Document 2qORK5d9qbpZJ2Dn6avg397G5

firrter (D &{ ^ ' UNION CA INTERNAL CORRESPONDENCE ) Z-A -*- IOE CORPORATION olo rdgebubv oq^, wjSu(?v^cT30i7 ULjflTu. ..V t jr To PtvKCn l_OCOW mm*SEE DISTRIBUTION SEP10 1984 Oae OrigNioeing Oaoc. P Jw August 29/ 1984 HS & EA i+s ^4.^. *'- C-*--ht"`- . | ^ K-< '{A*' W**~*HPy C Cse> sa Ar^a Suoice: Epidemiology* P2 NIOSH Texas City Case Oj"j Control Study 'f'X The NIOSH cane control study (Lefflngwell. S.S. et. al) of the Texas City brain tumor cases has been accepted for publication by the Journal of Neuroepidemiology^ Enclosed are the galley proofs of Dr. Leffingvell's article. It is interesting to note that he did alter Che abstract slightly at our recommendation, so that the reader knows that the effects found for duration of exposure were not statistically significant* He has also used our inhouse study jbo contrast some of his findings (p. 6). and has added a plot of residence^ of the cases and controls (p. 14). The article hould not be quoted from directly until it is published, Dr. Lefflngwell expects this to occur in October or November. Sincerely. ^e.jj A* Robert Schnatter Epidemiologist/Biostatistician ARS:gac Enclosure RECEIVED 198-4 R. L YELTCf! rrmSfc' - .1: PRIVILEGED AND "CONFIDENTIAL MATERIAL SUBJECT TO PROTECTIVE ORDER" Cf ' l'to ISi UCC 071762 PRIVILEGED AND "CONFIDENTIAL MATERIAL SUBJECT TO PROTECTIVE ORDER" UCC 071763 75/ 759 760 761 762 763 764 765 Running tule: Brain bliomu in Tests Chemical Plant eceiveo AUG 2 4 5984 AjR- SCHNATTER Neureepkkmioiogy. Case-Control Study of Gliomas of the Brain among Workers Employed by a Texas City, Texas Chemical Plant (2) Sanford S. Leffingwell, Richard Waxweiler, Victor Alexander, Howard R. Ludwig. Wiliam Halperin Nations! Institute for Occupations! Safety and ^ealth, Cu^uiati, Ohio, USA 5 W OJ 769 771 H 772 H 773 774 775 776 777 H 778 779 Keywords. ticals Brain tumors Occupation Residence Ahatract. A 7 petrochemical plant had elevated standardized mortality ratios for neoplasms of the brain] A case-control study examined possible associations between gii- mas of the brain and title, departmental employment history, chemical exposure his> tory.'geographic tocaiic within plant, dates ofemployment, and residence. The peatest apparent risks were ted with exposure to carbon dioxide, diethyl sul&tc. diethylene glycol, ethanol, ethyl tsopropanol, methane, tetiaethykne glycol, and vinyl acetate; with first cm the 1940s or early 1950s. and with residence in La Marque, Tex. JjitoJtJjLefySfl) R * fi / s jt i f \9 gsfiii No significant di sura to any of these between cases and controls were apparent in duration of expo- j f**iSi ///*// * 782 Introduction ! i 784 In February 1979, the National Institute for Occupational Safety and 785 Health (NIOSH) and OSHA, with the aid of the company management 786 began an investigation ofa duster of primary brain tumors at a chemicals 787 and plastics plan] in Texas City, Tex. Alexandera al. [I] have described the // t# '/ Ininis a*w ; (II Si' */ ? * td H 88 characteristics of the brain tumor cases and the methods for case identifi K 69 cation. Two retrospective cohort mortality itudies ofall male hourly work 790 ers employed at this plant from 1941 through 1977 have been reported [2, 791 3], Findings included an overall standardized mortality ratio [SMR * 100 792 (observed/expedted)] for benign, malignant and unspecified neoplasms of 793 the brain of 206 (p < 0.05). For those who worked more than 20 yean at 794 the plant (i.e., who began work before 1957), the SMR was 377 (p < O.OSf . /s 795 [2]. No other cancers were found significantly in excess ofexpected, and the 796 overall SMR was not elevated. We here describe a nested case-control study twlZjf 797 of primary brain tumor cases among the employees of that plant. 798 Austin andSchnatter (4] have recently published a parallel study which 799 showed no association between exposures in the plant and brain gliomas. 800 They used deceased employees whose deaths were known to the company 801 as controls for the same cases, employed an unmatched design and did not 802 include analysis of residence. They were able, however, to use multiple 803 comparison groups; one group excluded controls who had died of other 804 malignancies while another permitted such decedents as controls. Subsets H 05 of these groups including only hourly employees permitted further refine UCC 071764 806 ment. These jwo studies are thus different enough to be complementary, 807 not duplicative. 810 Methods Materials 812 13 H 114 13 16 St? SIS 819 20 21 22 23 824 825 H 826 27 828 29 830 31 32 Case Identi Itoti A total of 23 ible eases were identified: the company documented 12 cnees from death certificates ii their possession; lists of ell adult males who were residents of sur- rounding counties who died of malignant brain tumors between 1950 and 1977 were matched with comj ly employment records* yielding 4 additional cases, another 3 cues hrramr ill and di during the course of the study; and 4 additional deceased cases were found in the of completing the cohort mortality study. We were able to obtain medical records 20 nf the possible eases and tissue specimens for 10. The Armed Institute of Pathology (AFIP) reviewed the tissue specimens, Where different di were recorded for the same patient, AFIP reviews were tanked fim. autopsy second, surgical pathology reports third, diagnosis from the hospital chan fourth, and certificate diagnosis last; the highest-ranked ^ign*"* was used for this study. Since glkn are carcinomas, arising from ectodermal tissue of the neural crest, while meningn arise from mesodermal embryonic tissue, we followed the recommen- dations of Sehoei et al. (5), and limited consideration in this study to the 17 gliomas among the 23 fc employees with death certificate diagnoses of brain tumor. This eliminated 1 case ith a metastatic brain tumor from an unknown ate. 1 who had been thought dinicalty have a brain tumor but was found at autopsy to have a congenital Malformation and tumor, and 4 meningiomas. A summary of the eases excluded and included is given table I. 8J4 Selection ofControl Subjects 35 For each easel a pool ofmatched potential controls was drawn from the cohort ofall 836 people ever empkr ed at the plant Matching criteria were: race and sex matched the case, 837 year of birth was ithin 3 yean of the cnee's; date of first employment at the Texas City 38 plant for the contn I was before that ofthe case, but year offirst employment at the Texas 839 City plant was no earlier than 3 yean before the case's: the date the control was Inst 840 employed was liter than the case's Inst date ofemployment; the control, ifdead, must not 41 have died ofa malignancy. For each case. 6 controls were then drawn by random number 42 from the pool of employees meeting these criteria. No control was used for more than I 43 case. Some cases aid controls had prior experience in refineries or ebemical plants, but the information avails de was. in our judgment, insufficient for analysis. 846 H 847 48 49 50 51 52 53 854 855 56 5? 58 H 859 60 H 161 862 63 64 Data Collect#) For each case and control, plant personnel completed coding sheets containing demo- graphic data, date af each new job title or deportment code, job code, deportment code, dale ofeach layoff, date offinal termination (ifno longer employed at the plant), and vital status (when know a). NIOSH/QSHA researchers independently verified the accuracy of the coding. Plant personae!I provided translations for the job and department codes, indicated which depanmem codes formed larger major department groups, and provided a list of chemicals used, podui ced. or redistributed in each department group. A 'department group' is an operational unit ofthe plant manufacturing related products and within which employees often re nsined over fairly long periods. The departmental coding schemes used for accounting purposes within the plant have changed over the years, so a common list tracing the history of each department was prepared. Plant engi were aMe to characterise the chemical feedstocks, outputs, and inter- mediate products feach department through the years the plant has been iiv operation, Since mdunriil hygiene data were availsMr only for recent yean and for certain com- pounds, we cqua the presence of a chemical in a department with potential worker exposure. This assiempnon was dearly not always accurate, nor were exposures necessarily equal in two depafc ments using the same chemical. UCC 071765 466 Analysis 467 Foresch jol. department, major department group, or chemical exposure common to 46* si least 4 cam, \ odd* ratio was calculated and tested for possible statistical significance 169 by Mantel andHi e/usef's [6] procedure, using the matched data program* ofRothman and 470 Boice [7J. Analyi % were conducted for periods less that) IS yean before the death of the g7l case. 15 or more years before the death of the case, and any time before the death of the g72 case (a case expo1 ed for greater than 15 yean and dying soon after Iasi exposure would be H |73 counted in all (h ee periods). Since this approach is a multiple-significance testing sects* H |74 nique which shot Id lead by chance to the landing ofapproximately one `statistically signify H |75 icant' positive as ociation, using a 90% (two-sided) confidence interval, for every 20 inde- 476 pendent jobs, dr lartments. or exposures considered, h is used here as an exploratory or 477 hypothesis-pcner iting mechanism; the probability values and confidence intervals died 474 throughout the p iper are given only to show relative strength of --ociation*. 479 Only those i onions ofthe controls* work experiences which occurred during the rime 440 the cotTetpondin \case was employed were considered in this analysis, since the matching 441 criteria selected < ontiols with longer total wort histories than the cases*. H |42 Duration ofexposure was examined for chemical exposures for which (a) the tssoda- 445 lion with brain turnon reached statistical significance in the analyses described above or H 114 0>) previous repeats untried a possible relationship. For each chemical, cases' and con- 445 tn>ls' median me aths ofpotential exposure 1S years or more before death ofthe case were 446 tabulated and a rank-sum test was performed [I], Employees with no exposure were 447 excluded. 444 Workers in a department nominally unexposed to a particular chemical could be 449 exposed to toxic airborne vapors or dusts from an adjacent department To asses this 490 possibility, the j curs worked by each subject in each department group were tabulated, 491 showing which department groups had about the expected 14 ratio of case to control- 892 years, which groups had disproportionately more case-years, and which had fewer. This 495 information wasjplotted oo a map of the plant which was inspected for dusters. 895 Non~ Work Faaon 896 We con the possibility that the excess risk at the plant might be a reflection of 897 an excess in the i immunities around the plant rather than a problem intrinsic to the plant 898 A case-location was retained to determine past places ofresidence for the cues and 899 controls and itkxial information was obtained through review of medical records 900 retained at the rnt. Analyses were conducted for communities in which at least 5 cues 901 had lived, using division of'ever lived* versus 'never lived* in the community for periods 902 15 or more before the death ofthe ease, leu than 15 years before the death ofthe ease, 903 and any time be the death ofthe case. The analysis ofresidence used all addresses up to 904 the date ofthe cue's death. A two-sided 90% confidence interval was calculated to display 905 the strength ofa sociation. Addresses were also plotted on a map, which was inspected for 906 further dues to he epidemiology of this occurrence. 909 Results 9U 912 913 H 14 H 15 916 Demogrc pthics Date of ii rst employment ofthe control subjects at the Texas City plant avenged 18.Q months before that ofthe cases, and the control subjects were born, on avertge. 4 months before the respective case, with standard deviations 14.2anti 24 months, respectively. Only 5 controls meeting the matching criteria w rre available for one of the cases. UCC 071766 918 In-Piani Vork History 919 "Operator was the only job code represented by 4 or more cases: the N 20 Mantel-Hams Eel odds ratio for operators was 0.54, with fih-H * 1-S4 (not N 21 statistically!. 1 here was no apparent commonality between the departments 922 in which these operators had worked. H 23 When aru lyzed by department codes, only the maintenance depart N 24 ment had 4 oi more cases; the Mantel-Haenszel odds ratio was 0.32 with N 25 1.37 (|tot statistically]. Grouping the department codes into major 926 departmenu y ielded no new groups of 4 or more cases. H 27 When cat and controls were analyzed by potential chemical expo 928 sures, a new pi oblem became evident: maintenance men moved throughout 929 the plant and i *re exposed to many different agents in an irregular manner. H 30 Accordingly, have examined the dau in two ways: in the first, mainte 931 nance men wi te considered to have been exposed to every agent in the 932 plant; in the second, they were excluded from analysis. The first method 933 may be closes to reality; indeed, maintenance men may have received 934 higher exposu es to toxic chemicals than operators since, in maintenance 935 work, they m ist open pipes, reaction vessels, and pumps. On the other 936 hand, the assu mption of exposure for all maintenance men, whether cases 937 br controls, nay tend to obscure an elevated odds ratio that might be 938 present ifexac t dau were available. Ofthe 505 chemicals reviewed, ubte II H 39 lists all chemit als to which at least 4 cases were exposed (excluding main 940 tenance jobs). 941 Sutistics Tor distribution of "exposures* among cases and controls are 942 presented in u ble III. The ubles include the chemicals to which 4 or more 943 cases were ex >osed and which showed the strongest positive association 944 with brain tumors, plus vinyl chloride. Vinyl chloride monomer was 945 included since it has previously been associated with brain tumor excess 946 [9-UJ. 947 Results c f analyses by duration of exposure are summarized in 948 uble IV. No statistically significant differences between cases and controls 949 were apparent in duration of exposure to any chemical. H SO In the mapping analysis of work locations within the plant, no signifi 951 cant clustering of areas with proportionately more case-years than control- 952 years was detected. Analysis of work locations of cases and controls within 953 the plant offei ed no useful dues to the etiology of the tumors. 955 956 957 958 959 960 961 962 H 63 964 965 Resident!W Data Addressee could be determined for 16 cases. The remaining case was known to havi lived for 15 years in Texas City and in the nearby town ofLa Marque, but ^either exact dates of residence in each community nor street addresses cou.11 be determined. Complete listings of at least the community of residence re available for 93 of the 101 controls; partial listings, often inexplicit as to the exact date of moving from one community to another, were available for another 7 controls. No information at all was available for the reside! ce ofjbontrol. In most instances, streets or streets and numben could be fetermined. but there were many instances where informants could no longfcr remember street addresses of 40 years ago. IX I J ^ iy7,t4^ Ji UCC 071767 966 Table V shows the distribution of addresses among cases and controls 67 (note: some subjects appear in both the Mess than 15 years* and 'greater than 968 15 yean* columns). The odds ratio for Texas City differed markedly from 1. H 69 while that for Galveston is essentially equal to 1. La Marque had an appar H 70 ent excess oflcases over controls (maximum odds ratio * 5.86). It is a resi 971 dential community, without heavy industry, and lies southwest of Texas H 72 City, west of the plant, and north of Galveston. A chemical dump, estab 973 lished in !9!9 and listed by the Environmental Protection Agency as a H 74 high-priority site for emergency cleanup, is located at the southeast extrem 973 ity of La Marque. Vinyl chloride levels as high as 161 parts per billion have 976 been measured at the fence line of the dump (12). Figure I shows that the 977 known addresses for the cases who lived in La Marque appear to duster 978 toward the southern and western parts of the dty, not adjacent to either the 979 plant or dump nor downwind from them. Drinking water for La Marque 980 and Texas City comes from deep wells, tapping the same stratum, while 981 water for Jveston comes from the Brazos River. The average length of 982 residence in Marque (excluding those who never lived there) was 12.4 H 83 years for and 16.3 years for controls; this difference was not statisti- 984 cally signiffi 983 Highlan Bayou, a slow-moving stream draining into Galveston Bay, 986 runs along southwestern and southern edge of La Marque, upwind from 987 the community. In view of the large number of publications associating 988 glioblastomas with viral agents [ 13-20], a mosquito-borae virus seemed an 989 interesting etiological hypothesis, as did kerosene and various inseetitides 990 used to control mosquitos during and after World War II. 991 To separate the effects of exposures to chemicals and residence in La 992 Marque, we (Classified subjects into 4 strata for each chemical according to H 93 history ofexposure to that chemical (exp+ or exp-) and history of Lajilar- 994 que residence (LaM+ or LaM-). Each of 3 strata (exp+LaM+, exp-LaM+, 995 and exp+LaM-) was compared to cases and controls who had neither risk 996 factor (exp-LaM-k i.tj three 2X2 tables were analyzed for each chemical N 97 and the odds |ratios were compared, as suggested by Kleinbaum et al. [21]. 998 We were unable to maintain the matching in this step, since too many 999 empty cells |Tor use with the Mantel-Haenszel procedure would have N 00 resulted. Results are shown in able VI. The odds ratios generally are greater 001 for La Marque residence without chemical exposure than for exposure to 002 individual chemicals without La Marque residence and often are quite a bit 003 greater for exposure to both risk factors than for either one alone. During 004 the period longer than 15 years before the death of the case for analyses in H 03 which maintenance men were excluded, only ethanol and di- and tetrJrthy- 006 lene glycol show a stronger association with disease than does La Marque 007 residence. 008 In reviewing the death certificates obtained for case-finding, we found 009 that 7 (9.9%) of 71 adult male residents of Galveston County whose death 010 certificates wure coded as 'malignant brain tumor* and who died between 011 1949 and 1977 had La Marque addresses on the death certificate. Estimates 012 of La Marqie's population over the same period were obtained [22]. 013 Weighting the populations of the community and county for the calendar 014 years under consideration yielded crude death rates of 2.00 adult male 015 deaths per IOLOOO total person-years at risk in La Marque and 1.63 for the H 16 remainder of Galveston County. The difference was not statistically signif 017 icant (p * 0.33). J JP M0T ucc 071768 020 Discussion 022 The grate it tpparem risks were associated with exposure to carbon 023 dioxide, diethy l sulfate, diethylene glycol, ethanol, ethylene, isopropanol, 024 methane, teim thylene glycol, and vinyl acetate; with first employment in H 25 the 1940s or ei rly 1950s; and with residence in La Marque [2], The chem H 26 ical association i found may need further study, but are not convincing evi 027 dence, particul uiy in view ofAustin and Scknazier's 14] negative findings. 028 The associitioi with residence is somewhat stronger. 029 Among ih in-plant exposures considered, the relationship between N 30 gliomas ofthe 1 rain and carbon dioxide is statistically significant only when 031 all work histories are included regardless of latency. We would expect a 032 carcinogen to show stronger relationships when latency was considered; the H 33 absence of a latency effect, coupled with the fact that carbon dioxide expo 034 sures likely to be experienced by workers would have virtually no effect on 035 normal physiologic levels, eliminates CO2 from further consideration. Its 036 appearance on me list serves to illustrate the pitfalls of multiple significance H 37 testing mentioned above and to inject a note ofcaution in forming conclu 038 sions based on other associations in this report. Methane, with 6 cases 039 potentially exposed (excluding maintenance men), is similarly unlikely to H 40 be present in amounts capable of increasing exposure markedly over back 041 ground levels produced by intestinal flora. 042 We were unable to find reports ofcarcinogenesis or mutagenesis testing 043 on tetraethylenp glycol. Diethylene glycol was found to cause bladder stones 044 and tumors in one test using rats [23], but has not been carcinogenic or , 045 mutagenic in cither experiments [24, 23]. Excluding maintenance men, a 046 total of 4 cases definitely worked in departments where di- or tetiaethylene 047 glycol was pret enL Even if the association were causative, the fraction of 048 cases attribute >!e would be insufficient to explain the observed excess; this 049 positive findin \ may be the result of multiple significance testing. 050 Diethyl sulfate is considered a carcinogen and has caused brain tumors 051 in experimental animals [26]. With regard to other chemicals, a statistically 052 significant excfcss of brain tumors was observed in a British isopropanol 053 plant, although the numbers were very small [27]. Several other cancers 054 have been associated with isopropanol exposure, although pure isopropanol 055 is not usually Considered carcinogenic [28-30]. Ethylene, like vinyl acetate 056 and vinyl chloride, has an unsaturated two-carbon moiety. Although both H 57 ethylene and vinyl chloride are metabolized through a highly reactive epox 058 ide stage whicn may alkylate organic compounds [31. 32], ethylene is not 059 mutagenic in Ames tests [Njosh. unpubl. data] and was not found to have H 60 any effect in a 2*year exposure study using rats [33]. There is less informa 061 tion available on vinyl acetate; it is not mutagenic in bacterial assay [34. H 62 35]. Information on association with length of exposure, latency, and num 063 ber of cases Aho could be attributed to the chemical if it were a brain H 64 carcinogen do not point clearly toward any of these chemicals but obser H 65 vation of othet clusters or cohorts might indicate lift one of them is danger 066 ous. H 67 While soihe of the associations found could be considered weak evi ] 068 dence of carcinogenicity, none was conclusive. There are wide confidence 069 intervals aroujid all of the odds ratios given; therefore, differences between 070 odds ratios should be interpreted cautiously. 071 It is possiple that a critical exposure was more general than implied by 072 the department group analysis used here, and that use of in-plant controls 073 constituted owr-matching which might have obscured a significant finding. N 74 We know of no satisfactory way to test this possibility within the confines of 075 the present si idy. , - v*/1 ^ I I hz. UCC 071769 H 76 Risk associated with La Marque residence seems greater than that asso * *H 77 ciated with ihe tfhemicalt studied. U Marque and Texas City share a com 078 mon water sourice, but La Marque has a higher ratio of cases to controls 79 than Texas City, therefore a water-borne environmental carcinogen [36] 080 seems unlikely. Since there are no major chemical or other industries 081 upwind from thle community, airborne industrial carcinogens do not seem H 82 to be a likely explanation for the association observed. (The wind is domi 083 nantly out of t)ie south-southeast.) The cases* shorter average duration of^ 084 residence in La iMarque argues against a causal association with residence. 083 From the com >n oT crooe death rates for U Marque and Galveston' 086 County, La M; iue does not appear to be over-represented among brain 087 tumor deaths, it this comparison does not assure that a cohort who lived 088 there ja the pas some of whom have moved away, is not a greater risk. A 089 refinca^Sdjoini this plant also appears to have an excess ofbrain tumors 090 ana tsunder st ly by NIOSH. Examination ofthe residence histories ofthe 091 cases who worl at that refinery showed that only 3 of 8 had lived in La 092 Marque. 093 The type of mosquito which constitutes the greatest problem in the La 094 Marque area hap a range of 25*50 miles, so control measures, which relied 095 on DDT and y-benzene hexachloride until the mid 1960s, have always been 096 countywide and would not appear to explain localization to this area [37]. 097 Italian fanners are reported to have higher brain tumor rates than urban 098 worker^ a finding which is statistically significant (38). The clustering of 099 cases near the periphery ofthe community may be meaningful, but further 100 information is teeded to understand iu 101 Greenwold et al. [39] have suggested that superior w 102 lead to more fr quent diagnoses ofbrain tumors among employed workers 103 with good medical insurance programs. They presented evidence of more 104 Masticated diagnostic methods in Eastman Kodak employees with brain \U 105 tumors. Allho it is reasonable to suppose such sophistication would lead 106 to fewer mi diagnoses of brain tumors, their report has not achieved 107 universal tance [40, 41] and does not directly assess the question of 108 missed diagn in the comparison populations. Schoenberg et al. [42] 109 found that di nces between incidence rates in Rochester, Minn., where 110 the Mayo Cls c provides virtually all medical care, and in Connecticut 111 were mainly to better diagnosis at autopsy; most ofthe excess cases in 112 Rochester due to meningiomas: and the differences in rates were most 113 pronounced in older age groups. About 70% of Rochester decedents had 114 autopsies, co pared to an estimated 38% in Connecticut; in Rochester, 115 60% of meni omas were found at autopsy, compared to only 17% of 116 glioblastomas, None of the cases in this series were first diagnosed at H 17 autopsy and gl omas predominated in this group (85 v$. 40.3% in Roches H 18 ter). Our impression, from the medical records examined, is that the diag H 19 H 20 noses cated in this case series did not hinge tests, but we cannot determine on either multiple or highly sophisti the extent to which diagnostic sensi I *2^ 121 tivity bias mi| In be a factor in this cluster. 124 Ackaowle igments 125 H 126 127 128 129 130 131 132 We wish lo bunk the Union Carbide Corporation for freely opening their records to us and in particular Mr. Denton Engle. Plant Manager. Dr. David Glenn. GulfCoast Med ical Director. Mft. Dorothy Heyen. Direelor of Personnel Relations: Mr. Robert Frantz. Engineer and Mf*. Ftrthan Yaicinkayo. Industrial Hygienist. We are also indebted to the Texas Bureau of Vital Sutisiics and to M.D. Anderson Hospital and Tumor Institute for their help in case finding and vital status ascertainment. Messrs. Richard Miller. Ben Bate. and Davis Levnc sfthcOSHA Houston South Area Office, and Dr. / H ilitam Uoyd of the OSHA Hcadqua ten SufT provided invaluable advice throughout the study. UCC 071770 US Reference i 136 1 Alexander. LefTvnfwtll. S.; Lloyd. J.W.; Waxweiler. RJ.; Miller. R.L.: Brain 137 cancers in petrochemical worten - A cue series report. Am. J. ind. Med. /:115-123 131 (1980). 139 2 Wuwilff. tj.; Alexander. V.; LeflinrwII. S.S.; Htrine. M.; Lioyd. J.W.: Mortality 140 from brain umor and other causes in cohort of petrochemical worten. J. aaia. 141 Cancer Inst 70. 75-81 (1983). 142 3 Austin. S.G., Schnatter. A.R.: A cohort mortality study of petrochemical worten. J. 143 occup. Med 25 304-312 (1983). 144 4 Austin. S.G.; Schnatier. A.R.: A case/conirol study ofchemical exposures and brain 145 tumors in patroehemical worten. J. occup. Med. 25.* 313-320(1913). 146 5 Schoenberg |B.S.; Christine. B.W^ Whisnant. J.P.: The descriptive epidemiology of 147 primary mupcraniaJ neoplasms: the Connecticut experience. Am. J. Epidem. 104: 148 499-510 (If 149 6 Mantel. N.; laenszel. W.: Statistical aspects ofthe analysis ofdata from retrospective 150 studies of di) t. J. aata. Cancer Inst 22:719-748 (1959). 151 7 Rothman. Boice. J.D.: Epidemiologic analysis with a programmable calculator 152 (USDHEW: I1H. Washington 1979). 153 8 Hollander. Wolfe, DA.: Nonparametric statistical methods, pp. 68-75 (Wiley, 154 New York l| i. 155 9 Maltoni, C.: I iictive value ofcarcinogenesis bioassays. Ann. N.Y. Acad ScL 271: 156 431-437(191 157 10 Byren. D.; Er^pholm. G.; Enghind. A.; Westerholm, P.: Mortality and cancer morbid H 158 ity in a group ofSwedish VCM and PVC production worten. Environ. Health Per- 159 sped. 17:167-170 (1976). 160 n Waxweiler. I.: Stringer, w.; Wagoner. J.1C; Jones, J.; Falk. H.; Carter, Cj Neoplas- 161 tic risk am workers exposed to vinyl chloride. Ann. N.Y. Acad. Sri. 271:4(MS 162 (1976). 163 12 Texas Air C< itrol Board: Vinyl chloride monomer air sampling at Tc City *Wye* 164 waste pits. Marque. Tex.. August 5-6. 1980 (Austin 1980). 165 13 Beeker, L.E. Narayan, O.; Johnson. R.T.: Studies of human papovavirus tumor 166 antigen in experimental and human cerebral neoplasms. J. can. Sci. annul. 5: 105- 167 109(1976). 168 14 Birkmeyer. G.D.: Miller. F.; Marguth, F.: Oneorna-viral information in human glio 169 blastomas. J. neurae Transm. 33:241-254 (1974). 170 15 Bendheim. P.E.; Dinowitx, M.: Panicles resembling oncornaviruses: spontaneous 171 release from cultured meningioma cells. Archs Neurol 34: 105-108 (1977). 172 16 Cuatieo. W.; Cbo. J.-R^ Spiegelman. S.: Molecular evidence for a viral etiology of 173 human CN5 umor*. Acta nturochir. 33: 149-160 (1976). 174 17 London. W.lHoufT. $.; Madden. D.L.: Fuccillo. DA.: Graven. M.; Wallen. W.C: N 175 Palmer. AE. Sever. J.L.: Brain tumors in owl monkeys inoculated with a human 176 polyoma vin t (JC virus). Science 201:1246-1249 (1978). 177 18 Ida, N.; Ikam l Y.: Ogawa. K.; Takada. M.; Sugano. H.: Cell culture from a rat brain H 178 tumor induce t by intracerebral inoculation with murine sarcoma virus. J. nauf Can 179 cer Inst. 33: <31-447 (1974). ' 180 19 Tanaka. R.; I oprowskt. H.; Iwaaki. Y.: Malignant transformation of hamster brain 181 cells in vitro ry human papovavims BK. J. aata. Cancer Inst. 56:671-673 (1976). 182 20 Yung. W.K.; Hank. N.K.; Vick. NA.: 'Glioblastoma* - Induction ofa reproducible N 183 autochtonous tumor in rats with murine sarcoma virus. Neurology 26; 7M3 184 (1976). 185 21 KJembsum. D.G.: Kupper. LL Morgensiem. H.: Epidemiologic research: Principles 186 and quamiiai|vc methods, p. 407 (Lifetime Learning Publications. Belmont 1982). 187 22 City of La Marque. Tex.. Comprehensive Planning Program: Economic develop- 188 meat. capital improvements, environmental assessment (Wise. Dallas 1981). 189 23 Fiuhugh. 0.0.; Nelson, A.A.: Comparison of ihe chronic toxicity of triethylene gly 190 col with that of tetmethylene glycol. J. ind. Hyp. Toxicol. 28:40-43 (1946). 191 24 Weil. C.S.: Carpenter. C.P.: Smyth. H.F.: Urinary bladder calculus and tumor 192 response folk wing either repeated feeding of diethylene glycol or calcium stone 193 implantation. Ind. Med. Surg. 36:35-37 (1967). UCC 071771 194 2$ 195 196 197 26 19* 199 200 27 201 202 28 203 204 205 29 206 207 30 208 209 210 31 211 212 213 32 214 215 33 H 216 217 218 34 219 220 221 222 35 223 224 36 225 226 227 37 228 3* 229 230 231 39 232 233 234 40 235 236 . 41 237 238 42 239 Pfeiffer. E.H., Dunkcibcrg. H.. Mutagenicity of ethylene side and propylene oxide and of the glytoli and halohydrins formed from them during the fumigation of foods. Food Comm. Toxicol. 18: 119*1 II (1910). Drockiey. H.[ Kruse, H.: Preussman. IL: Ivankovic. S.: Landschutz. C.: Cancerogene alkyhcrendc pubstaiuen. til. Alkyllulogenidc. Suffetc. Sulfonate, und ringgnpennic Z. K/ebelbrach. 74:241 (1970). .; Rattan. Mortality ofworkers on an isopropyl alcohol plant and ing planu. Br. J. ind. Med. 37:1549 (19*0). ; Ulis, IL; Alexander, V.; Maltoni. C: Importance ofsequential expo* uction ofepiehlorohydrin and isopropanol. Ann. N.Y. Acad. Sci. Ml' I. yth, H.F.; Nate. T.W.: Quest for a suspected industrial carcinogen, occup. Hlth 5: 935*547 (1952). Occupational and environmental cancers of the respiratory system, tional cancers and their environmental counterparts (Springer, New ; Osterman-Golkar, S.; Scgerfaack, D4 Svennscm, Kj. Calleman. CJ.: Alkylaiion of haemoglobin after metabolic conversion of ethene to ethene oxide in vivo. MuttL Res. 43: 175-1*4 (1977). Green, T. Hathaway, D.E.: Hie biological fete of vinyl chloride in relation to its oncogenicity Chem.-bioL Interactions 11:545-562 (1975). GrmUa. EJ.: A twenty-four month inhalation toxicology study in Fischer*344 rats exposed to itmospheric ethylene (Chemical Industry Institute of Toxicology. Re- search T Park 19*0). Banach. H ontesano. R.: Prescreening ofenvironmental and industrial chtmicili in a series o: thon-term tests for the detection ofpotential carcinogens. Comm. Eur. Communi (Rep.) EUR: 1SS EUR 63S*. Environ. Res. Programme, pp. 269-373 (19S0). Mjinsky, W Andrews. A.W.: Mutagenicity ofvinyl compounds in Salmonella typhi- murium. Ti tog. Carcinog. Mutagen. 1:259-267 (19*0). Cantor. K. Hoover. R.: Mason. J.T.; McCabe. LJ.: Associations of Cancer mor- tality with methanes in drinking water. J. natn. Cancer Inst. 61: 979-9*5 (I97S). Mix. D.: Personal communication. Musicco. Ml; Fihpprni. G.: Bordo. B.M.; Melotto. A4 Morallo, G.; Berrino. F.: Glio* mas and occupational exposure to carcinogens: case-control study. Am. J. Epidem. 116: 712-799 (1992). Greenwmld. P.; Friedlander. B.R.: Lawrence. CX; Hearoe. T.; Earie. IL: Diagnostic sensttivity bias * an epidemiologic explanation for an apparent brain tumor excess. J. occup. Med 23: 690-694 (19*1). Gann. P.; Rosenman. K.O.: Conclusions questioned in brain tumor excess study. J. occup. Medj. 24:42* (19*2). GreenwakL P.; Friedlander. B.R.; Lawrence. C.EL: Heame. T.; Earie. IL: Author's raspoose. J. occup. Med. 24. 421-432 (19*2). Schoenberg. B.5.; Christine. B.W.; Whisnam. J.P.: The resolution ofdiscrepancies in the report* I incidence of primary brain tumors. Neurology 28:817^23 (1978). 243 Sanford S. LeffingweU. MD. MPH. 244 National Institute for Occupational Safety and Health. 245 4676 CoMnbia Parkway.^ 246 . Cincinnati. OH 45226 (USA)J d Ucc 071772 N 000 AN!:NNEP!0* XA.96 L00I Table I. Cua fl ori cohort study included in and excluded from cuecontrol study 002 Pifnosii Included Excluded 005 Oliomu 007 Glioblastoma multiforme (includes diagnoses of astrocytoma 008 Brides 111 and IV) 010 ThaJamk gjidblasioma suspected clinically; no tumor ON in biopsy or abnormal appearingnrtn of brain cones 013 Pituitary adenocarcinoma suspected clinically; radio 014 therapy followed by proven glioblastoma 12yean later 016 Astrocytoma grade II 018 Metastatic tumor, unknown primary 020 Brain tumor tu peeved clinically, not found at autopsy 022 Meningioma, n alignant 024 .Meningioma, b nign 026 Subtotal 029 Total 17 14 1 I 1 I? 23 |_ 034 Table IL Cheir ieals to which four or more ases were exposed 035 Acetaldehyde 036 Acetic acid 037 Acetone 038 Carbon dioxid 039 Oieihanotamiii e 040 Diethyl satiate 041 Dietbyicne glyiol 042 Ethanol 043 Ethylene 044 Ethylene dkhl iride 045 Ethylene |lyco 068 Hydrochloric acid Hydroaypropyl acrylate Uopropanoi Isopropyl acetate Isopropyl peroxydicerbonate Lubricating oil Methylisobutyl ketone Methane Methanol Methyl ethyl ketone Moooethanolamine Nonanc(s) Potassium hydros- <T3eJ ------ - Sodium carbonate Sodium hydroxide Styrene Sulfuric acid Tetruethykne glycol Toluene Triethylene glycol Vinyl acetate Vinyl chloride UCC 071773 * N 000 ANI:NNEM044> A.92 L0T2 TaMtm.Fi of exposure and odds ratio* for potential chemical exposure* 073 Chemical 075 077 Maintenance men excluded counted as exposed1 078 V 081 087 Carbon dkuudi 089 090 109 Diethyl sulfate 111 112 7 f j 7131 ' Diethylene fliycol 133 134 i 153 Ethanol / 155 1 156 1 175 Ethylene f 177 1 178 1 197 Isopropandl 199 1 200 j 219 Methane 1 221 ! 222 I 241 Tetraettylene glycol 243 J 244 j 263 vinyl ajbeiatc 265 I 266 I 285 Vinyl chloride 287 | 288 0-U 154 ever 0-14 154 ever 0-14 154 ever 0-14 154 ever 0-14 154 ever 0-14 154 ever 0-14 154 ever 0-14 154 ever 0-14 15* ever 0-14 15* ever E/UE 6/1 S/4 7/4 2/5 3/6 4/7 2/5 3/6 4/7 2/5 3/6 4/7 4/3 7/2 7/4 5/2 4/5 6/5 3/4 6/3 6/5 2/5 3/6 4/7 5/2 5/4 6/5 3/4 3/6 4/7 F<m-hi 90% a on R<m.h> 4.80 1.35 2.14 4.43 4.87 2.10 (2)* (3)* (4)* 3J4 4.03 2.26 1.17 4.03 1.32 4.91 0.87 1.78 1.46 3.08 1.80 3.70 (3)* (4)* 3.10 2.74 3.30 t.16 0.87 M3 1.42-16.20 0J2-5.60 0.72-6.35 0.54-36.20 1.05-22.53 O.J7-7.73 8.15-* 95.55-59.43-- 0.64-17.59 O.BS-fg.46 0.67-7.62 0.34-4.04 0.88-18.32 0.45-3.91 0.94-25.5? 0.20-3.95 0.45-6.98 0.39-S.56 0.79*12.02 0.60-5.43 0.57-23.87 20.24-4 3.0I-* 0.80-12.05 0.57-13.20 0.84-12.88 0.29-4.58 0.20-3.76 0.31-4.04 E/UE Rm-m 90% Cl on 12/1 12/3 15/2 8/5 9/6 11/6 8/S 9/6 11/6 8/5 9/6 11/6 10/3 13/2 14/3 11/2 10/5 13/4 *9/4 13/2 14/3 8/5 9/6 11/6 M/2 11/4 13/4 9/4 9/6 11/6 2.88 1.76 3.40 1.19 0.75 M3 1.26 1.05 US M3 0.84 1.19 1.17 1.69 1.23 2.75 0.76 1.73 1.38 2.12 2.01 1.24 1.19 1.58 2.67 1.89 2.4? 1.05 0.91 M2 0.83-10.01 0.58-5.35 1.01-11.38 0.45-3.15 0.27-2.12 0.45-2.81 0.48-3.33 0.36-3.06 0.48-3.25 0.41-3.89 0.30-2.38 0.47-3.01 0.36-3.78 0.42-6.82 0.40-3.74 0.80-9.46 0.25-2.31 0.59-5.07 0.44-4.39 0.66-5.32 0.66-6.19 0.49-3.11 0.38-3.72 0.57-4.42 0.85-8.38 0.62-5.75 0.88-6.94 0.42-2.65 0.38-2.16 0.49-2.59 N 107 * Mantel-Haenuel odds ratio estimate: 90% Cl 90% confidence interval; '(N) * infinite odds ratio. V 108 based on N tables suitable for analysis: E ** exposed: UE unexposed. 309 > Maintenance employees who alio worked in other departments without exposure counted as unexposed. 310 i 0*14 work less than 15 years before death ofcase: 15* work IS Of more yean before death ofcase. Cases lit witty work experience in both the 0- to 14-year and IS* year time periods are counted in each group. Id ir UCC 071774 1 N 000 ANI:NNEPI04);A.9| L 315 Table tV. Median month* ofexposure to selected chemicals (15 yean or more before death 316 ofease; employee* with no exposure excluded) 317 Time in maintenance department 311 excluded counted as exposed |319 cases/controls -zl caaes/cooirols i1 334 Carbon dioxide 323 Diethyl Millste 326 (Methylene flyeol 327 Ethanol 328 Ethylene 329 Isopropanol 330 Methane 331 Tetraethylene |)yo J 332 Vinyl acetate 333 Vinyl chloride 44/36 7/51 30/61.5 9/50 33/41 33/91 19/31 50/61.5 55/60.5 63/59 0.30 - 1.51 0.55 0.65 m 0.62 - 1.47 0.81 _ 0.36 0.42 0.24 45.3/39 50/80 65.5/10.5 69/36 7/59 55/57 36/47 81/84 63/84 88/82 1.14 0.11 0.46 0.63 0.36 0.29 -1.10 0.77 0J5 0.80 374 The expected vahn of the sum is at(n na+ 1V2 with variance einj(ai+n2+iyi2. 375 '((observed rank** m)-(predictcd iank-*um))/(pfedicted variance): s > 1.96 implies p < 376 0.05. L 380 TaMe V. Distribute >n among cases and controls of community ofresidence 381 Community 383 Work period' Cases lived ever/never Rw-hi 90% Cl on 389 La Marque 391 392 402 Texas City 404 405 415 Galveston 417 418 0-14 15+ ever 0-14 15+ ever 0-14 15+ ever 9/8 9/8 12/5 5/12 4/13 6/11 3/14 7/10 7/10 3.95 4.80 5.86 0.48 0.28 0.49 0.92 0.8? 0.13 1.60-9.77 1.61-14.26 2J5-I5J5 0.16-1.41 0.09-0.86 0.19-1.29 0.32-2.67 0.36-2.06 0.36-1.96 428 429 430 R<m.hi Mantcl-Hitpsxel odd* ratio estimate; 90% Cl 90% confidence interval. 1 0-14 b wort leu t|an 15 years before death ofcase: 15+ work IS or more years before death of case.  L434 4j. TaMe VI. Unmatc ed odds ratios for combinations of exposure to selected substances and La Marque retifence 436 Time Maintenance men 438 excluded counted as exposed 439 440 441 464 466 467 443 493 493 496 521 523 524 549 551 552 577 579 580 605 607 608 633 633 636 661 663 664 689 691 692 717 719 720 Carbon dioxide 0-14 15+ ever Diethyl sulfate 0-14 15^. ever Diethylene glycol 0-14 !5 ever Ethanol Ethylene Isopropanol Methane 0-14 15+ ever 0-14 15+ ever 0-14 15+ ever 0-14 13+ ever TetraeihyJene gjyt Ol 0-14 154. ever Vinyl acetate 0-14 15+ ever Vinyl chlohfe 0-14 15+ ever exp- Exp* LaM- LaM(A) (B) Exp- Exp* LaM* LaM* <o <D> Exp- Exp* LaM- LaM- (A) (B) Exp- Exp* LaM* LaM* (O (D) 1.00 7.11 1.00 2.54 1.00 3.67 3.20 4.71 8.57 32.00* 22.X X.00* - (LX)* 1.43 _ IX 1.78 1.00 I (L14)* (3.73)*"* 5.43 13.13* 2J3 17.50* 1.00 4.89* 4.40* o.x 1.00 3.67- 5.30* 3.67* 1.00 3.63 24.17* 5.80* LX LSI LX 0.36 LX 1.33 1.40 6.25* 12.10* 4.31 4.17 7.00* 1.00 7.67* 1.00 int 1.00 3.71* 3.13* 0.X 3.43 16.X 7.56* 17.00* LX 1.69 l.X 0.48 LX 0.78 5.79 3.75* 3.50* 4.91 7.00* 8.49* 1.00 1.00 1.X 1.X 1.X 1.00 l.X 1.X 1.X l.X 4.89* 7.67* 4.29* 0.72 (l.X)* 2.05 6.X 0.56 1.64 0.53 (I.X)* 3.17 4.40* 4.60* 11.75* 0.X 5.11* 7J0* 0.87 8.67 (1.01)* (4.18)*"* 7.50 X.00* 7.X 3.33* 11.50* 7.50* 3.X 9.20* 0.71 inf. (1.22)* (inf.)*** 9.50* inf.* LX LX LX l.X LX - LX LX LX LX 1.69 0.38 1J3 0.81 (LX>* 1.15 (IW 0.21 L02 0.59 (LX)* 1.78 8.10 5.20* 10.67* 2-83 (5.90>* 6.X (2.22)* 4.67* 5.40* 1.22 (3.26)* 5.25 4.15 4.67* 1.62* 3.40 (6.74)*.* 8.44* (2.7I)*-* 3.73 7.59* 3.67 (I.29)*-* 14.54* I.X 7.67' I.X 7.67' 1.X 2.75 I.X 7.11 I.X 0.67 1.X 2.36 4.60* 0 3.29* IS.33 8.80* 11.00* 6.40 0.89 7.X 16.00* 10.67* 17.33* LX LX LX I.X LX 1.69 0.45 0.74 (LX)* 0JS 1.27 6.75 3.63* 3.78 4.50 6.77* 7.56* (l.X)* (3,03)*"* 1.24 7.09* 3.44* 11.63* I.X LX IjQD 2.67 0 0.X 6.X 1.X 13.50* 0 4.00* 2JS LX 2^0 LX 0.15* LX 1.20 9.X 2^5 8.57* 6.X 4.50* 7.50* 743 746 747 748 749 750 731 752 V 733 Maintenance emi loyees who also worked in other departments without exposure counted as unexposed. Except where othJ raise indicated, alt comparisons are to the unexposed/never lived in La Marque category. Column B may bJ interpreted as the effect ofchemical exposure alone, column C as the effect of La Marque rtaifence alone, a ad column D as the effect of both. Exp- * never exposed to the chemical; Exp* ever exposed to the eftj emieal: LaM- m never lived in La Marque; LaM* ever lived in La Marque: inf. no controls in the cat rgory results in infinite odds ratio. Figures in parentheses are odds ratios with reference to never lived in La ttarque/exposed. * Differ? from col imn A: Fisher's exact test, p < 0.05. * No cates in the never lived in La Marque/never exposed category result in infinite odds ratio. V 757 Fig-1. Map if La Marque addresses ever held by cases and controls. UCC 071776 UCC 071777