Cancer risks among New Zealand meat workers

FRASER 1. Cancer risks among New Zealand meat workers. Scand J Work Environ Health 1989;15:24-29. The study reports a series of case-referent studies based on the New Zealand Cancer Register and involving 19904 male cancer patients aged 20 years or more at the time of registra tion during the period 1980-1984. For each cancer site, the registrations for the remaining sites formed the referen ce group. An increased risk for lung (OR 1.30,95 % CI 1.06-1.58) and laryngeal (OR 2.01, 95 % CI 1.19-3.39) cancer was found among meatworkers. It was confined to men aged <65 years at registration. The risk for soft-tissue sarcoma was elevated (OR 1.90,95 % CI 0.90-4.02). The risk esti mate for all types of leukemia was elevated moderately (OR 1.45,95 % CI 0.90-2.31), but cell type specific analyses revealed a greater effect for acute myeloid leukemia (OR 2.12, 95 % CI 1.09-4.12). This study adds to the evidence that employment as a rneatworker is associated with increased risk for several forms of cancer.

Recent studies in New Zealand (1-6) , the United States (7,8), and several European countries (9)(10)(11)(12)(13) have focu sed attention on meatworkers as an occupational gro up that appears to be at increa sed risk for several form s of cancer. Slaughterhouse workers, meat pro cessors, packers, and butchers are exposed to substa nces that are carcinogenic in animals or suspected to be carcinogenic in humans. Th e chlorophenol 2,4,6-trichlorophenol (2,4,6-TCP) has been used in the treatment of pelts in New Zealand slaughterhouses (1). Other chemicals found in meat packing plant s include nitrosamines, but ylated hydrox yan isole (BHA) , and but ylated hydroxy toluene (BHT) (7,14,15). In addition to chemical carcinogens , meat worker s may be exposed to oncogenic retroviruses such as the bovine leukemia virus (BLV), which causes lymphosarcoma in cattle (16) and is present in tissues and milk of infected animals (17). Despite these exposures, only limited information exists to assess the risk of cancer among persons employed as meatworkers.
Increased risks among meat workers have been described in studies of soft-tissue sarc oma (I, 2), non-Hod gkin 's lymphoma (NHL) (3,5), and acute myeloid leukemia (4) in New Zealand. The se previous casereferent studies included registrants with soft-tissue sarcom a fro m 1976 to 1982 (1, 2), non -Hodgkin's lymphoma from 1977 to 1981 (3,5) , and leukemia from Reprint requests to: Dr JS Reif, Depar tment of Environmental Health, Colorado Slat e University, Fort Co llins, CO 80523, USA. 24 1979 to 1983 (4). Lung, bone, oral, and pharyngeal can cer and Hodgkin 's disease were associated with meat work in a historical cohort study in the United States (7).
Increased risks for lung can cer in but cher s and slaughterhouse meatcutters have been reported from several European countries over varying time periods (9)(10)(11)(12)(13).
Cancer risks among meatworkers ma y be effectively studied in New Zealand due to its large meat export indu str y and the existence of a national cancer register. We undertook the current investigation in order to assess further the increased risks for lymphatic and hematopoietic cancer and soft-tissue sarcoma noted among meatworkers in previou s studies and to determine whethe r the risk for lung canc er reported elsewhere also occurred in New Zealand. Furthermore, it seemed appropriate to determine whether this occupational group experienced excess risks for other relevant cance r sites. Therefore, the current study constitutes an analy sis of risks across all cancer sites among New Zealand meatworkers.

Subjects and methods
Cancer cases reported for adult men in the period 1980 to 1984 were obtained from the national ca ncer register. The New Zealand Cancer Register obtains complete population coverage from publi c and pri vate hospitals throu ghout the countr y, as well as from death cert ificates and autopsy findin gs (18). Nearly all patient s who die o f cancer have been reported to the nation al cancer register prior to death . The data file included all 24 762 men aged 20 years or more at the time of registrat ion . Anal ysis was restri cted to the 19 904 men who had a recorded occupation, 80 070 of the total sample (table I). zation. The NZSCO category of butchers and meat prepa rers included slaughtermen, other slaughterhouse workers, freezing workers (employed in large packing plants preparing frozen meat for export), meat graders, and retail butchers. It did not include tanners, fellmongers (workers who remove wool from sheep carcasses), and pelt dressers.
Ana lyses were conducted with the statistical ana lysis system (SAS) (23). The odds ratios (OR) were estimated with the Mantel-Haenszel procedure (23), and the 95 % confidence intervals (95 % CI) were calculated with Miettinen's approximate method (25). All the ana lyses were adjusted for age in lO-year intervals.
The age-adjusted odds ratios for specific cancer sites are shown in ta ble 2. An excess risk among meatworkers was observed for two forms of respiratory tract cancer, ie, malignant neop lasms of the lung and larynx. Ot her smoking-related sites did not show consistently elevated risks. For example, an increase was noted for cancer of the esophagus (OR 1.38) and pancreas (OR 1.45), but the odds ratios for bladder cancer (OR 0.70), cancer of the oral cavity (OR 0.66), and cancer of the pharynx (OR 1.07) were not elevate d . The precision of these estimates was poor. There were no cases of cancer of the nasal cavity and sinuses or pleura among the meatworkers.
The finding of an odds ratio of 1.90 (95 % Cl 0.90-4.02) for soft-tissue sarcoma , based on seven exposed

25
Ty pe of cancer- The study design involved a series of case-referent studies, each based on the overall group of 19 904 cancer registrants. For each site, the registrants for other sites formed the reference group . This stud y design has been used in several previous investigations based on the National Cancer Register of New Zealand, and the relevant issues have been descr ibed in two recent papers (19,20). The main advantage of using persons with oth er cancers as referents is the minimization of information bias, as well as the minimization of selection bias since cancer registration is virtually comp lete (18). The main potential disadvantage is that selection bias may occur if the exposure under study is associated with an increased risk for other cancer sites. Any such bias is genera lly likely to be small in occupational studies (19), although more significant bias may occur if an expos ure is associated with a common cancer such as lung cancer. Even in this situation, the appro priate approach is not always clear, since excludin g exposure-related cancers from the reference group may actually increase the net bias (20).
In this instance a check on the potential for selectio n bias was provided by New Zea land cancer morta lity data for food and beverage workers (the general occupational category to which meatworkers belong). The overall age-standardized cancer mortality for food and beverage workers was 161 per 100 000 person-years at risk, as compared to 139 per 100 000 perso n-years at risk for all emp loyed men. The excess mor tality was due almost entire ly to an excess for lung canc er, and the overall mortality for other cancers in food and beverage workers was virtually the same as that for other emp loyed men. This information, together with the finding in the current study of an excess risk for lung cancer, but not for other major cancer sites, suggested that lung cancer should be excluded from the reference group .
Thus the 4 224 cases of lung cancer were excluded from the analyses of other sites. For example, in the analysis of the data for cancer of the buccal cavity [Intern ational Classific ation of Diseases (lC D) codes 140-149], the case group consisted of the 698 registrants with this cancer type, and the reference group comprised the other 14 982 cancer registrants for sites other than lung cancer. A similar procedure was followed for toe other cancer sites, except for the calculation of risk for subtypes of leukemia , for which other forms of leukemia were also excluded from the referents. For lung cancer, all other cases of cancer (N = 15 680) were included as referents .
Occupation was defined in the National Cancer Register as the individual's current or most recent occupation at the time of registration and was coded with the use of the New Zealand Standard Classification of Occupations (NZSCO) (21), a modification of the International Standard Classification of Occupations (22). Occupational information for cancer registrants is obtained mainly by interview at the time of hospitali-cases, was consistent with the previously described association of soft-tissue sarcoma with meat work in New Zealand (1,2). However, the association between non-Hodgkin's lympho ma and meat work described in two previous New Zealand case-referent studies (3,5) was no t seen in the cur rent analysis . Th e overall odds ra-  (table 3), the previous finding (4) of an elevated risk for acute myeloid leukemia (OR 2.12) was reproduced, although it should be emphasized that there was considerable overlap with the previo us study. The risk for chronic myeloid leukemia was similar (OR 2. 12) to that for acute myeloid leukemia, but the precision of the estimate was lower due to smaller numbers. Similarly, an apparent excess of meatworkers among cases of acute lymphocytic leukem ia (OR 2.54) was fo und but was based on only th ree exposed cases.
We examined the relationsh ip between an increased risk for cancer among meatworkers and age by separating the cases and referents into those aged 65 years and over and those less than 65 year s of age at the time of registration (table 4). For cancer s of the larynx and lung, the excess risk was confined to the younger age group . The highest risks were noted for those less than 40 years of age for both cancers , ie, lung (OR 2.94 , 95 0,70 CI 0.72-12.02) and lar ynx (OR 7.07, 95 0J0 CI 1.05-47.84). Similarly, among the subtypes of leukemia, the odds ratios tended to be greater for the younger men, although the estimates were based upon very sma ll numbers of cases, and the odds ratios were also elevated for the older age groups.

Discussion
The finding of an excess risk for lung cancer among male meatworkers in New Zealand corroborates the excess of lung cancer among butchers and meatcutters in national mortality data for different time periods in Sweden (9), England and Wales (10), and Denmark (II) and in historical cohort studies of male (7) and female (8) meatworkers in the United States. Johnson et al (7) noted that various plastics are used to wrap meats, eg, polystyrene, polyvinylidene, polyethylene, rubber hydrochloride, cellophane, and polyvinyl chloride (PVC). Exposure to fumes from the thermal decomposition of plastic film causes the acute syndrome known as "meat-wrapper's asthma" (26). When heat is used to cut and seal PVC during the wrapping or labeling processes, the decomposition products include benzene, phthalic anhydride, and phthalates (7,27). Benzene is an established human leukemogen (28), while the phthalates have been reported to be carcinogenic in animals (29).
Exposures to fumes liberated during the packaging and labeling of meat or to nitrosamines, BHA , or BHT were sugge sted as candidates to explain the excesses of lung cancer found among the cohorts of male (7) and female (8) meatworkers in Baltimore, Maryland. In the studies from the United States, the excess risks among male workers were found in packing plants rather than slaughterhouses. The women were engaged primarily in the wrapping and labeling of meat in packing plants, grocery stores, or supermarkets (8). In a recent Swedish study (13), the type of work or work area was not related to the risk of lung cancer among butchers and slaughterhouse workers.
Unfortunately, a separation of these exposures in the New Zealand study was not possible, as the occupational category in the NZSCO combines slaughtermen, other slaughterhouse workers, meat graders and preparers, and retail butchers. However, New Zealand workers are engaged primarily in slaughtering, meatcutting, packing, and freezing meat for the export market, rather than in packaging for the domestic retail trade . Thus it would appear that only a minority of workers are exposed to PVC decomposition products.
The lung and laryngeal can cer findings are difficult to interpret because of the potential for confounding by tobacco smoking, which is the strongest determinant of these cancer types (30,31) . Smoking was found to be the predominant risk factor responsible for the increased incidence of lung cancer among butchers and meatworkers in a recent case-referent study in Sweden (13).
To evaluate the potential role of smoking in the excess risks found among meatworkers in New Zealand, we examined data on smoking and occupation from the 1981 census (32). Food and beverage workers, the occupational group to which meatworkers belong, were more likely to be current smokers (46.8 070) or ever smokers (65.4 %) than the total full-time male labor force, among which 37.7 % smoked currently and 59.6 % had ever smoked . The potential extent of confounding due to cigarette smoking was evaluated according to the method of Axelson (33). The odds ratios attributable to the relatively high smoking rates among meatworkers were estimated to be 1.20 for lung cancer and 1.16 for laryngeal cancer. Therefore, the increased risks for lung and laryngeal cancer found in the current study are likely to be due, at least in part, to cigarette smoking.
Slaughterhouse workers may be exposed to various chemicals, induding 2,4,6-trichlorophenol (2,4,6-TCP) used in the treatment of pelts (34). This compound is highly skin permeable, and the treatment of pelts in New Zealand has involved manually removing the pelts from the treatment solutions and handling wet skins without protection. Studies with rats and mice have shown that 2,4,6-TCP can cause lymphomas, leukemias, and hepatocellular carcinoma (35). Epidemiologic studies in Sweden have found associations between exposure to chlorophenols and soft-tissue sarcoma (36,37) and malignant lymphoma (38), although previous New Zealand case-referent studies ha ve not supported the association between chlorophenols and these cancers (1, 3,5). In particular, the findings for pelt department workers were similar to those for other slaughterhouse workers, and in only 22 % of the cases among sla ughterhouse workers did the person report working in pelt departments.
The failure to find an association between non-Hodgkin's lymphoma and employment as a meatworker is in conflict with previous findings in New Zealand. A previous study (5) of 183 non-Hodgkin's lymphoma patients showed an association with employment as a meatworker, with similar estimates of risk (OR 1.8, 1.7) calculated for ICD 200 and 202. In the current analysis of 535 patients, odds ratios of 0.81 and 0.90 were calculated for the two categories of non-Hodgkiri's lymphoma. An obvious difference between the studies is that the earlier study collected complete occupational histories by personal interview, while the present study used the current or most recent occupation recorded at the time of registration. Patients with other forms of cancer, obtained from the National Cancer Register, comprised the reference group for both comparisons.
Some increase in risk for leukemia among meatworkers was noted (OR 1.4), but it was more pronounced for specific cell types, particularly acute myeloid leukemia. The previously described association between acute myeloid leukemia and employment as a meatworker in New Zealand (4) was also found in the current analysis (OR 2.1), although there was considerable overlap of the two case groups. Recent evidence suggests that the acute nonlymphocytic leukemias are the most likely to be associated with occupational or environmental exposures (39). The association between leukemia and employment as a meat-worker is consistent with the frequently described increased risk among farmer s (40) and suggests that common expo sures may be involved .
The hypothesis that a leukemogenic agent is present in slaughterhouses is supported by the finding that veterinarians employed as meat inspectors also have an increased risk for lymphatic and hematopoietic cancers (41). Epidemiologic evidence for the role of zoonotic viruses in human cancer is currently weak, although some studies have found associations between various hematologic malignancies and farming (40). One candidate, bovine leukemia virus, is an exogenous C type retrovirus which has been established as the etiologic agent of the adult form of bo vine lymphosarcoma (l6) and occurs in New Zealand (42). The virus is related closely to human T-cell leukemia-lymphoma virus type I, the cause of adult T-cell leukemia in humans (43), and is transmitted horizontally (44). Virus particles are found in tissues and excretions of infected cows (16) and in milk (45). Erythroleukemia developed in two of six chimpanzees fed milk from cows infected with bovine leukemia virus (46), and chimpanzees inoculated with bovine leukemia virus became persistently infected (47). Bovine leukemia virus is capable of infecting human cells and inducing syncytia in tissue culture (48). Attempts to detect evidence of human infection with bovine leukemia virus by serologic methods have been generally unrewarding (49)(50)(51)(52). However, the interpretation of serologic studies and attempts at identification of bovine leukemia virus in human cells remains inconclusive.
The major disadvantage of using register data for case-referent anal yses of occupational cancer risks is the relati vely crude nature of the exposure data. Occupational data are provided as the current or most recent emplo yment at the time of registration rather than as a complete occupational history as might be obtained through an interview. Therefore, the estimation of risk for an ever-never exposure to meatwork is precluded . However, the net effect of this bias is in the conservative direction since the misclassification introduced will bias the estimate of risk towards the null value (53). More importantly, the data do not permit the examination of dose-response relationships based on duration of employment in an industry, the assessment of induction time, or the estimation of risks for specific departments or processes.
A potential disad vantage of using other cancer s as the reference group ma y be the int ro duction of selection bias if the exposure being examined is associated with more than one form of cancer. In this instance the effect would be to reduce the estimate of relative risk. Conversely, some " other cancers" may be associated with other occupations and thus decrease the proportion of meatworkers in the reference group . The latter phenomenon would have the effect of artificially raising the estimate of relative risk. In this instance it was found that the age standardized cancer mortality rates for male food and beverage worker s in New 28 Zealand were elevated and that this increase was due almo st entirely to an increase in lung cancer mortality. Thus selection bias was minimized by the exclusion of lung cancers from the reference group. In an y case, only minor changes in the odds ratio estimates occurred , depend ing on whether lung cancers were included or excluded from the reference group .
Despite the aforementioned limitations, and the possibility that the lung and laryngeal cancer findings may be due to smoking, this study adds to the growing body of evidence that employment as a meatworker is associated with several forms of cancer. Additional studies are needed to characterize the exposures which occur in various departments of the slaughterhouse and in the processing of meat products and to identify which chemical or viral carcinogens may be involved.