Mortality of cadmium-exposed workers A five-year update

KAZANTZIS SULLIVAN Mortality of cadmium-exposed workers: A five-year up date. Scand J Work Environ Health 14(1988) 220-223. A cohort mortality study of cadmium-exposed workers initially followed to the end of 1979has been updated for a further five years. The update has confirmed the findings of the original study with a significant excess risk from bronchitis related to inten sityof exposure, although over the five-year period the excessmortality was no longer significant, a finding suggesting that the risk from bronchitis may now be declining in this cohort. In contrast, there is now a stronger indication of an excess risk from lung cancer related to intensity of exposure, significant for both the total and the five-year periods. There was again no increased risk from prostatic cancer, and from this and other studies it appears unlikely that cadmium, in the concentrations encountered in this and other recent studies, acts as a prostatic carcinogen. As in the initial study, there was no significant excess risk from hypertensive disease, nor any suggestion of an increased risk from cerebrovascular or renal disease.

The International Register of Potentially To xic Chemicals (lRPTC) of the United Nations Environment Programme (UNEP) includes cadmium in its listing of environmentally dangerous chemical substances and processes of global significance (11). Atmospheric emissions of cadmium from man-made sources are considered in the IRPTC to exceed those of natural origin by one order of magnitude. Cigarette smokers were observed to be potentially at risk because of the high absorption of cadmium from tobacco smoke. Certain compounds of cadmium have been shown to give rise to cancer in animal studies (5), but the role of cadmium as a human carcinogen remains uncertain. The International Agenc y for Research on Cancer (IARC) in 1982 considered the evidence for the carcinogenicity of cadmium to be sufficient in animals , but limited in humans (3). Their classification of cadmium at that time as a 2B carcinogen was proposed for reclassification to category 2A in 1987.
The study of groups with occupational exposure is of particular value in elucidating adverse effects, as not only is exposure likely to have been greater in the occupational than in the general environment, but also absorption is markedly more effective following the inhalation of cadmium oxide fume and cadmiumcontaining dusts than after the ingestion of cadmium in food or water. A cohort of almost 7 000 male 220 workers born before 1940and exposed to cadmium for mor e than one year between 1942and 1970was initially followed with regard to mortality experience to the end of 1979 (1,4). A statistically significant excess of deaths recorded as bronchitis was related to both intensity and duration of exposure, predominantly in the small group of men with heavy pa st exposure to cadmium. There were marginally more deaths from lung cancer than expected, but the excess was not related to exposure levels. No excess of deaths due to prostatic cancer, cerebrovascular disease, or renal diseases and no significant excess due to emphysema or hypertensive disease were observed. This cohort stud y has now been updated for a further five years to include all deaths to the end of 1984.

Material and methods
Details of the methods used for the cohort analysis have been given earlier (1) and are briefly summarized here. Deaths were coded by underlying and other causes according to the eighth revision of the International Classification of Diseases (lCD) (13). Expected numbers were calculated from mortality rates for the population of England and Wales corrected to the eighth revision ICD codes, and regional variation in mortality was taken into account by the use of the cause-specific standardized mortality ratio (SMR) for standard regions published by the Office of Population Censuses and Surveys for the period 1969-1973. The ratios of observed to expected deaths (SMR values) were calculated, in five-year age and calendar strata, with the use of the Oxford Man-year Program devised In addition to th e seven disease categories for which there was some a pr iori -reason to believe that a relation ship to cadmium exposure might exist , stomach cancer was now included for det ailed anal ysis becau se an excess mortality had been ob ser ved in a reanalysis of the 1979 data (6). Thus eight categories, together with "all causes" and "all neoplasms, " ha ve been included in the present repo rt. Deaths occurr ing in th ese 10 catego ries have been followed to the age of 85 years. For a further 32 disease categories a screening procedure has been used with a censoring age of 75 years.
Jobs were again classified according to the level of past cadmium expo sure into th e three groups high , medium and low , and the years at risk were divided on the basis of the se categories into th e three groups " ever high " for a minim um period of one year, " ever medium " for a min imum period of one year, and " always low." For those wor kers who were still employed aft er 1979, the ir employment status between 1980-1984 was updated .
Rescrutiny of the initial study population identified a small number of erroneous entries, which reduced the number of workers included in th e SMR analysis from 6 995 to 6 958 for th e five-year period. The mean duration of expo sur e to cadmium was 12 years, and the mean interval fro m first expos ure to the end of follo w-up was 29 years. Th e distribution of the worke rs in th e five principal industries involved was as in the initial study. By the end of 1984, 2 523 (36 % of the study populat ion) had d ied; 4 184 (60 %) were kno wn to be alive; 184 (3 % ) had emigrated; and 67 (l %) could not be traced . Th e reported tabulations refe r to 2 467 deced ents below the age of 85 years. The five years 1980-1984 gave an extra 21 857 per son-years to th e study with a to tal now of 173 961 per son-yea rs for th e period 1943-1984.

Results
Cause-specific mortality in rela tion to cadmium expo sure for the total study per iod is shown in table 1. With the additional five years of follow-up, all-cause mortality remained virtually un chan ged . All-cause mortality increased with length of follow-up, being significantly lower than expe cted for those follo wed for less than 10 years (SMR 79, 95 % CI 68-90) and significantly higher than expected for those followed for more than 30 yea rs (SMR 139, 95 % CI 104-174), a finding in keeping with the commonly obs erved " healthy worker effect. " Over the five-year period there were 26 deaths coded as bronchitis or chro nic bronchitis, this level ind icating a nons ignificant excess (SMR 132, 95 % CI 86-194) which was not related to intensity of exposure. However over the total study period the excess mortality from bronchitis was highly significant (SMR 132, 95 % CI 113-151), pa rt icularly in the small highexpo sure group (SMR 382, 95 % CI 203-654). There was a small but sign ificant number of deaths from emphysema over the five-year period, all from the lowexposure group (SMR 277, 95 % CI 102-603) which, for the total study period, approached a significant level, aga in all in the low-expo sure gro up (SMR 174, 95 % CI 99-282).
Morta lity from all neoplastic disorders was significantly rai sed for both the five-year and the total study period (SMR 124,95 % CI 106-142 and SMR 110,95 % CI 102-118, respectively), attributed mainly to an increase in lung and stomach cancer mortality . Th e SMR for prostatic cancer had fallen from 99 in the 1979 cohort to 90 in the total study period , with 30 ob served deaths in th e low-exposur e gro up and again no deaths in the medium-or high-exposure categories.
Th e five-year update contributed 75 additio nal cases of lun g cancer, now giving a significa nt excess mortal-ity both for the five-year period (SMR 134, 95 010 CI 103-164) and the total study period (SMR 115, 95 % CI 101-129), an increasing SMR with intensity of exposure occurring in both the five-year and the total study periods (SMR 359, 95 % CI131-780, and SMR 194, 95 % CI 100-339, respectively, for the highexposure group). However a statistical test for trend was not significant. The increased lung cancer risk occurred mainly for those employed before 1940, rising with length of employment and also with length of follow-up. In an examination of lung cancer mortality in relation to type of industry, the copper cadmium alloy plants included in the study showed an excess mortality in the five-year update (SMR 231, 95 % CI 115-413). However the pattern of mortality was determined largely by long employment in a primary production plant, starting before 1940, but with only low or medium exposure to cadmium.
There were 22 deaths from stomach cancer over the five-year period (SMR 179, 95 % CI 112-271) with 98 deaths over the total study period (SMR 139,95 % CI 111-166), seen mainly in the low-exposure category for men employed less then 10 years.
A small, nonsignificant excess mortality was observed for the category nephritis and nephrosis in the five-year update, but not for the total study period. There was no excess mortality from hypertensive disease in the five-year period, with a small, nonsignificant excess for the total study period. The SMR for cerebrovascular disease was significantly lower than expected in both the five-year and the total study period. Screening for the other 32 disease categories showed a significantly low SMR for diseases of the circulatory system (lCD 390-458) with an SMR of 83 (95 % CI 73-94) for the five-year period and an SMR of 89 (95 % CI 84-94) for the total study period. Within this broad category, ischemic heart disease (ICD 410-414) gave SMR values of 87 (95 % CI 74 -100) for the five-year period and 92 (95 % CI 85 -98) for the total study period.

Discussion
The five-year update of the 1979 cohort study confirmed and strengthened the findings of the original study with a marked excess mortality from diseases coded as bronchitis or chronic bronchitis, which showed a strong relationship to both intensity and duration of exposure. As in the initial study, there were no deaths coded as emphysema in the high-or mediumexposure groups, but now with a significant excess mortality in the low-exposure group. It is difficult to interpret these emphysema deaths in relation to cadmium exposure, as emphysema tends to be underreported as an underlying cause of death in Great Britain.
With regard to cancer of the prostate, there were again no deaths in the small high-exposure or larger 222 medium-exposure groups. These findings support the conclusion reached by Doll (2), in reviewing the evidence following the initial study (1), that cadmium should now cease to be regarded as a prostatic carcinogen. The Environmental Protection Agency in the United States has also concluded, from a review of existing studies, that there is insufficient evidence of a risk of prostatic cancer from exposure to cadmium oxide dust and fume (12).
Since the initial study, an unequivocal dose-related increase in the incidence of primary carcinoma of the lung has been shown in rats following the prolonged inhalation of a cadmium chloride aerosol (9). An update of a cohort study on American smeltery workers with heavy past exposure to cadmium oxide has shown an excess lung cancer mortality related to cumulative exposure to cadmium, but with certain assumptions having been made with regard to exposure to arsenic and cigarette smoke (10). However nickel cadmium battery workers exposed to cadmium oxide dust have shown no clear evidence of an increased lung cancer risk (8). In the initial cohort study, a significant excess of lung cancer deaths was shown only for men employed more than 10 years in the always low-exposure category, and it was considered unlikely that this excess could have been related to cadmium exposure. There is now a stronger indication of an excess risk from lung cancer, with a significant excess mortality in the cohort as a whole, and also in the small, high-exposure group. The increased risk was the most marked for men who were first exposed before 1940, with long exposure, and who have a long period of follow-up . However the majority of lung cancer deaths were from a large nonferrous smeltery which provided over 60 070 of the total study population, but with only low or medium exposure to cadmium. In a subsequent case-referent study in the smeltery, the excess lung cancer mortality was found to be related to length of employment in the smeltery, but not to cumulative exposure to cadmium (results of Ades & Kazantzis awaiting publication). A working group on the carcinogenicity of cadmium (14) found the epidemiologic data on lung cancer difficult to interpret because of confounding from multiple occupational exposures including other putative carcinogens and because of the difficulty with the absence of smoking histories in assessing the contribution from cigarette smoking. A further case-referent study of lung cancer in workers with heavy past exposure to cadmium, taking account of concomitant exposures, is in progress.
In the initial study decedents were included below the age of 85 years for those causes believed a priori to be related to cadmium exposure and below the age of 75 years in a screening procedure adopted for other causes. An increased mortality from stomach cancer was identified in the 1979 cohort after a reanalysis in which a censoring age of 85 years was taken instead of 75 years. This increased stomach cancer mortality has been confirmed in the current update. In the five-year period 22 additional cases of stomach cancer have been identified, giving a significantly raised SMR of 179 for the five years and an SMR of 139 for the total period of study. In contrast to lung cancer, however, no relationship has been identified with intensity or duration of exposure, the excess risk occurring in the low-exposure group and among men with less than 10 years' employment making it unlikely that cadmium could have been responsible.
With regard to other diseases of interest, the small excess mortality from hypertensive disease, which was not statistically significant in the initial study, has remained virtually unchanged, and similarly there has been no change in the mortality pattern in the category nephritis and nephrosis, which showed a nonsignificant deficit. However, as in the initial study, deaths from cerebrovascular disease were significantly lower than expected. While this finding is likely to be due to chance, it should be noted that a low mortality from cerebrovascular disease had previously been observed for cadmium-polluted areas in Japan (7). Furthermore, in a screening for 32 other causes of death, a deficit has also been found for diseases of the circulatory system and, within this category, for deaths from ischemic heart disease. While the overall pattern does not show cadmium to be a risk factor for the cardiovascular system , these observations of a possible reduced risk require further study.