Document MJBGVpL29xYxjr7YyEKmv7LXx

*' !TlV Industrial Haxards ol Plastic* and Synthetic Elastomer*, pages 45-78 ,i *.' -4 1984 Alan ft. LlaS, Inc., 150 Filth Av*,, New York, NY 10011 ' V:': U i.'.V TRENDS IN CANCER fe'Vi'l 1 POLYMERS INDUSTRY jN* " : '"'V/ MORTALITY : .. AMONG! WORKERS .V'.:^' ; ' , 't 1 , i-v'i.'i.i-.v- ..... ,,f, - : "V \ 1 f-e9SC< J,j > U` "HI:* IHTHE SYNTHETIC William J. Diane Tart Nlfcliolson, Paul ^K;` Henneberger and IfafTfifi It- Environmental Sciences Laboratory, Mount Sinai School ol Medicine New York, New York of City University of New York 10029, U.S.A, 1 t' , . *W \S , t;ci v.r 1 ; -''L . ; R&S 132309 iNT I. i;: i INTRODUCTION ;: The reactive double bonded structure of ethylene-like molecules allows a wide variety of chemicals ' to undergo polymerization. Unfortunately, this same structure h*e been found capable of traur.formation to an epoxide by the manvnal- ian mixed function oxidase system (Bonse And Henschler, 1976). ' These epoxides or their reactive . mctabolites ' can bind to cellular macromolecnles and may be' responsible for the carcinogenicity of the parent molecule. Epoxide forma tion has been suggested as an intermediate in the carcino genic action of vinyl chloride (Van Dmiren,'1975) and vinyl- idene chloride (Maltoni, 1977), and in the mutagenic action of styrene (Milvy and Cairo, 1976). The epoxides of ethyl ene, styrene and vinyl chloride have been Shown to be carci nogenic, as well as directly mutagenic in bacterial test systems without the need for activation. 'The potential for conversion of eLhylcne-1 i kt* molecules to the epoxides is greater for unsymmetrical structures such as vinyl chloride and vinylidene chloride than for symmetrical structures, such as ethylene, 1,2-dichloroethylene or tetrachloroethy- lene. It is beyond the scope of this review to discuss the Structure-activity relationships of- the monomers used In the plastics industry. Nevertheless, available data suggest , *: .`"ftSs,.# that carcinogenicity depends on the metabolism of these AonomerS to reactive intermediates And that these reactions Aay be non-linear. However, when the metabolism of a com pound is understood, a coherent picture of the dose and time-'Jit dependence of cancer should emerge. 1 . r i **'!' r* 5% S -o t ) 17 66 / Nicholson, Henneberger, tnd Tarr At this time, data are available on both experimental and human carcinogenesis from exposure to vinyl chloride '' r/ . (VC) and on its metabolism that provide information Import4' ' i- ' * ant for the understandihg-'of observed dose4response rela tionships. This paper will consider ' thesis ' data on . VC in detail as they provide estimates of the trends in future disease potential from past exposures and information on the .'-v* - efficacy of current occupational standards. As human and animal data accumulate on the effects of exposure to other monomers, the approach Suggested- by VC can be applied to their evaluation; , t't- DOSE-RESPONSE RELATIONSHIPS VC is one of the best studied chemicals in animal systems. The magnificent research by Maltoni and Associates (1981) on nearly 7,000 animals Over A ten year period is virtually unmatched in experimental carcinogenesis A principal feature of their results is summarized in Figure 1 which shows the dose-response relationship for the percent age of animals that developed hcmangioSarcOmA (USA) of the liver from A hr/day, 5 day/wk, 52 wk'exposures to different concentrations of VC. As can be Aedn, the relationship is a non-linear one with evidence of Saturation at '.high . con- f w- 10,000 20,000 so.ooo ADMINISTERED CONCENTRATION OF VINYL CHLORIDE (ppm) M V >V Figure 1. The percentage of rats developing liver hemangio- sarcoma from 52 wk exposures to VC for A-hr/dayK 5 doy/wk. 30 99 (J) u NJ 03 .... , 'V- " / j<V: Cancer Mortality / 67 centrations. However, at concentrations of VC less than 500 ppm* A reasonably linear ilose-response relationship obtains. Gehring et al (1978) have, explained the nort-linearity in terms of Michaelis-Hcuten kinetics, in which the trans formation of VC to a reactive Intermediate follows the equation; ' . r* `j1- ` V = VmS/(Km + S) O) V and V are the rate and maxi mum rite, respectively, for the biotransformation of VC; S is the concentration of VC in inspired air, and K , the Michael is constant, K was deter mined experimentally to be 860 pg/I and V to be 5,706 \i%/U hr. Figure 2A displays the dolse-responsc relationship between the percentage of animals with liver USA and the quantity of VC metabolized according to Eq. 1. As can be Seen, a direct liner relationship exists with no evidence of a threshold or altered slope at low doses. The possibility of a non-linear dose-response relationship from detoxifica tion kinetic steps has been postulated (Gehring and BlAu, 1977); and discussed in detail (lloel et al, 1983), but no evidence exists for such non-linearity in the data yet available. The unweighted least squares .regression equation for the dose-response relationship is % USA = -0.066 i 0.0039 V (2) 60 s n * < u 25 Z 10 "* - 4.0 OZ 40 *- <3 2 wI 11 *W IX- o.s ** i o.z log V ( : V {fjq of vinyl chloride melobolized/4 hr) Figure 2. Linear and prublt dose-response relationships for the quantity of VC metabol izeil/6 hr exposure (5 day/wk,52 wk).V t? / f :> 33 fl 0) croo co 68 t Nicholson, Hennebsrger, and Tarr Gchring et al (1978) fitted the early data of Maltoni and Lefcmine (1975) to A log-probit model. Figure 2B showt the log-probit plot using all available dati from the Stud ies from Maltoni ct al (1981,). The, unweighted least Square* regression line is . j' : Probit = 0.2A + 1.01 log V ' (3) While such a plot fits the observable data (r^ - 0.68), the linear d^se-response relationship fits the data somewhat better (r = 0.77). Further, there is very limited biologi cal rationale for the use of a log-probit relationship in carcinogenesis and its use as a means of extrapolation to predict effects at very low exposures would appear to be more au act of faith Llian of science. On the other hand, a linear dose-response relationship between the incidence of USA and the quantity of VC metabolized is biologically plausible and fits all available data. Its use is strongly suggested. ' i TIME COURSE OF CANCER Much of human cancer has been found to follow a power law relationship with age (Armitage and Doll, ,1961; Cook ct al, 1969), R = btk (A) where R is the incidence rate of cancer at a specific site, t is age, and b a`iul k arc constants specific to site. In general, 'k is between A and 6 for most epithelial malignan cies. While data for exposures to specific carcinogens arc limited, bronchogenic carcinoma from cigarette smoking and mesothelioma from asbestos exposure also follow a power law of time from onset of exposure with an exponent between 3 and 5 (Doll and Peto, 1978; Newhotise and BCrry, 1976; Peto et al, 1982). These findings have been interpreted in terms of a multistage model of carcinogenesis, the implications of which have been discussed by PotO' (1977), Whittemore and Keller (1978), and Day and Brown (1980), among others. Deviations from the above time course occur with exposures to carcinogens that interact synergistically, such as asbes tos and cigarette smoking in the production of lung cancer. This interaction can be incorporated in the multistage mo del, but a more complicated relationship obtains.' However, Cancer Mortality / 69 for a rare tumor, such iJs USA, interactive effects may not be important and a power law relationship Should adequately describe the time course of risk following exposure: Some data arc avail able from, the use of Thorotrast in Japan and Denmark that indicate the incidence rate of USA docs follow Eq. 4 (Mori et al, 1979a; Mori et al, 1979b; Faber, 1978). The material was used in these countries over a limited period of time, so the incidence per calendar year and estimates of the population at risk can be used to estimate incidence rates by time from onset of exposure. While the data are very limited, they are consistent with a power law dependence of risk and suggest an exponent of ap proximately 3. Three is also compatible with the incidence of liver USA in the mortality study of polymerisation work ers described elsewhere in this Volume (Nicholson et al, 1983). However, only nine cases are available for analysis. PROJECTIONS OF FUTURE MORTALITY FROM PAST VC EXPOSURE Sufficient data have accumulated on the pattern of mortality from past VC exposures to allow an estimate future mortality from these exposures of using- a linear dose-re sponse relationship and a Lime course for risk of death from liver USA given by Eij. U. Figure 3 shows the number of cases of USA according to various measures of time that have been identified in the United States, Western Europe and the world (NIOSH, .1982). The distributions shown in Figure 3 arc the result of the exposure to VC of various groups of individuals in different periods of time Since 1935. Equa tion U indicates that the Incidences (not incidence rates) according to calendar year, year of exposure, and year from onset of exposure, respectively, are: lj.t Fj(Mort) b = dj b t*? . K.(Mort) b * ci ih J-* J ,,. J l C FjtMort) ;; (5a) (5b) (5c) where i represents the quinquenium of exposure and j, the quinquenium of observation. i runs from 1 to 8, represent ing the years 1935-1974 and j from 1 to 9, extending the observations through 1979. The F.(Mort) are the appropriate age and calendar year adjustments : to the population in JK i<r 'X ' *Cirtfti cuiiopi iterr'/ ilc/ a Pa Ps a 1 ao ) ro 73 *o CoUndar ' Ttsr oI Otolh i 10 13 >0 IS 30 33 *9 Ytori Sine* Ont4t of EipotuX lo 0<S Figure 3. The number of eases of hemangiosarcoma of the liver in the U.S., Western Europe ami the world .according to Sevc*.' ral time criteria. qtiinquenium j from normal mortality. The C.'a are propor tional to the total population exposure, i.e.j the average number of workers exposed in a given time period times the average VC concentration. Since the dose-response relation* ship for both inspired and metabolized VC ii linear in the range of most worker exposures, the risk of USA is propor tional to the total population exposure; one need not know the number of workers exposed and their vinyl ...Chloride exposure separately. A Cancer Mortality I 71 Relative value!: for tlic C.'s can be determined from two sets of data. The first is tfie incidence of HSA according to calendar period of first exposure (I.). Here the C.'s are directly proportional to th,e incidence1 in given calendar period and available data are sufficient to establish rea sonable values of C. for the time period 1935-1955; Addi tional data on C. can be developed from published data on the production of VC monomer. Figure 4 displays the avail able information on production in the United States (S.P.I., 1975-1978; U.S. Tariff Commission, 1948-1968) and Western Europe (0.E.C.I1. , 1971). A first approximation to the population exposure in different years would be to consider the C.'s to be proportional to VC production. However, average VC concentrations changed over the years b concern (Table 1) and an adjustment for the different relative exposuresin different times must be made. This adjustment is indicated in Table 1 and on Figure 4. Further, an ad justment must be made to take into account the different number of workers required to .produce a metric ton of VC in different time periods. As it would be expected that more workers were employed per tonne of VC produced during ear lier years, an adjustment is required to account for produc tivity. Initial estimates of this factor are also indicated in Figure 4. The relative population exposure; taken to be the product of production, the workforce productivity ad justment, and the exposure adjustment is shown hy the solid Figure 4. The production of VC in the U.S. and Western Europe along with estimates of the population exposures to VC polye...^, erization workers in different calendar periods., ' H'. SJ flo C/> CroO co cn R8.S 132316 72 / Nicholson, Henneberger, snd Tarr Table l Measure.! an.l ost | mat ml exposures to vinyl chloride to po 1 ymiT 1 rat. Inn wirkon In various Clue periods Calendar Period Approximate Vinyl chloride exposures (ppm) relative Barnes (1976) Oct ot al (1975) Suclu et al (1975) exposure Before 1950 1950-54 1955-59 1960-64 1965-69 1970-74 1000 1000 400 - 500 300 - 400 300 - 400 150 - 300 100 - 400 100 - 400 30 - 80 20 - 80 200 - 900 40 - 50 40 - 60 5 5 3 2 1.5 1 lines across each qu i nqueiii uin. The dashed lines across each quinquenium during earlier years are those determined by fitting the observed MSA incidence to Eq. Sa and matched to the value estimated from production data in the quinquenitim 1955-1959- As cap be Keen, the comparison of the two sets of data suggested that the population exposure prior to I960 was slightly less in some quinquenia than that estimated by the use of the adjustment factors indicated in Figure 4. The procedure of estimating the relative values for C., particularly in the years alter I960, is clearly an approxi mate one. To ronsider how sensitive any projections of future mortality are to the choices of C.'s, alternate choices are shown by the light solid lines in Figure 4a. Any realistic estimates uf the C.'s must lie between the two lines. ' 1 Relative values of I., and 1. . were calculated using the relative values of C shown In Figure 4, values of k between 2 and 4, arid absolute Values determined by matching to the incidence ilala of USA found in Figure 3. In this calculation, the age distribution used for time of first exposure was: 15-19, 8.5%; 20-24, 26%; 25-29, 26%; 30-34, 15%; 35-39, 11%; 40-04, 7%; 45-49, 4%; 50-54, 2.5%. This distribution was that of 740 VC workers examined by Mount Sinai School of Medicine personnel during 1974. The pattern of duration of employment was assumed to be a decreasing exponential with an average employment time of 12 years. This corresponds to typical patterns of employment for long-term workers in the chemical industry (Nicholson et al, 1982; Wong, 1982). Separate calculations were made for the Cancer Mortality I 73 United States and Western Europe. The results of this procedure, combining Lite data for the United Stated and Western Europe, arc shown in Figtire 5. As can be seen I._.* the incidence according to years from onset of exposurj is best fit by a value of k = 2. A value of 3 is compatible with the data, but values greater than A can be ruled out. I. is relatively insensitive to the choice of k but a value oi 4 fits the data best. An interesting fenture of this calculation is that the separate determination of the C.'s for Western Europe and the United States indicates that (hie population exposures per tonne of VC produced were approximately four times greater in Western Europe than the United States. This would Sug gest that more intense exposures occurred in some European plants or that more workers were exposed per tonne of VC produced. R&S 132317 1 -------- 3S-J9 43-49 1a 33-39 S3 9 YEAR OF FIRST EXPOSURE i* 0-4 IO*M .iiIi 10-14 30-34 40-44 YEARS OF EXPOSURE YEAR OF DEATH YEARS SINCE ONSET OF EXPOSURE Figure 5. A comparison of the calculated incidence of hemangiosarcoma of the liver with that observed in the U.S. and Western. Europe according to several time criteria and models for calculation. ".i '; j ' R&S 132318 74 / Nicholson, Henneberger, and Tarr One set of data that differs Significantly from that calculated is the distribution of cases according to years of VC exposure. As mentioned previously, we assumed the distribution of employment Limes in the VC industry would be a decreasing exponential with a .mean employment time of 12 years. The significant deficit of cases with employment, times less than 10 years suggest thut the available informa tion on duration of VC exposure may not be correct, that our assumed employment distribution may be -in error, that there may be an iindcrascertainiuent of cases witli shorter exposures, or that there may be proportionately less risk for shorter exposures than would be predicted on a linear dose-response relationship. It should be mentioned that duration of em ployment is not an important variable in assessment df popu lation risk. Shorter employment Limes would have required more men to be exposed, but their average exposure would be proportionately lower. Using the values of C.'s determined by the preceding analysis and values of k lrom 2 to 4, the mortality from liver USA is calculated to the year 2040, using Eq. 4a. These data are listed in Table 2, separately for Western Europe and the United States. Also shown in the data for the United States are projections using values of C.'s indi cated by the solid curves of Figure 4 and projections assiiming that the risk of MSA will increase quadratically with age of exposure. This age dependence was suggested by expe rimental results of (iruth cl al (1981). ^We also considered a time course for IIS A that increased as t for only 45 years and remained constant thereafter. As can be seen, the pro? Table 2 Projections of mortality ln the United States and Western Europe to the year 2040 from exposures to vinyl chloride prior to 1975 Model Total Projected Horcallty Ab United States Western Europe , to 45 years Cron onset of exposure , upper exposure curve. Fig. 4 , lower exposure curve. Fig. 4 . + Age1 190 340 630 310 240 450 260 540 1190 27BO 1120 26 desths have occurred through 1979 39 deaths have occurred through 1979 t Cancer Mortality / 75 jected numbers of USA for the United States range from 200 to 600 and, for Western Europe, from 550 to 2,800. (The greater range for Europe is the result of the mote recent usage of pattern.) The most probable projection fbr future disease is felt to be that represented by -a power of 3, a choice suggested by Thoratrast data and the very limited mortality data on USA in the study by Nicholson et al (1983). I.ower values arc also reasonable, but the fit to the data would suggest that the use of a power of 4 may be inappro priate. Obviously, many caveats exist in the consideration of these projections. The estimates strongly depend upon a reasonable ascertainment of cases through 1979. The concerns for VOinduccd USA in recent years would suggest that ascer tainment was fairly good, at least for long term employees and pensioners. However, some cases in short term workers may have been missed. The projections also depend on the choices of the 0, and the k. We have projected mortality based on reasonable choices for these parameters. However, other choices cannot be absolutely excluded. While these uncertainties exist, the data indicate that, within a factor oi 2 or 3, future 1IRA mortality from exposures prior to 1975 will be about 350 deaths in the United States and 1,200 in Western Europe. further, these deaths will occur in a relatively small population. In the United States, the group at highest risk would lie comprised of fewer than 5,000 individuals. Among this heavily exposed group, USA may account for !0% ol all deaths (Nicholson et al, 1983). Clearly, any intervention techniques that might he developed to reduce this projected risk could be efficiently applied. OCCUPATIONAL STANDARDS FOR VC Nicholson et al (1983) have shown that liver USA accounts for at least 50% of all VC-induced malignancies. Thus, it would appear that overage exposures of 200-500 ppm in pre vious years will lead to 1,000-4,000 excess cancer deaths in all workers exposed to VC in Western Europe and the United States prior Lo 1975. If a standard of 1 ppm is met, the average exposure of all the workers would be between 0.2-0.5 ppm, 1,000 times less than that which existed previously. One would expect the VC-induccd malignant risk to be reduced by a corresponding amount. This implies that, if the VC industry complies with a 1 ppm standard, cancer from employ- fV<u 7fc ilr R&S 132320 Cancer Mortality / 77 R&S 132321 Gehring PJ, Watonabe PG, Park CN (1978). Resolution of dose-response toxicity data for chemicals requiring meta bolic activation: example - vinyl chloride. J Toxicol Appl Pharmacol 44:581-591. Groth DU, Coate WB, Ullond BH', Hornung, RW (1981). Effects of aging on the induction of angiosarcoma. Environ Health Persp 41:53-57. floel DG, Kaplan NL, Anderson HW (1983). Implication of non linear kinetics on risk estimation in carcinogenesis. Sci 219:1032-1037. Haltoni C, Lefemine G (1975). Carcinogenicity assays of vinyl chloride: current results. Ann NY Acad Sci 246: 195-224. Haltoni, C (1977). Recent findings on the carcinogenicity of chlorinated olefins. Environ Health Persp 21:1-5. Haltoni C, I.efcminc G, Cilibcrti A, Cotti G, Carretti D (1981). Carcinogenicity bioassays of vinyl chloride monomer: a model of risk assessment on an experimental basis. Environ Health Persp 41:3-29. Hilvy P. , Garro A3 (1976). Mutagenic activity of styrene oxide (1,2-epoxyethylbenzene), a presumed styrene metabo lite. Mutat Res 40:15-18. Hori T, Kato Y, Shimnmine T, Watanabe S (1979a). Statisti cal analysis of Japanese Thorotrast-administered autopsy cases. Environ Res 18:231-244. Hori T, Nortiyamc T, Kato Y, Tahahashi S (1979a). Epidemio logical follow-up study of Japanese Thorotrnst cases. Environ Res 18:44-54. National Institute of Occupational Safety and Health (U.S.) (October,1982). Reported cases of angiosarcoma of the liver among vinyl chloride polymerization workers. Newhouse ML, Berry G (1976). Prediction of mortality from mcsothelial tumors in asbestos factory workers. Brit J Indus Med 33:147-151. Nicholson WJ, Pcrkel G, Selikoff IJ (1982). Occupational exposure to asbestos: population at risk and projected mortality - 1980-2030. Am J Indust Med 3:259-311. Nicholson WJ, Hcnncbergcr P, Seidman 11. Occupational ha zards in the VC-PVC industry. This volume. ft Organization for Economic Cooperation and Development, Chemical Industry (1971). Quoted in: Levinson C. Work ha zard: vinyl chloride. ICF Geneva. Ott MG, Langner RR, Holder BB (1975). Vinyl chloride expo sure in a controlled industrial environment. Arch Environ Health 30:333-339. 78 / Nicholson, Hennebcrger, and Ttrr Peto R (1977). Epidemiology, multistage models and short- , term mutagenicity teste. In: Origins of Human Cancer (Eds. Hiatt IQf, Watson JD, Winsten JA), Cold Spring Harbor Laboratory pp. 1403-1430. The Society of the Plastics Industry, Inc (1975-1982). Facts and Figures of the U.S. Plastics Industry, New York, Suciu I, Prodan El, Poduraru A, Pascu L (1975). Clinical manifestations in vinyl chloride poisoning. Ann NY Acad Sci 246:53-69. U.S. Tariff Commission (1948-1968). Polyvinyl chloride and copolymer production data. Van Duuren B (1975). On the possible mechanism of carcino genic action of vinyl chloride. 246:258-267. Ann NY Acad Sci Whittcmore AS, Keller JB (1978). Quantitative theories of carcinogenesis. Society for Industrial and Applied Mathe matics Reviev 20:1-30. Wong 0 (1981). An epidemiologic study of workers potenti ally exposed to brominated chemicals: with a discussion of a multifactor adjustment. In: Quantification of Occupa tional Cancer (Eds. Peto R, Schnciderman H). Banbury Report 9 Cold Spring Harbor Laboratory pp. 359-378. 132322