Registration Dossier

Administrative data

Endpoint:
epidemiological data
Type of information:
experimental study
Adequacy of study:
supporting study
Study period:
1950-2002
Reliability:
1 (reliable without restriction)
Rationale for reliability incl. deficiencies:
other: see 'Remark'
Remarks:
Mulitcentre cohort study with clearly defined criteria for identifying and assembling the cohort, a large study base, a common strategy for exposure assessment, an almost complete follow-up and determination of causes of death, and a comparison of mortality with local and national rates.

Data source

Reference
Reference Type:
publication
Title:
Cancer Risk Among Tetrafluoroethylene (TFE) Synthesis and Polymerization Workers
Author:
Consonni D et al
Year:
2013
Bibliographic source:
American Journal of Epidemiology (in press)

Materials and methods

Study type:
cohort study (retrospective)
Endpoint addressed:
carcinogenicity
GLP compliance:
yes

Test material

Reference
Name:
Unnamed
Type:
Constituent

Method

Type of population:
occupational
Ethical approval:
confirmed and informed consent free of coercion received
Details on study design:
Study Population
After a feasibility study, four companies with seven production sites in the USA and Europe participated in the study. The six production sites which employed the entire European/USA TFE manufacturing population and were located in Gendorf (Germany), Dordrecht (The Netherlands), Spinetta Marengo (Italy), Hillhouse (United Kingdom), Bayonne (New Jersey, USA), and Washington Works (West Virginia, USA), were included in the study. A plant in Fayetteville, North Carolina USA that started operations in 1979 and employed only 31 workers (29 men, 2 women) in TFE processes was excluded. At three production sites (Spinetta Marengo, Hillhouse, and Bayonne), all workers of either gender who had ever worked in a plant with potential exposure to TFE were included. At the others, only workers with a minimum of one year (Gendorf and Dordrecht) or six months (Washington Works) of potentially exposed employment were included. In these production sites, this choice was motivated by uncertainties in the enumeration of short-term workers and because of incompleteness of their work history records. Short-term workers include part-time workers and those working in support roles before assignment. Since TFE production involves highly technical job assignments and extensive training, they are unlikely to have held a position in TFE divisions.
Exposure assessment:
estimated
Details on exposure:
Exposure assessment
Exposure assessment was undertaken for workers exposed to TFE. TFE is flammable and may present an explosive risk, and so production is mainly undertaken within closed systems. The main potential TFE exposure occurs from leaks, from opening autoclaves in the polymerization area or from decomposition of PTFE. The sparse measurement data available from the plants suggests TFE exposure levels could have been up to a few parts per million. In the absence of suitable measurements, a method of reconstructing exposure was developed. The final output was a job-exposure matrix (JEM) with yearly semi-quantitative estimates (in arbitrary units) of TFE exposure for relevant job titles at each production site, from the start of TFE production up to 2002. Plant-level information on potential exposure (yes/no) to asbestos, APFO (a perfluorinated surfactant) and to vinyl chloride monomer (VCM), was also collected.
The TFE sub-cohort from Washington Works in this multicenter study included 2,379 individuals with at least 6 months of employment in the TFE production unit in the period 1950-2002. Upon further review and categorization of the job titles by the exposure assessors, they were classified as exposed (1,554 workers), unexposed (820 workers), or having unknown exposure (five workers) due to missing work history records. The TFE sub-cohort is a subset of the entire occupational cohort of about 6,000 workers ever employed at the Washington Works production site in any of five production areas between 1948 and 2002. Plant-wide mortality had been previously evaluated by Leonard et al. (11).
Statistical methods:
Cause-specific observed mortality of each sub-cohort was compared with that expected according to national mortality reference rates for males in 5-year calendar periods and 5-year age groups. Rates for England and Wales were used for UK and rates for whites for the USA. Standardized Mortality Ratios (SMRs) were expressed as the ratio of observed to expected deaths, with their exact 95% Confidence Intervals (95% CIs) based on the Poisson distribution. For selected causes (all causes, all cancers, and lung cancer; cancer sites of a priori interest; cancers with SMR >1.00 in the overall analysis; liver and renal non-malignant diseases), analyses by length of exposure, time since first exposure, and cumulative exposure to TFE were performed. Two sided chi-square tests for linear trend were carried out by treating the variable as ordinal. Complementary SMR analyses were performed as follows: 1) using as reference regional mortality rates of Bavaria (Germany), Piedmont (Italy), Lancashire (UK), New Jersey and West Virginia (USA); 2) including only workers with complete job histories; and 3) by cumulative exposure to TFE after including plant workers never exposed to TFE. Statistical analyses were performed with Stata 11.

Results and discussion

Results:
This multi-centre study examined for the first time the cancer risk in the entire European and North American workforce exposed to TFE during synthesis and polymerization processes.

Increased SMR were estimated for cancers of a priori interest based on animal experiments, specifically liver, kidney, and leukaemia; however, SMR estimates were imprecise (large confidence intervals) and the exposure-response trends were not significant. Strengths of this study are the clearly defined criteria for identifying and assembling the cohort, the large study base, a common strategy for exposure assessment, the almost complete follow-up and determination of causes of death, and the comparison of mortality with local and national rates. Limitations were: the low power to detect a significant doubling of mortality for rare cancers (as calculated in the feasibility phase); an unknown degree of non-differential misclassification of TFE exposure estimates; and the semi-quantitative nature of exposure assessment, which precludes extrapolation of quantitative exposure-response relationships to other settings.

In summary, risk elevations, although with limited evidence for exposure-response relations, were seen for liver cancer, kidney cancer, and leukaemia, all outcomes suggested by experimental studies. Liver cancer showed an upward trend, although not significant, by increasing category of cumulative exposure. None of the potential confounders addressed emerged as possible alternative explanation for the findings.

The pattern of findings in this large study substantially narrows the range of uncertainty on the possible cancer risk entailed by working in TFE synthesis and polymerization, and justifies continuing efforts to minimize exposure, which has already dropped considerably over the years. However, the findings could neither conclusively confirm nor refute the hypothesis that TFE poses a carcinogenic risk to human beings. If a cancer hazard exists, then the risk is small, even in workers with relatively high exposure.
Confounding factors:
The study population includes all six current PTFE sites in Europe and North America. Since polymerization involves the use of ammonium perfluoro-octanoate (APFO), a compound with a carcinogenic potential (8-10) mortality in relation to exposure to this potential confounder was also examined.
Strengths and weaknesses:
Strengths of this study are the clearly defined criteria for identifying and assembling the cohort, the large study base, a common strategy for exposure assessment, the almost complete follow-up and determination of causes of death, and the comparison of mortality with local and national rates.
Limitations were: the low power to detect a significant doubling of mortality for rare cancers (as calculated in the feasibility phase); an unknown degree of non-differential misclassification of TFE exposure estimates; and the semi-quantitative nature of exposure assessment, which precludes extrapolation of quantitative exposure-response relationships to other settings.

Any other information on results incl. tables

The number of women eligible for inclusion was 778, of which 683 (87.8%) were employed at the Washington Works; numbers at the other sites ranged from 0 to 50. Among them, 16 deaths were recorded with no unusual mortality pattern (results not shown). In the following analyses female workers are not included.

After excluding 101 male workers because of missing date of birth and 21 lacking date of hire, the number of male workers eligible for analysis was 5,879 (Table 1). The termination of follow-up for individual sites varied between 2001 and 2008. Ascertainment of vital status was 98.8% complete. The number of unknown causes of deaths was small. The number of workers ever exposed to TFE according to the JEM amounted to 4,773 (81.2%) and that of never exposed to 1,081 (18.4%). The total number of person-years at risk was 148,483; TFE-exposed personyears amounted to 108,862 (73.1%). Only 25 workers (0.4%) had completely missing job histories. The number of workers with incomplete job histories was 306 (5.2%). The majority of these were in the German sub-cohort, in which the proportion of workers with incomplete job history was about one third, but that of working years in unknown jobs was lower (11.9%), indicating that many of those jobs were of short duration. Web Table 1 shows statistics for selected time-related variables for the 4,773 TFE-exposed workers. The oldest sub-cohort was in the UK. The sub-cohorts with the highest lengths of exposure were in The Netherlands and in Italy; in the Bayonne sub-cohort the median length of exposure was only 1.1 years. The overall average time since first exposure was 22.3 years (range: 16.6 to 28.8 years). Overall, cumulative exposure to TFE had a mean of 519 and a median of 191 arbitrary unit-years.

In comparison with national rates, mortality from all causes, all cancers, and many cancer and non-cancer causes among the TFE-exposed workers was lower than expected (Table 2).

Elevated SMRs were found for cancer of the liver, kidney, and leukemia, the cancer causes of a priori interest. Deaths from nephritis and nephrosis (three from chronic and one from unspecified renal failure), also of a priori interest, were below expectation. Liver cirrhosis was not in excess overall but was increased in the sub-cohort in Bayonne (15 observed, 3.6 expected). The SMRs calculated using regional rates for reference were quite similar to those calculated using national rates, e.g. 1.32 for liver cancer, 1.51 for kidney cancer, and 1.51 for leukemia.

Among those not exposed to TFE (including 26,515 person-years from 1,081 never exposed workers and 12,817 person-years accumulated by exposed workers before start of TFE exposure) a markedly lowered mortality from all causes was found (133 observed, 238.2 expected) and all cancers (38/61.1). There were no deaths from liver cancer (1.6 expected), leukemia (2.5 expected), and liver cirrhosis (7.5 expected). Three deaths were from pancreatic cancer (2.9 expected), two from kidney cancer (1.7 expected), and one from nephritis and nephrosis (1.7 expected). Seven deaths were recorded among the 25 workers (649 person-years) with unknown TFE exposure, with no unexpected pattern.

Table 3 shows SMRs by categories of cumulative exposure to TFE (in arbitrary unit-years). Among the 4,773 workers exposed to TFE, no definite exposure-response relationships were apparent for all cause or all cancer mortality. For liver cancer, there was a two-fold elevation of mortality among workers with the highest estimated cumulative exposure; the P value for trend, however, was high (P= 0.24). There was no trend with cumulative TFE exposure for kidney. cancer and leukemia mortality. Elevated mortality from liver cirrhosis in the low exposure category was attributable to the Bayonne sub-cohort (12 deaths against 2.4 expected) and underlay a clear negative trend in the whole cohort. No increases were observed for nephritis or

nephrosis. Mortality from pancreatic cancer was elevated in the medium and high exposure categories (P for trend = 0.21). Lung cancer mortality remained well below expectation in all categories. Esophageal cancer was above expectation only in the low exposure category. When the non-exposed person-years were added into the analysis, the only notable difference was aPvalue of 0.06 for test for trend for liver cancer. When person-years and deaths of workers with one or more jobs unknown (250 ever and 56 never TFE-exposed workers) were excluded, the results changed negligibly (results not shown).

Analyses by length of exposure to TFE for selected causes among workers ever exposed to TFE (Table 4) showed negative trends for all cancer and lung cancer mortality. For none of the other cancer sites/types were clear-cut patterns seen. The modest liver cirrhosis excess in the lower category was driven by the Bayonne sub-cohort, in which 13 deaths were observed (2.9 expected). Mortality by time since first exposure to TFE (Table 5) did not show any definite trend.

Applicant's summary and conclusion

Conclusions:
Tetrafluoroethylene (TFE), a compound used for the production of fluorinated polymers including polytetrafluoroethylene, increases liver and kidney cancer and leukemia incidence in rats and mice. This is the first time the cancer risk in humans is comprehensively explored in a cohort mortality study (1950-2008) which included all current polytetrafluoroethylene production sites in Europe and North America. A job-exposure matrix (1950-2002) was developed for TFE. Among 4,773 workers ever exposed to TFE, the authors found lower mortality from most causes, increased risks for cancer of the liver (SMR 1.27; 95% CI: 0.55, 2.51; 8 deaths) and kidney (SMR 1.44; 95% CI: 0.69, 2.65; 10 deaths), and for leukemia (SMR 1.48; 95% CI: 0.77, 2.59; 12 deaths). A non-significant upward trend (P = 0.24) by cumulative exposure to TFE was observed for liver cancer. This pattern of findings narrows the range of uncertainty on potential TFE carcinogenicity, but cannot conclusively confirm, nor refute, the hypothesis that TFE is carcinogenic to humans.
Executive summary:

Tetrafluoroethylene (TFE), a compound used for the production of fluorinated polymers including polytetrafluoroethylene, increases liver and kidney cancer and leukemia incidence in rats and mice. This is the first time the cancer risk in humans is comprehensively explored in a cohort mortality study (1950-2008) which included all current polytetrafluoroethylene production sites in Europe and North America. A job-exposure matrix (1950-2002) was developed for TFE and ammonium perfluoro-octanoate, a chemical used in the polymerization process. Standardized mortality ratios (SMR) and 95% confidence intervals (CI) were calculated using national reference rates. Among 4,773 workers ever exposed to TFE, the authors found lower mortality from most causes, increased risks for cancer of the liver (SMR 1.27; 95% CI: 0.55, 2.51; 8 deaths) and kidney (SMR 1.44; 95% CI: 0.69, 2.65; 10 deaths), and for leukemia (SMR 1.48; 95% CI: 0.77, 2.59; 12 deaths). A non-significant upward trend (P = 0.24) by cumulative exposure to TFE was observed for liver cancer. This pattern of findings narrows the range of uncertainty on potential TFE carcinogenicity, but cannot conclusively confirm, nor refute, the hypothesis that TFE is carcinogenic to humans.