Hydrazine

Epidemiological data

Administrative data

Endpoint:

epidemiological data

Type of information:

experimental study

Adequacy of study:

supporting study

Study period:

2010-2014

Reliability:

2 (reliable with restrictions)

Data source

Reference

Reference Type:

study report

Title:

Unnamed

Year:

2014

Report date:

2014

Materials and methods

Study type:

cohort study (retrospective)

Endpoint addressed:

carcinogenicity

Principles of method if other than guideline:

The aim of this mortality study was to search for possible excess mortality, due to cancers in particular, linked to work in a unit making use of hydrazine at the Arkema site in Lannemezan. Once the necessary authorizations had been obtained, the cohort was constituted from a list, provided by the medical service of the company, of all the men who had worked for at least six months in the hydrazine production building at the Lannemezan factory for whom a vital status could be established. The cohort thus included 190 people in total. These individuals were first exposed between January 1962 and September 2010.

The principal objective in this study of mortality was to provide elements useful for evaluations of the health risks linked to employment in a factory producing hydrazine on the Arkema site at Lannemezan. Its aim was to investigate the possible existence of excess mortality among the staff, and particularly cancer-related deaths. The specific objectives were:
1) To compare all-cause mortality and cancer-related mortality (all cancers and each type of cancer emerging from the study) for the employees with that of the French population, taking age, sex and time period into account;
2) To compare all-cause mortality and cancer-related mortality (all cancers and cancers at the sites emerging from the study) for the employees as a function of their exposure levels.

The necessary authorizations were obtained: CCTIRS authorization was obtained on May 12th 2011 and CNIL authorization was obtained on December 16th 2011.

GLP compliance:

no

Test material

Test material information

Constituent 1

Reference substance name:

Hydrazine

EC Number:

206-114-9

EC Name:

Hydrazine

Cas Number:

302-01-2

Molecular formula:

H4N2

IUPAC Name:

hydrazine

Details on test material:

not appropriate

Method

Type of population:

occupational

Ethical approval:

other: CCTIRS authorization was obtained on May 12th 2011 and CNIL authorization was obtained on December 16th 2011. (Consultative Committee on Information Processing for Research in the Domain of Health)

Details on study design:

This study is a continuation of the epidemiological surveillance of the employees of Arkema exposed to hydrazine (Arkema, 2003). It covers a longer follow-up period than previous studies, particularly for the workers most exposed to hydrazine, who were first exposed between 1962 and 1987, and also involves the integration of new workers into the cohort.


1) Study population
The cohort consists of all the male employees having worked for at least six months in the building in which hydrazine is produced at the Lannemezan factory. These employees were first exposed to hydrazine between January 1962 and September 2010.
At the Lannemezan site, 191 employees, all men, were found to have been exposed to hydrazine for at least six months. Only one employee was excluded from the cohort because his vital status was unknown. The final cohort thus consisted of 190 men, corresponding to 5403 person-years. All the descriptive results are presented for the whole cohort and/or according to exposure level, for classifications I and II.



The following items of information were obtained about these workers from the factory medical service:
- Surname
- First name
- Sex
- Date of birth
- Place of birth
- Administrative area (département) of birth
- Country of birth
- Year in which joined the company
- First year of exposure
- Year in which left the company
- Posts occupied and the dates of occupation of each post.

2) Reference population
The reference population for comparisons with the cohort consisted of all the men living in mainland France and aged 15 years or over between 1962 and 2009.


3) Cumulative level of exposure
The cumulative levels of exposure to hydrazine, reconstituted for the entire career of each employee for the previous INSERM studies were used in this work. Health and safety managers and previous Arkema workers had defined annual levels of exposure according to the post occupied and the years in which these posts were occupied. This semi-quantitative coding was implemented on the basis of metrological data. The levels of exposure presented in table 1 range from 0 to 4 for the years 1962 to 1979 and from 0 to 2 for the years 1980 to 2011.
Table 1: Annual level of exposure as a function of the post occupied and year
Post 1962 -1979 1980-1987 After 1987
Packing (transfer of hydrazine to containers) 4 2 1
Production: manufacture of hydrazine 3 1 1
Maintenance 3 2 1
Supervisory personnel and engineers 1 1 1
Desalting (1970-1979) 4 - -
Laboratory (until 1981) 3 1 1
Team leaders and their deputies 3 1 1
Derivatives: BAG, ATA, AZO, Mechanical maintenance for the DH, DH team leaders 1 0 0
DAIB drier, others 0 0 0

The total or cumulative level of exposure is the sum of exposure levels for all the posts occupied multiplied by the time for each post was occupied during the career of the employee. It can be calculated as follows:

Exposure classes were defined in the previous studies; classification I includes three classes as previously defined:
- A low exposure group, with an exposure level no higher than 3,
- An intermediate exposure group, with a level of exposure of 3 to 8 inclusive,
- A high exposure group, with levels of exposure strictly higher than 8.

A new classification (classification II) was created for this study, based on the distribution of the variable “exposure level” (see graph n°6). This classification also includes three groups:
- A low exposure group with levels of exposure no higher than 3,
- An intermediate exposure group with levels of exposure of 3 to 37 inclusive,
- A high exposure group with a cumulative exposure level greater than 37.

In some analyses, the low and intermediate exposure groups were combined for a clearer comparison of the individuals “least exposed” and “most exposed” to hydrazine.

4) Vital status and causes of death
Vital status was checked in the RNIPP, and cause of death, for those employees who had died, was determined from CépiDc data. Such checks were made up to December 31st 2011.

We transmitted the surname, first name, sex, date of birth (day, month and year), place and administrative area (département) of birth to the RNIPP. For only one person from the cohort, missing information prevented the determination of vital status.

Requests for information about vital status from the RNIPP were followed by a search for the cause of death (CépiDc data) for those who had died. The CépiDc supplied the alphanumeric codes of the International Classification of Diseases (ICD) corresponding to the various causes of death. This classification has been updated a number of times (resulting in different correspondences between codes and diseases). For this study, we used the two most recent versions: ICD9 and ICD10.

Deaths were then classified into 65 major groups of causes according to Eurostat criteria (shown in appendix 1). This European classification groups causes of death according to ICD codes, with harmonization of the three most recent versions of the ICD. A cause of death in the Eurostat classification corresponds to several ICD codes, but an ICD code corresponds to only one cause of death in the Eurostat classification.

5) Definition of the variables of interest
In addition to variables such as age, age at death or end of follow-up, date and cause of death and exposure level, a number of other variables were considered in the data analysis to be of interest and are presented below.
Year of first exposure
The year noted was the year in which the employee took up the first post in which he was exposed.
Age at first exposure
Age at first exposure was defined as the age of the employee at the time of his first exposure to hydrazine. It was determined by subtracting the year of birth from the year in which the employee took up the first post in which he was exposed.
Duration of exposure
The duration of exposure was defined as the time spent in the company from the first exposure to hydrazine.
Duration of follow-up
The duration of follow-up was defined as the time spent in the cohort between the date of first exposure and the date at which the cohort ended (the end of 2011) or the date of death if before the end of 2011.
Age at exit from the cohort
This is the age of the employee at the end of the cohort (end of 2011) or the age at death for those who died before the end of 2011.
Year of exit from the cohort
This is the year in which the cohort study ended (2011) for employees still alive at the end of 2011 or the year of death for those dying before the end of 2011.

Exposure assessment:

estimated

Details on exposure:

• Materials and methods
Cumulative levels of exposure to hydrazine were reconstituted by INSERM from the annual levels of exposure for the various posts concerned and the years during which these posts were occupied. According to semi-quantitative coding, exposure levels were between 0 and 4 for 1962 to 1979, then between 0 and 2 from 1980 to 2011. Cumulative exposure to hydrazine during the entire career of the employees ranged from 1 for individuals with very low levels of exposure to 71 for the most exposed individuals. A quarter of the employees had a level of exposure of no more than 3, and more than half the employees had an exposure level of no more than 8. A subgroup of individuals with a much higher level of exposure (>37) was identified, and accounted for 10% of the cohort. Two classifications were created on the basis of this cumulative level of exposure.

- Classification I (already used by INSERM) into three groups: low (1 to 3), intermediate (4 to 8) and high (9 to 71) levels of exposure,
- Classification II into 3 groups: low (1 to 3), intermediate (4 to 37) and high (38 to 71) levels of exposure.

Statistical methods:

The data were analyzed with SAS version 9.3 software, and the significance threshold for statistical tests was fixed at 5%. The statistical analysis methods used in this study were those routinely used for studies of mortality in occupational environments. They were based on comparisons of mortality by indirect standardization with an external reference (the national population) and comparisons with an internal reference, between exposure groups, carried out with Cox models.

Results and discussion

Results:

In total, 43 people in the cohort (22.6%) died before the end of 2011. Age at death for these individuals was between 48 and 85 years, with the highest proportion (10 deaths) occurring between the ages of 60 and 64 years. The two principal causes of death in the cohort were tumors (16 cases) and circulatory system diseases (16 cases). The tumors most frequently implicated in the deaths of cohort members were lung cancers (5 cases) and bladder cancers (2 cases).

Significantly fewer (27% fewer) deaths “from all causes” were observed in the cohort than would be expected by reference to the general population in France. Differences of mortality were also observed for tumors (32% fewer) and circulatory diseases (17% upper) but these differences were not statistically significant

Cox models, taking age at first exposure and calendar period into account, were developed for comparisons, within the cohort, of the group of individuals with high levels of exposure with that of individuals with low and intermediate levels of exposure together. With classification I, no difference in mortality was found between the two exposure groups considered either for deaths from all causes or for deaths from circulatory diseases. By contrast, deaths from tumors were slightly overrrepresented among those with the highest levels of exposure (not significant). With classification II, we observed a significant excess mortality for all tumors and a non-significant excess mortality for deaths from all causes, lung cancers and bladder cancers in the group with the highest levels of exposure. By contrast, circulatory diseases were slightly underrepresented in this group.

This analysis provides no evidence of excess mortality in this cohort with respect to the general population of France. Similarly, comparisons within the cohort taking into account the classification of exposure used in previous INSERM studies provided no evidence of a significant difference in specific mortality between the exposure groups. With the classification comparing the subjects with the highest exposure levels (only 10% of the subjects) to the less exposed subjects, a statistically significant increase in the risk of death from all tumors was observed, but this result is not robust because it relies in the highest exposure group on only 4 deaths, of 4 different types of cancers. For lung cancers, which are among the tumors found potentially to occur in excess in other studies of employees exposed to hydrazine, the risk of death was more frequent but not significant in this same group with high levels of exposure. Similarly, the risk of death of bladder tumors were more frequent but not significant, with only two cases (one in the high exposure group and the other in the less exposed group). The results of this study should be interpreted with caution, given the imprecision of measurements of exposure and the small number of subjects in the cohort, decreasing the robustness of the study.
This study does not currently provide conclusive evidence of a causal relationship between hydrazine exposure and cancer. Now that this cohort has been established, we recommend its continued follow-up in the years to come.

Strengths and weaknesses:

Limitations

One of the key limitations of this study is the semi-quantitative nature of the evaluation of exposure, which therefore only partially reflects the true exposure of the employees. In addition to the imprecision inherent to this type of evaluation, it was also possible for a worker to occupy two different posts in the same year without it being clear what proportion of his time had been spent in each of these two posts. In such cases, we arbitrarily allocated a time of 50% to each post. The errors resulting from such approximations would be expected to tend to decrease the strength of the link between exposure and mortality. In addition, these decisions may not have been identical to those taken in the previous studies carried out by INSERM.

The small size of the cohort resulted in a limited statistical power, and it was therefore possible to detect only large differences in mortality due to particular causes of death, as demonstrated by the calculation of detectable SMR values (appendix 3). This small size also resulted in a lack of robustness of the results, which varied in a non-negligible manner according to the number of categories chosen for the explanatory variables. This lack of robustness was also aggravated, particularly as relevant to lung and bladder tumors, by the absence of information about smoking habits.

The “healthy worker” effect favors an underestimation of the effects of hydrazine on mortality during comparisons of the cohort with the reference population. This effect is not relevant to the comparisons of different groups of workers, but for these comparisons, it must be borne in mind that the reference group is an exposed group, even though the level of exposure is low.

Strengths of the study
One of the strengths of this study is that the principal source of cohort exposure was hydrazine. By contrast, other professional cohorts may be subject to exposure to multiple factors, such as the hydrocarbons of fuels containing hydrazine.

The identification of deaths was carried out with the RNIPP, unbiased, and in the same way as for the French reference population. By contrast, the previous studies assessed morbidity and mortality from the database available for the occupational health service, and this is associated with a possibility of under-declaration.
Indirect standardization was used to take age and calendar period into account during evaluations of the link between exposure and mortality. The Cox models also took these two variables into account, especially age at death, by comparing those who had died with those of the same age who were still alive, as age is the principal factor linked to death. The use of Cox models also made it possible to adjust for age at first exposure. The number of adjustment variables nevertheless remained small due to the small size of the cohort and the correlation between many variables linked to exposure, such as duration of follow-up, duration of exposure and age at first exposure. The use of Cox models also made it possible to take into account interactions between the explanatory variables. However, these interactions were not found to be significant and were therefore not retained in the final results presented.
Two classifications of exposure level were tested in this study, and the use of a classification isolating a group with the highest levels of exposure (<37) made it possible to demonstrate, for all tumors, an excess mortality that was more marked than with the classification used in the previous INSERM studies. All the members of this high exposure group were first exposed to hydrazine between 1962 and 1979.

Applicant's summary and conclusion

Conclusions:

No excess mortality was detected in the cohort with respect to the general French population. This study does not currently provide conclusive evidence of a causal relationship between hydrazine exposure and cancer.Now that this cohort has been established, we recommend its continued follow-up in the years to come.

Executive summary:

·       Conclusion and Discussion

The principal objective in this study of mortality was to provide elements useful for evaluations of the health risks linked to employment in a factory producing hydrazine on the Arkema site at Lannemezan. Its aim was to investigate the possible existence of excess mortality among the staff, and particularly cancer-related deaths. The specific objectives were: 1) To compare all-cause mortality and cancer-related mortality (all cancers and each type of cancer emerging from the study) for the employees with that of the French population, taking age, sex and time period into account; 2) To compare all-cause mortality and cancer-related mortality (all cancers and cancers at the sites emerging from the study) for the employees as a function of their exposure levels. The necessary authorizations were obtained: CCTIRS authorization was obtained on May 12th 2011 and CNIL authorization was obtained on December 16th 2011.

This study is a continuation of the epidemiological surveillance of the employees of Arkema exposed to hydrazine. It covers a longer follow-up period than previous studies, particularly for the workers most exposed to hydrazine, who were first exposed between 1962 and 1987, and also involves the integration of new workers into the cohort. At the Lannemezan site, 191 employees, all men, were found to have been exposed to hydrazine for at least six months.

This analysis provides no evidence of excess mortality in this cohort with respect to the general population of. Similarly, comparisons within the cohort taking into account the classification of exposure used in previous INSERM studies provided no evidence of a significant difference in specific mortality between the exposure groups.With the classification comparing the subjects with the highest exposure levels (only 10% of the subjects) to the less exposed subjects, a statistically significant increase in the risk of death from all tumors was observed, but this result is not robust because it relies in the highest exposure group on only 4 deaths, of 4 different types of cancers.For lung cancers, which are among the tumors found potentially to occur in excess in other studies of employees exposed to hydrazine, the risk of death was more frequent but not significant in this same group with high levels of exposure. Similarly, the risk of death of bladder tumors were more frequent but not significant, with only two cases (one in the high exposure group and the other in the less exposed group).The results of this study should be interpreted with caution, given the imprecision of measurements of exposure and the small number of subjects in the cohort, decreasing the robustness of the study.

This study does not currently provide conclusive evidence of a causal relationship between hydrazine exposure and cancer. Now that this cohort has been established, we recommend its continued follow-up in the years to come.