Document r3z30NDoQ9QbqqB3nEBYE1XE

Toxic Substance Exposure and Multiple Myeloma: A Case-Control Study `s2 peter 0. M o d s M.D., Thomas 0. Koepseil, M.D., Janet R. Oallng, Ph..D., 4 5 John W. Taylor, M.D.. (I Jo-ph L Lyon, M.D.. C. Marie Swanson, Ph.D., 6 Margad Child, M.D., and Noel S. Weis, M.D. 4,51Q ~ 8 m A c T d my eans ot a population-based, multicenter caseinvedgation. cenain toxic substances warn evaluated as wk facton for rnultiole myeloma Interviews worn completed on 698 subpcm with newly diagnosed multiple myeloma. and 1,683 mntrals were selected from the same geographic areas as those of me cases. Respondents were asked if thy had ever been "highly" ~ p o s e d "to one or more of a list of toxic substances or to other s ~ ~ ~ s t a nncoet son the list. With the aid of a toxicologist. responses w r e then categorrrad into 20 exposure groups. Thw who mported past exposure to pesticides had an estimated relative risk of 26 for multiple myeloma (95% confidence interval (C1)=1.54.61. Subiec:s exposed.to a variety of compounds commonly us@ by painten had an estimated relative risk of 1.6 (95%CI=l.l-24). An increased risk also was found for those who were exposed to source%of cardon monoxide (relative risk= 1.8. 95% Cl= 1.032). Associations of borderline statistical significance were found for metals andiorganically high polymen (plastia and elastomers). No statistically significant associations were soan for expoaum to fertilizam.dyes and inks; alkalies: acids othef caustic substances: chemical asphyxiants: aliphatic, chlorinated. or aromrfic h y d m carbons aldehydes and ketones: &em esten: oils: d u s t s or UhSOS.--JNCI 1986: 76:987-994. comprising Metropolitan Adanla (GA).This analysis dcait with data obtained &om 698 subjecu with newly diagnosed multiple rnycioma and kom 1,683 controls drawn from the general popularion of h e study a r u s Cascr.-AlI persons under 80 y u r s of age with multiple myeloma diagnosed between July 1. 1977, and June 30. 1981, were identified by the cancer regisvies seniing the four arms. (All rcgismn w e e participants in the SEER Program of the National Cancer Institute, Bethada. MID.) To be eiigible for h e study, a case was rquircd to reside in one of the study areas at the time of diagnosis. Case were disuibuted by age as follows: Two percent were under 40 y e a n of age, 6% were between 40 and 49 years of age. 2%were between 50 and 59 years oE age, 37% were between 60 and 69 y c m of age. and 33% were 70 or more years of age. Most of the cases were derived from Detroit (SI%), with smalIer proportions ABSII~ATZOPU us=: CI =codidaxe intant(s): SEER=S w e i l h n a . Epidcmiolog. and End b u l u From clinical and epidemiologic investigations. some preliminary evidence exists that multiple myeloma m y be associated with certain occupational exposures. Farmers may be prone to the development of this ancer. as suggested by several studies (2-5). In one of these reports. an increased risk was related to use of pestiddm (5). Myeloma may occur excessively in workm exposed to mcraIs (I, 6, 7), plastics (8).rubber (9, IO), pevoleum Products ( I f - l 3 ) ,and asbestos ( I , 14-17). Other SWVCYS have found an excess of myeloma among wood workers (1. 12. I&, l a t h e r workers (12. 19). and paintm (20). The following analysis was performed to further evaluate possible associa~ionsbetween exposure to some of h e substances and the risk of developing multiple myeloma. SUBJECTS AND METHOOS The results in this paper come from daw, generated by a population-based, collaborative case-control study of multiple myeloma. The investigation was a m i d out in four parrs of the United Stata: 1) King County and Pierce County (WA); 2) Davis County, Salt h k c County. and Weber County (UT); 3) the three counties compris`nS Metropolitan Detroit (MI); and 4) the five counties `Rcccivcd April 24. 1985; rnixd November 7. 1985: accepted Orrnnba26. 1985. *Supported in part by Public H a l h Savia qnnr I-N01Ch-15679 'from the Division of E x m m d hcliviua. National G n m InsticuteDcparunrntd Family Pndce. Univusicy of T w Health Science Cmta at b n Antonio. 1701 Floyd Curl Or.. Sur Antonio. Tx 78234. `Department oi E p i d a n i o l q , School of Public Health and Corn- muniw Y&ne. Univmicy Oi Washington. S a d e . WA 98195. 'Oivisian oi Public Hairh Sciences. F d Hutchinton Caner R c s a r d ~Gnca. 1124 Columbia SL, S a d e . W h 98104. Addrerr r e q u u u 10 Dr. DaIing ~l this addrar Depuunenc of EpidemioloJn. Sd~oalof Public Halh and Cammunity Mediane. University of Washington `Division oi Epidcmioloe, Department oi Family and Community Ycdidne. University of U u b Medical Gnrp. SO!? h r k Buildhg, 50 N o d M e d i a l Dr.. kk kke city. UT 84112 `Division of Epidemiolon. Michipn G n c a Foundauon. I10 Eut W ~ r r nAvc. h i t . MI 48200. 9 A t l j n u Canm Surveillance Center, Emory Univmity, 246 S y a more SL. Suite 100. Deatur, CA SOOSO. Io We arc gntdul to Or. David t t o n . Dtpuancnt of Environrnmcal Health. School of Public H a l & and Community Mediane. Uni-icy of Washington. for his advice on the toxicologic ~ p c c ouf b e study: to Cvol Ure. Pmjrn Coordinator. for witting in the study; to J o r u h n M. Liif tor computer programming assiscane. and to Carolyn Bums and Anne Petenon for their help in the prepantion of the manlwipc Data from h e G n c a Surveillance System and the yrisunce ot iu suif arc ~ p p & d 987 J N U . V O L 76. NO. 6. JUNE 1986 988 -Morris, KoepseII, Dallng, et al. coming from Washington State (26%). Atlanta (13%), viewed, a family member was located to provide and State of Utah (10%). Of all the cases of multiple inlotmation. myeloma ascertained at the four SEER Program centers Of the 698 case interviews, 68% were obtained directly and fulfilling these criteria, interview data were obtained from the subject. whereas 99%of the 1,683 control inter- from 698 (89%). Information was not obtained for the views were provided by the subject. Because the reported following cases: 23%due KO physician refusal; 33%due to frequency of several exposures differed between self- patient refusal; and 44% due KO miscellaneous reasons, respondents and proxy respondents, separate analyses e.g., had moved or could not otherwise be contacted by were conducted according KO respondent status. The the SEER Program center. Of the Detroit cases 6% principal results were shown both for comparison of all refused to participate, while the proportions in Washing- cases with controls and for comparison of self-respon- ton State, State of Utah, and Atlanta were lower (3, 2, dent cases with controls. and 1% respectively). The interviews focused on lactors hypothesized as Controls.-While the control group consisted of indi- being relevant to the etiology of multiple myeloma viduals selected at random from the same geographic (chronic antigenic stimulation, allergies, drug histories, areas as those of the cases, the method of conuol selec- tion varied according to the study area. In Washington State, a population survey using standard area sampling methods was conducted in 1979. King County and TABLE l.-Subataeur included in erposurr Pierce County were divided into 175 contiguous geographic areas of approximately equal population size. Exposure cstegory Major substance3 included Two sampling units averaging four households each were randomly selected from each area. Members of the household were then asked to provide a household census. Of persons 50-64 years of age, 1 in 9 served as a Acids Aldehydes and ketones Aliphatic hydrocarbons Muriatic, hydrochloric. sulfuric. chromic, nitric, and acetic acid; other miscellaneous acids Formaldehyde. ketones. acetone Gasoline. diesel, butane, propane. control, while all persons 65-79 years of age were ethylene. acetylene. kerosene interviewed. In the other three areas, random-digit telephone- dialing techniques were used to identify controls. In Alkalies Aromatic hydrocarbons Asbestos Ammonia. sodium chlorate, caustic soda. lime. cement Benzene, toluene, xylene. creosote Asbestos insulation, asbestos on Utah. random samples of working telephone numbers brake shoes, asbestos shipyard in the study area were called. If a number proved KO be Carbon monoxide Diesel, jet fuel, and automobile I residential, a census of the household by age and sex was exhaust; coal fumes: smoke taken. When a household member of eligible age was Chemical asphyxiants Hydrogen cyanide, hydrogen sulfide identified, a second random-sampling procedure was Chlorinated hydrocarbons Carbon tetrachloride, trichloroeth- applied to determine whether that person would become a potential control, with the selection probability de- pending on the estimated overall age and sex disuibution needed among controls to resemble the Utah cases. Dusts ylene, methylene chloride ' Cotton. textile. rock. wood. silica. coal, plant. and animals dusts: fiberglass dust: insulation material dust For each myeloma case, 2 controls were later drawn at Dyes and inks Dyes, beautician and/or hairdress- random from among the potential controls of comparable age (same 5-yr age group) and sex. In Atlanta and Detroit, random-digit-dialing techniques based on the Waksberg method (21)were used. For each myeloma case, a series of telephone numbers to be called was r Esters Ethers Fertilizers Metals ing products. inks Nitroglycerine, phosphate esters Ether Fertilizers, nitrous soda fertilizer Cast iron, lead vapors and liquid, arsenic, gold, manganese. brass generated by appending random numbers to the initial digits of the case's telephone number. The resulting numbers were then telephoned until a person of the fumes, copper powder and fumes, molybdenum fumes, mercury vapors and liquid, tin. silver, beryllium. cadmium same age (5yr), sex, and race was identified. In all Oils Oils,greases, grinding fluid three study areas for which random-digit dialing was Organically high Plastic and rubber compounds used, up to six to nine calls were made to a telephone number selected at random before the number was abandoned, including attempts during daytime, evening, and weekend hours. polymers Other caustic substances Paint and paint-related Chlorine. sulfur. fluorine, and nitrogen compounds: phosgene; ozone Paints, paint thinners and . Overall, 1,683 controls (83%) completed an interview. Of the controls not interviewed, 79%refused and 21% did not participate for miscellaneous reasons, e.g., had products and/or other organic solvents removers. lacquers, lacquer thinners, varnish, varnish removers, shellacs. glues. adhesives, other solvents moved or could not be contacted by the SEER Program Pesticides Pesticides, insecticides, organophos- center. Interviews.-Trained interviewers administered a stan- dardized questionnaire to each study participant. For individua\s who had died or were too ill to be inter- phorus and organochlorine compounds, arsenical insecticides, pyrethrum, herbicides, rodenticides JNCI. VOL 76. NO.6, JUNE 1986 II TOXICSubstances and Multiple Myeloma 989 mdiation and chemical exposures, and occupation) in the basis for matching (or frequency matching) in at addition to demographic variables such as race, marital least one study area and since they were assodated with status, education, and religion. The question pertinent disease status a priori. Odds ratios adjusted for these fac- to this report read as follows: "In your daily life, at tors were then calculated for each study area. With one home. at work, or elsewhere, was there ever a time when exception (noted below), these area-specific odds ratios ..you were highly exposed to products or fumes such did not differ from each other to a statistically signifi- as . 3" This question was followed by the possible cant degree for any of the exposures shown in table 2; responses: "Gasoline," "turpentine," "alcohol (other accordingly, only summary estimates that combined than liquor)," "cleaning solvents (carbon tetrachloride, results across all four areas were shown. Study area, etc.)," "oil (type)," and "other (specify)." If this ques- however, was treated as a fourth confounder because of tion was answered affirmativeiy for any exposure, the the varying ratio of cases to controls among the four interviewee was then asked to recall the year in which areas. All four confounders were controlled for simul- the exposure first occurred. taneously by using the Mantel-Haenszel stratified analy- Fornation o/ exposure categories.-Without knowl- sis procedure (22). We then computed the 95% CI. for the edge of case-control status, all of the responses to "other resulting Mantel-Haenszel pooled odds ratios, by Miet- (specify)" were hand tabulated and grouped into 20 tinen's method (23).The Mantel-Haenszel chi-square exposure categories with the aid of a toxicologist. (See statistic was also calculated for the adjusted odds ratios table 1 for a listing of the major substances included (22). Adjusted odds ratios were also calculated for indi- in each category.) Affirmative responses to the sub- viduals first exposed 1-19, 20-39,and 40 or more years stances gasoline, cleaning solvents, benzene, and oil were before interview, in comparison to individuals who did included in the aliphatic hydrocarbon, chlorinated not record an exposure. These associations were assessed hydrocarbon, aromatic hydrocarbon, and oil-exposure , by Mantel's test for trend (24). categories, respectively. Certain responses indicating As described below, stratified analyses suggested the substances or occupations in which several exposures possible importance of pesticides: paint, paint-related were possible were placed in more than one category. products, and/or other organic solvents; metals; and For example. persons reporting exposure to coke oven organically high polymers. For these exposures, logistic fumes werq classified as being exposed to carbon regression was also used to estimate adjusted odds ratios monoxide. 'aromatic hydrocarbons, and fhemical (25). Adjustment was made for age, sex, race, study site, asphyxiants. The responses requiring placement into and education as a measure of socioeconomic status. more than one exposure group are detailed in "Ap- Since the results from that analysis agreed closely with pendix." the results for the stratified analysis procedure, only the Analysis.-Three variabies (age, sex, and race) were results from the stratified analysis procedure were pre- identified as potential confounders, since they had been sented. TABLE 2.-Risk of multiple myeloma accwding to substance ezposurc. in comparison of all cases with controls Exposure category4 .- Percentage reporting exposure among: All cases. n=698 Controls. n=1.683 Adjusted odds ratiob Estimate 95%CI` Acids Aldehydes and/or ketones Aliphatic HC Alkalies Aromatic HC Asbestos Asphyxiants Carbon monoxide Chlorinated HC Dusts Dyes and inks Esters Ethers Fertilizers Organically high polymers Metals Oils Other caustics Paints and/or solvents Pesticides 2.9 1.2 15.6 2.4 2.7 1.0 0.4 3.9 13.2 3.3 1.3 0 0.6 0.6 2.0 5.6 14.9 0.9 7.3 4.0 2.5 1.0 0.6-1.9 1.2 0.7 0.2-2.1 16.3 0.9 0.7-1.2 1.8 1.0 0.5-1.9 4.3 0.6 0.3-1.0 0.6 1.3 . 0.5-3.1 0.7 0.7 0.2-2.5 3.4 1.8 1.0-3.2 14.0 0.9 0.6-1.1 3.5 1.0 0.6-1.8 - -0.7 1.9 0.8-4.6 0.2 0.4 1.2 0.3-1.3 0.1 3.0 0.6-15.3 1.2 2.0 0.9-4.1 4.3 1.5 1.0-2.3 16.9 1.0 0.7-1.3 1.8 0.5 0.2-1.3 4.8 1.6 1.1-2.4 1.5 2.6 1.54.6 *`HC=hydrocarbons. Mantel-Haenszel summary odds ratio estimate, adjusted for age (10-yr groups), sex. race (white. black, other), and study site. -=-O. `Test-based 95%CI for adjusted odds ratio. JXCI. VOL 76. NO. 6. JUNE 1986 RESULTS Table 2 shows that 4% of all persons with multiple myeloma were reported to have been exposed to pesticides, in contrast to only 1.5% of controls (adjusted odds ratioz2.6, 95% CIz1.5-4.6). As for most other exposures, the adjusted odds ratio was slightly higher (2.9) when self-respondent cases were compared to controls (table 3). The adjusted odds ratio was also found to vary among study areas as follows: Atlanta, 1.8; Detroit, 2.2: State of Utah, 15.5; and Seattle, uncomputable for l a d of any exposed controls. There was no clear evidence of a wend toward an increasing or a decreasing risk according to time since first exposure, however. Table 4 shows the specific pesticides reported by persons with such exposures. Because of the small number of subjects in each category, no firm conclusions can be drawn concerning the particular class or classes of pesticides responsible for the overall increase in risk. Subjects exposed to substances in the category that included paint, paint-related products, and other organic solvents also had an increased risk of myeloma (adjusted odds ratio for all cases=1.6,95% CIz1.1-2.4). Although the results are not shown in detail, there was little variation in the degree of excess risk according to time since first exposure. There appeared to be modest positive associations between myeloma and exposure either to paint and paint-related products or to other organic solvents (table 4), although firm inferences are precluded by small numbers of subjects with these exposures. For carbon monoxide, a positive association was apparent if responses of xlf-respondenu only were considered (ad. justed odds ratio=1.8,95% CI=l.O-S.P). Evidence was also found of a modest increase in mye. loma risk for persons reporting exposure to various metals (adjusted odds ratio for all ca~es=1.5, 95% CI= 1.0-2.3), although this assodation was of borderline statistical significance. A similar suggestive result was obtained for organically high polymers (adjusted odds ratio for all cases=2.0, 95% CI=0.9-4.1). There was a decrease in risk with increasing time since first exposure for metals and organically high polymers, but these trends fell short of statistical significance. Findings shown in table 4 indicate that no single subcategory of metals or of organically high polymers accounted for the apparent excess risk for persons reporting exposure to either class of substances. However, results not shown suggested that the excess risk associated with metal exposure varied according to race: For whites, the adjusted odds ratio was 2.1 (95%CI=1.3-3.4), while for blacks it was 0.6 (95%CIs0.2-1.4). Table 2 also shows increases in myeloma risk for those exposed to substances in two other categories (dyes and inks, fertilizers), but the associations might well have arisen by chance. There was little evidence of any increase in risk for those exposed to alkalies, acids, other caustic substances, chemical asphyxiants, aliphatic hydrocarbons, aldehydes and ketones, ethers, esters, chlorinated hydrocarbons, aromatic hydrocarbons, oils, dusts, or asbestos. Additional analyses (not shown) indicated very little change in the results shown in tables 2 and 3 if only TABLE 3.-Riak of multiple myeloma according to d s t a m e esposure. in comparison of self-respondentcases with corrtmb Percentage reporting exposure among: Adjusted odds ratid* Exposure category' Self-respondent cases. n=477 Controls, n=1.683 Estimate 95%CI' Acids Aldehydes and/or ketones Aliphatic HC Alkalies Aromatic HC Asbestos Asphyxiants Carbon monoxide Chlorinated HC Dusts Dyes and inks Esters Ethers Fertilizers Organically high pc.,fmers Metals Oils Other caustics Paints and/or solvents Pesticides ~~~ 3.9 2.5 .. 1.5 0.8-2.8 1.5 8- 1.2 17.8 16.3 1.1 0.4-3.6 1.1 0.8-1.5 2.9 1.8 1.2 0.6-2.5 3.4 4.3 0.8 0.5-1.4 1.0 0.6 1.3 0.4-3.8 0.4 0.7 0.7 0.2-29 5.0 3.4 1.9 1.1-3.2 14.7 14.0 1.0 0.7-1.4 3.8 3.5 1.2 0.7-2.3 - -1.5 0.7 1.9 0.7-5.3 0 0.2 0.6 0.4 1.1 0.3-4.9 0.6 0.1 3.2 0.5-19.5 2.1 1.2 2.0 0.9-4.6 6.3 4.3 1.8 1.1-2.9 15.7 16.9 1.0 0.8-1.4 1.0 1.8 0.7 0.2-21 . 8.2 4.8 1.8 1.2-2.7 4.4 1.5 2.9 1.5-5.5 ~ ~~~ ~~~ ~ ~ a HC=hydrocarbons. -=NbMo.antel-Haenszel summary odds ratio estimate, adjusted for age (10-yr groups), sex, race (white, black, other), and study site. Test-based 95%CI for adjusted odds ratio. J N U , VOL 76. NO.6. JUNE 1986 Toxic Substances and Multlple Myeloma 991 Type of e x p u r e in each category No.(%) for: Cues. ~ 6 9 8 Controls. n=.l.683 Pesticides Pesticides in general: paticides. pest sprays, mosquito sprays. cotton poison, bug sprays Organophosphonw compounds: dinzinon. mdathion. Chlorthion, parathion . Organochlorine compounds: DDT. chlordane. benzene hexachloride. heptachlor. toxaphene Arsenical pesticides Herbicides: weed killen. defoliage gaa used in Vietnam, 2.4-D weed killer, agent orange Paint and paint-related products and/or other organic solvents Paint and paint-related products: paints. paint thinners and removers, lacquers, lacquer thinners, varnish, varnish removers. shellacs. refinishing materials, glues and/or adhesives Other ormnic solvents: variem of unspecified solvents I Metals Welding: welding fumes, welding gas. steel welding, arc welding. electric arc welding Soldering: soldering fumes. lead soldering. soldering breadboards for computers Metal plating: acid plating, plating shop chemicals. fumes from acid electroplating 4 Foundry dust, fumes, smoke . Other metal exposures` 0rg;inically high polymers Plastics: epoxy resins, plastics. vinyl materials, polystyrene. polyethylene Elmtomen: rubbers 9 (2.7) 2 (0.3) 9 (1.3) 3 (0.4) 4 (0.6) 40 (5.7) 14 (2.0) 10 (1.4) 3 (0.4) 3 (0.4) l(O.1) 24 (3.4) 10 (1.4) 4 (0.6) 12 (0.7) 5 (0.3) 8 (0.5) 5 (0.3) 2 (0.1) 69 (4.1) 13 (0.8) 21 (1.2) 3 (0.2) 4 (0.2) 12 (0.7) 34 (2.0) 15 (0.9) 6 (0.4) Iron. lead vapors and liquid, anenic. gold. manganese. b r a g fumes. copper powder and fumes. molybdenum fumes, mercury vapors and liquid, tin. silver, beryllium, cadmium, toxic metal solution, metal frames,metal shavings, and metala at work. exposures beginning 5 or more years before the interview were considered. For the eight toxic substances listed in table 2 to which at least 3% of cases reported exposure, analyses were also performed to determine the strength of association between myeloma risk and a history ,of exposure to a combination of two such substances. Although limited by small numbers of exposed cases, the results of these analyses suggested that those combinations involving pesticides were associated with approximately a threefold increase in myeloma risk. This increase was quite similar to that shown in tables 2 and 3 for any pesticide exposure, suggesting little modification of any toxic effectof pesticides by joint exposure to other substances considered. DISCUSSION Numerous potential limitations to this study should be noted. First, the positive results may have been due in part to recall bias, particularly since exposure information was derived from an open-ended question rather than from a separate question for each category. As de- scribed by Sackett (26),cases of a given disease may have enhanced recollection of exposure to a potential cause because of "rumination" about possible explanations for their having been stricken by the disease. However, if this form of questioning did result in such bias, one might expect increased odds ratios for most, if not all, of the exposures. In fact, only a few exposures were found to be positively associated with disease. Recall bias mav also have resulted from the higher proportion of proxy respondents for cases compared with controls. Proxy respondents were probably less likely to be aware of or to recall a significant exposure. The most likely consequence of this inequality would be an underestimate of the actual disease-exposure relationship when all cases were compared with controls. In fact, the results demonsuated just that: Most of the adjusted odds ratios were slightly higher when self-respondent cases were cornpared with controls. Another potential source of error was the classification of responses into exposure categories. It is unlikely that all subjects accurately recalled all relevant exposures. Furthermore. judgments had to be made for several responses as to the most likely substances encountered in each instance (see "Appendix"). For example, a person who stated exposure to welding fumes was considered to have been exposed to both metals and phosphine. While it was probable that most subjects who listed welding fumes encountered phosphine at certain times, there were surely some cases or controls who were nonetheless falsely classified as exposed. However, with JNCI.VOL 76. NO. 6. JUNE 1086 992 Mods, Koepsell, Daling, et al. the assumption that these errors occrrrred randomly and were independent of case-control status, the most likely consequence of such misclassification would be to diminish and not to increase the apparent degree of association between disease and exposure. Systematic bias on the part of the investigators was avoided by classifying exposures without knowledge of case or control status. In addition, these results came from a large data set containing responses to many questions. When one examines a large number of disease-exposure relationships, there will be some statistically significant associations found by chance alone. It is possible some of the positive associations were due to this phenomenon. Similarly, the finding of statistically significant heterogeneity of effect among study areas for pesticides may have emerged on this basis, since this finding was the only instance of such heterogeneity among 20 exposures considered. For this reason, further study of these positive associations is warranted before firm conclusions can be reached. Pesticides Paints and Paint-Related Products and/or Other Organic Solvents T h e results of this study suggest that subjects exposed to paints and solvents have an lncreased risk of myeloma. Products in this category usually contain a mixfure ofcompounds includmg aromatic, chlorinated. aud amhatic hydrocarbons; ketones; ethers; and esters. With -the' possible excepuon of some of the Organic solvents, these products are used frequently by painters. When the solvents were eliminated from consideration, there were still more cases than controls who were exposed to the substances in this category. Only two studies were found in the literature in which the association between painters and hematologic cancers were examined (20, 29). On the basis of census reports of occupation and cancer for England, Adelstein (20) calculated a standardized mortality ratio of 126 for multiple myeloma among painters and decorators. Viadana (29) studied 1,345 white males with leukemia and 1,237 controls from New York State, Baltimore (MD), and Minnesota. Painters working in the construction industries had an incidence rate of leukemia estimated to be three times the rate for nonpainters. The largest increased risk was for subjects who acknowledged past exposure to pesticides. The results of Metals several studies suggest that 1 occupational group with high exposure to pesticides, farmers, may be at increased The overall association between metal exposure and risk for the development of multiple myeloma (2-5). In myeloma risk was not statistically significant in this two studies, significantly increased proportionate mor- study; but stratification by race showed that whites had tality ratios were found for white male farmers in a significant twofold increase in risk, while no increase Washington State (I)and the State of Iowa (2). Four of was found for blacks. Whereas this finding could signify the studies involved comparison of the death certificate an inherent racial difference in susceptibility to the statement of occupation for men who died of myeloma effect of metals, the potential exists that, with's0 many and for men who died of other causes, and each found race-specific contrasts possible, this result emerged by the former group to have a higher proportion of farmers chance alone. , than the latter group (3, 5 ) . In a more recent case-con- Insufficient numbers of exposed subjects precluded trol study in British Columbia, a significant increase in analysis according to type of metal, thereby making myeloma risk was found for subjects working in agricul- comparisons with the literature difficult, since most stud- ture (4). The present study also provided suggestive evi- ies have attempted to look at specific metals. In a pro- dence of such an association: With restriction of con- portionate mortality study in Washington State, MiI- sideration KO self-respondents only, the adjusted odds ham Jr. (I)reported a slight excess of myeloma deaths ratio for myeloma in relation to the subject having lived among white male smelter workers. Most of the indi- on a farm was 1.3 (95% CI=l.O-1.6). viduals studied worked in a large copper smelter in Also, one of the studies based on death certificates ( 5 ) Tacoma (WA). A similar study showed a Significant showed a significant increase in risk for farmers who increase in mortality for males employed as composing had lived in counties in which high use of both insecti- room workers in Washington (DC) (6). Composing cides and herbicides was experienced, but it showed no room workers are mainly exposed to inorganic lead, significant increase in risk for those who resided in both particulate and vapor. A case-control study con- counties in which use of these compounds was low. ducted at a Swedish copper smelter demonstrated an The carcinogenic potential of pesticides has been odds ratio of 4.6 (P=.02) for exposure to arsenic, but the documented in other work. Many commonly used pesti- cases included persons with leukemia as well as persons cides are known to be carcinogens in experimental with myeloma (7). animals, and several others are suspected of being car- cinogens (27). Furthermore, pesticides that are not themselves carcinogenic may have an additive that is carcinogenic (27);and several of the pesticides have the Organically High Polymers potential to react with nitrite to form carcinogenic Exposure to these compounds may involve their N-niuoso compounds (28). unreacted constituents, a variety of additives, degrada- JNCI.VOL 76. NO. 6, JUNE 1986 d e- K- id th $9 le re le in ic -1s in 3r aid ')# In :d id is id se fy le 'Y 'Y d g 1- 1- 1- IS 1- n It g g 1, 1- n e IS d 7 r. iLa I 1 3 i Toxlc Substances and MulUplo Myeloma 993 Lion products, or fully reacted polymers. Since subjects did not specify the circumstances under which exposure occurred, little can be said about the agent(s) responsible for the increased estimated risk. One way to approach this difficulty is by a separate analysis for plastics and elastomers, since different additives are used in the pro- duction of these two types of compounds. However, in this study both plastics and elastomers appeared to be related to disease. The results may support other findings in the litera- ture. Mason (8) found higher age-adjusted myeloma death rates for white males in U.S. counties in which plastic and synthetic fiber manufacturing took place rela- tive to deaths for white males in control counties. A cohort analysis of white male rubber workers revealed an excess of lymphoreticular cancers, which included both lymphatic cancers and myeloma cases (9).An addi- tional follow-up study examined the cancer mortality oE female workers in a rubber-manufacturing plant. There was a slight excess of multiple myeloma deaths, but the analysis was based on small numbers (10). 4 Other Exposures i There Are no previous reports of the association found between myeloma and the carbon monoxide-exposure category. The substances included involved exposure to several combustion productions in addition to carbon monoxide. Major pollutants were probably carbon dioxide, nitrogen oxides, sulfur oxides, hydrocarbons, and particulates. The increased risk could be due to carbon monoxide, one of the other combustion products, or any possible combinations. Preliminary surveys suggest that myeloma mortality may be excessive in U.S. counties involved in petroleum manufacturing (11) and metalrelated occupations with cutting-oil exposure (12). In another report. proportionate mortality rates for myeloma increased as years of membership increased in the Oil, Chemicai, and Atomic Workers Union in Texas (13). However, in the present study no positive associations were found for <he hydrocarbons or oils. The positive associations lor dyes and inks, a n G l i z e r s . were based on extremely small numbers and are therefore difficult to interpret. Multiple myeloma has been noted to occur in men occupationally exposed to asbestos (Z4-27). but the odds ratio of 1.3 observed here supports only a weak relationship at most. CONCLUSION On the basis of this study's findings, special attention might be paid to multiple myeloma in occupational epidemiologic studies of workers exposed to pesticides, paint and paint-related compounds, and carbon monoxide. Further research also appears warranted to assess the strength of any association between myeloma and particular types of pesticides, especially those in current use. REFERENCES ( I ) MILHAMS JR. ~CCUpatiOdmorulity in Washington State. 1950-71. Washington. DC: US. Covt Print Off, 1976 [DHEW publication No. (NIOSH) 76-1751. (2)B U R M E I ~LRF. Cancer mortality in Iowa w e n , 1971-78. JNCI 1981;66:~i-m. (3) MILHAMS JR. Leukemia and multiple myeloma in farmers. Am J Epidemiol 1971;94:307-310. ( 4 ) GALLAGHERRP, SPINEU JJ. ELWOODJM. et 11. Allergks and agricultural exposures as risk factors for multiple myeloma. Br J Cancer 1983;48:853-857. (9)B C R M E I ~LRF, EVER^ GD. VAN LIERSF. et al. Selecred cancer mortality and farm practices in Iowa. Am J Epidemiol 1983: I18:72-77. (6) GREENEMH. HOOVER RN. ECK RL. et al. Cancer mortality among printing plant worken. Environ Res 1979 2066-73. (7) AXEUON 0.DAHLGREN E,JASSSONCD, et al. Arsenic exposure and mortality: A case-referent study from a Swedish copper smelter. Br J Ind Med 1978:35:8-15. (8) MASON TJ. Cancer mortality in US. counties with plastics and relaced industria. Environ Health Pmpcct 1975; 11:79-84. (9) MONSON RR. NAKANO KK. Yomlity among rubber workers. I. White male union employes in Akron. Ohio. Am J Epidemiol 1976 103:284-296. (10) ANOJELKOVICHD. TAULBJ,EBELUMS. Mortality of female workers in a rubber manufacturing plant. J Occup Med 1978: ?0:409--113. (11) BLOT WJ. Gncer mortality in U.S. counties with petroleum industrja. Science 1977: 19851-53. (12) DECOUFLE P. STANISUWUYKK. HOUTENL. et al. A retrospective survey of cancer in relation to occupation. Washington. DC: U.S. Govt Print Off. 1977 (DHEW publication No. (NIOSH) 77.1781. (13) THOMAS TL. D ~ c o u Pd, M o u R f - E m R Monality among workers employed in petroleum refining and petrochemid plants. J Occup Mcd 1980 2.97-103. (14) LIEBENJ. Malignancies in asbestos workers. Arch Environ Health 1966; 13:619-621. (13) CERBEMRA. Asbestosis and neoplastic disorders of the hematopoietic system. Am J Clin Pathol 1970 53:204-208. (16) ROBERTSON MA. HARINCTOJHS. BRADSHAWE. The cancer pat- tern in African gold mincn. Br J Cancer 1971;2!5:395-402. (17) CAN E. JACOBSON RJ, YEWG K, et al. Asbestos-associated neoplasms of B-cell lineage. Am J >led 1979;67:325-330. (18) BRINTON LA. STONE Bj, BLOT WJ, et al. Nasal cancer in U.S. furniture industry counties. h n c e t 1976 2268. (19) DORKENH. VOLLHER I. The epidemiology of multiple myeloma. Investigation of 119 cam. Arch Geschwulstforsch 1968: 3I: 18-38. (20)AOEWEIN AM. Occupational mortality: Cancer. Ann Occup Hyg 1972: 1553-57. (21) WAK~BERCJ. Sampling methods for nndom digit dialing. JASA 1978;73:40-46. (22) MANTEL N. HAENSELW. Statistical aspects of the analysis of data from retrospective studia of disease. J Natl Cancer Inst 1959:22719-748. (23) MIEITINEN OS. Estimability and estimation in case-referent studies. Am J Epidemiol 1976 103:26-235. (24) MANTEL N. Chi-square tests with one degree of freedom: extensions of the Mantel-Haenszel procedure. JAsA 1963:58:590-700. (25) KLEINBAUMDG. KUPPER LL. MORCENSTERNH. Epidemiologic research. Principles and quantitative methods. Belmont. CA: Lifetime Learning Publ. 1982. (26) SACKETTDL. Bias in analytic research. J Chronic Dis 1979 32:51-68. (27)MOSESM. Pesticides. In: Rom WN. ed. Environmental and occupational medicine. Boston: Little. Brown & Co.. 1983: 547-571. (28) KEARNEYPC. Nitrosamines and pesticides: A special report on the occurrence of nitrosamines as terminal residues resulting from agricultural use of ctrwin pesticides. Pure Appl Chem 1980: 52499. D JNCL VOL 76. NO.6 . JUNE 1986 994 Morri"~K, oepsell, Drling, et rl. (29) VIADANAE. BROS ID. Leukemia and ormpUon* Pm hied 1972: 1513-521. APPENDIX RESPONSES REOUlRlNG PLACEMENT INTO MORE THAN ONE EXPOSURE CATEGORY Responses indicating substances or occupations that can result in more than one exposure were placed in more than one exposure category. These responses are listed 1-13and are followed by the exposure categories in which they were placed. The potential exposures of the substance or occupation follow each appropriateexposurecategory. 1) Coke oven fumes, smoke from coke ovens a) Carbon monoxide: carbon monoxide b ) Aromatic hydrocarbons: benzene c ) Chemical asphyxiants: hydrogen sulfide and hydrogen cyanide 2) Welding. welding fumes. welding gas, steel welding a) Metals: brass. lead, chromium b ) Other caustic substances: phosphine 3) Arc welding, arc welding fumes a) Metals: variety of metals b) Other caustic substances: ozone 4) Electric arc welding a ) Metals: lead, manganese, molybdenum, zinc b) Other caustic substances: ozone. fluorine. nitrogen dioxide 5 ) Fumes from acid electroplating a) Metals: arsenic, barium, chromium, cobalt, lead, mercury, molybdenum. nickel, selenium, zinc 6) Alkalies:ammonia, lime, sodium hydroxide, potassium hydroxide c) Acids: formic, nitric, sulfuric acids d ) Other caustic substances: carbon disulfide, fluoride compounds, nitrogen dioxide, ozone e) Chemical asphyxiants: hydrogen cyanide f ) Chlorinated hydrocarbons: trichloroethylene g) Dusts: graphite dust 6) Metal plating, acid plating, chemicals: plating shop a) Metals: molybdenum, zinc b) Acids: variety of acids 7) Galvanizing a) Metals: arsine, arsenic, brass, lead, zinc b ) Alkalies: ammonia c) Acids: anhydrous hydrochloric acid, sul- furic acid d ) Chlorinated hydrocarbons: trichloroeth- ylene 8) Soldering fumes, lead soldering. soldering breadboards for computers a ) Metals: arsine, cadmium, copper, lead b ) Chemical asphyxiants: hydrogen cyanide 9) Metal extractor a) Metals: variety of metals 6) Alkalies: ammonia 10) Vulcanizing tires a ) Metals: barium. tellurium b ) Alkalies: ammonia c) Other caustic substances: sulfur chloride, tetramethylthiuram disulfide d ) Chemical asphyxiants: hydrogen sulfide ' e) Aromatic hydrocarbons: amino compounds of benzene 11) Foundry dust. fumes, smoke a) Metals: aluminum. nickel, tellurium. tita- nium, zirconium b) Acids: phosphoric acid c) Other caustic substances: sulfur dioxide, fluorine compounds d ) Carbon monoxide: carbon monoxide e) Aliphatic hydrocarbons: acetylene j ) Aldehydes and ketones: formaldehyde ~ ,. g) Aromatic hydrocarbons: creosols 1 h ) Dusts: graphite dust 12) Dry cleaning fumes a) Other caustic substances: carbon disulfide b ) Aliphatic hydrocarbons: naphtha c) Ethers: dichloroethyl ether. ethyl ether, cellusolve d ) Chlorinated hydrocarbons: carbon tetra- chloride, dichloroethylene, ethylene dichlo- ride, methyl chloride, methyl chloroform, perchloroethylene, propylene dichloride, trichloroethylene 13) Laundry fumes a) Acids: acetic, formic, oxalic acids b ) Other caustic substances: chlorine, fluorine I ! I F a r r' [: S c W IT ir' "n IC ai ar tc 0: er tr tlJ V2 tc h. h. a.-.. I H C r: F 5; i 7 r: I i I i JNU. VOL 76. NO. 6. JUNE 1986