Objective The aim of our study was to estimate the incidence of digestive cancers within a cohort of asbestos-exposed workers.
Methods Our study was based on a cohort of 2024 participants occupationally exposed to asbestos. The incidence of digestive cancers was calculated from 1 January 1978 to 31 December 2009 and compared with levels among the local general population using Standardised Incidence Ratios (SIRs). Asbestos exposure was assessed using the company’s job-exposure matrix.
Results 119 cases of digestive cancer were observed within our cohort, for an expected number of 77 (SIR=1.54 (1.28 to 1.85)). A significantly elevated incidence was observed for peritoneal mesothelioma, particularly in women. Significantly elevated incidences were also observed among men for: all digestive cancers, even when excluding peritoneal mesothelioma (SIR=1.50 (1.23 to 1.82)), oesophageal cancer (SIR=1.67 (1.08 to 2.47)) and liver cancer (SIR=1.85 (1.09 to 2.92)). Concerning colorectal cancer, a significant excess of risk was observed for men with exposure duration above 25 years (SIR=1.75 (1.05 to 2.73)).
Conclusions Our results are in favour of a link between long-duration asbestos exposure and colorectal cancer in men. They also suggest a relationship between asbestos exposure and cancer of the oesophagus in men. Finally, our results suggest a possible association with small intestine and liver cancers in men.
Statistics from Altmetric.com
What this paper adds
Asbestos can cause cancers in non-pulmonary localisations (laryngeal and ovarian cancers, mesothelioma), and certain digestive cancers, especially colorectal cancer, may be related to asbestos exposure.
In our study, colorectal cancer was associated with long-asbestos exposures (more than 25 years) in men, and oesophageal cancer was associated with asbestos exposure in men, especially for those having a Cumulative Exposure Index above 80 fibres/mL×years.
We also observed a link between asbestos exposure, and small intestine and liver cancers.
Our study provides data in favour of a link between asbestos exposure, and colorectal and oesophageal cancers.
Until recently, the medical consequences of asbestos exposure were considered to essentially concern the respiratory tract: benign pleural pathologies, asbestosis and malignant pathologies (pleural and peritoneal mesothelioma, primary bronchopulmonary cancer).
In a recent International Agency for Research on Cancer (IARC) monograph, laryngeal and ovarian cancers were also linked to asbestos exposure, with sufficient evidence in humans.1
With regard to digestive localisations, Selikoff, followed by Miller, were the first authors to suggest a link between asbestos exposure and certain digestive cancers.2–5 The results of later studies, often based on mortality data, diverge.6–20 Colorectal and stomach cancers are, nonetheless, considered by the IARC as potentially associated with asbestos exposure, despite limited evidence in humans.1 Owing to the presence in the Calvados department of an asbestos processing plant, a cohort was set up and specific monitoring implemented, including precise knowledge of asbestos exposure for each worker, thanks to a company-specific job-exposure matrix (JEM) and reliable cancer identification via a population-based registry. In a previous report on this cohort, we observed a significant association between asbestos exposure and cancer of the small intestine in men, and cancer of the oesophagus in men.21 The objective of the present study is to investigate the risk of digestive cancers among asbestos-exposed workers over an extra 5-year follow-up period for the same cohort.
Materials and methods
The retrospective cohort comprised participants who had worked in an asbestos reprocessing plant (production of textile materials and friction lining) located in the south of the Calvados department in Normandy, France. All participants meeting the following three conditions were included: alive in 1978, having worked for at least 1 year in the plant and having resided in Calvados during at least part of the study period.
Precise knowledge on sociodemographic data (name, surname, gender, date and place of birth) was available for each of the participants in the cohort thanks to the consultation of files held in the employing company’s occupational health department. Vital status at 31 December 2009 and the date of any death or relocation outside the Calvados department during the study period were collected from several data bases: consultations at the Caen University Hospital’s occupational and post-occupational pathology department, electoral roles obtained from the council of the last known place of residency and letters addressed to local councils in the subject’s place of birth.
Data on professional history and occupational exposure to asbestos came from both medical files held in the employing company’s occupational health department as well as from a company-specific JEM: date of first employment, date of departure from the company, exposure sector (textile/friction), type of asbestos handled (chrysotile or both chrysotile and amphibole) and Cumulative Exposure Index (CEI) for asbestos expressed in ‘fibres/mL×years,’ calculated thanks to the company’s own JEM.
Given that Calvados has its own digestive cancer registry since 1978, any case of cancer occurring between 1978 and 31 December 2009 (point date) in a participant from the cohort after the beginning of his/her exposure was recorded and precisely documented (date of diagnosis, anatomic site, histological type). The incidence of digestive cancers was thus estimated for each site (International Classification of Diseases for Oncology (ICD-O) 3 coding), excluding haematological tumours. All cases of peritoneal mesothelioma were validated by an expert pathologist from a national college of experts in pleural and peritoneal mesothelioma diagnosis, and recorded in the French multicentre mesothelioma registry.
The JEM was developed thanks to annual dust accumulation measurement data collected by specialised laboratories, since 1960, in numerous points of the company's workshops. Methodological aspects of its development have been described in detail elsewhere.22 Briefly, the first dust accumulation measurement device (used from 1960 to 1974) was manufactured by ARM (Avy-Raillère-Martin), which provided the total number of particles per litre of air. A second device (CASELLA) was used from 1974 onwards, based on a membrane filter method, the results being expressed as fibres per ml (one fibre corresponding to a length of at least 5 µm, a diameter of less than 3 µm and a length/diameter ratio superior to 3). A conversion factor between the two devices was established based on simultaneously measured data in 1974, enabling measurement results from the ARM device to be converted into fibres per mL. No measurement was available before 1960; consequently, for this period, an exposure model was developed by applying coefficients, based on production reports and retired executives’ testimonies, to the first available measurements. The final exposure level for each individual corresponded to the cumulated exposure (in fibres/mL×years) at the end of his/her period of exposure.
The participation of each individual started at the date of employment in the asbestos plant if prior to 1978, or in 1978. Individuals participated in the cohort until their date of death (if applicable), date of relocation out of Calvados (if applicable), date of last news (if lost to follow-up), date of first digestive cancer (if applicable), or 31 December 2009, whichever occurred first.
In order to obtain Standardised Incidence Ratios (SIRs: ratio between observed cancers and expected cancers), the number of expected cancers for the 1978–2009 period was estimated for each year and each anatomic site, using year-by-year annual cancer incidence levels calculated from Calvados Cancer Registry data and standardised for age (using 5-year age categories) and sex.
Cancer sites included: colon/rectum (ICD-O-3 topographic codes C180 to C20X); oesophagus (C150-C159); peritoneum (C481-C482); stomach (C160-C169); small intestine (C170-C179); liver (C220-C229); bile ducts (C230-C249); pancreas (C250-C259) and anal canal (C210-C218).
Analysis of cancer incidence was performed for three different exposure indicators: CEI expressed in fibres/mL×years and corresponding to the cumulative exposure throughout the participants professional career; exposure duration expressed in years; and exposure levels expressed in fibres/mL, equal to the ratio between CEI and exposure duration. For each exposure indicator, two equally numbered groups were formed on each side of the median, which corresponded to 80 fibres/mL×years for the CEI, to 25 years for the duration of exposure and to 4 fibres/mL for the mean atmospheric level of exposure to asbestos.
The cohort included a total of 2024 individuals: 1605 men (79.3%) and 419 women (20.7%). Six hundred and forty-five participants (31.9% of the population) died and 154 (7.61%) were lost to follow-up between 1978 and 2009; 489 (24.2%) left the department. The total number of person-years (all digestive localisations included) was 44 555. Table 1 provides details about the characteristics of the study population.
One hundred and nineteen cases of digestive cancers were registered from 1 January 1978 to 31 December 2009, among which 89% were men and 11% were women.
Table 2 provides the SIR for each cancer localisation.
In male participants, a significantly elevated incidence was observed for peritoneal mesothelioma, liver cancer, oesophageal cancer, and for all digestive cancers combined, including or excluding peritoneal mesothelioma. In female participants, the only localisation with a significant excess of risk was peritoneal mesothelioma.
Table 3 provides analysis of results based on the CEI in men.
For the lowest exposure class (CEI ≤80 fibres/mL×years), no significantly elevated incidence of digestive cancer was observed, all localisations considered.
For the highest exposure class, a significantly elevated incidence of peritoneal mesothelioma was observed in both sexes, and SIRs were four times higher in women than in men. Within this exposure class, localisations with significant higher risks among men were: liver (number of observed cases n=15), small intestine (n=3), oesophagus (n=20) and all digestive cancers combined, including or excluding peritoneal mesothelioma.
Over and above peritoneal mesothelioma, an elevated risk of colorectal cancer was observed (n=19) in men with the highest exposure duration (in excess of 25 years; table 4). The excess of risk of liver cancer was only significant in men with the lowest duration of exposure (n=11). In women, an increased risk was observed exclusively for peritoneal mesothelioma. Elevated risks were observed for two other localisations in men exposed to the highest atmospheric levels (table 5): small intestine (n=3) and liver (n=17).
The risk of cancer was significantly associated with asbestos exposure for many digestive sites in our study.
The leading digestive cancer site was peritoneal mesothelioma, which was constantly associated with increased SIRs. Depending on the exposure criterion considered, SIRs varied from 13.3 to 28.6 in men, and from 42.4 to 74.8 in women. Moreover, for a given criterion, the SIR in women was approximately four times higher than in men, indicating that women may be more likely to develop peritoneal mesothelioma than men. This has already been reported in the literature and our findings confirm the well-established link between asbestos exposure and peritoneal mesothelioma.23
We also observed an excess of risk of colorectal cancer among men with the longest exposure duration (more than 25 years). This is consistent with other findings in the literature.10 15 16 19 In particular, Smailyte et al16 observed a significantly increased SIR for colorectal cancer in male asbestos-cement workers for a duration exposure of 10 years or more. In contrast, Offermans et al19 observed an elevated risk of colorectal cancer in workers ‘ever highly exposed’ to asbestos. Our results are also consistent with the recent IARC statement, which indicates that positive associations between colorectal cancer and asbestos exposure are most likely to appear in studies with a long-exposure duration.1 However, no duration threshold has been repeatedly investigated in the literature; it is consequently difficult to specifically define a ‘long exposure duration’. When considering the CEI, the SIR was elevated but non-significant among men with the strongest exposure (CEI above 80 fibres/mL/years).
Over and above these two localisations, for which a link with asbestos exposure has previously been established (peritoneal mesothelioma) or suspected (colorectal cancer), our study reports significant associations for a number of other sites.
A significant excess of risk of oesophageal cancer was observed in men in general (SIR 1.67 (1.08 to 2.47)), and in men with the strongest CEI (SIR 1.90 (1.16 to 2.94)). The SIR was also elevated and close to significant among men exposed to the highest asbestos-dust atmospheric levels (SIR 1.63 (0.98 to 2.55)). Certain authors have reported increased mortality from oesophageal cancer in cohorts of asbestos-exposed workers.24 25 In the aforementioned studies, standardised mortality ratios (SMR) had the same magnitude as the SIR in our own cohort: SMR 1.61 (1.13 to 2.40) among American insulators;24 and SMR 1.87 (1.09 to 2.99) among workers from an asbestos-textile plant.25 However, a recent and well-conducted meta-analysis showed neither a link nor a dose-effect relationship between asbestos exposure and oesophageal cancer.26 In our population, the relatively high number of observed oesophageal cancers (25 cases) and the elevated risk in men with the strongest CEI are criteria in support of the validity of our results.
We also report a significant increased risk of cancer of the small intestine among men exposed to the highest asbestos-dust atmospheric levels (SIR 5.88 (1.18 to 17.2)) and those with the strongest CEI (SIR 6.50 (1.31 to 19.0)). To the best of our knowledge, such increased risk has never been reported in another cohort, possibly because this localisation has rarely been investigated. However, since very few risk factors are established or suspected, it is unlikely that our results are due to a confounding factor. The low number of observed cases among men (3) prompts two remarks. On the one hand, this could be due to pure chance; in support of this hypothesis, the overall SIR was not statistically significant, with a wide CI (SIR 4.47 (0.90 to 13.1)). On the other hand, given the rarity of cancer of the small intestine,27 and given the expected number of cases (0.67), this could well suggest an explicative factor other than pure chance. Longer follow-up of the cohort may help us determine the most appropriate explanation.
Finally, we report an association between liver cancer and asbestos exposure in men, in general (SIR 1.85 (1.09 to 2,92)) and for the upper class of two exposure indicators: CEI (SIR 2.18 (1.22 to 3.59)) and mean level of asbestos-dust exposure (SIR 2.26 (1.31 to 3.61)). In contrast, a significant increased risk of liver cancer was observed in the lower class of exposure duration. We found no correlation likely to explain this phenomenon (eg, correlation between short-duration exposure and high-intensity exposure).
A few cohort studies dealt with the risk of liver cancer in relation to asbestos exposure; their results were contradictory.21 25 28 29 In an exploratory case–control study, Brandi et al30 reported a significantly elevated risk of intrahepatic cholangiocarcinoma in Italian workers exposed to asbestos. However, liver cancers in our population were essentially hepatocellular carcinomas (11 of 18).
Biological findings may well support the hypothesis of an association between asbestos exposure and liver cancer: asbestos fibres have already been found in liver,31 32 and a translocation pathway has been proposed.33 It has also been suggested that, since asbestos impairs the immune system, it may favour the development of liver tumours.34 However, to date, no epidemiological data has been able to confirm this hypothesis. Our findings have narrow CIs, which is a consistent characteristic. Moreover, the fact that standardisation was based on the local population should reduce confounding bias, in particular concerning alcohol consumption, but we cannot ascertain it formally. Thus, our results should be considered with caution, and potential links between liver cancer and asbestos exposure should be investigated by studies with appropriate design.
Our study was limited by a lack of statistical power, particularly for women.
We nevertheless wish to emphasise the strengths of our study: first, the existence of a company-specific JEM, based on the measurement of atmospheric fibre concentrations, enabled us to precisely quantify exposure for each participant included in the cohort.
The reference population used for incidence ratio standardisation was local and very likely to be similar to the cohort, in particular with regard to confounding factors for each cancer (eg, alcohol and tobacco consumption for oesophageal cancer, alcohol intake for liver cancer, dietary habits and body mass index for colorectal cancer). Although we were unable to formally verify this hypothesis because of the lack of data for the main confounding factors, the choice of this reference population should consequently reduce the risk of comparison bias. Moreover, highly accurate and year-by-year incidence data were available for standardisation.
Using incidence data from two specialised registries (respectively recording digestive tumours and mesotheliomas) ensured the most comprehensive case compilation possible; moreover, all cases were confirmed histologically. In particular, all cases of peritoneal mesothelioma were validated by an expert pathologist from the ‘mésopath’ group (national panel of experts), hence avoiding potential confounding with non-specific peritoneal carcinosis.
Our study provides new data in favour of an association between asbestos exposure and colorectal cancer in men, which is consistent with the most recent data in the literature. We also observed an association for oesophageal cancer, also in men. The literature is more divergent for this localisation; however, by avoiding major confounding bias and employing incidence data from a specialised registry, our own study provides reliable data. Our results suggest potential associations for small intestine cancer and liver cancer, both in men.
The authors would like to acknowledge Patricia Marion (DIRECCTE, Hérouville-Saint-Clair, France), Blanche Bazin (Service de Santé au Travail, GISTAF, Condé sur Noireau, France), Nolwenn Le Stang (Service d'anatomopathologie, CHU de Caen, Caen, France).
Contributors BC was involved in supervision and is the guarantor. MB performed data collection. BC, LG and GL were responsible for methodological guidance. FM participated in data analysis. MB was involved in article drafting. VB, BC, FG-S, LG, GL, M-FM, FM, CP and CR were responsible for article review and correction.
Competing interests None declared.
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.