Document G6O1QNogNzQ15kELLkGerdq1r

fj CW. r*ic Vol. 2i. No. 2. pp. I4T-20Z \W fnnud io Gre*t Bnuin. Ail n*bu rc"Tvai 027S-4? 13/8 7 S3 00 f C OO Copyright < 19*7 ^tTfixnon 3on/a*U Led Review Section VINYL CHLORIDE: AN ASSESSMENT OF THE RISK ' OF OCCUPATIONAL EXPOSURE* l. F. H. Purchase Central Toxicology Laboratory J. Stafford Plastici and Petrochemicals Division and G. M. Paddle Central Medical Group, Imperial Chemical Industries pic, Alderley Park, Macclesfield, Cheshire, England {Received 14 December 1983; revisions received 13 January 1986) Introduction Vinyl chloride monomer (VCM), more properly named monochiorethane, is a'colourless gas normally handled under pressure as a liquid which boils at -- 14C at normal pressure. Discovered around 1835, VCM's commercialization did not begin until the 1930s and did not reach high volume until after 1945. Present manufacture is around 12 x 10* tonnes per annum, nearly all of which is used to make the polymer polyvinyl chloride (PVC). Until the 1960s, VCM was regarded as a material of low human toxicity and the main concerns were related to the compound's narcotic effect. Indeed there are many reports of employees exposed to VCM monomer in polymer plants becoming dizzy and unconscious. Because VCM was considered to be relatively innocuous, it had a threshold limit value (TLV) of 500 ppm, 8-hr time-weighted average (TWA) for many years (ACGIH, 1974; Lester et at. 1963; Torkelson tt at. t96l). Measurements of em ployee exposure were infrequent, since most mea surement and warning systems were designed to ensure that plant atmospheres were beyond the ex plosive limits, fire and explosion being the main hazards of VCM. Retrospective estimates (Bames, 1976) of typical TWA personal exposures (in ppm) for polymerization workers have been cited as: 1000 in 1945-1955. 400-500 in 1955-1960. 300-400 in 1960-1970. 150 in raid-1973 and 5 in 1975. However in some jobs, particularly in the cleaning of the autoclaves in which VCM is polymerized to PYC, very much higher exposures, in thousands of ppm, were undoubtedly experienced for short/medium pc- A tongcr version of this paper has been published in Toxicotogieai Risk Assessment, edited by D. B. Clayson, D. Krewslci and I. Munro and published by CRC Press. Inc.. Boca Raton. FL (1985). Abbreviations: AOL - acro-osteoiysis; ASL - angio sarcoma of the liver; PVC - polyvinyl chloride; TLV threshold limit value; TWA - time-weighted average; VCM - vinyl chloride monomer. riods, since in some plants operators became faint and unconscious from time to time. ' The first clear indication of chronic health prob lems associated with VCM arose in the 1960s in men who entered VCM polymerization autoclaves to re move build-up of polymer from the walls. Some of these men developed acro-osteolysis (AOL; Cook et at. 1971; Harris Sc Adami, 1967; Suciu et at. 1963). Modification of working practices led to a reduction in the incidence of AOL cases in autoclave cleaners. Although AOL is occasionally seen in people not exposed to VCM (Meyerson St Meier, 1972; Wilson et at. 1967) it is a rare disease. In the late L960s, studies in rats involving exposure to high concen trations of VCM for long periods (Viola, 1969) failed to produce AOL but showed an increase in the incidence of tumours at various sites. Further studies (Maltoni et at. 1980 & 1981; Maltoni & Rondinclla, 1980) showed the rare tumour angio sarcoma of the liver (ASL) in exposed rats, and confirmed VCM as an animal carcinogen. Three ASL cases in employees at a PVC polymerization plant (Creech & Johnson, 1974) confirmed VCM as a human carcinogen. Other known aetiological agents for ASL in man were thorium dioxide, arsenic and, possibly, anabolic steroids (Maltoni et at. 1980). Since 1974. the health hazards of VCM have been the subject of many investigations, scientific papers, seminars and other presentations (Conference to Reevaluate the Toxicity of Vinyl Chloride Monomer, Potyfvinyl Chloride) and Structural Analogs, 1981; Gauvain, 1976; IARC Working Group, 1979; Selikoff, 1975; Szadkowski & Lchnert, 1982; US DHEW, 1980). The plethora of information (and misinformation) now available suggests that an ob jective historical case study of VCM would be of value. Experimental and human data Experimental studies The principal effect seen in the acute and subacute studies is anaesthesia, which occurs at relatively high 187 AP00055821 I8& 1. F. H. Purchase et al. Tible 1. Lownl MDcentralions or doiei at which a uprtficani exccu or virioui tumour type-i w*j ofcnervad in rt carcinogcnuiy nudie*____________ Tumour Concn (ppm) Dc*e (nuAll Foreslomactl papilloma Zf-mbtl-jUnd carcinoma NeuroblaHoma Ncphfoblisiomi Liver injioiireoma Mammary-gUnd adenocarcinoma 30.000 10.000 10.000 250 (ferrule) too (male) 200 50 S (ferrule 1 SO (mile) 16.65 (female) Dau from Maiioni et at. (1981). doses (7-10%) in both animals and man. The doses responsible for acute toxicity are about 1000-fold higher than the minimum dose for carcinogenicity and there is frequently no sign of overt organ toxicity prior to the development of the carcinogenic re sponse. VCM is mutagenic in a variety of test systems including Salmonella typhimurium (Rannug et al. 1976). Saccharomyces (Loprieno et al. 1977) and Drosophila (Vcrburgt &. Vogel, 1977), usually with some form of mammalian microsomal metabolizing system to convert VCM into its active metabolites, chloroethylene oxide and chloroacetaldehyde. The data on the mutagenicity of VCM provide useful qualitative information on its mode of action and metabolism, but are not suitable for the quantitative estimation of risk to man. The most useful experimental data are derived from long-term animal carcinogenicity studies. An extensive series of 17 studies (Maitoni et al. 1981) gives a useful database for risk assessment. Other studies (Feron et al. 1981; Lee et al. 1978) tend to confirm the findings of Maitoni. Carcinogenic effects were observed in mice, rats, and hamsters. A complication in the selection of these data for risk assessment is the variety of tumour types observed (Table ]}. Some of these occurred at very high exposure levels, but mammary adenocarcinoma in females and ASL in both sexes of both rats and mice occurred at 50 ppm or less, exposures simitar to those believed to have occurred on manufacturing plants (Barnes, 1976). Epidemiological studies Several major epidemiological studies on workers exposed to VCM have been reported (Table 2). The main organs that have been associated with higher incidences of cancer in workers exposed to VCM arc the liver, lung and brain. Increases in the standard ized mortality ratios of cancers in the buccal cavity and pharynx, of lymphomas and of cancers of the lymphatic and cardiovascular systems have been re ported in one or two studies. The analysis of cancer of the respiratory system is often confounded by smoking, making quantitative analysis of the con tribution of VCM difficult. The excess of liver cancers is due to an excess of ASL in many of the studies. An analysis of the statistical power of various studies for association between VCM exposure and cancer of the lung, liver and brain (Beaumont & Breslow, 1981) concluded that the results for liver were consisient with an aetiological role for VCM. For brain cancer, where three out offive studies had statistically significant findings, the results were more variable, positive findings occurring in the studies with the greatest statistical power. The most reason able interpretation was that the data were consistent with a causal association between VCM exposure and an excess of brain cancer. Infante (1981), in reaching the same conclusion, points out chat the relative risk for brain cancer is much lower than that for liver cancer. Only two out of eight studies on lung cancer (Beaumont & Breslow, 1981) yielded statistically significant results and, because studies with a high power were negative, a causal association was consid ered unlikely. ASL is the most suitable endpoint for analysis of the risk of exposure to VCM for a number of reasons, it is a rare cancer in unexposed populations, making attribution to VCM exposure on the basis of work history a reasonable approach. ASL occurs in both animals and humans exposed to VCM and it is unlikely that any other carcinogenic effect of VCM will be found to occur at lower exposures than the lowest exposures that induce ASL. For these reasons, most work on the quantitative risk assessment of chronic exposure to VCM has used ASL as the endpoint to study. Cose register The availability of data from a comprehensive case register of ASL cases with a history of occupational exposure to VCM provides an opportunity to identify risk factors for the induction of ASL. Persons potentially exposed to vinyl chloride Current manufacture and use of VCM and PVC results in the potential exposure of four groups of the population. The highest exposure category covers the workers involved in the manufacture of VCM, its polymerization to PVC and certain other industrial uses of VCM. Within this group, certain occupations, particularly autoclave cleaning, involve higher poten tial exposure than others, although all groups would now be expected to have exposures complying with hygiene standards of 1-5 ppm. The next category covers those exposed as a result of using the PVC. Workers in the compounding and fabrication of PVC products are exposed to residual VCM released from PVC on heating (but PVC does not decompose to VCM when heated). In general the exposure levels for these workers are very low in comparison to those for PVC polymerization workers (from 10 to 100 times lower). Consumers who eat food and drink beverages that have been packed in PVC may ingest unreacted VCM AP00055822 Vinyl chlondc--risic ajjcjjmcni 189 which has migrated into the food or beverage. Since 1974. the amount of VCM in PVC has been reduced to less than 1 mg/kg with the result that the maximum human daily intake of VCM in food and drink is 0.1 *jg/day (Ministry of Agriculture, Fisheries &. Food, 1978). The fourth group with potential exposure to VCM are those who live in the vicinity of VCM or PVC manufacturing or fabricating factories. The levels in ambient air around a factory are very low (in the parts per 10' range) but much larger population groups, which include all age groups, are involved. For the workers in VCM manufacture and PVC polymerization and fabrication, the route ofexposure is by inhalation. Much of the animal carcinogenicity data are based on inhalation exposure and the human epidemiology is predominantly of populations ex posed occupationally by inhalation. Thus an assess* meni of the risk factors and the quantitative risk of inhalation exposure is the main objective. For the consumer exposed to VCM via food and beverages the route is by ingestion. Relatively few experimental studies have used oral administration and only one study used a comparable exposure pattern (Feron et o/. 1981). Similarly there are no specific epi demiological data on ora) ingestion. Risk assessment for exposure via the oral route must rely on the existing animal data and on extrapolation from epi demiological and experimental studies of inhalation exposure. After administration by gavage or inhalation, part of the dose is exhaled unchanged and the remainder is excreted or retained in the carcass. A general scheme ciy I c =o I CHCHjSfHjCHj , NH OH ,JF I V=O I CHCHjSCHjCOjH NH Risk assessment'from experimental animal data Clu Clu Assumptions In carrying out a risk assessment on the basis of animal data, a number of assumptions have to be made. The first of these relates to the overall dosi metry. Experimental animals are exposed to concen trations of vinyl chloride or dosed with amounts of vinyl chloride that allow an estimate of the amount to which they have been exposed. It is possible to calculate a correction factor for these quantities so that they are applicable to man. However, rats and mice live for relatively short periods of time (up to 2 years) during which they develop cancers of a type similar to those seen in man. The latent period for the same tumours in man may be between 20 and 40 years. It is therefore assumed that the lifetime of man is equivalent to the lifetime of an experimental animal species even though the chronological time is sub stantially different. Strictly speaking, mathematical extrapolation of risk on the basis of experimental animal data pro vides an estimate of the risk at low doses to the experimental animal under consideration. A variety of factors, particularly inherent biological sus ceptibility and differences in metabolism, render the extrapolation of the data from animals directly to man subject to numerous errors. It is at this point that scientificjudgement is required to decide whether these data are applicable to the human situation. Metabolism In rats, VCM has been shown to be metabolized extensively, producing a range of excretion products. COjH CHCHjSCHjCHj HH(Ac) OH () COjH CHCHjSCHjCOjH NHj tr> COjH I C------ CHjSCHjCOjH II O S(CH^CO,H), <) Fig. [. Scheme showing the metabolism of vinyl chloride monomer (VCM) in rats to -containing metabolites. VCM (a) ii converted to chloroethylene oxide (b) which is trans formed spontaneously to ehloroacetakiehyde (c). These two metabolites are mutagenic and hence are considered to be the proximate carcinogens. The urinary excretion products A/-acetyl-J-(2-hydroxyeihyl)cysteine (e) S-(carboxyraethyl)* cysteine (f) end thiodtgiyeollic acid (g) are derived from these mutagenic metabolites via (d). Gly and Glu are the glycine and glutamate residues of glutathione. (After Green A Hathway (1977)). i!. F. H, P 'S-CHass ei a!. AP00055824 Reference Monson tl a/. (1974) Tebcrshaw A Gifley (1974) Duck rt at. (1975) Nicholson tl el. (1975) Oit tt at. (1975) Byrcn <1 at. (1976) ORG (1976) Rein) A Weber, 1976; Reirtl tt at. I97t, Weber rt of. 1911 WjLxwcilcr it ol. (1976) Foa A Collier (1977) FrcUel-Beymc tt at. (t97l) Bertara tl at. (1979) Bufller tl at. (1979) Chime A Ferenoe (1981) Chime tt at. (1910) Beaumont A Bralow (1911) Tabic 2. Epidemiological studies of cancer associated with exposure lo vinyl chloride monomer nC. Ifi Mtiuj4 -----(/. follow up) Increase 7 8J|4 (85%) Brain Lung Liver, including ASL Buccal cavity and pharynx Respiratory system Unknown xtu Lymphoma Angiosarcoma 2120 257(99%) 594 (99%) 771 (97%) 10.173 (95%) 11,028 (90%) 1(51 7409(99%) None ASL All tumours? Uvcr/pancrcai Cerebrar? Cardiovascular Digestive tract Malignant liver Lymphatic system Cl tract Brain Respiratory tract Lymphatic system ASL Primary liver ASL 1618(95%) 5441 (16%) 464(100%) JW7 Colon/stomach Proslalic hyperplasia All tumours Respiratory system Digestive system liver Brain Site* (or luffloun) with change* in SMR - ---- * i -- No increase i Comment* Genital Di|eative organs Urinary tract Leukaemia Brain Stomach Brain Lymphatic and hacmopoiclic system Significant SMR not significant but increases with exposure and time Some criliciim of conduct of study Arsenical* involved Significant increase (I ASL) Increase not significant PMR study Related lo duration of exposure Mixed exposure, not VCM related Not significant Significant Breast Respiratory tract PMR study or female and male fabricators Increase in PMR not confirmed by case-controlled study Review of nine studies S Vi'nyi chloride--risk assessment 191 nin i x 2 a. of VCM metabolism in rats is given in Fig. I. On the basis of this scheme, the highly reactive intermediates in the metabolic process (particularly chloroethylene oxide) react with cellular macromolecules, including DNA to produce the critical lesions leading so mu tation or the induction of cancer. Studies on the quantitative aspect of VCM metab olism have shown that there is a dose dependency in the rale of metabolism. After administration of ,4C-labelled VCM by gavage at doses between 0.5 and 100 mg/kg to Wistar rats, the amount of '*C excreted in the urine and faeces and retained in the carcass was estimated over 72 hours (Watanabe Sl Gehring. 1976). As the dose of VCM was increased, the proportion exhaled increased and that excreted in the urine and faeces decreased (Fig. 2). The proportion retained in the carcass also decreased. The same general trend occurred after administration by in halation, although the magnitude of the differences in retention and excretion was less (Watanabe Sc Gehring. 1976). Studies of the amount of non-volaiile material retained in the carcasses of rats exposed to various levels of "C-labelled VCM for 6 hours demonstrated that the metabolism of VCM appeared to be in accordance with Micbaelis-Mentcn kinetics (Gehring el at. 1978). The constants for maximum velocity of metaboiism (Vm in /ig metabolized/6 br) and the Miehaelis constant (K.,, in fi% VCM/litre air) accord ing to the formula: K ,,FS (where V * velocity of metabolism in pg/6hr and i S = concentration of VCM being inhaled) were Vm 8558^g metabolized/6 hr and Ka 860/ig i i p II 1 I * ? VCM/litre air. Thus there was a considerable change in the ratio of administered dose to metabolized dose as the exposure concentration increased (Table 3). At the higher doses a smaller proportion of VCM was metabolized than at low doses. } Review of earlier calculations of risk There have been a number of attempts to calculate L the risk, of ASL development on the basts of extrap olation from experimental data. These have been reviewed by Barr (1982) and an adaptation of his H data is presented in Table 4. The introduction of biotransformalion data into the estimation of risk increased the level of exposure calculated to cause a 10'* lifetime risk, from parts per billion to in excess of one part per million. A further refinement of the technique using DMA binding as 9 the measure of dosimetry (Anderson el al. 1980) provided a similar estimate of the exposure. 1 A variety of mathematical models can be used for g extrapolating below the experimental dose range, and it is not possible to select from amongst these math ematical models on the basis of goodness of fit to experimental data. Attempts to do so have shown that most of the models fit the data equally well (Gehring tt al. 1979). It is equally difficult to select amongst the models on the basis of the assumed mechanism of action of VCM. Thus a comparison of the lifetime risks calculated using the Armiiage-Doll AP00055825 192 I. F. H. Purchase et al. Table 3. Vinyl chloride dote and incidence of hepatic angiojareoma in Sprague-Dawley rati ei posed on 5 dayi/wk for 52 wk* Conen (ppm> Amount meiiboliied u%J4 hr (tout) Anposareoma incidence (V>) Mile Female Mean Eipmt no. 30.000 10.000 6000 2500 500 250 200 150 ICO 50 25 10 5 1 0 5647 5521 540) 5030 3413 2435 2129 1761 1309 739 395 169 84 17 0 1.47 x 10` 16.6 43.3 t.44 x to* 10.0 13.3 1.41 x 10* IJ x 10* 10.3 20.0 33.3 2J.3 8.8 x 10' 0 20.0 6.3 x 10' 3.4 6.7 5.5 x to* tt.7 8.3 4.6 x 10* t.7 8.3 3.4 x 10* 0 1.7 1.9 * 10* l.l 7.2 1.0 x I01 4.4 x 10* 1.7 0 6.7 1.7 2.2 x 10* D 0 4.4 x 10' 0 0 0 00 30.0 tt.7 22.0 21.7 10.0 5.1 too 5.0 0.1 4.2 4.2 0.8 0 0 0 BT 67 Btl BT 1 BT l BT 1 BT 1 BT2 BT 2 BT 2 BRt.9 BT IS BriJ BT 15 BT 15 BT1.2, 9, 15 `After Malioni ft tl. (1981). t Experiment BT 6 ended after only 65 wk, while the rest were all approximately HO wk: therefore the percentage of tumours in BT 6 is probably low relative to the rest because of the short latency period available. multistage model by the Food Safety Council (1980) and by Gaylor & kodell (1980) showed that for the same 10"* lifetime risk, the Food Safety Council estimated the dose as 2 x 10'J ppm whereas Gaylor ie. Kodell estimated the dose as 5 x I0"`ppm. The difference between these two estimates was due to alternative assumptions on the value of the expansion of the exponential term used. In general, calculations based on the amount of material metabolized or on human data have pro duced exposure values of about l ppm for a 10 lifetime risk. All the other studies have produced exposure values in the ppb range. A large variable appears to be the selection of the mathematical model applied to the experimental data. In the following section two models are used to calculate the exposure for a 10"* risk from a variety of experimental animal data applying the correction for metabolism used by Gehring er at. (1979). Calculation of exposure for 10 risk A summary of the crude ASL incidence rates for inhalation studies in Spraguc-Dawley rats is given in Table 3. Similar data for Wistar rats exposed by Table 4. Summary of quantitative rik assessments for vioyl chloride monomer' Reference Species Exposure for 10"* lifetime risk (ppbf) Comment! Schoetdcrmaa ti of. (197$) Rat Kuanaelc 4k McCaughy (1975) Gehring tt al. (1979) Food Safety Council (1980) Rat, man Rai. man Rat Rat Andcnou tt al. (1980) Gaylor H Kodell (1980) CarJborg (1981) Barr (1982) Thii paper (Table 9) EPA (1980) NAS (1980) Crump A Gueu (1980) After Birr (1911). fExccpl where stated otherwise. Rat, mao Rat Rat Man Rat Mouse Man Rat Mouse Man Rat Rat Man Rat By WaUcioa 73 119 2 14 140-1400 >1000 <10->1000 20 20 2.1 x 10"' 3.9 x 10"' > 1000 0.7 0.5 2.5 x tO-' >100 9.025-9.Ifi\ 2 x tO"1* / 0.63-90 2 x IO*'-2 x 10"* 6 x I0"*` 0.067-4.14 By lagtttJea 4>g/day 3 x 10** mg/fcg/day 0.7pg/diy 0.5jrgfday Probit (slop* -1, Mantel) ' Logit (slop* 3.45) Lopl (slope 2.3. one-hit) Linear through.zero Log-prohit Biotraniformilion data ioeluded Linear or log-probit Depends on mathematical model used One-hit A/miiage-Doll Wcibull Multi-hit DNA binding used for doilmetry Upper 97.5% confidence limit of linear model Armittge-Doll Wcibull Derived from Barr*i negative epidemiology Log-probit Log-probit including bioirimfonnauen data for man Wetbull Weibull including bioiransformatioit for man Food or water Water Applying worker data to water Upper 93% confidence limits Vinyl chloride---riilc assessment 193 inhalation (Table ji) Tor rats exposed orally (Table 6) and for mice exposed by inhalation (Table 7) arc also presented. Data from experiments with various ex posure periods of short duration are given in Table 8. For calculating the amounts of the dose metabo lized in rats in the inhalation experiments, the con stants calculated (Gehring et al. 1978) have been applied. For Wistar rats, the K* and Vm values derived for Sprague-Dawley rats have been used. These estimates of metabolized dose have been in cluded in the tables. For the experiment in which VCM was given by gavage, the data from Fig. 2 were used to estimate the amount of VCM exhaled unchanged. As the t|^ for exhalation of VCM was 14 minutes, these data based on a 72-hour period give a good estimate of the fraction of VCM exhaled in the 24 hours between doses. It has been assumed that the VCM not exhaled was metabolized, an assumption similar to the one used for estimating metabolized dose in the in halation experiments. Green & Hathway (1975 & 1977) showed that VCM administered by gavage to Wistar rats was exhaled and metabolized in a similar manner to that in the Sprague-Dawley rats, and the VB and K.m values derived for Sprague-Dawley rats have been used. In the experiments by Feron et al. (19R)), who used Wistar rats, the same assumptions about V,, and Km have been made. The quantity of VCM administered has been dealt with as if it had been administered by gavage. Fig. 2, Summary ofdose-dependent urinary and pulmonary excretion of vinyl chloride monomer (VCM). Urinary excre tion (#) represents metabolites of VCM, while pulmonary elimination (A) is unchanged VCM. [After Watanabe & Gehring (1976)}. For mice, the data have been combined in Table 7. The estimation of the dose metabolized in mice has been calculated using values for V* that have been adjusted on the basis that, for a chemical requiring metabolism to its active form, the quantity metabo lized will be proportional to the body surface area and must be expressed in terms of metabolized dose/kg body mass. This technique has also been used by Gehring et al. (1978) for estimating the dose metabolized by man. Table 5. Vinyl chloride dose and incidence of hepatic angiosarcoma in male Wisur nu exposed on $ days/wk for 52 wk Cancn (ppm) Amount metabolized Angiosarcoma P/ hr Mf (tout!) incidence (V,) Expmt no. 10.000 6000 2300 300 250 50 1 0 3521 5403 5030 MU 2435 739 n 0 t.4 * 10* 1.4 x I0` 1.3 K 10` S.S x 10' 6.J x 10* 1.9 x 10' 4.4 x 10' 0 29.6 11.5 12.0 10.7 3.7 0 0 0 BT7 BT7 BT7 BT7 BT7 BT7 BT 17 BT7, 17 Table 6. Vinyl i:htoride (VCM) dote end incidence of hepatic angiosarcoma in rats given VCM by gavage or ingestion Dose - * (mg/kg) 50| 16.65 3.33 1.0 0.3 0.0) 0 3001 14.|f 5.0 1.7 0 Amount (% of dose) SO 3S 10 2 1.7 t.4 -- 80 32 16.3 2 69 Amount metabolized pg/doset 6250 2703 750 3245 74 7.4 0 13.000 2390 (040 420 0 HI (total) 1.6 x 10* 7,0 x 10J 2.0 x 10' 7.26 * tO* 2-16 x 10* 2.16 x 10' 0 6.2 x 10* J.6J x |0* 7.25 x 10* 19 x JO1 0 Angiosarcoma incidence (*/) Male 20 10 0 1.3 0 0 0 49 49 to 0 0 Female 22.3 15.1 0 2.7 1.4 0 0 53 16 4 0 0 Mean 21.2 12.5 0 3.0 0.7 0 0 51 32 7 0 0 Expmt no. BT 11 BT II BT 11 BT27J BT 17 BT 27 ST 11, 27 Feron at at (1910 'Calculated from data derived from Wattoabe A Gehring (1976) pieacntcd is Fig. 2. tAuujning a 250-g rat. tSpraguc-Oawli-y rats doted by gavage with VCM in com oil 5 times/wk Tor 32 wk. {BT77 doud for $9wk. IWitter rats used a control! by Feron tr et. (1911) and doted for 83 wit, IWitiar mi receiving diet containing VCM diitolved Js PVC. AP00055827 194 l. F. H. PURCHASE et al. Table 7. Vinyl chloride dose and incidence of hepatic angiosarcoma in mice Conen (ppm) Amount meubclued Vll* hr fig (loul) Angiosarcoma incidence (*/.) Male Female Mein no. 10.000 6000 2500 1000 500 250 250 30 1 0 11.245 11,007 10.246 *699 6952 4959 4959 1506 1506 0 1.7 * 10* t 7 * 10* 1.5 x 10* 3.4 x 10* 1 0 x 10* 7.4 x 10* 7.4 x 10* 2.2 x 10' $.9 x |0J C 3.8 6.7 20.7 39.4 20.(1 30.Q 24.0 3.3 10.3 0 30 36.7 J3.3 30.0 26.7 30.0 47.0 0 0 0 I7.a 21.7 27.1 44.7 23.3 30.0 36.5 1.7 5.2 0 BT4* BT 4 BT4 Lee tt al.t BT 4 BT4 Lee et al.T BT 4 Lee et al. t BT 4 A Lee tt al. Swiss mice. ll>tvk experiment. dosed for 30 wk. tCD| mice, J2-wk eaperimem. 6 hr/day exposure (Lee et al. 1978). These results have not been included in (he calculations for Table 9 because the experimental design incorporated interim kills. Thus: 0 011 m2 -5706,,g/4 hr x^il - 1395 fig/4 hr The values of 0.045 mJ and 0.011 m* are Ihe body surface area of a rat and a mouse, respectively. Since toxicity is a function of the concentration of the toxic metabolite in the tissue, the amount transformed must be normalized for mass to estimate an equiv alent response. Thus Vm must be adjusted on the basis of the body weights of a rat (0.2S kg) and a mouse (0.03 kg) by dividing by 0.03/0.25 = 0.12. The Vm for the mouse on a mass-equivalent basis is therefore: 1395 11625pg/4hr This value of Vm has been used in calculating the total amount of VCM metabolized (Table 7). From the variety of models (or mathematical ex trapolation techniques) used for low-dose risk extra polation (Table 4), an arbitrary choice of models has been made to test the robustness of the extrapolation from the different animal studies. A log-probit analysis of the dose that would be expected to produce a lifetime risk of ASL of 10'`is presented in Table 9. This calculation can be carried out on the basis of (he concentration inhaled, the daily dose metabolized or the total quantity metabo lized during the whole experiment. There is a wide variation in the estimated dose depending on the database used for the calculation. The largest vari ation between doses derived from the rat experiments is 360-fold (0.025 ppb v. 9.1 ppb) when exposure in ppb is considered, but this decreases to 100-fold for other estimates of dose. The results from mice are substantially lower when expressed in ppb (2 x 10_l!ppb) but the difference is less for other expressions of dose. Similar calculations of the dose expected to give a 10"` lifetime risk of ASL have been based on a Weibull analysis (Table 9). This is a more `conserv ative' mathematical model and the estimates of dose are accordingly lower. The variation in estimates of dose is, if anything, larger than that observed with the !og-probii analysis (for example, a 10"J difference between the. S values derived from Wistar and Sprague-Dawley rats). The doses for mice are so much lower than those calculated for rats or man that the assumptions used in their calculation must be suspect. A further calculation to derive the human dose likely to produce a risk of 10_` is given in Table 9 (S calculated for man). These calculations are based on aV, for man of 1675 gig/8 hr based on corrections for body surface area and mass. The values are substan- Table 8. Vinyl chloride (VCM) dote and hepatic anposarcoma incidence in Sprajue-Dawley rati exposed to VCM by inhalation Conen (ppm) Schedulef dose* Amount metabolized^ U(/4hr pg (total) Angiosarcoma incidence (V*) * Expat Male Female Mean no. 10.000 I 260 5521 1.4 x 10* 10 13.3 U.7 BT 1 10.000 II 85 5521 4.7 x10s 0 0 0 BT3 10.000 111 25 5521 1.4 x 10s 1.7 0 0.1 BT 10 10.000 IV 100 IJ79 1.4 x |0> 1.7 0 0.8 BT 10 10.000 V 25 5521 1.4* 10' 0 1.7 0.8 BT 10 6000 1 260 5403 1.4 x 10* lOJ 33.3 22.0 BT 1 6000 It 85 5403 4.6* 10' 0 3.3 1.7 BT3 6000 III 25 5403 1.4 x 10' 0 0 0 BT 10 6000 IV 100 1350 1.4 x 10* 3.4 1.7 2.5 BT 10 6000 V 25 5403 1.4 k 10j 0 1.7 0.8 BT 10 After Maltoai tt ol. (I9SI). tSchedule* 1--4 hr/day. S diyt/wk (or 52 wk; 11--4 hr/day, S dayi/wk for 17 wk; HI--4 hr/day. 5 diyi/wk for 5 wfc tv-- l hr/day, 4 dayi/wk for 25 wlc V--4 hr/day, | day/wk Tor 25 wk. tAifiouai metaboliaed (v) in 4 hour derived from the formula: V O^l/bi) x S/K. 4- S where Va ii 4ft of the A hr value. Vinyl c ilo rid e -- risk u s c u m s r.t AP00055829 O' 6000 t -01 x 1 HI iWO X|tM XB <p*isl poulquicn nei a~S PU< JUS'Ml pooqtpqil uinuHKFui Sukh paifiuiifg) jil g/3rt {^91 si uetu ioj *a `A - 5Z91/D9I A*S :*l"WJoj M|| fuMTl ([ UIUX|M Ul) A UCJJ pO||H3]T3 UtllodljJ 'JSIWLttqtO poms sdaoia. /3|m((3-3nltJd5 -- Q-s JMI| tf> JO tUIOSJCTOlluV " 1SV 610 two , 01 * l 1000 000 tOOOO 19 C ztwo .01 x C 11.100 , .01 *z Mop/Iui COO'O op/tui(C0 wop/SP.-OI *C ZJl Cl 00 -- ' (.01 x 9 <..01 " 9 */*" ,-01 t *V,U`-01 "9 *sj/uiw_0l * 6 <-oi *: ,.CI " z ucHKitwur uiii*uou$ 'tiej a-< UOjlC|C1|UI *p?uiquJ03 t|El Q-$ pt* JtltlM U0l|l|vyui {S`09-- a-s-- ir)Si/A--uoiioloi *tir}| uonqequi 'Xjuo qiui 'titi iC|tAV uont|eqii| *tiu a-s 0 5> n 9 S -- ZLO too -- 06 90 9ft 5(0 9000 IB'O ZO III Ct toco WOO IM 090 3Mp/*UI0fl asop/Iui 4| j Ttop/tus 69 0 651 -- IfO'O ,1-01 * z !/**,-01 *9 V*w,-0l *6 ,.o| x ( 916 CM intd lq Z0'0 uoiici*siui uu>i-uoui *tiu n-c uourjrtju) `pxiiquico i*J a-S pu* J*l!M uaiiqiqui "ww JMUXk-- a-s-- "on0*0! '*i*n wi*iequt `/(uo qrut "(in s*n>M *t)*i Q-s 1 5> It 9 5 * JO| p3|ejfl3(tt s| A way psiqnqca (qiid) 9jfisod>3 (Sui WjJ Siuapoi Xq p3i;ioqe|>ui lunouic |t]OX (jq9/1rf a) tivspai q jq 9 oi pnrtoqnsuj ivnouy Uqdd s) Sltlspoj IOJ unsodia ci*p jnu3(uuadi3 %ov tq*j. U>;inquiip unqij^ * jo (KA|ut iiqojd-ftq jo (f*q aqi uo J(>P t-01 JO 1SV J 9U,!PI!I * *!* ol paiqnqo spuojqa |kuu jo uojitJiiiXHJcx> jo lunoiuc 3i|i sc pmsjdo pu* np Xipiualoupjca jcuiiuc qqqttAC way pMUp suojituiitsa :q*u *Mj*i>iu*n{) 'g qq*x ' 196 I. F. H. Purchase et al. tially higher than Lhose calculated for the rat and mouse and there is still a range of over 100-fold in the estimates derived from the different rodent experi ments. When this amount of variability occurs in the extrapolation of the risk of low-dose exposure to VCM based solely on different experiments in the same species, the reliability and hence the utility of these procedures is open to question. The general relationship between the dose adminis tered and the incidence of angiosarcomas derived from 52-week exposure docs not apply to exposures of shorter duration (Table &). In all experiments a total metabolized dose in excess of 5 x IOs/rg was required to produce an incidence of angiosarcoma in excess of 1--2%. This relationship was seen in both rats and mice and in experiments in which VCM was administered by gavage or by inhalation. In long term inhalation studies, a total metabolized dose of 5x 10' pg is equivalent to about 200 ppm adminis tered over 52 weeks and represents a practical thresh old for this series of experiments. In conclusion there is a wide variation in the estimates of dose for a 10"` lifetime risk. This vari ation is due to the type of mathematical model that is applied, to the assumptions that are made and to the particular experiment that is used to provide data for the extrapolation. A high level of confidence cannot be placed on low-dose extrapolations when variables that would not be expected to alter the expression of risk have a profound effect on the estimated risk. In addition, the interspecies extrapo lation from experimental animals to man is largely intuitive. It is dear that estimates of risk should take into account all available data, including epi demiology, to provide a degree of reliability. Risk assessment from human studies Register of ASL eases Since 1974, lists of reported ASL cases attributable to VCM exposure in the VCM/PVC industry have been kept by NIOSH (Spirtaj Sl Kaminski, 1978), by IAR.C and by the VCM Committee of the Association of Plastics Manufacturers in Europe (APME). Details of 99 cases in the APME register at Table 11. Clustering of ASL tua in individual PVC pUnu Plane* no. i 2 3 a 1 2 3 1 2 i Country Woura Evnpe Weil Germany West Germany Wen Germany Wen Germany France France France UK UK Sweden North America 1 Canada 1 USA 2 U5A 3 USA Rot ef World 1 Japan t YugoiUvia 1 CaectioilovaJua No. of ASL caeca . Total... 10 4 2 2 3 i 2 i l 5 42 Total.. 10 n 9 4 34 2 4 2 Total... 1 'For the purposes of thii tau study, it is sot necctttxy to identify the precise ownership and location of these plants. the end of 1982 have been analysed by country and by manufacturing company and plant. The cases have been recorded from all major VCM/PVC manu facturing countries (Table 10). but the incidence has not necessarily been in proportion to the PVC pro duction capacity now or prior to 1962. In the absence of data on the number of workers employed, pro duction capacity is the only available indication of the numbers of people potentially exposed. The majority of the ASL cases are PVC autoclave cleaners or men who have worked in or around autoclaves. There are ASL cases among men who manufactured VCM and a few cases were involved both with monomer and with polymer production. Only one case suffered from both acro-osteolysis and ASL. The ASL cases tended to occur in larger numbers in some plants than in others (Table 1 i). Of the total of 39 ASL cases recorded in North America, 34 have occurred at four PVC plants, while over 40 Table 10. Distribution of ASL cue* by country PVC production nameplate capacity (kilotonna/yr) Country ASL caw 1952 1W2 USA 29 193 704 Wol Germany 2! 22 260 France Canada 14 11 176 to 5 22 UK 7 27 177 Swedes 5 3 20 Yugoslavia 4 3 1 Italy I 9 212 Caechoilovaii* 2 1 23 Japan 2 12 Jt4 Beliium Norway l 3 25 t 2 20 Total,,, 99 Wettern Europe 52 12 951 North America 39 191 726 Real of World a 31 709 Total... 99 331 2316 ASL - Angsotafcoma of the liver 1972 2090 ItJJ 627 as 302 103 60 77S 4$ 1699 195 65 3950 2I7S 3334 9462 Vinyl chloride--fills assessment Tible II ASL cut number* by year of death and leogriphical location (excluding rTOI*) Yi:*r t>l dcith Wiem Europe ASL csxetf in: North Amnio Reel of world publication! 1955 Cl 6 _ 7 C2 8 9 I960 l uss 2 C3 3 4 USJ Ci2 6 7 a 9 1970 I 2 3 4 5 ii 7 1 ? 1980 1 FI Gl Swl G2 Nl. Sw2, UKI. Ii2 GH G4, GJ. UK) F2, F3, G6. G7, 08. Il3 Bl. F4. FJ. F6. F7. Sw3 FI. F9, GiO. Gil. G12. Sw4 FIO. Fll, 09. 013, 013. CI6, 017 FI2. F13. UK4, UKJ. G18 UK6. UK7, G19. SwS, G20, G2I U4, FI 4. UK*. G22 C4. CJ, US4. US7. USIO US12, US 16 USI i C6. US2 C7 C*. USI, USJ. US23 C9, US13 US6, US9, USI8, US26 US19, US20. US22 CIO. US2I. US245 US27, US2I USI?. US29, US30. USJ2 Yl. Y2, Cxi Japl Jap2. Y3 Y4 Viol* Mxltoni Creed* & Jotinjon Total.. 52 3*1 1 ASL m Anposareoma of (he tivar luluncueOI wti not a typical ASL; hit primary tumour war probably of the pericardium. Thu mao wai icngaged in extrusion of PVC sacks. fS &elpum, G -- W. Germany; Sw --Sweden; C -- Canada; It -- Italy; UK United Kingdom; Cx-- CzedioiiaviUa; Jap - Japan; Y - Yugoslavia; F - France; N > Norway; US - USA. Thus G9 case mo. 9 in West Germany. Cases UK2, GI4. USI4, USI5 and US25 were shown not to be associated with VCM exposure and hence withdrawn from the list. {Aerosol can filler. fChoUnpourcoma. IDoes not include US31 (still alive). 197 North American PVC plants have not recorded an ASL case so far. The average latent period between starting work in an occupation involving VCM exposure and death from ASL for the 99 cases is 21,9 years (in France, Sweden and the USA between 24 and 25 years, in Germany about IS years). It is still too early to predict whether the annual number of ASL cases amongst VCM workers has reached a peak. ASL cases appeared earlier in North America than in Western Europe and while the occurrence is tending to decrease in North America (Table 12), it is still high in Western Europe. On the basis of the data in this case register, it is possibie to draw certain conclusions about risk fac tors associated with ASL. The large number of ASL cases in some factories and the absence of ASL cases in others of similar age indicates that variations in manufacturing p ractices between factories may be the cause. These variations may reflect both differences in the types ofjob carried out by individual workers and differences in engineering practices. The bulk of the cases have occurred, however, in highly exposed autoclave cleaners, with relatively few in other PVC or VCM production jobs. So far no wellauthenticated cases have occurred in PVC com pounding or fabrication where many more people have been exposed but to a much lovyer dose. Prediction offuture ASL coses os a consequence of pre-1974 exposure The causal relationship between VCM and ASL is proved beyond doubt by the specificity of the tu mour, the high relative incidence of that tumour in highly exposed workers, the consistency of the excess in different parts of the world, the time relationship between exposure and diagnosis and the doseresponse relationship. An intensive analysis of the pre-1974 cohorts should establish the dose-response curve for ASL after VCM exposure and predict the likely outcome for the future. It will be impossible to collect a complete data set on which to calculate risks of ASL for the whole world, but within a single company there may be closer definition of the cohort, the number of cases and the pattern of exposure. Using these data and averaging across the worldwide population exposed to VCM, it is possible to calculate the future inci dence of ASL using relatively crude assumptions which can only be tested in time when the prediction can be judged against the final outcome. AP00055831 19ft I. F. H. Purchase tt oi. Tabic IT ASU Yer of *- number* by year of ftrst eapoeure and leofrxptvicil location (eiciudinj ITOl") ASL <axxt in: exposure 1939 40 1 2 J 4 S i Weitern Europe Frit Fr!4 UK1 Sw2 Frl, FfJ. Sw4 1 % 9 1930 t 2 3 4 J 6 7 t 9 1960 1 2 3 4 i 6 7 I 9 1970 Sw5 Fr9 Frl2. Fr4 FrT. Nl. UKS Swl.UK.3 G3 CIS. IO G7. GS. U1C4. GI9 Gtl.Gt6.GU FrtO, G1 FrS. G4. Il2. G2 Fr6. 01 Fr2. ti4 Gi, G13 G9, G10, G12. G17, G20, G22 C6. UK6. G21 Fr) 3. UX7 Sw5 FrS UK1 Nonh Amend US24t a, US27 US13. US29 C2. U5I9 C). CS, USJ. US7. TJS28 C4. US3. US9 C7, C9, USS. USII. U511. LTS3t{ C6, US22, US25 ust US 12 US16 US1Q, US32 US4 CIO USIJ US2, US17, US20 US2J Rtl of world - Y2, Cl2 Jpl, Yl Y3 Cxi Jp2, Y4 evenu Cl USS U530 Visit 2 3 Milioni Tottl... 52 39 t ASL - Anpoiarcomt of the liver 1(01 ii not continent wtlh olhef ASL cues: the primxry tumour may have been of the pericardium. The man extruded PVC uekt. tFor explanatory key, tee Table 12. tCholanpotarcotTva. JUS31 it itill alive. lAerotol can filler. The data required are: (1) Animal populations of employees classified by age; (2) Annual exposure estimates for each person in (1); (3) An exposure-response latency model for ASL induced by VCM. The data under item (l) are available in the UK as a result of the data extracted from the relevant occupa tional records (Fox & Collier, 1977). Exposure data for item (2) are more difficult to obtain, but can be gleaned from the records that are used to define the occupational population. The problem of oc cupation changing, which occurred frequently, has been dealt with by using the principal employment category or the highest exposed employment cate gory. The estimation of time-weighted average ex posures for the least exposed employees is straight forward, as the exposures were essentially continuous and constant, but for autoclave cleaners, mainte nance workers and laboratory workers, exposures could vary, from zero to near narcotic levels. In the calculations described below, it has been possible to avoid using the exposure data directly by relying on the similarity in exposure levels in differing locations. The exposure/response/latency data indicated under item (3) can be derived from established cases. The key data for these procedures are the set of cases worldwide, together with the descriptive data (Tables 12-14). It has been possible to calculate aa incidence rate for each latency period for each ex posure level for each age group (on the basis of the UK data and assuming that it is representative of the worldwide population) and to use these rates to derive a simple model of dose-response latency that can be applied to the population data. The broad conclusions are that most cases have a latency of about 20 years and cases will continue to occur for the next 10 years. In the calculation used to estimate the future number of ASL cases (Table 15) an assumption has been made that when exposures were reduced to low levels, the future risk of ASL became negligible. Two dates at which the negligible risk levels were attained have been selected: 1964, when levels were reduced to hundreds of ppm and 1974 when the levels were reduced to below 10 ppm following the discovery of AP00055832 Vinyl chloride--risk assessment Table 14. Annua) incidctjex of ASL cun (date of death) by geographical are* No. or ASL case* dj-inj in: Year Weturn North Europe America Rett of world Annua) total Cumulative lots! Key *vqU 1955 7 1961 2 4 7 8 9 1970 t 2 3 4 3 6 7 % 9 1980 1 2f Tout... l t l 1 11 i 5 i2 t1 !2 41 14 3 3i 64 1 63 2 63 2 S 64 1 4 00 0 S2t 33* 8 11 12 13 I4 26 17 3 12 3 15 2 17 Viola 3 20 5 25 8 33 Miltoni 3 38 Goodrich 11 49 It 60 9 69 to 79 s 84 to 94 4 98 0 98* ASL - Angiosarcoma of (he liver Doe* not include USJI (sli)l alive in 1982). 7At time of compilation. tloetuda G03 (aerosol can filler) but amiu ttOI (h| extruder). 199 ihc association txtween ASL and VCM exposure. A hypolheticai exposed population of 100,000 has been used, but this is unimportant (see (a) below). An estimate of the age distribution within the hypolhet icai 'total' exposed population of 100,000 has been based on UK data (Fox & Collier, 1977). For persons already exposed during the whole of the various latent periods, the numbers with a latency of 30 years or more form only a small proportion of the total. The numbers of persons at risk in the future are calculated by advancing time in 5-year periods taking account of the age-dependent death rates in the population at large. Death rates for an intermediate year for the male population of England and Wales have been used in this calculation and the future cases (column 10) have been obtained by multiplication. The incidence figures for long latent periods (>1S years) are unreliable or non-existent but those for latencies of 15--25 years are fairly constant and values of 0.5 and 0.5 cases/1000 persons have been used for all latency periods over 15 years to calculate the expected number of cases for the 1964 and 1974 assumptions. The calculation is unrealistic in many respects but the simplifications are unlikely to affect the estimate of future cases by more than a small factor. For example: (a) The population size used for the calculation ` is probably larger than the exposed population, but the calculation depends on the ratio of "'person-years to come*' and "person-years ex perienced" and this ratio is the same for any population size. (b) Exposure level has been ignored. The calcu lations are based on the overall risk to the cohort and although the incidence figures for sub-cohorts could be higher, the estimate of future cases will change very little. Similarly duration of exposure has been ignored. Table 15. Hypothetic*! calculation of future ASL eases turns two different assumptions about the due at which the levels became free of risk Calculations usuminj no risk after 1964 Calculations assuming no risk after 1974 Latency W 1-5 6-10 11-15 16-20 21-25 26-30 31-3$ 36-40 41-43 46-50 5116- Cases to date Perwnt at risk to date 0 100.000 1 91.250 H 95.500 28 84,750 21 61.400 11 36.750 6 21.230 6 6750 0 600 00 0 0 J-yr incidence 0.00 0.01 0.12 0.33 0.46 0.49 0.21 0.89 > t * 0.50 Future persons at risk 0 0 0 6750 24.550 41.750 48.100 SI.750 45.750 34.500 47.600 300.750 Future cases O 0 0 2 11 20 13 46 7 7 .7 ISO Persons at risk to date 100.000 94,500 78,000 46.500 28.750 18,750 10.850 3500 310 0 0 S-yr incidence 0.00 0.01 0.14 0.60 0.97 0.96 0.55 1.71 > * 7 0.80 Future persons at risk 0 3750 17.500 45.000 57.200 59.750 58.500 55.000' 46,350 34.500 47.600 403.900 Future eases 0 0 2 27 57 57 32 94 * 4 T 323 For details of the asiiumptioai aod methods tee mt {pp. >97 A 198). AP00055833 JW 1. F. H. Purchase et al. fc) The UK is not typical of the worldwide growth in the exposed population. (d) No account has been taken of plant im provements occurring prior to 1964 and hence fewer cases may occur in, for example, the 1980-2000 period than are estimated from the 1940--1930 experience. An assumption that the risk of ASL ceased in 1964 rather than in 1974 results in a considerable reduction in the estimate of future cases. For either assumption, the number of new cases observed annually should soon begin to decline and the rate of decline will indicate which assumption is nearer to the truth. There have been two other predictions of the number of cases of ASL likely to result from previous exposure to VCM. Nicholson et at. (1984) suggest that there will be a further 1500 cases of ASL, while Forman ei at. (1986) conclude that a farther 150-200 deaths might be expected over the next 30 years. Our estimates rely on a more sophisticated model than the latter estimate and on a larger data set than the former. Nevertheless, the conclusions of Forman ef at. (1986) are similar to ours. Only the experience of the next few years will show which is the best estimate, Summary tod conclusions There is little doubt that exposure to high levels of VCM as a consequence of occupation can result in an increased incidence of ASL. A review of 20 epi demiological studies involving about 45,000 workers occupationally exposed to VCM showed that neo plasms of the liver showed an increase in incidence in the majority of studies. For brain cancer the associ ation between exposure to VCM and an increased' incidence was less clear because of the lower relative risk. Neoplasms of the respiratory tract, digestive system, lymphatic \nd haemopoietic system, buccal cavity and pharynx, cardiovascular system and colon/siomach were reported to show an increased incidence in one or more studies, but to show no increase, or in some cases a decrease, in incidence in other studies. In view of the increased incidence of breast neoplasms in rodents exposed to VCM, (he studies or Chaizze et at. (1980), who did not confirm these findings in humans, are of importance. The register of ASL cases now contains records of 99 persons with confirmed ASL and occupational exposure to VCM. The average latent period between first exposure to VCM and death from ASL is 21.9 years. The majority of cases occurred in autoclave workers, who are recognized as having been exposed to extremely high levels. Although precise estimates of exposure are not available for the periods of most interest, the pattern of cases roughly suggests that ratremely high exposures were necessary for the induction of ASL. For example, ASL cases tended to occur in larger numbers in some plants than in others, a finding that can be explained most easily by differences nn exposure patterns. There is an extensive series of animal studies on the carcinogenicity.f VCM. Some of these precede the epidemiological studies confirming the association between VCM exposure and ASL in man. ASL and neoplasms of a number of other organs have been induced in laboratory rodents by VCM. Estimation of the exposure levels likely to cause a lifetime risk of ASL of 10_` on the basis of these data give extremely low levels (down to 3.9 x 10"' ppb) which appear to be unrealistic estimates for man. Part or the reason for this is that laboratory studies have shown that VCM is metabolized in the liver (and elsewhere in the body) to the reactive metabolites chloroethylcne ox ide and chloroacetaldehyde. The rate of convcrsion.is limited at high levels of exposure giving inaccurate estimates of the slope o\ the dose-response re lationship, It has not been possible to estimate the rate of conversion in man, and hence extrapolation of these low-risk dose estimates is conjectural. The second part of the problem of extrapolation at low risk is the selection of (he most suitable mathematical model for extrapolation. Using Maltoni's data from rats (Maltoni ef al. 1981), there is a substantial range (up to I0`) of low-risk dose estimates, depending on the mathematical model and the assumptions used in applying the models. Using the same (probit and log-dose) mode) and different sub-sets of experi mental data, a large range of estimates is again obtained, even after correction for the non-linear kinetics of metabolism at high dose (which reduces this range to about )0J). Larger differences are obtained with calculations using the Weibull analysis as a basis of low-dose estimation, suggesting that this is a problem with the use of mathematical models rather than one associated with the log-probit anal ysis. Although there was considerable variability in the dose-response relationship in the different experi ments reported, in all cases a total metabolized dose of 5 x 10}fig (equivalent to inhalation of 200 ppm) was required to produce an elevation in ASL inci dence. This dose represents a practical threshold in rodents. At this stage in their development, mathe matical models for low-risk dose estimates are not sufficiently reliable or reproducible to engender confidence in their use. , Using negative epidemiological studies of popu lations living in the vicinity of VCM production facilities, an estimate of the dose for a I0~` lifetime risk in man may be made (Barr, 1982). The value (100 ppb) is similar to the highest estimates derived from animal data and taking biotransformation data into account, is substantially larger than the lowest estimates, which are up to 10IB lower (3.9 x I0'7 ppb using a multi-hit model). The higher estimates are compatible with occupational experi ence and suggest that the current hygiene standard of around 1 ppm is sufficiently low to protect the health of VCM/PVC workers. The estimates also give a considerable safety factor for the general public consuming PVC-packed food and drink or living near VCM/PVC facilities. It has been possible to provide a crude estimate of the number of cases of ASL that may occur in the future from exposure to VCM prior to 1974. Using the age structure of employees in one company, the total number of cases of ASL reported to date and the mortality pattern expected from a normal popu lation, the possible future number of ASL cases has been estimated as in the region of 15D-300. Vinyl chloride--risk assessment 201 Acknowledgements--We thank Dt M. Thoma* for his help with the calculations and Dr D. M. Conning for his help with the manuscript. REFERENCES ACGIH--American Conference of Governmental Indus trial Hygjertim (1974). Threshold Limit Values for Chem ical Substances and Physical Agents in the Workroom Environment with Intended Changes for 1974. ACGIH, Cincinnati, OH. Anderson M. W., Hoel D. G. dt Kaplan N. L. (1980). A general scheme for the incorporation of pharmacokinetics in tow-dose risk estimation for chemical carcinogenesis: example--vinyl chloride. Toxic, appl. Pharmac. 55, 154. Barnes A. W. (1976). Vinyl chloride and the production of PVC. Proc. R. Soc. Med. 69, 277. Barr J. T. (1982). Risk assessment for vinyl chloride in perspective. Presented at the 75th Annual Meeting of the Air Pollution Control Association, New Orleans, USA. June 1982. Beaumont J. J. Sc Breslow N. E. (1951). Power consid erations in epidemiologic studies of vinyl chloride work ers. Am. J. Epidem. 114, 725. Bertazzi P. A.. Villa A., Foa V., Saia B.. Febri L., Mapp C.. Marcer C.. Manno M.. March! M. Sc Bottasso F. M. (1979). An epidemiological study of vinyl chloride ex posed workers in Italy. Archo higrada toxicol. 30, 379. Buffler P. A., Wood S., Eifler C., Suarez L k Kilian D. J. (1979). Mortality experience of workers in a vinyl chloride monomer production plant. J. occup. Med. 21, 195. Byrcn D., Enghol G., Engiund A. Sc Westerholm P. (1976). Mortality and cancer morbidity in a group of Swedish VCM and PVC production workers. Envir. ffhh Perspect. 17. 167. Carlborg F. W. (1981). Dose-response functions in carcino genesis and (he Weibull model. Fd Cosmet. Toxicol. 19, 255. Chiazze L. k Ference L. D. (1981). Mortality among ' PVC-labricating employees. Envir. Hltft Perspect. 41. 137. Chiazze L.. Warn O., Nichols W. E. Sc Ference L. D. (1980). Breast cancer mortality among PVC fabricators. A occup. Med. 22, 677. Conference to R.eevaluate the Toxicity of Vinyl Chloride Monomer, Poly(vinyl Chloride) and Structural Analogs (1981). Conference sponsored by NIEHS/NIOSH/OSHA at N1H, Bethesda. March 1980. Envir. filth Perspect. 1981. 41, 1-231. Cook W. A., Grever P. M., Dinman B. D. & Magnuson H. J. (1971). Occupational acro-osteolysis II. An indus trial hygiene study. Srchi envir. filth 22, 74. Cooper W. C. (1981). Epidemiological study of vinyl chlo ride workers: mortality through December 31, 1972. Envir. Hlth Perspect. 41, 101. Creech J.L4 Johnson M. N. (1974). Angiosarcoma of the layer in the manufacture of PVC. J. occup. Med. 16, 150. Crump K. S. St Guess H. A, (1980). Drinking Water and - Cancer. Report no. PB81-I28167. NTIS, Washington. Duck B. W., Carter J. T. k Combes E. J. (1975). Mortality study of workers in a polyvinyl chloride production plant. Lancet U, 1197. EPA (1980). Ambient Water Quality Criteria for Vinyl Chloride. Environmental Protection Agency Report, EPA 440/5-80-075 (October). Feron V. J,, Hendriksen C. F. M,, Speek A. J., Til H. P. Sc Spit B. J. (1981). Lifespan oral toxicity study of vinyl chloride in rats;. Fd Cosmet. Toxicol. 19, 317. Filatova V. S.. Anlonyuzhenko V. A.. Smulevieh V. B,, Fedotova 1. V,, Kryzhanovskaya N. A.. Bochkareva T. V., Goryacheva L. A. k Bul-bulyan M. A. (1982). The blastomogenic hazard of vinyl chloride (a clinico-hygiene and epidemiological study). Git. Truda prof. Zabol. 26(1), 28. Food Safely Council (1980). Proposed System for Food Safety Assessment. Final Report. FSC, Washington, DC. (Also in Fd Cosmet. Toxicol. 1980, IB, 7] 1). Formats D,, Bennett B , Stafford J. dt Doll R. (1986). Vinyl chloride and angiosarcoma of the liver--a report of the register of cases. Br. J. ind. Med. 42, 750. Fox A. J. k Collier P. F. (1977). Mortality experience oT workers exposed to vinyl chloride monomer in the manu facture of polyvinyl chloride in Great Britain. Br. J. Ind. Med. 34, I. Freuel-Beyme R.. Schmitz T. St Thiess A.M. (1978). Mortalitatsstudie bei VC/PVC Arbeiteme der BASF. Aktiengesellschaft, Ludwigshafen am Rhein. Arbeitsmtd. Sozialmed. Preventivmed. 13, 218. Gauvain S. (1976). Vinyl chloride. Proc. R. Soc. Med. 69, 275. Gaylor D. W. k Kodeil R. L (1980). Linear interpolation algorithm for low-dose risk assessment of toxic sub stances. J. tnvir. Path. Toxicol. 4, 305. Gehring P. J., Watanabe P. G. & Park C. N. (1978). Resolution of dose-response toxicity data for chemicals requiring metabolic activation: example--vinyl chloride. Toxic, appl. Pharmac. 44, 581. Gehring P. J., Watanabe P. G. Sc Park C. N. (1979). Risk of angiosarcoma in workers exposed to vinyl chloride as predicted from studies in rais. Toxic, appl. Pharmac. 49, 15. Green T. Sc Hathway D. E. (1975). The biological fate in rats of vinyl chloride m relation to its oncogenicity. Chemico-Siol. Interactions 11, S4S. Green T. Sc Hathway D. E. (1977). The chemistry and biogenesis of (he S-comaining metabolites of vinyl chlo ride in rats. Chemico-Biol. Interactions 17, 137. Harris D. K. Sc Adams W, G. F. (1967). Acro-osteolysis occurring in men engaged in the polymerisation of vinyl chloride. Br. med. J. 3, 712. 1ARC Workieg Group (1979), Monographs on the Evalu ation of the Carcinogenic Risk of Chemicals to Humans. Vol. 19. Some Monomers, Plasties and Synthetic Elas tomers and Acrolein, p. 377. International Agency for Research on Cancer, Lyon. InJante P. F. (1981). Observations of the site specific carcinogenicity of vinyl chloride to humans. Envir. Hlth Perspect. 41, 89. Kuzmack A. M. dt McGaughy (1975). Quantitative Risk Assessment for Community Exposure to Vinyl Chloride. US EPA Report, 5 December. EPA. Washington. DC. Lee C. C., Bhandari J. C., Winston J. M,, House W. B.. Dixon R. C. k Woods J. S. (1978). Carcinogenicity of vinyl chloride and vinylidene chloride. J. Toxic, tnvir. Hlth 4, 15. Lester D., Greenberg L. A. k Adams W. R. (1963). Effects of single and repeated exposures of humans and rats to vinyl chloride. Am. ind. Hyg. Ass. J. 24, 265. Loprieno N., Bavale R., Baroncelli S,. Bartsch H., Brouzetli G., Gammellini A., Coni C., Freza D.. Nieri R.. Leporini C-, Rasellini D. & Rossi A. (vt. (1977). Induction of gene mutagens and gene conversions by vinyl chloride metab olites in yeast. Cancer Res. 36, 253. Milloni C., Lefemine G., Ciliberti A.. Cotti G. Sc Carretti D. (1980). Epidemiologic animate et epidemiologic hu mane: le cas de chlorure de vinyl monomere. In XXe Reunion de dub de Cancirogenise Chimique. ISBN 2.86315.007.3. p. 15. Publications Essentielles. Paris. Maltont C., Lefemine G,, Ciliberti A.. Colli G. k Carretti D. (1981). Carcinogenicity bi-assay* of vinyt chloride monomer a model of risk assessment on an experimental basis. Envir. Hlth Perspect. 41, 3. Maltoni C. St Rondlnelta R. (1980). Hepatic angiosarcoma in workers exposed to vinyl chloride in Italy. Acta Oncologica I, 35 (in Italian). Meyerson L. B. dt Meier G- C. (1972). Cutaneous lesions in acro-osteolysis. Archs Derm. 106, 224. r.CT, 2J1-F 202 I. F. H. Purchase ti al- MLmstry of Agriculture. Fisheries and Food (1978). Survey of Vinyl Chloride Content of Polyvinyl Chloride for Food Contact a.nd of Food*. HMSO. London. Monson R. R.. Peters J. M. & Johnson M. N. (1974). Proportional mortality among vinyl chloride workers. Lancet U, 397. NAS (National Academy or Sciences) (1980). Drinking Water end Health. Vol. 3, p. 38. National Academy Press, Washington. Nicholson W. J,, Hammond . C,, Seidman H. St SclikolT I. J. (19721). Mortality experience of a cohort of vinyl chloride/polyvinyl chloride workers. Lancet U, 1197. Nicholson W. J., Henneberger P. K. & Tarr D. (1984), Trends in cancer mortality among workers in the synthetic polymers industry. In industrial Hazards of Plastics and Synthetic Elastomers, p. 65. Alan R. LisJ, New York. ORC (1976). Mortality data collected by ORC concerning the effects of vinyl chloride exposure in PVC fabrication. Cited by Barr (1982). Otl M. G.. Ungner R. R. St Holder B. H. (1975). Vinyl chloride exposure in a controlled industrial environment. Arch envir. Hhh 30, 333. Rarmug V., Gothe R. St Wachtmeister C. A. (1976). The mutagenicity of chlorocthylene oxide. chloroacetaldehyde, 2-hloroethanol and ehloroacetie acid, conceivable metabolites of vinyl chloride. Chemico-Biol. Interactions 21, 25i. Reinl W. St Weber H. (1976). Stand der epidemiologischcn Forschung uber die Vinylchlorid-krankheit. Zentbl. Arb- Med. ArbSchutz 16, 97. Reinl W,, Weber H. St Greiscr E. (1978). Epidemiology, study of the mortality of workers exposed to vinyl chloride in FRG. Paper presented at the 19th Inter national Conference for Occupational Health. Dub* rovnre, September. Schneiderman M. A., Mantel N. & Brown C. C. (1975). From mouse to mao--cr how to get from the laboratory to Park Avenue and 59th Street. Ann. N. Y. Acad. Set. 246. 237. SelikofT I. J. (1975). Toxicity of vinyl chloride/polyvinyl, chloride. Ann. N.Y. Acad. Set. 246. Spinas R. A. Kaminski R. (1978). Angiosarcoma of the liver in vinyl chloride/polyvtnyi chlonde workers. J. occup. Med. 20. 427. Suciu J.. Drejman I. it. Valaskii M. (1963). Contributions to the study of disease by vinyl chloride. Mtd. interna 15. 967 (in Italian). Szadkowski D. A Lchnert G. (1982). Vinylchlorid all KnnkheiUuruchfi. Eine Bibliographic. VKE, Frankfurt. Tabershaw J. R. St Gaflcy W. R. (1974). Mortality study of workers in the manufacture of vinyl chloride and its polymers. J. occup. Med. 16, 509. Theriault G. (1982). Cancer mortality of Canadian workers exposed to VCM; a three-year follow-up. J. occup. Med. 24. 730. Theriault G. St Allard P. (1981). Cancer mortality of a group of Canadian workers exposed to vinyl chloride monomer. J. occup. Med. 23, 671. Torkelson T. R., Ogen F. St Rowe V. K. (1961). The toxicity of vinyl chloride as determined by repeated exposure of laboratory animals. Am. ind. Hyg. Ass. J. 22, 354. US DHEW--Department of Health, Education and Wel fare (1980). Selected abstracts on (he carcinogenicity of VCM. DHEW. Washington, DC. Yerburgt F. G. St Vogel E. (1977), Vinyl chloride muia* genesis in Drosophila me/anogaster. Mutation Res. 48,327. Viola P. L. (1969). Pathology of vinyl chloride. Proceedings of the 16th international Congress on Occupational Health, Tokyo. Watanabe P. G. & Gehring P. J. (1976). Dose-dcpcndenl fate of vinyl chloride and iu possible relationship to oncogenicity in rats. Enoir. Hhh Penpect. 17, 145. Waxweiler R. J., Stringer W., Wagoner J. K.. Jones J., Falk H. & Carter C. (1976). Neoplastic risk among workers exposed to vinyl chloride. Ann. N.Y. Acad. Sci. 271, 40. Weber H.. Reinl W. St Grieser E. (1981). German in vestigations on morbidity and mortality of workers ex posed to vinyl chloride. Envir. Hhh Perspect. 41, 9S. Wilson R. H., McCormick W. E-, Tatum C. F. St Creech J. L. (1967). Occupational acro-osteolysi$. J. Am. med. Ass. 201, 577.