Registration Dossier

Toxicological information

Epidemiological data

Currently viewing:

Administrative data

Endpoint:
epidemiological data
Type of information:
other: historical cohort study
Adequacy of study:
key study
Study period:
1942 to 2000
Reliability:
2 (reliable with restrictions)
Rationale for reliability incl. deficiencies:
other: See Bukowski review in 7.10.2-003. While most aspects of the study were of good quality, some control of confounding factors/bias was missing or incomplete
Cross-reference
Reason / purpose:
reference to same study

Data source

Reference
Reference Type:
publication
Title:
Unnamed
Year:
2007
Report Date:
2006

Materials and methods

Study type:
cohort study (retrospective)
Endpoint addressed:
carcinogenicity
Test guideline
Qualifier:
no guideline followed
Principles of method if other than guideline:
Historical cohort study re-evaluating data from four chloroprene-production sites, two in USA and two in Europe. 12430 workers with some exposure to chloroprene between 1942 and 2000 were investigated with special emphasis placed on identification of primary liver and lung cancers. Vital status determinations completed and cause ofdeath established for 95% of cases. Standardised mortality ratios were computed, taking account of demographics, work history and exposure factors. Worker pay type (white or blue collar worker) was used as a surrogate for smoking history.

GLP compliance:
not specified

Test material

Reference
Name:
Unnamed
Type:
Constituent
Details on test material:
- Name of test material (as cited in study report): chloroprene (2-chloro-1,3-butadiene)
- Substance type: monomer used exclusively for production ofsynthetic rubber/latex
- Physical state: Not stated
- Analytical purity: Not stated - liable to variation between sites
- Impurities (identity and concentrations):Not stated - liable to variation between sites
- Purity test date: No information

Epidemiological reports were initiated for chloroprene following reports of liver angiosarcoma among workers exposed to vinyl chloride, a case report for a worker exposed to chloroprene but not vinyl chloride who also developed liver angiosarcoma and the stuructural similarity of chloroprene and vinyl chloride. Subseuently case studies were conducted among chloroprene exposed workers at sites in Russia, China, Armenia, the USA and later France. In 1978 IARC classified chloroprene as Group 3 (not classifiable as to its carcinogenicity to humans). In 1999 following review of two inadequate epidemiology studies IARC reclassified chloroprene as Cat 2B (possibly carcinogenic to humans). The current study provides a more robust and rigorous approach to human epidemiology of chloroprene exposure intended to address the mortality risk , particularly posed by two known risks - liver and lung cancers

Method

Type of population:
occupational
Ethical approval:
not specified
Details on study design:
HYPOTHESIS TESTED (if cohort or case control study): SMR <1 or alternative hypothesis SMR >1

METHOD OF DATA COLLECTION
- Type: Record review / Work history/ Death certification:
- Details: All workers with potential chlloropene exposure at each of four production sites were included in the final cohort, n= 12430

STUDY PERIOD: Data reviewed from plantset up to 1999 or 2000. Four plants (L,M, P and G) involved in chloroprene production as follows
L: 1942 - 1972
M: 1969 - present (2006)
P: 1960 - 1998
G: 1966 to present (2006)


STUDY POPULATION
- Total population (Total no. of persons in cohort from which the subjects were drawn): all workers at each plant
- Total number of subjects participating in study: L = 5507; M = 4849; P = 1357; G = 717
- Sex/age/race: predominantly white male blue collar workers Ammericans, Irish and French
- Smoker/nonsmoker: Surrogate assessment of smoking history based on correlation between smoking prevalence and pay type
- Total number of subjects at end of study: L = 3095 56%; M = 4414 91%; P = 1225 93%; G = 630 88%


COMPARISON POPULATION
- Type: State registry / Regional registry / National registry : National Registries used where and when available

HEALTH EFFECTS STUDIED
- Disease(s): mortality and primary cause of death - liver angiosarcoma, liver and bile passage cancers and lung cancers
- ICD No.: liver and lung malignancies assessed according to ICD 9th revision


OTHER DESCRIPTIVE INFORMATION ABOUT STUDY:
Vital status assessed at end of study (2000). 3002 deaths were identified in the total cohort, the cause fof death was established for 2850 or 95% of the cohort participants. Lost- to- follow up rates were between 0 and 3.5%
Exposure assessment:
not specified
Details on exposure:
The duration of exposure was used as a factor in the determination of mortality pattern. Workers were subdivided into exposed and non-exposed groups, by duration of employment (<5 years; 5 to 19 years and >20 years). Co-exposure to vinyl chloride was investigated with details of production processes likely to result in VC production included in the paper.
Statistical methods:
A priori the statistical power expected for the study was the probablitlity of obtaining an SMR significantly greater than 1.00 at the 0.05 level for a onesided test, assuming no excess risk and estimated numbers of expected deaths. The statistical power to detect a two fold or greater excess in lung cancer was 0.87 to 0.97 for Plants G and P and essentially 1.00 for the cohort as a whole. Statistical power for the liver cancer assessment was less - 0.25 for Plants G and P but 0.97 to 0.99 for Plant L or the whole cohort.
Total and cause specific mortality were analysed. Cohort analyses were performed using modified lifetables. Person-years at risk were classifid by race, sex, age group, calendar time, duration of employment and time from first employment. Expected numbers of deaths were calculated using person-years at risk multiplied by annual average standard population death rates for age, race, sex and time. Local death rates were used for the American plant data and national rates for data from France and Ireland.
SMRs were calculated for each plant, for all subjects and for some subgroups

Results and discussion

Results:
Plant L:
Statistically significant deficits in deaths for all causes were apparent when compared to local county registries (SMR = 0.74) and for all cancers combined (SMR = 0.75). Significant deficits were also apparent for most of the malignant and non-malignant cause of death categories. Respiratory stystem cancers, had a 25% deficit based on observing 266 deaths, 252 of which were attributed to cancer ofthe bronchus, trachea or lung (SMR = 0.75). A 10% deficit was noted forthe other cancer of particular interest - liver cancers, as defined, "cancer of biliary passage and liver" were lower based on 17 mortalities. The authors cite the prevalance of smoking in the localcounty region - around the time of study conduct Kentucky had the highest smoking rate of any state in America - as a possible reason for the disparity between the chronic disease categories



Plant P:
A significant 47% deficit in all causes of death combined (SMR = 0.53) andfor all cancers combined (32%; SMR = 0.68) was calculated. Elevated SMRs were noted for several cancer sites but the number of deaths observed were low and none f the SMR values were statistically significant. SMRs for all non-malignant cause of death categories wer less than 1.00 and some ofthe deficits were significant. For respiratory system cancer a 38% deficit , based on observation of 12 deaths (SMR = 0.62). Ten of these deaths were due to cancer of te bronchus, trachea or lung - the deficit for lung cancer was therefore 45% (SMR = 0.55).
No deaths from liver cancer were noted at Plant P

Plant M:

Statistically significant deficits in deaths (for all causes of death combined) were estimated using national comparison figures for Northern Ireland, SMR = 0.60 for all deaths = 40% deficit and SMR = 0.68 for all cancers combined or a deficit of 32%. Deficits, often statistically significant, were also noted for almost all of the malignant and non-malignant cause of death categories. The deficit for respiratory system cancer was not significant but was 21% based on results from 48 deaths, SMR = 0.79. 90% of the deaths were attributed to cancers of the bronchus, trachea or lung. Only one liver cancer (unspecified) related death was recorded for Plant M.


Plant G:
Statistically significant deficits in deaths (for all causes of death combined) were estimated using national comparison figures for France, SMR = 0.65 for all deaths = 35% deficit and SMR = 0.59 for all cancers combined or a deficit of 41%. Elevated SMRs (non-significant) were noted for some of the cancer sites and some non-malignant cause of death categories. The deficit for respiratory system cancer was not significant but was 15% based on results from 10 deaths, SMR = 0.85. 40% of these deaths were attributed to cancers of the bronchus, trachea or lung. Only one liver cancer (unspecified) related death was recorded for Plant G.


Aggregated figures for all cancers combined, RSC and liver cancer show that there was a 27% deficit for all cancers combined (SMR = 0.73) based on 834 deaths in the whole cohort. In fact for almost all ofthe study variables and subcategories examined there were deficits in deaths across the four plants following occupational exposure to chloroprene. The reduced death incidences compared with anticipated mortality were generally statistically significant. None of the case of elevated SMRs were significant and none were related to the a priori cancers - lung and liver tumours.
The majority of deaths occurred (806/834) among chloroprene exposed individuals but this was to be expected since the entire cohort was drawn from workers in chloroprene production sites and therefore all were potentially exposed. However, the plant specific SMRs for non-exposed subjects (somewhat imprecise due to low numbers), were actually higher than the SMRs for subjects with some chloroprene exposure.

For plants L and M chloroprene exposed subjects were also potentially exposed to vinyl chloride and yet the SMR (0.60) showed a 40% deficit in all cancer deaths combined (based on 143 mortalities). The SMRs for exposed and non-exposed workers in each plant with VC exposure were less than 1.00.
For respiratory system cancer the aggregate SMR (= 0.75) was significantly lower (25%) based on 336 observerd deaths. For nearly all of the study variables and sub categories investigated, including chloroprene exposure and surrogate variables for chloroprene exposure, there were significant deficits in deaths across the plants. None of the elevated SMR values were statistically significant. With the exception of a slight excess in RSC among workers not exposed to chloroprene in Plant M, the SMRs for workers exposed and unexposed to chloroprene and vinyl chloride were less than 1.00.

17 of the 19 total liver cancer deaths occurred among white male blue collar workers in Plant L. This was reflected in a 28% deficit in deaths for the whole cohort.






Confounding factors:
Cohort subgroups were defined by race, gender, worker pay type, duration of employment (short/medium or long term), time period, year of hire, service type and chloroprene or vinyl chloride exposure.
Smoking history was approximated by use of a surrogate variable

Attempts were included to address the confounding issue of tobacco-smoking history of subjects that died from respiratory system cancer (RSC)- however the low availability (only 28% of subjects at one plant and 54% from a second actually had a reliable history), meant that a case-control study for RSC was unfeasible. Two aspects were included that were considered to address the confounding issue of smoking in some degree - use of local mortality comparisons allowed a geographical adjustment for variability in tabacco use. Secondly, workers were categorised by pay type (blue/white collar workers) as a rough surrogate for education/socioeconomic status which is highly correlated with smoking prevalance. The exposure response analyses used this estimate to adjust for smoking related confounding issues.
Strengths and weaknesses:
The major strengths of this investigation were identified as diversity of location and production processes, length of observation periods, substantial proportion of worker cohort employed for >20 years, almost total cohort enumeration with cross-referencing, vital status tracing, cause of death determination, detection of two-fold or greater overal mortality excess for all cause of death categories of prime interest with powerful statistical analysis, a rigorous, innovative, chemical process-based exposure reconstruction for both chloroprene and vinyl chloride; use of national and local mortality comparisons and robust statistical modeling of internal cohort rates. These factors contributed to an overall rigorous and robust assessment of mortality following chloroprene exposure.
Attempts were included to address the confounding issue of tobacco-smoking history of subjects that died from respiratory system cancer (RSC)- however the low availability (only 28% of subjects at one plant and 54% from a second actually had a reliable history), meant that a case-control study for RSC was unfeasible. Two aspects were included that were considered to address the confounding issue of smoking in some degree - use of local mortality comparisons allowed a geographical adjustment for variability in tabacco use. Secondly, workers were categorised by pay type (blue/white collar workers) as a rough surrogate for education/socioeconomic status which is highly correlated with smoking prevalance. The exposure response analyses used this estimate to adjust for smoking related confounding issues.

Applicant's summary and conclusion

Conclusions:
No increased mortality risks were observed among the cohort subgroups. The authors conclude that persons exposed to chloroprene (alone or in conjunction with vinyl chloride exposure) at levels enountered in four production sites did not show any elevated risk of mortality from the causes of death investigated. In particular the cancer sites of interest - lung and liver - showed no increase in incidence or mortality for chloroprene exposed workers.

The study is a definitive assessment of human carcinogenic potential following long term exposure to chloroprene. The results show that exposure to chloroprene at the four sites evaluated was not associated with increased general mortality, no with elevated risk of mortality from chloroprene exposure and that there were no excesses in all-cause cancers nor the two cancers of primary interest - lung and liver cancer
Executive summary:

This historical cohort study covering four production plants in EU and USA represents the most rigorous, largest and most comprehensive study of long term health effects of chloroprene exposure conducted to date. The combined cohort was 12430 subjects or greater than one-third of a million person-years of observation with 41% (151691 person-years) consisting of workers with more than 20 years exposure. Follow-up to 2000, resulted in 3002 observed mortalities including 834 from all cancers combined. Two causes of death were of primary interest - respiratory system cancer and liver cancer (including biliary passage cancers), of which 336 and 19 were observed respectively.

The major strengths of this investigation were identified as diversity of location and production processes, length of observation periods, substantial proportion of worker cohort employed for >20 years, almost total cohort enumeration with cross-referencing, vital status tracing, cause of death determination, detection of two-fold or greater overal mortality excess for all cause of death categories of prime interest with powerful statistical analysis, a rigorous, innovative, chemical process-based exposure reconstruction for both chloroprene and vinyl chloride; use of national and local mortality comparisons and robust statistical modeling of internal cohort rates. These factors contributed to an overall rigorous and robust assessment of mortality following chloroprene exposure.

Attempts were included to address the confounding issue of tobacco-smoking history of subjects that died from respiratory system cancer (RSC)- however the low availability (only 28% of subjects at one plant and 54% from a second actually had a reliable history), meant that a case-control study for RSC was unfeasible. Two aspects were included that were considered to address the confounding issue of smoking in some degree - use of local mortality comparisons allowed a geographical adjustment for variability in tabacco use. Secondly, workers were categorised by pay type (blue/white collar workers) as a rough surrogate for education/socioeconomic status which is highly correlated with smoking prevalance. The exposure response analyses used this estimate to adjust for smoking related confounding issues.

Statistical power to detect excess liver cancer was 99% but lower in all plant except the plant that included 17/19 liver cancer deaths. However since the majority of SMRs were <1.00 the issue of statistical power only applied to the alternative hypothesis that SMR>1.00.

In plants L amd M subjects were potential exposed to a range of other chemicals in addition to chloroprene and vinyl chloride. However the nature of exposure to other agents was considered not to have impacted on the health effects for long term workers and so, while recognised, there was no attempt to characterise cross-exposures.

The length of the follow-up times may have mitigated the influence of short-term worker (<5 years) mortality - it was noted that about half the cohort were short term workers but that they did not exhibit the normal differential mortality pattern often associated with increasing mortality for malignnt and non-malignant disease.

Total and cause specific mortality patterns were generally consistent between plants, generally showed a statistically significant reduction in mortality risk from all causes combined, all cancer sites combined and for many of the other disease categories investigated (malignant and non-malignant). These reduced risks were evident in all cohort subgroups. The risk reduction was also maintained for the various variables used as surogates for chloroprene exposure. The authors conclude there may be a "healthy worker effect" - favorable mortality patterns, particularly for chronic disease, influenced by the relative absence of deleterious risks in relation to employment combined with positive benefits of long term health care and quality of life improvements in the workplace, i.e. reduced cancer incidence due to generally better standard of life.

For this study it was of particular importance to note the absence of elevated mortality risks for all cancers combined or for the two sites of primary interest, the lungs and liver. The lack of an increased risk of liver cancer was interesting (the authors note that acetylene was used in the manufacturing processes for Plants L and M in the production of chloroprene with vinyl chloride generated as a by-product. VC is an established risk factor for a specific liver cancer - angiosarcoma and is linked to risks of hepatocellular carcinoma, brain tumours, lung tumours and malignant effects on lymphatic and haematopoietic systems. Other studies (see those from China, Moscow, Armenia cited in 7.10.2.-003) evaluated by Bukowski, appear to show excess liver cancers but the inherent methodological limitations of these studies raise questions regarding their significance for human cancer risk assessment.

In this study liver cancer was examined within the braoder range of "biliary passages and liver". No evidence of an increased risk of death was found in the total cohort or in cohort subgroups (one plant accounted for 17/19 liver related mortalities). Of the 19 deaths only 8 were classified as pimary liver cancer. No evidence was found for increased mortality risks for other sites linked to vinyl chloride exposure - but this may be explained by the relatively low historical VC exposures in Plants L and M. A full evaluation of the occurrence of VC-related liver angiosarcoma was not possible (angiosarcoma did not receive an ICD code until 10th revision and so can only be roughly identified in studies using earlier coding systems, as was the case here).

A comprehensive death certificate review was also not possible since limited records were available from the sites under investigation.

17 of the 19 deaths occurred at Plant L, only four of these pre-dated 1979 (introduction of the US National Death Index) but a death certificate was only available for one person, citing liver cancer as the cause of mortality. Two of the causes of death were cited onthe certificates as cancer of the liver but the use of ICD 10 codings meant that is was not possible to evaluate accurately the possible incidence of angiosarcoma in this subgroup.

The study found no excess risk of respiratory system cancer (although 90% of the malignancies were cancers of the bronchus, trachea or lung i.e. lung cancers).This result differed from the previous cancer incidence report that looked at the same data set (see French cohort study in 7.10.2 -003) but the incidence summary did not include women, employees with less than 2 years exposure or workers that left the area before 1979. The current mortality investigation included all of these in the overall cohort. The basis for comparison also changed - in the French cohort investigation the cancer incidence was assessed using the cancer registry for the area of interest, the current investigation used the French national death registry that records informatiion from across all France. The absenceoof excess lung cancer risk was consistent with information from animal studies which show variability in sensitivity to chloroprene induced lung cancer and flag up difficulties in extrapolation to human lung cancer risk.

In conclusion the study is a definitive assessment of human carcinogenic potential following long term exposure to chloroprene. The results show that exposure to chloroprene at the four sites evaluated was not associated with increased general mortality, no with elevated risk of mortality from chloroprene exposure and that there were no excesses in all-cause cancers nor the two cancers of primary interest - lung and liver cancer