primary digestive cancers without any other site cancers. There were two cases of secondary cancer (one oesophageal cancer and one stomach cancer) that were excluded from the analysis. Underlying causes of death from cancers were further verified in the local cancer registry. Detailed employment data, including job titles, employment duration and date of retirement, were obtained. Smoking information was collected from workers or from their spouses if deceased. The study was approved by the Human Subject Committee of the Chinese University of Hong Kong.
Asbestos exposure assessment
As described elsewhere,19 the analysis of asbestos samples from the mine showed that no tremolite was detected with the limit of detection of 0.1%. Periodic data of asbestos dust concentrations in different workshops were available since 1984. Total asbestos dust was sampled based on the national criterion for dust measurement.21 Specifically, fixed point samplings were applied to collect dust in various workshops. Dust samples were collected from the breathing zone with a flow rate of 5 L/min (±0.1) for 4 h. For extremely high dust concentrations in some workshops, the sampling only lasted 15 min. The total suspended dust concentration measured in 1984 was 800 mg/m3 on average. Although the average concentration of total dust decreased to 140 mg/m3 in 1995 largely due to the introduction of wet processes and engineering controls, it remained much higher than the regulated maximal limit of 2 mg/m3 that was applied until 2002. The dust concentrations were further lowered in 2006 (table 1) but still exceeded the updated national standard of 0.8 mg/m3. The concentrations were particularly high in the two milling plants.
We used historical dust measurements at different workshops to estimate individual cumulative dust exposures. Cumulative exposures were formed in a time-dependent manner so that each increment of person-years was assigned to the exposure category to which a death would be assigned if it occurred at that time. The individual cumulative dust exposure was expressed in mg-years/m3 as the product of dust concentration in a specific workshop and the duration of work for individuals in the workshop, summed across all job titles.
We used life table method to calculate person years and compared mortality from specific sites of digestive cancers in the cohort with the Chinese general population. In doing that, we first used the Chinese national population data to calculate SMRs for digestive cancer, expressed as ratios of observed deaths to expected deaths in the cohort. The expected numbers were calculated based on person-years multiplied by the reference mortality rates. Person-years at risk for each cohort member were accrued from the date of entry into the cohorts until the date of death or the end of follow-up, and stratified into 5-year age intervals and two calendar time periods of 1981–1995 and 1996–2006. National male-specific mortality rates obtained from two nationwide censuses during 1990–199222 and 2004–200512 were used as the reference mortality rates corresponding to the two calendar times of 1981–1995 and 1996–2006. CIs (95% CI) of SMRs were calculated with an assumption of a Poisson distribution for the observed number of cancer cases.23
In the data analysis, age was used to define the risk set for each cancer death. Then we estimated cumulative exposure as a time-dependent variable after the risk sets were created by reference to the exact date that each person in a risk set reached the attained age of the index case.24 We categorised the cumulative exposures with a lag of 10 years into three levels, levels 1–3, by partitioning the cumulative exposures of the deaths from each kind of digestive cancer into tertiles.25 The deaths from colorectal, pancreatic and gallbladder cancers were not included in the stratified analysis due to a very small number. Furthermore, we applied Cox proportional hazard analysis to estimate HRs and 95% CIs for mortality from stomach, oesophageal and liver cancers in relation to the cumulative exposure levels. In the Cox models, the cumulative exposure variable was treated as time-dependent, the lowest exposure level (level 1) was a reference category and smoking status was included as a potential confounding factor. Smoking status was a dichotomous variable coded as never or ever smoker. In addition, tests for trend in the SMRs were conducted using a Poisson trend test for monotonic dose–response relation described by Breslow and Day.23 Tests for linear trend in HRs were performed by fitting models with median values of each exposure category as a continuous variable. Data analyses were carried out with the Statistical Package for the Social Sciences Software V.18.0 for Windows and Microsoft Office Excel 2007.
Table 2 presents characteristics of the cohort that was followed from 1981 to 2006. A total of 34 736 person years at risk were observed among 1539 workers. At the time of entry, mean age of the workers was 36 years and the duration of employment was 14 years. Eighty-five per cent in the cohort were ever smokers. Individual cumulative dust exposure was 109 mg-years/m3, on average, ranging from 2 to 3614 mg-years/m3. There were 428 deaths due to various causes, accounting for 28% in the cohort.
Table 3 shows the observed and expected deaths at each site of digestive cancer and corresponding SMR. Altogether, 51 deaths from digestive cancers were observed, with 82% from stomach, oesophageal and liver cancers. Deaths from other sites included four colorectal, three pancreases and two gallbladders. No death from peritoneal cancer was identified. Mortality from stomach cancer was 50% higher than expected (95% CI 0.91 to 2.47). In addition, mortality from colorectal, pancreatic and gallbladder cancers was also higher than expected, though with a small number of deaths. Mortality from all digestive cancers was increased by 45%, with statistical significance (95% CI 1.10 to 1.91).
Table 4 displays mortality from stomach, oesophageal and liver cancers by cumulative exposure level. There was a clear exposure–response trend for stomach cancer mortality, in which mortality at levels 2 and 3 of exposures was significantly higher than expected, with SMRs of 2.39 (95% CI 1.02 to 5.60) and 6.49 (2.77 to 15.20), respectively. On the other hand, significantly elevated mortality for oesophageal and liver cancers was only seen at level 2. No evident trend of mortality from oesophageal and liver cancers was seen with the cumulative exposure levels.
Table 5 shows HRs associated with the exposure levels in multivariate analysis. A highly significantly increasing trend was observed for stomach cancer when smoking status was included in the model. When compared with the reference group, those at levels 2 and 3 had a nearly 4-fold and 11-fold higher risk, respectively, for death from stomach cancer. For oesophageal and liver cancers, a significantly higher risk was seen at levels 2 and 3. When individual cumulative exposures were treated as a continuous variable in the models, the risk associated with exposure to every 10 mg-years/m3 was 1.011 (95% CI 1.010 to 1.013) for stomach cancer, 1.014 (1.011 to 1.016) for oesophageal cancer and 1.015 (1.013 to 1.017) for liver cancer, also indicating that the risks were significantly associated with asbestos dust exposure. No significant association was observed between smoking and mortality from any of the cancers.
We previously observed and reported significantly elevated mortality from lung cancer, with SMR of 3.59 (95% CI 2.76 to 4.66) in this asbestos miner/miller cohort.19 In this analysis, we observed excess mortality from digestive cancers among workers who were exposed heavily to asbestos in the cohort.
Among 51 deaths from digestive cancer observed in this cohort, the majority of cancers were at the site of stomach, oesophagus and liver. The analyses showed no excess mortality from any of these cancers at exposure level 1. However, the workers who were exposed to higher levels had excess mortality from each of the cancers, particularly for stomach cancer. An exposure–response gradient was most clearly seen for stomach cancer, in which the workers in the highest exposure level had a 6.5-fold increased mortality rate than that of the general population. Furthermore, the highest exposure level had the highest relative risk of mortality from stomach cancer. The result suggested a close relationship between exposure to chrysotile mining dust and stomach cancer.
It has been suggested that stomach cancer is the most likely result of non-pulmonary outcomes of asbestos exposure.6 ,14 Inhalation of asbestos dust in the air is the major route for workers’ exposures, whereas it is estimated that one-twentieth of the inhaled dust would be ingested.26 Most reports on a positive association between asbestos and stomach cancer were from relatively earlier studies.27–30 This could mean a positive result was observed in heavily exposed workers that were more commonly seen several decades ago. For example, in a British study, excess mortality from gastrointestinal cancers was only detected among workers with high asbestos exposure during the 1930s and 1960s.31 A slight but significant increase in proportionate mortality ratio from gastric cancer was reported in occupations subject to high asbestos exposure.14 Another study observed a significant association between the diffuse subtype of stomach cancer and asbestos-related occupations (OR=3.71, 95% CI 1.40 to 9.83).32 A study conducted in Quebec chrysotile miners found excess mortality from stomach cancer at a higher exposure level (300 mpc·y).33 In this Chinese miner/miller cohort, workers from milling workshops were heavily exposed to asbestos dust (table 1), particularly during the earlier periods before dust control measures were introduced. Most of the workers who had a higher exposure level were from milling workshops. This result was consistent with the abovementioned studies. IARC recently concluded that positive associations were observed between exposure to all forms of asbestos and cancer of stomach with limited evidence in humans.7 In addition, US IOM reported that the evidence was suggestive but not sufficient to infer a causal relationship between asbestos exposure and stomach cancer.10 The current data added additional evidence for a positive association with chrysotile asbestos exposure.
There are a limited number of studies that reported oesophageal cancer as another outcome of asbestos exposure. Nine deaths from oesophageal cancer were identified in the cohort. A significantly higher mortality rate was also found in the workers at the exposure of level 2. The results were consistent with some previous studies4 ,28 ,34 ,35 In a case–control study conducted in Spain, a 3.5-fold increased risk for oesophageal cancer was observed in workers with high asbestos exposure in comparison with workers with low exposure.34 In another study conducted in 3072 chrysotile textile workers from South Carolina of the USA,35 investigators reported a significant excess mortality from oesophageal cancer (SMR 1.87; 95% CI 1.09 to 2.99). A significantly elevated incidence of this cancer (SIR 1.85; 1.08 to 2.96) was also observed at cumulative exposure exceeding 80 f-years/mL among male workers in an asbestos reprocessing plant in Normandy, France.4 However, sample sizes in the available studies, including the present one, were generally small, limiting study power to detect a clearer association between asbestos exposure and oesophageal cancer. Apparently, further studies with larger sample size are required to clarify this association.
Previous reports on asbestos and liver cancer were sparse. We observed an increased mortality from liver cancer in the workers with high exposure level 2. This was consistent with results from a 33-year cohort study in chrysotile asbestos textile workers in China,16 in which there was a significantly increased mortality rate of liver cancer (SMR=2.30, p<0.01). Previous studies found that asbestos fibres or asbestos bodies were present in histological samples of oesophageal tissue,36 gastric tissue,37 biliary tract38 and colon tissue.39 It is possible that asbestos fibres can migrate to the liver and induce primary liver cancer. This possibility was confirmed by discovery of an asbestos dust-laden giant cell in the human liver.38 Unfortunately, we had no data to show whether the subjects with liver cancer had fibre-containing liver tissue to further verify that fibre could transfer to liver in this study. The result suggests an association between heavy asbestos exposure and liver cancer, although it does not necessarily mean a causal link, given that potential confounding effects, such as alcohol consumption and chronic hepatitis infection, could not be ruled out in these workers.
The consistent results from an external comparison with general population and an internal comparison within exposure levels suggested that the observed associations between asbestos dust exposure and these cancers were not likely to be a chance finding. Unlike previous studies that used job titles15 ,32 or mortality from lung cancer or mesothelioma as surrogates of asbestos exposure,2 ,6 this study used quantitative estimates of individual cumulative exposures, which might better reflect workers’ exposure levels. However, several limitations in the study should be noted. First, the individual cumulative exposures were estimated based on area sampling measurements, rather than individual samples, which might have led to exposure misclassification. Yet, we observed exposure–response relationships in either external or internal comparison, especially with stomach cancer mortality, which indicated that the estimated individual exposures reasonably reflected the workers’ exposure levels. Furthermore, we applied the same estimation method to assess the exposures of all workers in this cohort, and the assessors were blinded to the disease outcomes when estimating the workers’ cumulative exposures. Thus, the misclassification could be non-differential. Another limitation is the small number of digestive cancer deaths in this study, which limited the study power and widened the CIs. In addition, individual data except for smoking and age, such as alcohol consumption, Helicobacter pylori infection and hepatitis B and C virus infections, were unavailable. These factors might be confounding factors for the associations of interest. Nevertheless, the difference of these factors within the cohort and even at different exposure levels might not be big because the subgroups of workers in the cohort were comparable in socioeconomic status, living conditions and dietary habits. Finally, there might be a possibility of misclassification of metastatic lung cancer.20 ,40 However, all disease outcomes included in the analysis were exclusively primary digestive cancer without any other site cancers, which minimised the misclassification of outcomes.
In conclusion, this study observed excess mortality from stomach, oesophageal and liver cancers in the chrysotile asbestos miners/millers who were heavily exposed. There was a clear exposure–response trend with stomach cancer, which supported a positive association between exposure to chrysotile mining dust and stomach cancer.
The authors wish to thank Qinghai Provincial Center of Disease Control for providing some exposure data, and Prof Wang Mianzhen and Dr Du Lili for their participation in data collection and administrative staff in the asbestos mine for their cooperation. They also thank Prof Leslie T Stayner at University of Illinois at Chicago for his assistance in data analysis strategies.
Contributors SL and XW developed the survey and study, supervised all aspects of its implementation, analysed the data and drafted the article. EY and IY helped conceptualise this article. YL organised the field survey. MNC was involved in analysing fibre types of the asbestos samples. DCC refined the article and contributed to the English editing. All authors contributed to the study, interpreted the results and reviewed the draft.
Funding This work was supported by the Pneumoconiosis Compensation Fund Board, Hong Kong SAR, China (PCFB 2008-2010).
Competing interests None.
Patient consent Obtained.
Ethics approval The study was approved by the Human Subject Committee of the Chinese University of Hong Kong.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.