Document xd16My7v4DY9JEkboMx2VeRG6

cc!t't\ C. Meyer J. Zack Summary A historical prospective mortality study of 82 vrorkers exposed to PCB's in their manufacture from 1945-1966 at Monsanto's W. G. Krummrich plant failed to reveal any causes of death, including malignant neoplasms, which were significant. Observed deaths were compared with expected deaths from the U.S. standard population and the St. Clair County population. The observations were based on somewhat small numbers, but the study does not support the carcinogenicity of PCB's in the liver, lung, pancreas, or skin. HONS 010811 ROUGH DRAFT A HISTORICAL PROSPECTIVE MORTALITY STUDY OF WORKERS EXPOSED TO POLYCHLORINATED BIPHENYLS Background The term "polychlorinated biphenyls]Vj(PCB's) describes a group of irtuv^ theeretloally- two hundred nine_difchlorinated biphenyls which --, vary in consistency from oils to sticky resins in direct relation to the degree of chlorination of the phenyl rings. PCB's are manufacturedvby the chlorination of biphenyl with anhydrous cAlorine in the'presencdof a fer>ic chloride orviron filing catalyst. ijy v.vj'aaw-v (JiA/wj&V Mixtures of PCB's have been in use since 1929Ain transformers, capacitors, lubricants, "carbonless" duplicating paper, fireproof sealants, plastics, adhesives, and may other products. Their-resistance to biodegradation, photodegradation, oxidation, acids, bases, aid other chemical agents made PCB's invaluable in scores of industrial uses. However, these properties ajj,so make PCB's more stable as residual pollutants when re leased, -iffto the environment. Current interest in PCB's began in the midrl960's, when technological developments in gas chromatography and mass spectrometry made environmental detection of low PCB levels possible. The first report of PCB's in tjie environment came from Swedepf'where Jensen identified residues of PCB1s in fish from various SWedish waters, Risebrough g_t al^ found/PcB's and DOT in marine ecosystems of the Pacific Ocean. Detectable levels of PCB's were also found in 8 out of 39 human milk samples-in rural Colorado. The first report of possible PCB effects on humans occurred in 1968, when over 1,200 Japanese were involved in an epidemic of Yusho, an acnelike skin disease which was later traced to the consumption of a rice-oil contaminated with PCB's from a leaking heat-exchange unit. However, it was later shown that the rice oil also contained a high level of polychlorinated dibenzofif&ans (PCDF's), possibly due to the high temperature of the heat exchange unit. j^CDF's were shown to be relatively more concentrated in the liver than PCB's, and these findings clearly indicated the necessity to pay greater attention to PCDF's for clarification of the nature of Yusho. \ While acute effects from relatively high PCB exposure levels;are known and can be avoided by appropriate control measures, attention has mamcrssentlr been focused on the chronic effects resulting,from -long-term low levels of exposure to PCB's. considers PCB's to be suspect carcinogens, but animal tests have not ' given any definitive answer to their suspicions. In extensive two-year chronic oral toxicity tests of Aroclor 1242, 1254, and 1260 (42, 54, and 607 chlorinated Monsanto PCB products) conducted -by. Industrial-Biatoats for the Monsanto Company, no carcinogenic effects were observed in the rats and beagles tested. Hyperplasia, hypertrophy, and several benign tumorB were observed in the livers of the rats, but there was no evidence -- of malignancy in any of the test animals. Other investigators have lilSfiiiiver tissue effects ranging from hyperplasia to well-differentiated hepatocellular carcinomas in rats or mice fed with various PCB-containing diets. For example, Kimbrough et el HONS 010812 -2- fed rats with Aroclor 1254 and 1260, and observed hypertrophy, lipid accumulation, and ade^ofibrosis, but no malignancies, in the livers of the test animals. In a later experiment the same investigator fed Sherman rats 100 ppm of Aroclor 1260 for 21 months and reported the induction of hepatocellular carcinomas and neoplastic nodules in the test animals' livers, whereas control animals showed little or no effects. Sie ieUeiit National Cancer Institute report of iUfes^carcinogenicity tests on Aroclor 1254 concluded that the substance was not carcinogenic in Fischer 344 rats under the conditions of the bioassay* However, their study was. not definitive, since they alluded'-to the non-significant ' ce of gastrointestinalNtract carcinomas and Kepatoce'i^ulafv to the ative lesions ^B~containing iinn tthhe^iirr ttfeesstt .aanniimmaallss diet. iXltWv>A aaVs'~~ppoossss.iith}llyy^jbbeginB -2 "dfcejljalatteedd *nuSi , g^.^WeA.VJOtXhAiAN.rjJ WAT'A. OKt Wqv, *MV-4tK ,1/ Were have Been Very few epidemiological Studies of human populations exposed to low levels of PCB's over a long period of time. The Mobil Oil Company released a study in 1976 conducted by Ot--Anita. Bahn*in which a significant number of melanomas and pancreajftic cancers were found in PCB-expor.ed chemical workers and research personnel. However, the study has been severely critized due to deficiencies in the identification of the study cohort, failure to iSSSIctJI other potentially carcinogenic chemical exposures, and other oroblems which left the results of the Mobil study highly in doubt.7 *0<^.-- /K preliminary study of PCB-exposed workers was begun at the Monsanto (Company^hortly after the Mobil studyyiiT response--to the nee<l~fgr~a-~ vWIJtrepidemiological.- study-of the ohronic effects of- PCB- exposure. on humans. The mtndy^focused on workers at the W. G. Krummrich plant in Sauget, Illinois, where PCB's were manufactured from 1936 until pro duction ceased in 1977. This preliminary study failed to support the Mobil study results. No melanomas or pancreatic cancers were found, but an excess occurrence of lung cancer deaths was found in tie study cohort. However, due to the preliminary nature of the study, no follow-up was done on over 30* of the study cohort who had left the company, and further definition of the exposed population was needed. The study being reported here is an attempt to complete the epidemiological evaluation of the worker population at Monsanto's H. G. Krummrich plant, and evaluate the chronic effects of PCB-exposure on humans. The mortality experience of workers exposed to PCB's in their manufacture has been followed and evaluated. Study Population ; The Monsanto plants involved in United States manufacture of PCB's were the W. G. Krummrich plant in Sauget, Illinois, and the Anniston, Alabama plant. Since the preliminary study, 'upon which this study is based, involved the Krummrich plant alone, the present study deals solely with the Krummrich worker population. The study population from the Krummrich plant consists of all male, waged employees who worked at least six months in department 246 where PCB's were manufactured between January 1, 1945 and December 31,1965. This population includes active, retired, and terminated employees. Through interviewing plant personnel, it was determined that exposure levels of workers involved in PCB manufacture did not vary considerably; therefore, all male, waged workers in department 246 are considered to have a common exposure level. HONS 010813 -3- Methodology Work histories for all past and present Krummrich employees were obtained from the plant and screened for department 246 work. In order to verify the completeness of this search, two other sources were researched. Interviews were conducted with supervisory per sonnel from department 246 and any names which they recalled were recorded. After this, a plant developed computer listing of all department 246 workers was obtained, and the results of the work history search and plant interviews were crosschecked with the computer listing in order to insure completeness of the cohort. It should be noted that some workers, e.g., maintenance and salary, may have had significant exposures to PCB's, but were not included in this study due to exposure verification difficulties. By using these methods of identification, 82 male, waged workers (57 white, 25 non-white) were identified as meeting the criteria for inclusion in the study cohort. For each worker, the following information has been obtained from company records: employee name, clock number, social security number, race, sex, date of birth, date of hire, date of termination (if applicable), and as detailed a work history as was available. The employment and vital status information on the 82 workers is shown in Table 1. Vital status information was obtained by use of standard follow-up techniques such as worker interviews. Veteran's Administration files, Polk City Directories, and local telephone directories. All active employees and retirees currently receiving pension checks as of December 31, 1977, were assumed to be alive, and all persons known to have died on or after January 1, 1978, were assumed to be alive for the purposes of this study. For the 28 workers known to have died, a state death certificate has been obtained and coded by an experienced, trained nosologist for the underlying cause of death according to the International Classification of Diseases, Seventh Revision. Results Person-years of observation contributed by each worker was determined over five-year age and time intervals and summed for the total cohort. The mortality experience of the PCB-exposed study group was compared to the age, sex, race, and time-specific death rates of the standard United States population using the method of indirect standardization. The expected cancer deaths generated by this method were also^adjusted to reflect the cancer mortality experience of St. Clair County (in which most of the Krummrich workers reside) by use of data found in U.S. Cancer Mortality by County:1950-69. ' A standard mortality ratio (SMR) was calculated by the following formulae Observed Deaths Expected Deaths (By Cause) (By Cause) x 100 Statistical significance of the SMR's was measured by the Mantel-Haensycl chi-square statistic: ("j'^.Observed--^Expected | " - 0.5).* Expected HONS 010814 -4- The Poisson assumption is made in computing this statistic - the variance and mean of the expected are assumed to be'equal. The observed deaths by cause and the SMR's associated with each disease category are shown in Table 2. Twenty-eight deaths were observed among the 82 members of the study cohort. The average exposure to PCB's in department 246 for the 28 deceased workers was 33.6 months (vs. 34.8 months for the 54 living). The average age at death was 57.0 years, and the average time between first exposure to PCB's and death was 17.5 years. SMK values for the causes of death listed are not significant at the .05 level. The Only category approaching significance was "all malignant neoplasms" in the comparison of observed deaths vs. expected (U.S.) deaths. However, St. Clair County's expected deaths are higher than U.S. experience, so that the county-adjusted SMR is decreased from 206.5 to 190.3. Tables 3 and 4 summarize the mortality experience-of white and nonwhite males, respectively. Seventeen deaths were observed among the 57 white male workers. The average age at death was 59.2 years. These workers were exposed an average of 36.0 months (vs. 37.2 months for the 40 living whites), and lived an average of 20.8 years after their first exposure to PCB's in department 246. As in the case of all males (Table 2), no cause of death category showed significance at the .OS level. The high SMR's in the "all other cancers" category resulted from an observed-expected distribution of 2-0.398 (U.S.) and 2-0.378 (St. Clair). The two observed cancers in this category were a generalized carcinomatosis (ICD 199) and a malignant neoplasm of the lung specified as secondary (ICD 165) with no primary site listed. The high SMR's for this category are not significant at the .05 level, and the numbers involved are too few to draw any definitive conclusions. Table 4 outlines the mortality experience of the nonwhite male subset of the study cohort. There were 11 deaths among the 25 nonwhite study members. The average age at death was 53.5 years, with an average of 12.3 years between first PCB-exposure and death. The 11 deceased workers were exposed an average of 30.0 months (vs. 28.8 months for the 14 living). One category, lymphatic and hematopoietic system cancers, showed significance at the .05 level, and this category only showed 1 observed death. The category of "all malignant neoplasms" had an excess number of deaths which was not significant at the .05 level. As in most of the causes of death in Table 4, th^'numbers involved are too small to draw conclusions about health effects of PCB's from the SMR values of this table. Discussion The workers involved in this study are unique in that they represent the largest cohort available in the United States who were exposed to PCB in its manufacture. However, these workers were also exposed to other chemicals in the process of manufacturing PCB, such as hydrochloric acid fumes, trichlorobenzene, tetrachlorcbenzene, biphenyl, and small amounts of chlorine gas. Furthermore, all of the workers were exposed to other chemicals during their employment at the plant, such as oil additives, dyes, rubber additives, benzene, toluene, aniline, etc. The smoking, drinking, and prior work histories of the study cohort is also unknown. Therefore, in evaluating the effects of PCB's av'iq HONS 010815 -5- workers exposed to these compounds, the investigator must consider not only PCB exposure, but also a host of other potentially harmful agents. Another factor to consider is the route of entry into the body. The workers' main exposure to PCB's was through skin absorption, since PCB's have relatively low vapor pressures. However, nost of the chronic animal testing involved low dose ingestion studies, and environmental contaminants which enter the food chain would ultimately be ingested by man. It is unknown whether long-term, low dose PCB ingestion and long-term, low dose PCB skin absorption in humans can be considered as equivalent phenomena in terms of the potential health effects. With these factors in mind, results from the analysis of the mortality experience of the PCB-exposed workers indicate no statistically significant excess in either total mortality or deaths due to almost all of the other categories - including malignant neoplasms. The results of this study do not support the Mobil study in which melanomas and pancreatic' cancers were associated with PCB exposure, nor do the results uphold translating to man the outcome of animal tests, in which hepatomas have been associated with long-term PCB exposure. In the study cohort, the only digestive system cancer which occurred was a cancer of the colon which was diagnosed in a nonwhite study member. The significant excess of lung cancer observed in the preliminary Monsanto study is not supported by this study. While 3 respiratory cancer deaths were observed and 1.3 to 1.6 were expected, the difference is not statistically significant, and other factors, such as cigarette smoking, would have to be taken into account. It is important, however, to note that the small size of the study cohort could have prevented the detection of infrequent conditions which had increased in probability of occurrence due to the PCB exposure. A larger sample size would have increased the likelihood of identifying such a condition. However, since this worker population is probably one of the largest occupationally-exposed groups which has been identified and completely followed-up on, the ongoing concern regarding PCB's as possible human carcinogens warrants the continued follow-up of this group's mortality experience. The results of this study, while not conclusive in denying PCB as a health hazard, do fail to support other studies which had internal design problems, and put PCB's in a better perspective in terms of their effects on human health. / HONS 010016 TABLE 1 Follow-up Status of Workers Exposed to PCB's 1. Still Employed 2. No Longer Employed a. Alive, Employed Elsewhere b. Alive, Retired o. Deceased 22 60 10 22 28 HONS 010817 ' Cause TABLE 2 SHR''s Among All Males (n 82) for Selected Causes of Death ICD Code 7th Revision Observed SMR v County-Adj usted SMR 9T80T0 SNOW All Deaths All Malignant N'ecplasms Buccal Cavity and Pharynx Digestive Organs and Peritoneum Respiratory System Genito-Urinary Tract Lymphatic and Hematopoietic Tissue All Other Cancers Vascular Lesions of CNS Circulatory System Diseases Respiratory System Diseases Digestive System Diseases All External Causes Other* 140-205 140-148 150-159 160-164 177-181 200-205 330-334 . 400-468 4 70-527 53Q-587 800-998 28 134,0 8 206,5 190.3 0- 1 88,8 81.0 3 231.5 189.8 1 236.4 250.0 1 326,8 ' 341.3 2 346,0 367.0 14 136.9 l a- 2 178,9 1 233,0 3 133.0 *"Other" ICD codes are all those not listed above. &T80T0 SNOW TABLE 3 SMR's Among White Hale* (n Cause ICD Code 7th Revision All Deaths All Malignant Neoplasms Buccal Cavity and Pharynx Digestive Organs and Peritoneum Respiratory System Genito-Urinary Tract Lymphatic and Hematopoietic Tissue All Other Cancers Vascular Lesions of CNS Circulatory System Diseases Respiratory System Diseases Digestive System Diseases All External Causes Other+ 140-205 140-148 150-159 160-164 177-181 200-205 330-334 400-468 470-527 530-587 800-998 57) for Selected Causes of Death Observed SMR County-Adjusted SMR 17 134.8 4 165.5 146.4 0- - 0- - 2 0 !o 2 10 241.1 - 502.8 158.1 184.2 - 529.1 1 0 1 0 2 137.2 187. 7 +"0ther" ICD codes are all those not listed above. h HONS 0 1 0 8 2 0 Cause TABLE 4 SNR' s Among Non white Hales (n 25) for Selected Causes of Death ICD Code 7th Revision Observed SMR-. County-Adjusted SMR All Deaths All Malignant Neoplasms Buccal Cavity and Pharynx Digestive Organs and Peritoneum Respiratory System Genito-Urinary Tract Lymphatic and Hematopoietic Tissue All Other Cancers Vascular Lesions of CNS Circulatory System Diseases Respiratory System Diseases Digestive System Diseases All External Causes Other+ i ' 140-205 140-148 150-159 160-164 177-181 200-205 330-334 400-468 470-527 530-587 800-998 11 132.9 4 274.5 271.7 0 1 209.1 191.9 1 214.1 202.0 1 539.3 619.4 1 1098.9 1234.6* 0 4 102.6 0 1 256.8 1 113.7 1 84.0 +"Other" ICO codes are all those not listed above. * p <0.05 References X. Jensen, S: Quoted in 'Report of a new chemical hazard.* New Sci. 32:612, 1966. 2. Risebrough, R.W.,Reicho, P. , Peakall, D.B,, Herman, S,G,, Kirven, M.N: Chlorinated hydrocarbons in marine ecosystems. Nature 220:1098-1102, 1968, 3. Savage, E.P., Tessari, J.D,, Malberg, J.W., Wheeler, H.W., Bagby, J.R.: A search for PCB's in human milk in rural Colorado, Bull. Environ. Contam. Toxicol. 9:222-226, 1973. 4,il, Nagayama, J., Kuratsune, M. , Masuda, Y: Determination of chlorinated ' dibenzofurans in JCanechlors and "Yusho oil," Bull. Environ. Contam. Toxicol. 15:9-13, 1976, Kimbrough, R.D. , binder, R.E. , Gaines, T.B. : Morphological changes ' in liver of rats fed with polychlorinated biphenyls. Arch. Environ. Health 25:354-364, 1972. '6,', Kimbrough, R.D., Squire, R.A., Linder, R.E. , Strandberg, J.D,, Montali, R.J., Burse, V.W.: Induction of liver tumors in Sherman strain female rats by polychlorinated biphenyl Aroclor J. Natl. Cancer Inst. 55:1453-1456, 1975, J. Natl. Cane. Inst. 547-55S, 1974. 1260, 53: National Cancer Institute: Bioassay of Aroclor'1254 for possible carcinogenicity. Technical Report Series No, 38, DHEW Pub. No. (NIB) 78-838, 1978. $, Lawrence, QL^PCB? and melanoma, N. Engl. J, Med, 296:108, 1977. (y? L lV<c 1' ) HONS 010021