Document gr0YQMq7ZRnR44x0ERqg3g6a

echnical Report N 31 r and carcinogenicity 0F v,NYL CHLORIDE : A HBSTdRICAL review and assessment ISSN ii - 8072 - 31 SPI-00678 Brussels, 28 July 1988 (C) ECETOC copyright 1988 ECETO; Technical Report No.31 Copyright - ECETOC (European Chemical Industry Ecology and Toxicology Centre), 250 Avenue L)uise (Bte 63), 1050 - Brussels, Belgium. No part of this pub ication may be reproduced in any form, by print, photoprint, microfilm , or any other means without written permission from the Director. This document has be n prepared and reviewed with all possible care by experts on behalf of ECETOC. It is provided solely for information and shall -form no part of any contract with you or your customers. It is not to be taken as a warrh nty for which we take legal responsibility. SPI-00679 -u- TABLE OF CONTENTS SUMMARY A. INTRODUCTION AND HISTORICAL REVIEW 1. VC and PVC Production Processes 1.1 Dispersion Proces 1.2 Bulk Process 2. PVC Conversion Proceslses 3. PVC Packagings for Foo d and Drink 4. Exposure to Vinyl Chi ori de 5. Effects of Vinyl Chlolr ide on Health 5.1 Effects on the Cei tral Nervous System (CNS) (roughly from 1930) 5.2 "Vinyl Chloride D isease" (roughly from 1957) 5.3 Carcinogenicity ( L970 and onwards) 6. Evolution of Occupati nal Exposure Limits B. EXPOSURE OF HUMAN BEINGS TO VINYL CHLORIDE 1. Odour Threshold 2. Occupational Exposure 2.1 Locations 2.2 Measuring and Mon toring Methods 2.3 Levels of Exposure 3. Exposure of the Gener 1 Public 3.1 By Inhalation 3.2 By Ingestion 3.3 Summary and Risk Assessment EFFECTS ON HUMAN HEALTH 1. Carcinogenicity 1.1. Cohort Studies of Cancer Morbidity and Mortality in Vinyl Chloride and Polyvinyl Chloride Producticr Plants 1.2. Studies of Workers in Polyvinyl Chloride Processing 1.3 Evaluation of Epidemiological Studies 1.4. Register of Angiosarcoma of the Liver (ASL) Cases 1.5. Angiosarcoma of tie Liver in the General Population n -i v 4 4 5 7 7 7 7 8 9 9 6 10 11 11 11 11 12 13 16 16 19 23 25 25 25 45 49 54 55 SPI-00680 T -in - 1.6. Conclusions 2. Clastogenicity and Mutagenicity 2.1. Studies 2.2. Conclusions 3. Other Effects Relevant tjo Carcinogenicity or Mutagenicity Diseases 3.2. Acro-osteolysis (ACL), Raynaud's Phenomenon and Effects on the Skir 3.3. Conclusions 57 57 57 59 59 59 61 52 D. EXPERIMENTAL TOXICOLOGY 1. Carcinogenicity 1.1 Exposure by Inhalat; on 1.2 Exposure by Routes (jther than Inhalation 1.3 Special Studies of inyl Chloride Carcinogenesis 1.4 Comments 1.5 Conclusions 2. Mutagenicity and Genotoiicity 2.1 Mutagenicity and Genotoxicity In Vitro 2.2 Mutagenicity and Genotoxicity In Vivo 2.3 Mutagenicity of VC-netabolites In Vitro and In Vivo 2.4 Conclusions 3. Other Toxicological Datp Relevant to Carcinogenicity and Mutagenicity 3.1 Acute Exposure by Inhalation 3.2 Conclusions 63 63 63 70 72 77 80 81 81 81 84 86 86 67 87 E. METABOLISM AND RELATED STUDIES 1. Animal Studies 2. Toxicokinetics of Inhaled Vinyl Chloride in Man 2.1 Retention of Inhaled Vinyl Chloride in the Lung 2.2 Clearance of Vinyl Chloride from Air by Inhalation 2.3 Biological Monitor! ng 2.4 Comparison of the fate of Uptake of Vinyl Chloride from Air in Man and othe r Mammals 89 89 91 91 91 92 94 F. CONCLUSIONS 95 SPI-00681 -tv- bibliography 97-106 TABLES 107-136 FIGURES 137-140 APPENDICES 1. Properties c-f Vinyl Chloride 2. Tests for Significant e of the Increased Incidence of Melanoma in Maitoni' experiment BT8 3. Estimation of the Exf osure Concentration of VC Corresponding with 50% of the Maxi njum Metabolic Conversion Rate in Humans 4. Members of Task Force Vinyl Chloride 5. Members of the ECETOC Scientific Committee 141 142 144 146 147 SPI-00682 T -1- THE MUTAGEN ICI Y AND CARCINOGE N I C I T Y VINYL C H L OR I D E : A HIST 0 R I C A L REVIEW AND ASSESSMENT OF SUMMARY In 1974 the first evidence that occupational exposure to vinyl chloride (VC) could lead to a rare type of liver caicer in man, angiosarcoma, was published. Since then an enormous amount of epidemiol Dgical, clinical and toxicological research has been carried out and a large volume )f data on occupational exposure and exposure of the general public has been collec ted. Th's information is scattered in many types of publications by numerous authc s. The purpose of this report is to present under one cover a coherent picture o': the most important aspects of the toxicology of VC (with the emphasis on carcinogtnicity and mutagenicity), and the risk it poses for human health at current levels of exposure. A brief Introduction and Hist rical Review is given in which the processes for producing VC and its polymer ipolyvinyl chloride, PVC) are described; the sources and types of exposure to VC a r ? given; the various diseases arising from excessive exposure are noted, and the evn lution of exposure limits is summarised. Human occupational exposure is predominantly via inhalation at plarts where VC and PVC are manufactured and at factories where PVC is processed to give various fabricated articles. In practi :e skin exposure from the liquid phase is negligable. The development of methods fjr determining VC concentrations in air is briefly noted, and the historical love's of VC at the various types of plants are reviewed. By far the highest levels (hundreds of ppm) were experienced periodically at PVC-production plants, but by mid-197^ they had been reduced to, typically, around 50 ppm. Progress in lowering the levels continued, and since 1978 they have been at a few ppm or lower. The general public is exposed to very small amounts of VC : i) from inhalation of ambient air in urban areas typically in the order of 5Ug/person/day with higher amounts in the vicinity of V( and PVC plants, ii) by ingestion of food and drinks packed in VC-containinc polymers as films, cartons or can-linings from which residual VC can migrate into the packed material and iii) by inhalation of tobacco smoke (a few ng/person/day). Since 1578 emissions from VC and PVC producing plants have been reduced cor.sidere bly. Between mid-1973 and mid-1975 the amount of SPI-00683 i -2- residual VC in polymers wa; also drastically reduced, typically from more than 10 ppm to about 1 ppm, and the concentrations in food and drink fell from around 100 to about 2 ppb over the sajnie time period, and data obtained in 1986 show typical values of 0.1 ppb or less. The carcinogenic, mutagenic clastogenic and related effects of VC on humans are reviewed. Twenty-six papers describing epidemiological studies are critically assessed and it is concl uded that, apart from the well-established fact that exposure to the high levels of VC experienced in the past can cause angiosarcoma of the liver, there is no con' incing evidence that it causes cancer at other sites, nor is there any evidence )f teratogenic or heritable mutagenic effects in man. Clastogenicity was only see n at higher occupational exposure levels experienced before the marked reduction in exposure in the mid-197Cs. Various non-carcinogenic efi<ects of chronic occupational exposure to VC have been described under the collecv ive heading "Vinyl Chloride Disease", and these are briefly reviewed. There have been no reports of such effects in workers exposed to VC after the early 1970s whei VC levels were lowered to a few ppm. The experimental studies on the carcinogenicity, mutagenicity and genotoxicity of VC are also reviewed. From the extensive work on carcinogenicity, much of it by Maltcni, it is concluded :hat VC is a versatile carcinogen in animals when administered by inhalation or orally. Jr the three species tested (rat, mouse and hamster) it induced hepatic Ipapmaciosarcomas, Zymbal gland tumours, nephroblastoma, pulmonary and mammary gland tumours, and forestomach papillomas. The minimum dose ?t which compound-related tu nours were induced by inhalation was 10 ppm for rats, 50 ppm for mice ard 00 ppm for hamsters. When VC in PVC powder was administered orally to rats, the minimum effective dose was 1.7 mg VC/kabw per day. When it was administered as a solution ir water to the same species, the minimum Effective dose was 25 ppm. SPI-00684 VC is mutagenic and clestoge nic in vivo. It is also mutagenic and clastogenic i_n vitro but only in the prqsence of appropriate metabolic activating systems, Chioroethylene oxide appears to be the most potent mutagenic metabolite. The liver is the prirery sit ; of structural damage by VC, and it appears that the effect is due to toxic metaboll ites whose formation i? induced by the mixed function oxidase system. The fact the t, in experimental animals, tumours were induced in -3- organs which were not adversely affected in acute and sub-acute studies, and that tumours developed in the liver at- dose levels well below those found to cause hepatotoxicity, is consistent wi th a genotoxic mechanism of action. The major pathways in the metabolism of VC have been established. In experimental animals the key metabolic Stef is its conversion by mixed function oxidase into chloroethylere oxide. Although this process is readily saturable, the ability of chioroethylene oxide to bind covalently to OKA at levels which do not induce saturation probably accounts fo r the induction of tumours at such low levels. From studies on the clearance rat s of inhaled VC in rats, mice and humans it is concluded that rats and mice exposed to VC produce more of the carcinogenic metabolite per kgbw than do humans As to the current risks, our conclusions are that although it is not possible to set definitely safe levels o' exposure for genotoxic carcinogens, the evidence presented in this report suggests that occupational exposure at current levels, as would be achieved when in comp iance with, for example the EEC limit of 3 ppm, does not present any significant ri k to health. At the atmospheric levels to which the general public is exposed, in the order of 5 Ug/person/day with higher values in the vicinity of VC and PVC pic nts, and less elsewhere, the risk of adverse health effects is even less. The content of residual VC in PVC and copolymers in food ard liquid packaging materials has fallen, typically from more than 10 ppm in the early 1970's to below 1 ppm in subsequent years. There has been a corresponding reduction in the VC content of packed food and drink from about 100 ppb, to below 2 ppb from about 1977 onwards. Conservative estimates have indicated that the current intake of VC from food and drink presents a negligible risk of cancer. SPI-00685 -4- A. INTRO D UCTION AND HISTORICAL REVIEW Since about 1974 vinyl chlor de (VC) has become one of the most intensively studied of all industrial chemical; from the point of view of its toxic effects. An enormous amount of clinical and toxicological research has been carried out and a large amount of data relating to occupational exposure and exposure of the general public has been collected. about 1985 many epidemiological or animal studies had been completed and reported Most of the information necessary to understand the toxic effects of VC have beeh obtained. As this information is widely scattered it is difficult to have a coherqn t view of the results and developments which form the background of control measure s currently in force. This is necessary for all who are concerned with the volunt ary and regulatory control of VC. As a consequence ECETOC set up a Task Force composed of experts to assess the carcinogenicity and mutageni ity of VC and their significance for the health of exposed workers and the gener al public. Emphasis was placed on carcinogenicity' and mutagenicity because measure taken to reduce the risk from these effects would automatically minimise, and probably eliminate entirely, the risk of those VC-related diseases and concitions, which develop only after exposure to "high" concentrations. These other \]c related diseases are described briefly for the sake of completeness. The report is set out as fol ows. The short historical review which completes this Chapter is followed by an acc ount of human exposure (Chapter B) and the effects of exposure tc VC on human heal 1 h (Chapter C). Experimental studies are then assessed under the headings: experiment al toxicology (Chapter D), and metabolism and related studies (Chapter E). Conclus ons from the whole review are drawn in Chapter F. It is emphasised that only the se publications relevant to the task in hand are considered. 1. VC AND PVC PRODUCTION PP.0C:SSES SPI-00686 Vinyl chloride (VC) is a col jurless, inflammable, explosive gas, slightly soluble in water, and soluble in fats and organic solvents (Appendix 1). It was first synthesised by Rlegnault in 1835. Research on the polymerization of VC wes carried out in Germany \o provide a substitute for rubber during the First World War, but it was not u rjtil a century after its discovery, i.e. in the 19?0s, -5- that the production of VC and its polymerisation to polyvinyl chloride (PVC) was developed industrially. After t(i e Second World War production expanded considerably and PVC became one of the most important synthetic resins produced by the plastics industry. Worldwide annual prod|u ction of the polymer in 1985 was about 12 million tons. Apart from its use in the produ :tion of PVC and the manufacture of copolymers with monomers such as vinyl acetate Dr vinylidene chloride, VC still retains some minor use as a raw material for the manufacture of 1,1,1-trichlcroethane and monochloroacetaldehyde. Certain of its less important uses have been abandoned since the early days and are lot considered further, eg aerosol propellant and general anaesthesia. There are two methods for the industrial production of VC. Formerly it was made from acetylene and hydrogen chi oride but nowadays ethylene and chlorine are mainly used. They are reacted to give 1,2-dichloroethane, which is thermally cracked to produce VC and hydrogen chi or de. There is very little exposure to VC, since production is carried out as a continuous process in plant closed to the outside atmosphere. Descriptions of VC polymerization processes have been given by Cook et al (1971), Cohan (1975), Barnes (1976), Bormefoy (1977) and Stafford (1977). There are two main processes : the dispersion process and the bulk process. Because most of the adverse effects of VC .on hunan health occured at plants in which PVC was manufactured by the dispersion process, this is described in some detail below. The bulk process was introduced onl^ in the later 1960s and a much shorter description of this is given. 1.1 Dispersion Process SPI-00687 As VC is a gas at ambient temper atures polymerisation is carried out in autoclaves at 40-7CC. The reaction is exo thermic, and the most common method of dissipating the heat is to disperse the mclinomer in fine droplets in an approximately equal quantity of water. The reaction takes place in a three-phase system; the solid polymer is precipitated in the m snomer droplets, which in turn are dispersed in the eouecus phase. Despite more than 30 years of research, no way has yet been found to prevent a film of PVC forming on the inside wall of the reactor. This film interferes with the transfer of heat between the reactor and its contents, and the -6- process has to be interrupled periodically to allow the reactor to be cleaned. The fact that it is technically impossible to polymerize VC by a continuous process in a permanently-closed systen lies at the origin of the occupational pathology of autoclave cleaning personnel Over the years, the size of the autoclaves has gradually been increased. Whereas the first had a capacity o only a few cubic metres, those in current use have a volume of several tens of cubic metres, and even larger autoclaves are being designed. In a typical process, dei Dnized water is first introduced into the autoclave, followed by emulsifiers, catalysts, surfactants, buffers, etc. Air is purged from the reactor, and the liqu d VC is then pumped into the sealed autoclave under pressure. The autoclave is heated to between 50 and 60C in order to initiate polymerisation which then c ontinues exothermically. Once polymerisation has ended, the autoclave charge is emp ;ied into degassing tanks, and the non-polymerized-VC is degassed and pumped into a gasometer where it is compressed and then stored under refrigeration in pressurised spheres. The wet PVC is transferred to mixers in which the charges from several reactors are combined in order to make the product uniform. It is then transferred to driers, and the resulting dried pov/der is sent either tc bulk storage silo: , or to hoppers for bagging. The autoclave, after being emptied, is opened, rinsed < nd washed either with solvents or by means of automatic high-pressure water jets. Nowadays, many precautions are taken to guarantee the best achievable sealing involving the complex syst^im of piping, tanks, pumps and valves in which the VC flows. This even applies to the autoclave which only needs to be opened occasionally. A combination of technologies are used for autoclave cleaning all of which involve washing betwe n individual, or a limited number of batches, thereby reducing the formation of c usts or deposits resulting from polymerisation. In the past, the autoclaves were cl eaned manually ; gloves were worn, whilst the inside of the autoclave was scrapped with a spatula, or sometimes a hammer and chisel, to remove the encrusted polyme r adhering to the walls of the vessel and the mixing devices. Lumps of polymer often released monomer when broken, resulting in dangerously high concentrati ons in the autoclave. The usual practice up until about 197C was not to go into the vessel, in which the atmosphere was still lader with monomer, until an explosime t e r had been used to check that the autoclave contained less than about 400 ppm, i a. a safety margin of around two orders of magnitude SPI-00688 T -7- below the lower limit of txplosivity of VC. Nowadays much more sensitive instruments are used to ensure that entry into autoclaves occurs only when the VC level is at or below 10 ppm. 1.2 Bulk Process Liquid VC under pressure, pi us a free-radical catalyst and other additives, are charged to an autoclave wher^ pre-polymerisation occurs. When 8 to 12% of the monomer has been converted in polymer this "seed" material is passed to the main polymerisation autoclave wherfe further VC, catalyst, etc. are added and the polymerisation is allowed to go substantially to completion. The remaining VC is removed and recovered, and the dry polymer is then screened, and passed to storage. In the period between its p reduction and use, PVC is stored in warehouses for periods of several days or we :ks and can lose significant amounts of its residual monomer by diffusion into the tmosphere. 2. PVC CONVERSION PROCESSES PVC is converted into a wide Ranee cf products by various processes, including dry blending. Most of which invol/e heating until it softens or melts. It can then be formed into solid articles (by extrusion, thermo-forming, or rotational moulding), or rigid or flexible film (ty extrusion or calendering). During the processes, part of the residual VC is expelled from the PVC. The industrial operations described above can lead to oicupational exposure to VC. 3. PVC PACKAGINGS FOR FOOD AN) DRINK Exposure of the general pub! c can arise from the packaging cf a wide variety of foods and drinks in contaire o or film made of PVC or VC co-polymers. Residual VC in the polymer may migrate into food or drink and minute amounts can thus be ingested by the consumer. 4. EXPOSURE TC VINYL CHLORIDE An inventory of the popul tion groups liable to be exposed to VC shows steeply-decreasing degree of exposure, in the following order (Bonnefoy, 1977): SPI-00689 -8- 1. Polymerisation unit workers the most heavily exposed because of the many manual operations, the frequent opening of equipment and the fact that the ins ta 1,1 at ions are often < ituated inside buildings. 2. Workers in monomer product ion units - only slightly exposed since the continuous process under pressure Necessitates sealed systems, and the installations are situated in the open air. 3. Workers in plants where If VC is converted into manufactured articles. The levels have always been very lo* , as the only source of VC is traces released from the resin. 4. People living near plants 5. Consumers of food and dr nk packed in PVC or VC copolymers containing residual monomer. More details about the 1 avels of exposure of these groups are given in Chapter B. The concurrent publication n 1974 of papers on angiosarcoma of the liver in VC exposed experimental animal* and man stimulated a major technological effort to reduce the exposure at worcplaces. The result was that exposure was markedly reduced and atmospheric cone entration limits in the workplace were lowered, e.g. from 500 ppm (ACGIH, 1970) t 3 a "harmonized" long-term Limit Value of 3 ppm as an annual average in the European Communities (EEC, 1978), cf table 1. At the same time there was a sustained ejffort to lower the amount of residual monomer in PVC, and hence in food and drink packaged in PVC-containing materials. In 1978 the EEC set a limit of 1 ppm of resi dual VC in PVC polymers, and a migration limit of 10 ppb in food or drink resulting from the packaging (EEC, 1978). 5. EFFECTS OF VINYL CHLORIDE 3N HEALTH There are many reviews of th( toxicclcgy of VC, eg. Schottek (1969), Viola (1974), Foa et al (1974), CIRC (1974 Haley (1975), Hublet (1975), Truhaut et al (1975), Szadkowski (1976), Heuse ( l|)78), Binns ( 1979) and Veltman (1980). In addition Cavigreaux ( 1975), Heiman et al ( 1975), Warren et al ( 1975) and Szadkowski et al (1982), have produced bibliog -aphic inventories. Our understanding of the texi eclogy of VC developed in three historical stages as set out below. SPI-00690 -9- 5.1 Effects on the Central Ne ~vous System (CNS) (roughly from 1930) During the early period cf thi? industrial development and use of VC and PVC, the effects of VC on the CNS, it what would now be considered to be very high concentrations, were recognise j in experimental animals and man (Patty, 1930). The symptoms include euphoria, head! aches, dizziness and loss of consciousness which are manifested in man at concenti ations of several thousand ppm. See, for example, Mastromatteo et al (1960), Copdier et al (1966), Berod et al (1972), Lange et al (1974), Mouiir, et al (1974), Lilis et_al (1975), Truhaut et al (1975), Walker (1976) and Delorme et al (1978 5.2 "Vinyl Chloride Disease" roughly from 1957) In the 1950s a variety of other1 effects became recognised as resulting from chronic exposure to VC at concentrati dns of (probably) several hundred ppm. Some of them were first described by Fi1 ate va (1957) and were later collectively called "vinyl chloride disease". The effect;, included a sclerotic condition of the connective tissue of the fingers accompan ed by a thickening of the dermis, and from fibrosis of the liver tissue and s Pi een. Another effect is acro-osteolysis, a rare bore-disease resulting in de- ca lcification of the terminal phalanges of the hand and affecting largely the 1 jutoclave scrapers" (Cordier et__al_, 1966). Less commonly, osteolytic lesions a re observed at other sites in the skeleton. Acro-osteolysis is frequently preceded by a Raynaud-type phenomenon in which there is a reversible constriction o the arterioles cf the fingers leading to numbness, pallor, and cyanosis of the fi rlaers. 5.3 Carcinogenicity (1970 and onwards) SPI-00691 While attempting to reproduce , cro-osteolysis in rats exposed to VC by inhalation, Viola (1970) and Viola et el ( 1971) found an increased incidence of tumours of the skin, lungs and bones. A few y ears later Mel tori et al (1974) confirmed this, and identified in addition an incrnased incidence of angiosarcoma of the liver. A most significant finding was also m ade in the same year when Creech and Johnson (1974) reported that a search of the medical files at a Goodrich plant in the USA had revealed three cases of death from angiosarcoma of the liver (a very rare form of cancer) among the deceased wor ers. It was recognised that the cause was likely to be inhalation at high levels (Irobably a few hundred ppm) of VC over long periods. -10- This finding, in combination with results from experimental studies, initiated an urgent and radical worldwic e revision of measures for protecting the health of groups of people exposed to 1C, and simultaneously led to extensive epidemiological and animal studies which are described in the following chapters. 6. EVOLUTION OF OCCUPATIONAL EXPOSURE LIMITS In parallel with the discovi; ry of the toxic properties of VC, the development of technology to lower the co rjcentration in occupational atmospheres, and improved analytical capability, the various recommended exposure limits were lowered, as summarised in Table 1. l r SPI-00692 -11- B. E X P 0 S U R E 0 F H UMA i_BEI N GS TO VINYL CHLORIDE 1. ODOUR THRESHOLD The odour of VC could be detect ed in the neighborhood of autoclaves when they were opened and while they were be ing ventilated prior to cleaning. Various authors have given differing values of the odour threshold. Viola (1974) and Patty (1966), who carried out experiments wi th human volunteers, put the threshold at 5000 and 4000 ppm respectively. Cook et al (1971) put the lower odour threshold at 400 ppm. Bauer (1974] reported that the range of perceptible concentrations was between 200 and 2000 ppm, while Baretta et al (1969) reported that one of the experimental subjects recognised the odour )f VC at 250 ppm, although the sensation faded very quickly. Lefevre, quoted by Hufclet (1975), puts the threshold at as low as 150 ppm. 2. OCCUPATIONAL EXPOSURE 2.1 Locations Occupational exposure tc VC by inhalation can occur at three different locations - VC production plants, - VC polymerisation (PVC-production) plants, - PVC processing areas. It is important to distinguish between them because, especially in the past, they correspond, to markedly diffe ert levels of exposure. VC concentrations and the incidence of various adverse effects or. health were undoubtedly highest in PVC-production plants and tp e measuremert/nonitoring of exposure levels has consequently been much more extensive at these plants, data from the other locations being scarcer. Skin exposure to VC in its lipui d phase is of no practical consequence in view of the closed manufacturing process SPI-00693 -12- 2.2 Measuring and Mom'torinc Methods During the early industrial Manufacture and polymerisation of vinyl chloride only limited measurements at worcplace atmospheres were carried out by simple grab sample methods based on : in which the colour change in a flame was observed when it was supplied with air containing VC. Workplace air samples were taken into a syringe or evacuated flask and tested in the laboratory, the limit o detection being around 500ppm, the ACGIH Threshold Limit Value established n 1959. The method is not specific for VC; - explosimeters were used for the (non-specific) detection of VC in the 1950s and 1960s. Initially, the limit of detection was around 200 ppm, but it was progressively lowered to about 40 ppm. Gradually more sophistics ted techniques became available, such as - colour-indicator test tu:)es of the Draeger or Gastec type, with a detection limit of down to 1 ppm iepending on the measuring range. The tubes are not specific for VC, sine? other unsaturated or halogenated hydrocarbons interfere; - GLC or infra-red absorpt on techniques specific for detecting vinyl chloride in workplace air samples, which were taken into syringes, evacuated vessels or "gas bags". This techn que became available in the 1960s. Its limit qf detection depends on samp e size, but is usually well below 1 ppm; it has been developed for continuous irea monitoring. - various portable direct reading equipment such as infra-red analyzers with absorption at specific rrequercies has also been described ( Wilkes and Lavery, 1974; Ketterer, 1974; Thain, 1975). non-specific direct-reading instruments, e.g. various flame-ionisation detectors, or more recently pheto-ionisation detectors, used since the mid-1970s to determine atmospheric VC at the ppm level. Reference has alsc teen n ade to application of paper-tape analyzers (Denbera and Mi Her, 1974). SPI-00694 Early measurements of Tine Wei rhted Average (TWA) exposures were often calculated on the basis of the time spent in given areas by the worker and the VC content of grab samples cf workplace air taken supposedly nearby. By the early 1970s more sophisticated methods had beco1 e available and were increasingly used, permitting TWA measurements to be made by personal monitoring in the worker's breathing zone, Air from the breathirc zero is drawn by a small light-weight pump through a carbon -13- absorber at a pre-set rate. The absorbed vinyl chloride is subsequently analyse GLC after solvent or thermal desorption (NIOSH, 1975; Hill, 1974; Zac: Rasmusson, 1974; Ketterer, 197^ ; Thain, 1975). Recently, passive sampling ce. (Perkin-Elmer, 1980), i.e. wi h no pump, have been used for this purpose further experience is necessary before their performance can be fully validate: In the early 1970's locatior monitoring predominantly in VC producticr polymerisation plants was dev eloped to give continuously-operating, seauer measurements at a number of ixed locations in the workplace, particular!., plants inside a building. The se extensive monitoring systems scan the worn: many times during each shift, providing in addition a near-instant review c* workplace concentration of shif t, weekly and monthly exposures. Since the early 1970s, persona exposure concentrations (TWAs) have been meas with portable equipment carried on the worker. 2.3 Levels of Exposure 2.3.1 VC-production plants There are little or no pub!ishe d data before the mid-1970s. Oelfke (1974) re:: that 8-hour TV/As for personal < ixposure at two VC plants were in the range or 45 ppm for all employees. Th< highest exposures were associated with tcr> filling, and levels were subs? quently reduced by careful purging of the 7:; hoses before disconnection and better techrioues for warning of the complet". filling. From 1977 onwards tf e 8-hour TWA was below 8 ppm (Jones, 1981; > 1985). In general, exposure levels i n VC production plants, in which the proce:' continuous and the systems clos ad, are lower than those in PVC-manufacturinc : where the process is discontinu dus and parts of the plant (e.g. autoclaves) h?' be opened periodically. 2.3.2 VC-polymerisation plants SPI-00695 There are few data on the expc sure of workers in polymerisation halls befc' carcinogenic role of VC was dis :overed in 1974. In 1957, Filatova et al (see et a 1, 197?) reported workplace atmospheric levels of VC of 20 to 315 pppmm.. -14- et al (1972) quoted average values of 10 ppm and stated that some workers were exposed to more than 300 pprr TWA. Cook et al (1971) found concentrations of 50 to 100 ppm in reactors after ventilation, i.e. when the autoclave cleaner went down into the autoclave, but levels of 600 to 1000 ppm could be detected around the workers' hands during scraping. Lefevre (cf. Hublet et__al_, 1977) carried out measurements in 1967 which showed that the average exposure of autoclave scrapers at that time was in the reg on of the TLV, i.e. 500 ppm. In the epidemiological study carried out by Ott et al (1975) (see also section C.l.l.a), very precise exposure profiles were give i for workers carrying cut different tasks in the polymerisation units of an American company where continuous monitoring of the atmosphere had been carried out from 1959. These values were significantly lower than those reported above. Several authors estimated av qrage exposure levels prior to the mid-1970s from the rather limited data availabl (see Baretta et al, 1969; Ott et al, 1975; Siciu et cH, 1975; Barnes, 1976; Ha steen et__al_, 1978). By about 1974 onwards, when accurate personal monitoring necame possible, the levels were much lower (Barnhardt et al, 1975; Mi 1 by, 1977; Jon s, 1981). The various author's estimates are not very different, and there is gen sral agreement that historical VC concentrations in polymerisation plants were br 3adly as in Table 2 (Dcwrick, 1974; Barnes, 1976). Until the mid-1970s, peak e xposures, probably of greater than 1000 ppm, often occurred during autoclave cl ar.iro. The data in Table 2 are reasonably consistent with the following statement Dy Sittig (1978): "Since early occupational lealth studies eftpr reported acute toxic effects (dizziness, heads :hes, nausea, etc.), it can be assumed that peak exposure levels of s veral thousand ppm were experienced at times, Air monitoring data ir cr group of PVC plants during the period 1950 to 1959 indicate that 6 hour TWA exposures in these facilities were in the range 120 tc 385 ppm. Thi may not be typical of exposure in all PVC plants. Peak exposures p 'obably exceeded 1,000 ppm". SPI-00696 The progress made in lower irg the weekly-jverage VC concentrations in sever polymerisation plants in the UK during 1974 and 1975 can be seen from Table 3. This trend has ccrtinuc-d to the present day, current average concentrations in plants located ir EEC member states falling well within the reouirements of the -15- European Communities Directive EEC, 1978) on the health of worke-.- e/: ' .<<! to U: i.e. 8 ppn over 1 h, 7 ppm over] 8 hs, 6 ppm over 1 week, 5 ppm cxe- 1 i-^nth and 3 ppm over 1 year. 2.3.3 PVC-processing plants Atmospheric VC-concentrations a these workplaces were, even ir **'ly 1970s, generally below 2 ppm (Dowrick, 1974; Eecker, 1974), although irte -- - -_*> -1 personal exposures of 60-120 and 1.5-2.2 ppm at the site of mixing and t'e-;-", operations at 70-110C and 46-68C, respect ively, were recorded by Oberg (19'^'. v^gnavita et al_ (1983) made grab sample meast. remen ts at a single PVC-processing rTa'* fror` ^72 onwards, finding atmospheric cc ncentrations of 5C ppm or more ir. 19/''. falling generally to 1 ppm in 1974 (with occasional peaks at 10 ppm) and tc lew-' levels in subsequent years with the introdlj ction of degassing procedures. Because the level o^ residual v|c in PVC has fallen to 1 ppm or less '' order t0 comply with regulatory requiremei ts, eg. EEC (1978), there car be Mzz'* doubt that current levels in the atmospho res of PVC-processing units are cc' 'rspondTr.gly lower. 2.3.4 Discussion Although this brief summery is b lieved to be a true reflection cf his ~y,''C6l vinyl chloride exposures, it must be Remembered that at some factories w z- conditions nay have existed than in those ,for which data are given above. Ir s -j-- countries there is a remarkable clustering of cases of angiosarcoma of the 'Hvr' '> a small number of factories, suggesting that working conditions and starca*-:ft of hygiene varied between factories. ThuR , cases of ASL in PVC-productic'" y arts were restricted to 1 out of 16 Japan se factories and 5 out of 28 US fact. O''*- s '23 out of 33 cases were in just 2 plant R). At a US chemical PVC plant in Mic' USA, the average concentrations were much] lower than those given in Table J - V''' 1975) AS stricter a consequence the results of a limal studies by Torkelson et_al ( in-house standards were introdi ced, and until now there have bea angiosarcoma of the liver at the plant despite the fact that it sta~ cases cpere!1t1 in 1950 (Bennett, 1985). SPI-00697 It is clearly difficult to quan tify exposures retrospectively, pa.'-v . ,rl, ** , ((;stigatl there are no, or only limited, contemporaneous observations. : r -16- attempting to relate the inci dence of disease to exposure, it is essential that the specific exposure data of the study population be used rather than the General data in the above survey. It can be seen from the fore going review that, in relation to the occupational exposure of humans, the inhal ation toxicology of VC is of interest in the range of up to about 3 ppm (the EEC 1 year average limit) for current exposures, and up to several thousand ppm for expo ures prior to the mid-1970s. 3. EXPOSURE OF THE GENERAL PUfLIC 3.1 BY Inhalation The general public may be ex oosed to VC present in ambient air not necessarily originating from VC plants Lahmann et al (1978) measured the concentrations of VC at 3 sites in Berlin - a residential suburb; at the si ie of a road with a high density of traffic; and in the middle of an industrial a rea . Fifty samples were taken at 15-min. intervals at each site from January to Jul y 1977. The average VC concentration at all 3 sites was between 0.11 and 0.15 ppt C.3 to 0.4 Mg/m3). Exposure of the public to VC c Iso arises from smoking tobacco. In the "mainstream" smoke of a typical 85 mm-long cigarette, 12.2 ng of VC per cigarette were found, the analytical date suggesting that the total inorganic chloride content of tobacco determines the amount of VC fo rmed (Hoffmann etal, 1976). 3.1.1 Plant emission levels The US Environmental Protect! c n Agency (EPA, 1976) reported that the average VC concentration in air around pc iymerisation plants was 0.017 ppm (44 Mg/m'). Gordon and Meeks (1977) sampled air a : randomly-chosen locations in industrial regions of Houston, Texas, where 40* of t|he total US viryl chloride capacity is situated. In 100 grab samples, 8? containec levels of VC below the detection limit (probably C.CG1 ppm,, 2.6 Mg/m3). The hi best concentration in the remaining 18 samples was 1.2 ppm (3.2 x 1C3 Mg/m3). SPI-00698 A survey of atmospheric VC 1 pvels at housing areas within 1 of VC-production plants in the UK| was performed to assess the to wl inhabitants were exposed (Aror. 1978). Preliminary results fryr that 97* of the readings were low 0.034 ppm, and that the hig`-*-2 0.12 ppm, compared with the UK Occupational exposure standard of 1 time. In a follow up study con quoted in 1984 when data was obta '^ from factory sites (Turner et al, 1S| 84) lower ambient VC levels were whined (measured at 100 to 1050 m ft om source), except during per- mi in, malfunction. For the entire two month sample period details were 'a Site No. 1 2 3 Distance from emission point (m) 670 570-1000 680 100-1050 180-540 Overall 24 hr mean, PPi" 1976-78 1984 0.015 0.028 0.008 0.020 0.029 <0.005 <0.005 <0.005 0.088 0.020 24 hi read. - a tom 1976-'* 1984 O.IC 0.23 O.IC o.i- 0.37 0.01] C.009 0.036 rep. tko - '3K Tte m The data for Site 4 are not ex|actly comparable, production cape, : -j had inornni'. and the only pratical sampling point was close to the plant (ICC Dimrrick (1981) reported the analyses of samples taken i' VC-producing, PVC-producing an<f PVC-processing plants. The loci not very clear, but seems to h ve been on the factory property buildings or area. The finaingd were as follows: e vicinity * of st { de of n, ^ VC plants : the arithmet c mean of grab samples taken at from the plant (the distance was not better defined in t' ppm (3.4 Mq/m3). The maxinum 24-hour average concentratic rr.^e -thou \(ft yaper) w*t 0,( 10-' iff it IQ3 Mg/m3), probably in t ie direct vicinity of the plant. - PVC production plants : tie arithmetic mean of grab sampler!. morb m from, the plant was 0.044 ppm (11.5 Mo/m3). - PVC-processing plants he highest 24-hour average concar -nation iff nplant was 0.007 ppm Mg/m3) Out of 313 samples, c/i`^our M concentration oi 224 of trerr was below the limit of detec , ie. fl,( (1.3 Mg/m3). SPI-00699 fcf; years despite ma intaining capacity, emissions have been lowered, for t- Dutch Ministry of Physical Planning and Environmental Housing quote /els of VC in the Netherlands to be : year air emission (tons) 1975 1978 1981 4,073 1,340 356 rie van Volkshuisvssting, Ruimtelijke Ordering en Mi 1ieubeheer, 1984). i. t'/jve Dutch Criteri. Document an estimate is given of the inhalation la the ,,> the Dutch populat on. The average concentration in the Netherlands as a Mosurt' Jated from the emi ssion and dispersion data is 0.00001 ppm (0.2 Pg/m3). +>U c* hig(lo t concentrations were calculated to be 0.001 ppm (2 Pg/m3) (annual - > '.lose to VC and PVf plants calculated average concentration ranges from ' Ittrage) f.CC? M" (S Pg/w3) at a d stance of about lKm to 0.0005 ppm (1 Pg/m3) at a distance ,,t 5 Kn. fim rogi>.nal rnal5S an^ PPU^ altion density data it seems reasonable to assume that 01 S "r the Dutch populat on is exposed to average levels of 0.0C2 ppm (5-6 ^,3^ n,u4t to levels of 0 0015 ppm (4-5 Pg/r3), and 0.05 % to 0.0013 ppm (3-4 ig/3) 1' iculated daily expo: ure therefore ranges from 4 Pgr per person per day on mrage I" inore than lJ5r * or the upper 0.01 % (5-6 Pg/m3) of the population. fcother n'ample of the reduc ion measures over recent years can be found on page 31-r c' ll,,: Dutch Criteria Dc cumert : tel-*. ill 52C kg VC/hour wa dramatically reduced after 1976/77 to 20 kg/hr. This has reduced the su rounding ambient air concentrations, as illustrated SPI-00700 -19- Year 1976-77 pose 1976-77 Distance : rom plant > 600 o 600 m 0-200 o 200-500 m >500 r. Samplcs average concentration maximum concentration average concentration average concentration average concentration CConcent t n I g/m I-1'* i. 210 i, - 600 I, ' 86 i. : 55 i, 18 3.1.2 Legislation regarding iiihalatien The Task Force is not aware of official and enforceable regulation1ambient air concentrations for VC in Western Europe. 1`r,c* In The Netherlands an unoffi cia 1 target concentration of maximum 1 '' " ' ^0f ambient air is being used. Th|i s value is based on an estimation thi' i /' * i r.uous exposure to 0.001 mg VC/m2 (3 8 ppb) corresponds to an additional car*'t f '' f` ^ risk of 1 in 10^ per lifeti ne for the general population (Nethef 1n>`' * ; Council, 1987). 3.? By Ingestion Many consumer goods are packaged in various physical forms of PVC or V The general public is thus pot antially exposed to very small amounts ' consumes cr uses the goods be causes residual VC in the polymer or migrate into the packaged conti1 rts. .' when 3.2.1 Sources of ingestion PVC, and VC copolymers, are wi Jely-used in food packaging in the form ' - bottles produced by Moi /-moulding for containing liquid food' cooking oils, vinegar, et - rigid film (either calenfl ered or extruded) which is converted <y* shaped containers by subs quent vacuum- or pressure-forming, fc' cf butter, rye-bread, swe ts, biscuits, salads, etc.; - flexible film, manufactu red by blowing or calendering, whic>applied to wrapping soiLl id foods such as cheese, meat, ^ vegetables and sometimes :heese; - coatings in metal cans. SP1-00701 -20- Another important use of PVC is in pipes and auxiliaries for the transport of potable water. Ir a number of countries a considerable part of the water network consists of plastic - mainly PVC - pipes. Two other sources of potenti al oral exposure to VC are via PVC-based packaging for pharmaceuticals and toys, In comparison with the ether areas of application mentioned above, however, exposure to VC originating from these sources is negligible. 3.2.2 general aspects of inq estion via migration Migration is the mass tran sfer of a substance to a medium in contact with it. Potential health hazards ex i st when substances migrate from a packaging material into food in significant qua ntities. Because so many different types of food are wrapped in PVC, migration i nto food and drink has been established by the use of simulants (EEC, 1980, 1981, 198 5, FDA, 1986). Ethyl Corporation Inc, in a series of reports to the FDA dating from 1975 have aeveloped a model of the nigration of VC from PVC into food simulants (FDA, 1986).This accurately predicted the extent of migration from bottles containing from 80 to 330 ppm of VC. Tie FDA later confirmed the accuracy cf the model with PVC sheet containing much lo4er concentrations of VC : sheets containing 0.44 and C.2C ppm of VC, in contact with 50% aqueous-ethanol,for 19 days at 49C cave concentrations of 2.4 and 1. : ppt in the simulant respectively (Diachenko et al, 1977). The model predicts the. migration will occur even if the VC level is below 1 ppm, as confirmed by the above figures. It has teen argued that in PV C containing less than 1 ppm of VC the monomer the latter is bourd to active si te s and will not migrate (Kontominas et al, 1985). The FDA (1986) did not support tlr s hypothesis. 3.2.3 Typical values for VC ip PVC articles and in food SPI-00702 Evidence has been produced shewing that the VC content of a fabricated article is influenced by each link in tie production chain (UK-NAFF, 1978). The initial VC content of the newly-produced polymer used for blending influences the residual VC content since monomer is lost during intermediate storage. The VC present in the powdering blend is influenced by the processing conditions, which ir, turr affect -21- the VC content of the fabricated article. The amount of VC detected in food simulants has been shown to be clearly related to the level in the packaging material, the storage time, an< the temperature. A number of investigations on the residual VC content in PVC packaging materials and in food or drink in contict with them have been reported in the literature during the past few years (UK -MAFF, 1974; Fuchs et al, 1975; Rosli et al, 1975; Ehtesham-Ud Din et al , 1977; UK-KAFF, 1978; van Lierop, 1979; Codex Committee, 1984). Because in this report ve are concerned with the potential cancer risk from VC in food and drink at curr ;nt levels, only the more recent data on these art discussed be lew. a) Levels of VC in polymers. A significant reduction in these levels has beer achieved since 1974. The UK-KAFF (1978, 1984) reported that whereas in 1974 only about ZC% of PVC bottles analysed contained less than 10 ppm of residual VC, by early 1977 all contained less than 1 ppm. Substantiation of this re duction has been provided by reports from the U.S. Society of the Plastics Industry (SPI, 1987) of VC levels of 10 ppb in PVC bottles and from Union Cart ide (1980) on an estimate 100 ppt VC in car coatings. b) Levels in food and drink These have also fallen substantially since 1974 as shown by the UK-KAFF (197$ , 1984). VC levels in concentrated fruit drink and cooking oil fell from eve 100 ppb in 1974 to less than 2 ppb (the limit ,if detection of the method of analysis) by 1977. A recent review (Codex Commi ttee, 1984) demonstrated, that in general, residual ^ levels in food and drink are well below 10 ppb, a maximum value which is hig^ than some of the earlier values noted above. The UK-KAFF (1987) have provided the results of a three laboratory -survey residual VC levels in food and drink undertaken during 1986. 50 samples analysed using methods of ana lysis with limits of detection ranging from 0.1 pPfl ^ 2 ppb. VC was detected in cnly 5 of these samples, the highest recorded lev being 0.04 ppb (sic) in sunflower oil and 0.74 ppb in orange drink. SPI-00703 Kontominas et al (1985) stud ed the migration of VC from PVC bottles into sel^ food simulants; distilled wa|ter, acetic acid, n-heptane and olive oil exaggerated storage conditic* s (4 months at 35C), migration of VC from the bottles containing the highest resid jal level (121 ppb) resulted in concentrations of 1.8 ppb in r-heptane and 1.7 ppb in olive oil. 3.2.4 Estimated average intake of VC by ingestion By considering the typical diily consumption of various foods and drinks by human beings, and their VC content, the average daily intake of VC can be estimated. The maximum VC intake/person/day in the UK thus was calculated to be 0.1 Mg in 1976 (UK NAFF, 1978) but less than 0.0:! Mc/d in 1978 (MAFF, 1984). The FDA (1986) arrived at similar values. They calculated the intakes from : oil, liquor and wine bottles; food packed in PVC, or VC-VDC co-polymer, film, and other sources. The upper limits of these were summed to give t total maximum daily intake of 0.025 Mg (25 ng). 3.2.f Evaluation of risk to V C by ingestion Til et al (1983) have estimat ;d the cancer risk in humans from the oral intake of VC, based on the results of a study in which rats were fed a diet containing VC as residual monomer in PVC. They used a linear model to extrapolate their results to man and estimated that at an ?ral intake of 0.4 Mg of VC per person per day the risk cf cancer was 1 in 10^. They concluded or the basis intake values at 0.1 Mg the risk of cancer was further lowered by a factor of 4. They noted that the linear extrapolation model is conservitive and concluded that : "translated into practical tkrms this means that the cancer risk of a maximum daily intake of Q.l Mg of VC/person/day is small enough to be practically neglected". The FDA (1986) also used the results of the animal studies by Til et al (1983), and together with their own estimated daily intake of 0.025 Mg/person/day, to calculate an individual lifetime risk of below 1 in 10. The linear proportional model used was said by the FDA to exaggera e the risk, and they added that "Because cf numerous conserv;tisms in the exposure estimates, 1ifetime-averaged individual exposure is expected to be substantially less than 25 ng/day". SPI-00704 -23- ECETOC (1982) has expressed its reservations about the validity of mathematical models for estimating the ri ;k of cancer in humans by : extrapolating data from animal experiments. It shou Id be emphasised that linear extrapolation models over-estimate the risk at low dose and this gives more weight to the conclusions cf Til et al and the FDA that the cancer risk from consumption of foods and drinks containing VC at present leve s is negligible. 3.2.6 Legislation affecting Ihe use of PVC for food packaging The EEC ( 1976) has published a General Directive defining health, safety and organoleptic requirements for food packaging materials. In its 1978 Directive the EC established a 1 ppm lim t for residual VC in PVC used for food contact applications, and a 10 ppb li it in the food. Further Directives, EEC (1980, 1981) describe the analytical methoc s tor the determination of VC ir. the polymer and food respectively. In the United States the Foe d and Drug Administration (FDA, 1986) has recently published new more stringent regulations on the use of PVC in contact with food, The residual VC content in the polymer should be limited to : 5 ppb in VC homo- or :o-polymer films and coatings, and plasticised PVC bottles; 10 ppb in rigid PVC; 0 ppb in water-pipes anc vinyl chloride-vinylidene chloride copolymer fi1ms. 3.3 Summary and Risk Asses sine rrt The exposure to the general public is primarily from ambient air, 0.2 Pc VC/m3 (Lahman et a 1,1978) up to 18 Ug VC/m3 in the immedicate vicinity of VC and PVC plants. Calculation of daily inhalati on rates indicates the amount inhaled ranges from 4 Mg/person/day on average to nj'ore than 100 Pg/person/day for a very small part of the population living imned'c ely adjacent to VC and PVC plants. Less than 50* of the inhaled amount is retairec and metabolised. Exposure from food sources is less than 0.1 Pg/d. SPI-00705 -24- Risk assessment should therefore concentrate on inhalation exposure, and reference is therefore made to the repcrt of the Netherlands National Health Council (1987) who estimated that continuou; exposure to 0.001 mg VC/m3 (corresponding to an inhalation dose of 20 Pg/persin/day) leads to an additional cancer mortality risk of 1 in 106 per lifetime for the general population. SPI-00706 i C. E F F E TS ON HUMAN HEALTH 1. CARCINOGENICITY This section surveys epiden iological studies of cancer incidence and c-' mortality, in occupationally OX'..posed workers. Certain terms used in reports clearly defined or consisten between publications. The following definition used in this report and, whe re necessary for clarity, the terms used by the s have been replaced by the fol owing : a) Exposure period : the per od during which workers were occupationally e/v when employed in a VC or PI C plant. b) Follow-up period : the pe riod over which the expected morbidity or nor*-' rate is calculated. It is e x pressed in person-years. c) Latency period: the perio{i from the beginning of exposure until the 6 '-` becomes apparent or death Ifrom the disease occurs. d) Observation period : the riod from the beginning of exposure of an indi / ' to the final date cf his fc How-up. e) Study period : the period from the beginning of exposure cf the first exposed in the cohort to tt e final date of the study. 1.1 Cohort Studies of Cancer Morbidity and Mortality in VC/PVC Production PI*'' Creech arc Johnson (1974) fir; t recognised the relationship between exposure and an excess incidence of rciosarcoma of the liver (ASL) in a PVC-prod'^ plant at Louisville, Kentucky . Foilcv/ing this, several cohort studies on ' exposed to VC were initiated il/orldwide. In most studies insufficient atter.ti'" paid to the collection cf inf 3nnation, the level and duration of exposure, ** the observation period was in sufficient to allow for detection of cancers w long latency period, especial ly for ether organs than the liver. Each stud, examined and judged on the- irr cmvtion provided in these areas and on the ev y/ of a relationship between VC- dxposure and cancer mortality. Tc be objective sufficier.t data are available the observed Standardised Mortality Ratios (SMF' different dose groups were tested for homogeneity and trend by the met*"/ Breslow etal (1983). SPI-00707 -26- 1.1.1 USA a) Ott et al (1975) studied a population of PVC-production employees who worked between 1942 and 1960 in a reas of potential exposure to VC. Nearly all those who had worked for at least 1 year, and about half of those who had worked for less than 1 year, were traced, group of 594 people were identified, of whom 72 were omitted because they had a so been exposed to arsenic compounds. The mortality of the remaining 522 was sttudied in relation to their exposure to VC. The C-hr TWA exposures were : from 1950-1959, almost certainly below 500 ppm (peak exposures, 4000 ppm); from 1960-1966, below 2|50 ppm (peak exposures, 500 ppm); from 1960 onwards, less than 50 ppm. The 522 workers were divide|d into 4 groups with average exposures for at least 1 month to : - more than 20C ppm (high group); - 25-200 ppm (intermediate group); - less than 25 ppm (low group); - unmeasurable (low group). There was a total of 206 in the two low groups, of whom 166 had been exposed to VC for less than 1 year. The total number of years of exposure was not given by Ott et al, but is estimated to be 4,000 from Table 8 in their publication. The start and end of the observjtior and follow-up periods were not stated but were probably 1542 (start) and 573 (er.d) for both. The number of person-years of follow-up was not presented by Ott et al but is estimated to be 13,000 based on the expected mortality. Of the 522 workers exposed to VC the mortality figures were : SPI-00708 Cause of death Observed All causes All cancer Respiratory cancer Cancer of digestive organs and peritoneun Cancer of all other sit 79 13 4 No liver angiosarcoma were i ound. Expected 89.1 16.0 5.2 4.7 6.1 .x#- This cohort was also strat ifieri or intensity, and duration observation period. Four e>posure categories were distingu'v levels : - High level exposure > 200 ppm) - Intermediate exposure (>25, <200 ppm) - Low exposure (< 25 pfm) - Unmeasured exposure. The cancer mortality data o these sub-groups are summarised v* All cancer deaths stratified on level of exposure Observed Expected High exposure group Regaining groups 5.1 10.9 The observed cancer mortaliljy in the remaining exposure grout-" lower than expected. All cancer deaths high exposure group with observation period > 15 Observed Expected Exposure duration Exposure duration Total < 1 year > 1 year 3 5 8 1.3 1.9 3.2 ** m f Throughout Section C stat stical significance testing is one-tailed poissor distribu ion on the observed number of cr-' / P-value <0.05 is considered as not significant (n.s.). //fl * SPI-00709 -ze- From the table above a clear influence of the duration of exposure on the observed/expected ratio (2.3 versus 2.6) in the high exposure group is not apparent, but might perhips rot be expected, as the numbers involved are small. b) Nicholson et al (1975, 1584) carried out a mortality follow-up study of workers in the USA with at least 5 years of exposure to VC in PVC plants. The follow-up period started 10 years <fter the onset of exposure so that the focus could be on long-term effects. Thr;e cohorts were studied : i) 256 workers in a p ant at Niagara Falls, with a follow-up period from Jan. 1956 to !ec. 1981; ii) 40 workers at the same plant, with a follow-up period from April 1974 to Dec. 1981; iii) 195 workers in a plant at South Charleston, with a follow-up period from Dec. 1966 to D9ac. 1980. The three cohorts were c orbined to one. The total number of person years of follow-up was 7062. The mo tality data are presented below : Cause of death Observed Expected Statistical Significance (P) All causes All cancer Lung cancer Brain cancer Lymphoma Liver cancer 80 28 4 1 3 10 85.6 19.7 7.3 0.76 1.14 0.42 n.s. 0.05 n.s. n. s. n.s. 0.001 Nine of the ten liver cancer deaths were caused by liver angiosarcomas. The fact, that the follow-ip period in the Nicholson study started 10 years after the onset of exposure must be taken into account when comparing with other studies in which the start of the observation and follow-up periods were much closer. At least 4910 perscr -years have to be added, to give a total of 11,972 years of observation. The av erage number of years of observation per worker was 24.3 years, a period long enough to allow for detection of cancers with a long latency period. SPI-00710 1 At the Niagara Falls plant no detailed data on exposure levels were available before 1972, measurements F>" ior to this being made solely to ensure that the explosive limit of VC in ai (greater than 3C.000 ppm) was not exceeded in plant operations. Nicholson eti 1 (1975) reported that over 50^ of the workers examined had experienced syr ptoms of dizziness, headache or euphoria during work periods, and 14 had suffe red loss of consciousness, suggesting that peak exposures may have exceeded 10,000 ppm. c) Waxweiler et al (1976) stud ied a population from 4 PVC plants in the USA which had been operating for mor^ than 15 years and had a sizeable workforce. Only people with at least 5 yearts exposure to VC, and at least 10 years employment before 31 Dec. 1973, were irtc luded in the cohort which comprised 1294 workers of which only 7 were lost to ollow-up. The follow-up period (not to be confused with the observation per|i oci of at least 10 years) represented 12,720 perscr-years. Of the cohor: of 1294 workers the mortality data are presented below : Cause of death All causes All cancer Brain cancer Respiratory cancer Lymphatic and haematopoietic sys tem cancer Biliary and liver a Observed 136 35 3 12 u 7 Expected 126.3 23.5 0.9 7.7 Statistical Significance (P) n.s. 0.05 n.s. n.s. 2.5 n.s. 0.6 0.001 When the cancer mortality c f workers with more than 15 years of observation was considered, the statistical significance of the above findings was increased as shown belcv: : Career bv site Observed All cancer Brain cancer Respiratory cancer Lymphatic and haematopoietic sv stem cancer Blliary and 1iver c ancer 31 3 11 3 7 Expected 16.9 0.6 5.7 Statistical Significance (P) 0.001 0.05 0.05 1.7 n.s. 0.1* 0.301 SPI-00711 The statistical signifi (fance of the increase of cancer mortality grows with increasing observation period except for cancer of the lymphatic and haematopoietic tissues. d) Buffler et al (1979) conducted a cohort mortality study of 464 white males employed in a VC production plant in the USA since 1948. The cohort comprised employees who had worked in the VC department for at least 2 onsecutive months between 1 Aug. 1948 and 25 Sept. 1975. All of the members of the cohort were traced for follow-up, and the follow-up period represented 5313 person-years. Expected number of deaths for the study population were calculated by applying 1950-59 end 1560-69 age-cclause specific death rates for white males in Texas to the observed distribution of person years of observation, categorised into five age groups. The mortality date ere summarised below : Cause of death All causes All cancer Respiratory cancer Observed 28 8 5 Expected 31.6 5.19 1.73 Statistical Significance (P) n.s. n.s. 0.05 Four of the 5 who died f rom respiratory cancer had a history cf smoking, but this information was rot a vailable for a large proportion (27.6%) of the cohort of 464 workers. In a subgroup of 314 work irs with a minimum observation period of 5 years the mortality data were : Cause cf death All causes All cancer Respiracory cancer Observed 22 6 4 Expected 25.18 4.34 1.49 Statistical Significance (P) n.s. n.s. n.s. This subgroup was further s tratified on level and duration of exposure and on a so-called overall exposure index, based on level and duration together. The mortality data are summaris d below for respiratory cancer. A high or low level was not precisely defined. SPI-00712 -31- Subgroups Lt>w level exposure High level exposure Respiratorv Cancer Observed Expected 1 0.82 3 0.68 Statlstlcal Significance n. s. 0.05 Exposure duration <2.29 year 0 0.45 Exposure duration >2.29 year 4 1.05 n.s. 0.05 Low exposure index High exposure Index i 0.56 3 0.94 n.s. n.s. Both a longer duration and a higher level of exposure during the first five years of observation were a ssociated with a statistical significant excess of respiratory cancer. However, when duration and level of exposure were combined in an overall exposure index the results were not significant. The authors conclude, that i i view of the absence of a significant dose-response relationship, the observed excess in respiratory cancer deaths may not be attributable to exposure to inyl chloride. However, the fact that excesses were seen in the croup with the linger duration of exposure in the initial five years suggests that a relationship may exist. e) EEh (1576). Equitable Environmental Health Inc. carried out a cohort mortality study for the US Manufacturing Chemists Association on workers employed for at least 1 year, between 1936 <ir,d 1572, in 37 VC/PVC production plants in the USA. The cohort included those fr om the other US studies described in a) - d), above. Of the 10,173 workers, 9,67 (95.1) were traced for follow-up; 2008 had had 15 or nore years exposure, and 1001 more than 20 years. The follow-up period ended on December 31, 1972, and w^s equivalent to 120,203 person-years. The mortality figures were : SPI-00713 -32- five S of nned ionse t be were ars ility t at ; usa. ; bove... ad 15 ended ality Cause of death All causes All cancer Cancer of the buccal cavity and pharynx Cancer of digestive crgans and peritoneum Liver angiosarcoma (fSL) Liver cancer (including ASL) Respiratory cancer Cancer of the genital organs Cancer of the urinary organs Leukaemia and lymphoma Miscellaneous cancers (I.C.D. 190-19S inc|l. brain) Brain cancer Observed 707 139 29 5 10 45 4 8 20 28 12 Expected 795 141.4 5.2 40.8 0.01 1.7 44.3 7.2 6.9 17.0 20.2 5.9 Statistical Significance (P) 0.001 n. s. 0.05 0.001 0.001 n.s. n. s. n.s. n.s. n.s. 0.05 The high incidence of liver cancer and liver angiosarcoma reported from the earlier investigations was c dnfirmed. The authors concluded that their results gave some support to the hypothesis that exposure to VC slightly increases the mortality due to cancer at sites other ther the liver. Thus, t he incidence of respiratory cancer and those cf TCP classification 190-199 (mi cellaneous cancers) appeared to be higher in individuals with higher levf 1 of exposure or longer periods of exposure. The miscellaneous tum.Gurs inclu bed 12 brain tumours versus 5.9 expected. The mortality of these was not ri: lated to either intensity or duration of exposure, The mortality from respiratorjy cancer was not significantly related tc the total integrated exposure dose. In addition, the Task Force examined the SMP.'s for respiratory cancer and leukaemia and lymphoma for lomogenicity and trend (by Breslow et al,19S3)in subgroups with different leve Is of exposure or with different integrated doses of exposure. No significant deviations from homogenicity nor any significant trends were found. SPI-00714 -33- Environmental Health Associates (EHA) (1986) updated this study. Loss of worker identification numbers and nor- P'i rticipation of some plants reduced the cohort to 9,200 workers and of these 72E (7.88%) were not traced. Of the 1,536 deaths, 97 death certificates (6.3% of al 1 deaths) were not available. These deaths were included in the "all death" fiour ?, but not in cause specific death number. The follow-up period of the updat ; was from January 1, 1973 to December 31, 1982 and contained about 80,000 pers on years. With the EEH period of follow up (120,203 years) the total foil >w-up was about 200,000 years; this is not mentioned in the update. The av e ra ge period of follow-up per worker was thus about 20 years; the average peri (d of employment per worker was reported to be 16 years. The mortality figures were as foV ows Cause of death Observed All causes All cancer Cancer of buccal cavity and pharynx Cancer of digestive organs Liver angiosarcoma (ASL) Liver cancer excluding ASL Biliary tract cancer Respiratory cancer Bone cancer Skin cancer Prostate cancer Bladder cancer Kidney cancer Brain and CNS cancer Lymphatic and haematopoietic system cancer Emphysema 1536 359 12 99 15 15 7 115 2 6 5 5 31 23 37 41 Expected 1705.3 341.7 11.6 89.2 0.02 3.1 2.7 122.3 1.8 7.4 15.2 8.5 9.1 12.8 36.3 22.8 Statistical Significance (P) O.OOl n. s. n.s. n. s. 0.001 0.001 0.05 n.s. n. s. n.s. n.s. n.s. n.s. 0.01 n.s. 0.001 sp,_ In addition to the effects upon he liver the mortality increase from biliary tract cancer was significant. This is the first report that biliary tract cancer appeared to be related to VC-expo ;ure . No increase of respiratory cancer or of cancer of the lymphatic and haema :opoietic tissues was found. The power of the updated study had an 8C% chance of detecting a relative risk cf 1.24 for respiratory career and of 1.45 f >r cancer of the lymphatic or haematopoietic tissues at the 5% level of signifi :ance. ... . . t' " "1 .. rTi , ____ .......... . -- .. -34- In the update, no evaluation of the historical exposure level was made, because the authors considered it subje :tive and not reliable because of lack of enouch personal monitoring data. The cohort was stratified on : - length of exposure; - time between start and erjd of follow-up; - age of first exposure; - year of first exposure; - type of plant (VCM.PVC). The authors found the SMR's in :he various strata to be significantly different from 1 at the 5% level and evafl uated the trend from visual inspection of the data. The SMR for liver cancers incre ised with increasing duration of exposure, with increasing period of follow-up and with decreasing age and calendar year of first exposure. Only two plants (VC polymerisation), involving 3,533 employees, contributed 14 ASL-cases to a to tal of 15. The authors felt this might indicate that exposure at these plants wa s different from that in others. The SMR for brain cancer was si gnificar.tly increased only in workers with more than 20 exposure years (6 versu 1.55 expected). However, brain cancer was not related to time of follow-up and the incidence increased with increasing age and year of first exposure, the reve rse of what is seen for liver cancer. What this means in terms of human risk re Tiains unclear, according to the authors. Twelve brair cancer deaths came from th a same two plants as the 14 ASL-cases. The SMR for emphysema increase^ with decreasing length of exposure and with increasing age at first exposu e, but did not change in relation to tine of follow-up or to calendar yea of first exposure. The SMR was higher in PVC-plants than VC-plants. The overall conclusion of the authors was that the study confirmed that VC workers experienced significant mortality excess from angiosarcomas, cancer of the liver and biliary tracts an| cancer of the brain and other central nervous system. The study also showed a significant higher mortality from emphysema but no excess of respiratory or lymphlatic or haematopoetic cancer. SPI-00716 T -35- This Task Force tested the SMR ; for the various strata by the Breslow et al (192) method. Only liver cancer^ showed significant trend (PC0.01) with length of exposure, time between start rid end of follow-up and age at first exposure, No other significant trend was fc|u nd. 1.1.2 United Kingdom a) Duck et al (1975) conducted a cc ort mortality study on 2,100 workers who had had any exposure to VC between 19|4 8 and 1975 v.'hile employed in VC or PVC plants, Only 7 could not be traced for follow-up and the follow-up period comprised 23,052 person-years. Of this co iort, 136 had died (expected, 142.2) and the number of deaths from cancer was 35 (36.4 expected). No case of ASL was found during the study period but 1 occp rred just outside it. There was nc increase in mortality from the common malignah t diseases. The observation and follow-up periiod (about 11 years) were too short to draw any conclusions about target organs fbjr cancer in relation to VC-exposure. b) Fox and Collier (1977) studied the mortality of workers exposed to VC at PVC plants. There was no criterion o exposure for admittance to the cohort. Of the 7561 persons identified as havi ndj started work between 1944 and 1974, 85 could not be traced for follow-up and information was lacking for a further 72. The cohort of Duck et al (1975) Was included. The number of person-years of follow-up was 75,000. Only 8l of the workers had been employed for more than 20 years. Analysis of the levels and dura tier of exposure revealed that of the-7,409 workers included in the follow-u 3, 385 were constantly, and 486 occasionally, exposed to high and medium level of VC for more than 10 years (high level TWA >200 ppm, medium level TWA > 2 < 200 ppm). The mortality figures were as follows : SPl-00717 T -36- Cause of death All causes All cancer Stomach cancer Liver cancer Lung cancer Brain cancer Lymphatic and haematopoietic cancer Observed 393 115 14 4 46 2 Expected 521.2 126.8 15.3 1.64 57.2 3.66 9.01 Statistical Significance (P) 0.001 n. s. n.s. n. s. n.s. n.s. Two of the 4 liver cancer deaths were due to liver angiosarcoma. There was no indication of excess mortality from other types of cancer. This is not surprising since the average period of follow-up was about 10 years. This study has been updated by Jones et al (in press) to provide a further 10 years of follow-up. A criterion Df 1 year of employment between 1940 and 1974 in a job or jobs with potential exposure to VCM for at least 25" of the working week was introduced, and this had the effect of reducing the male population to 5560, of whom 5498 (98.9%) were traced. As only 105 female employees qualified for inclusion, analysis was restricted to the male employees. The study period has been extended to the end af 1984, follow-up is started one year after commencing employment, and the t ota1 number of person-years is 104,000. Expected figures were calculated from ma le mortality rates for England and Wales. The mortality figures are summarised below : Cause of Tf.ath Observed Expected Statistical Significance (P) All causes All cancer Stomach cancer Liver cancer (including AS! ) Liver angiosarcoma Respiratory cancer Brain cancer Lymphatic and haematopoietic cancer Skin melanoma 780 235 26 11 7 81 4 16 2 894 228.6 23.9 1.94 0.1 92.1 6.2 12.4 1.7 0.001 n.s. n.s. 0.001 0.001 n.s. n.s. n.s. n.s. As part of this update, each em Dloyee 's work history was classified into job titles, and these job titles wer further categorised into 4 groups, autoclave workers, baggers ard driers, cr ftsmen, and other workers. All 7 cases were autoclave workers (6 from one pi ant), and the mean latency was 25 years. The SPI-00718 -J/ - V date were analysed by occu p It ion for lung cancer, brain cancer, lymphatic cancer, malignant melanoma and! cancer of the thyroid. No evidence was found of any increase in mortality ft cm these causes, that was attributable to VCM exposure. There was no eviden c e of increased mortality from non-malignant liver disease (5 observed versus .9 expected), and the incidence of deaths from respiratory disease was low (78 observed versus 105 expected) and v/as not affected by PVC dust exposure. A significant excess of blade er cancer (14 observed versus 8.0 expected, P < 0,05) was found in the total cohort. Further analysis showed that this excess could rot be considered to ble associated with VCK exposure because mortality from genito-urinary cancer was heavily concentrated in the low exposure workers at 3 of the 9 factories. According to the authors this update cerfirms the relation between exposure toi VC and liver cancer morta' ity. There was rc evidence for other cancer mortalities in relation to VC exposure. 1.1.3 Federal Republic of Gerrriary a) Frentzel-Beyme et al (1978) s ludied a cohort of 1618 workers exposed to VC while employed in VC or PVC plants, There was no exposure criterion for inclusion in the cohort. Of the German and non-German workers, 95.5 and 60% respectively were traced for follow-up. The 3 13 non-Germans all had their first exposure after 19CC. The follow-up period w. is from 1952 to December 1975, and the number of perscr-years was 19,767. Stand ard mortality rates were estimated from the vital statistics of the population of the Rhinehessia Palatinate over the years 1970-1975. The mortality figu es were : Mortality from Oberserved Expected All causes All cancer Cancer of the digestive organs St on,a cl. cancer Colon cancer Respiratory cancer 81 18 12 3 4 6 90.8 16.8* 8.8 2.4 1.1 5.1 + estimated figures from tables, net presented by the authors. Statistical Significance (P) n.s. n.s. n.s. n.s. 0.05 n.s. SPI-00719 -38- The absence of deaths from mSL may, according to the authors, be due to the relatively low levels of expo ure to VC at the plants concerned. b) Reinl et al (1979) and Greis er et al (1982). A mortality follow-up study was carried out on all workers ei" ployed before 31 December 1974 in 11 VC and PVC plants in the FRG. In parall el , a cohort of workers with no exposure to VC and another of workers from PVC processing plants were studied. The first group comprised 7021 German and Au striae workers whose combined follow-up period represented 73,734 person-yea rs up to 31 December 1974 (882 workers of other nationalities, of whom S ha died, were omitted from the group). Standard mortality rates were estima ;ed from the vital statistics of the national population from I960 to 1974 (R einl et al,1979). The mortality figures were : Cause of death All causes All cancer Cancer of Che digestive organs Stomach cancer Colon cancer Liver cancer Respiratory Cancer Bladder cancer Brain cancer Cancer of lymphatic and haematopoietic tissues Observed 414 94 45 18 6 12 24 1 "I 15 Expecced 434.7 90.6 32.7 14.4 5.8 0.9 26.6 2.9 1.3 7.7 Statistical Significance (P) n.s. n. s. 0.05 n.s. n.s. 0.001 n. s. n.s. n.s. 0.05 There were cases of ASL among hose of liver cancer but the number could not be stated because of a lack of adequate identification in the early years of PVC production. The statistical probability that the increased mortality from liver cancers was due to chance is 3 x lO'10, providing further evidence for a casual relationship between liver caln cer and exposure to VC. Mortality from liver cancers was elevated in the ot|i er two cohorts, but to a much lesser extent. The increased mortality from cance of the lymphatic and haematopoietic tissues is also statistically significant (P < 0.05). In the other two cohorts, deaths from this cancer were lower than exple cted. Although r.o information on 1 =vel of exposure was provided in this study, Marsteller et _aj H?75) bad reviously reported that workers in German PVC plants, especially autoclave c eaners, had often suffered pre-narcotic effects due to exposure to VC. SPI-00720 -39- Greiser et al (1982) presented observed and expected mortalities from career for subgroups of the VC/PVC-cohor|t exposed to VC during <1, 1-5, 5-10 and >10 years. ECETOC tested the SKR of liver cancer, of respiratory cancer and of cancer of the lymphatic and ha ^matopoietic tissues for homogeneity and trend by the Breslow etal (1983) method. The SFR' s did not show any significant deviation from homogeneity nor any significant trend with exposure-duration (P < 0.05). 1.1.4 Italy Bertazzi et al (1979) studied male workers employed in VC and PVC plants since they started operating in 1952, Of the 5^41 workers with at least 6 months exposure to VC, 4777 (88%) were traced for follow-up. The follow-up period was from 1952 to 1975 and is estimated by the pres ant authors to represent 23,000 person-years. The mortality figures were : total deaths, 62 (ltl e.xpected); deaths from cancer, 30 (30.9 expected). Of the deaths from cancer, 8 we re from liver cancers. These included 3 from ASL, 2 of which were of workers et PVC plant and 1 of a worker in a PVC-processing plant (his employment history was uncertain). The average follow-up per worker was 4.2 years, a period too sf ort to allow a study of cancer of other organs, This paper was essentially a di scussion of individuals clinical symptoms and the above data were extracted from the narrative. A partial update of this study was presented by Belli et al (1986). From the VC-PVC plant at Ravenna (sta tfted 1959), the VC plant at Rosignano (started 1953), and the PVC plant at Fe *rara (started 1953) a cohort of 1263 workers was identified, who had more than 6 months exposure to VC. Seven workers were not traced. The expected mortality was calculated from the statistics for the total Italian population. The follow up period was from the start of production until December .31 , 1983 for the Ra\|e nna plant and until December 31, 1984 for the Ferrara and Rosignano plants, The number of person years of follow-up was 22,3 and so the average period of fcllow-up per worker was 17.7 years. SPI-00721 There were C3 mortalities (102 6 expected) and initially 36 cancer deaths (26.? expected). Of these cancer dea :hs 2 were due to liver cancer (C.d expected) and 12 to lung cancer (7.6 expect'|d). After careful examination by the authors of -HU- the available clinical and pathological evidence these figures changed to 39 for all cancer deaths, 5 for liver oancer deaths and 13 for lung cancer death. These liver and lung cancer mortali ties represent a significant excess at the 5" level. This excess in all anc specific cancer deaths arose mainly from the cohort of the PVC-plant at Ferr jra, all cancer 26 (14.5 expected), liver cancer 4 (0.1 expected) and lung cancel- 10 (4.1 expected). The careful examination of the clinical and pathological evidence in the study population introduced observer )ias since the same evaluation could not be made for the reference population. Fji rther, the lung cancer mortality in the area of Ferrara seems to be higher than in the general Italian population. Finally, the report does not provide informa :ion on VC exposure levels, on possible exposure to other carcinogenic compounds or on smoking habits. The study does confirm the relation between liver cancer a r d VC-exposure. 1.1.5 Sweden Byren et al (1976) studied a cohor t of 771 workers employed in VC/PVC production from the early 1940s to October 1974. There was no exposure criterion for admittance to the cohort. Twenty-ore workers with a very short period of employment could not be traced and were omittqd The total time of employment of the remaining 75C workers was 6300 person-years and the follow-up period amounted to 12,000 person-years. The mortality findingjs were : Cause of death O iserved Expected Statistical Significance (P) All causes All cancer Lung cancer Liver, pancreas cancer Brain cancer 58 11 3 4 2 Not provided Not provided 1.78 0.97. 0.33 n.s. 0.05 0.05 Two of the 4 cases of liver and pa ncreas cancer were liver angiosarcoma. Only 112 workers in the study group had beer exposed for more than 10 years and conclusions about cancer in other target organ s than the liver cannot be derived from such a small group. SPI-00722 T -41- 1.1.6 France Pierre et al (1979) studied the mortality and cancer incidence in a population of workers in VC and PVC plants in lavaux. Of the 1,482 workforce, 160 were lost to follow-up and together with the 1 women in the population were excluded from the follow-up analysis. The remaining 1311 workers were followed from 1953 until 31 December 1976. There were 15,458 )erson-years of follow-up. The number of workers exposed to VC for more than 5 years was 872. They had been heavily exposed (81 for more than 5 years and 44 for more :han 15 years). The mortality findings were : total deaths 25 (46.7 expected); - deaths from cancer, 8 (8^63 expected). There was 1 death (0.0015 estima tied expectance) from ASL of an autoclave cleaner with more than 15 years of high exposure. The rather low observed and expected mortality in this cohort is attri :uted to the low average age of the population, which makes it difficult to draw conclusions from this study. 1.1.7 Canada Theriault and Allard (1981) performled a cohort mortality study on 451 workers at a PVC plant with more then 5 years exposure to VC between 1 January 1948 and 31 December 1972. The follow-up per iod ended cn 31 December 1977. The average exposure-period of the cohort was 17.4 years and the average observation period 22.6 years. Mortality was as folio'; Cruse of death All causes All cancers Cancer of digestive organs Liver cancer (all ASL) Respiratory cancer Cancer of bone, skin, and connective tissue Cancer of eye, CN'S Leukaemia, lymphoma Observed 59 20 14 8 2 2 0 1 Expected 71.1 16.4 5.4 0.14 5.80 0.38 0.60 1.67 Statistical Significance (P) n.s. n.s. 0.01 0.001 n.s. n.s. n.s. n.s. SPI-00723 Although the average exposure per od and the average observation period were much longer than in nearly all other pr esented studies, there was r.c evidence of excess mortal'ty from cancers other than of the i ver. The authors suggest the possibil ty that some of the men who died of liver cancer, might have developed lung cance r had they lived longer. However, the average latency time of the liver cance rs was 19.6 years and this seems long enough to allow for the development of canc er in other target organs. Theriault (1982) presented an update of this study, in which the follow-up ended on January 31, 1981. Eleven men :ould not be traced. The mortality figures are summarised below : Cause of death All causes All cancers Cancer of digestive organs Liver cancer (all ASL) Respiratory cancer Cancer of bone, skin, and connective tissue Cancer of eye, CNS Leukaemia, lymphoma Observed 73 23 15 8 4 2 0 1 Expected 91.1 21.5 0.16 0.16 7.70 0.50 0.57 2.07 Statistical Significance (P) 0.05 n. s. 0.01 0.001 n.s. n.s. n.s. n.s. The results of this updated stud. t confirm the association between VC-exposure and liver angiosarcoma incidence, but does not support an association betv/een VC-exposure and the incidence of pespiratory or of brain cancer. 1.1.8 Norway Heldaas et al (1984) studied ijhe overall mortality from all causes and the incidence of cancer in 454 male employees who had worked for more than 1 year between 1950 and 1969 at a VC/PVC plant. The follow-up period was from 1953 to the end of 1979 and involved 8676 per:; on-years. During this period 50 workers h d died (59.3 expected). This difference between observed and expected mortality wis not statistically significant. This is the only cohort study, which presents ca ncer morbidity instead of cancer mortality. The cancer cases arc summarised below SPI-00724 -43- New cases of cancer All sites Lung Skin melanoma Colon Thyroid gland Liver angiosarcoma < >b served 23 5 4 3 2 1 Expected 20.2 2.8 0.8 1.44 0.16 0.001 Statistical Slgnlf1cance (P) n. s. n. s. 0.01 n. s. 0.01 0.001 The task force noted that the excess of different cancers was seen in the group with highest estimated exposur level and this observation might suggest an association between vinyl chloriijfe exposure and these types of cancers. However, a significant trend of cancer incii Jence with level of exposure, years of observation or integrated exposure dose coulM not be established by ECETOC for any cancer type according to the method of Breslcw et al (1983). The absence of a significant trend may be caused by the small cohori. and the small observed incidences. Recently, Heldaas et al (1987) presented an update of this study, in which the' follow-up period was extended w th the period 1980-1984. The study population in this second period consisted o|f 430 male workers. A number of 1809 years of follow-up were added tc the pre ious 8183, resulting into a total of 9992 years, The new cases of cancer for the otal follow-up period were as follows : New cases of cancer Lung Skin melanoma Colon Observed 7 6 5 Expected 4.0 1.1 2.1 Statistical Significance (P) n.s. 0.001 n.s. All sites (subgroup) 31 24.0 n.s. The Task Force tested again the cancer incidences on trend in relation to level of exposure and in relation to the number of years from first employment. A significant trend could not be e >tablished. 1.1.9 Japan a) Masuda (1979) studied a group of 305 workers who had been exposed to VC at a VC/PVC plant for ?.t least 1 y ear in the period 1949-1975. The author noted that after 1961 the maximum level of exposure of autoclave cleaners was below 250 ppm, but may have beer much higher before this. Only 1 person was lost to follow-up, and the number of person-years of follow-up was 4,777. Masuda found : SPI-00725 -44- 27 deaths (26.5 expected); 8 deaths from cancer (5.8 expect ed), but no significant increase in any specific type of cancer nortality. b) Nakamura (1983) conducted a (fohort mortality study on male employees who had worked for at least 1 year befb re 31 December 1964 ir one of 25 plants which had begun to produce VC and/or P/ C before 1965. The follow-up period was from 1 January 1950 to 31 October 1975 . This gave a cohort of 4,524 workers of whom 29 were lost to follow-up. It is not clear whether the Kasuda cohort (see above) was included. Nakamura's cohorlt was divided into PVC-workers (2546) and others, the former group representp ng about 41,500 person-years of follow-up. Twenty-eight percent of this g|roup was observed for more than 20 years. In the group of PVC-workers the morta ity data were as follows : Cause of death All causes All cancers Stomach cancer Liver cancer Pancreatic cancer Lung cancer Other cancers Obsei ved 128 37 16 6 3 2 10 Expected 147.6 26.9 11.9 2.54 1.05 2.33 9.04 Statistical Significance (P) n.s. 0.05 n.s. 0.05 n.s. n.s. n.s. The SMR of liver cancer rose wi th increasing duration of exposure, but that of all the remaining types of canc er did not. Of the 6 cases of liver cancer, 1 was diagnosed as ASL and it is hi gily probable that there was one other. A further case, recorded as death from n unspecified cancer, was also probably one of ASL. There was no correlation between deaths from liver cancer and length of employment as an autoclave clea ner. In the group of non-PVC workers there was no increase in deaths from any s pe: ific cause. 1.1.10 Soviet Union Filatova etal (1982) and Fedotove (1983) reported a retrospective cohort study of all Soviet workers exposed to VC between 1939 and 1977. The cohort of 3,232 was dividied into 3 groups : - those exposed to more than 1 15 ppm VC, and often to thousands of ppm. This group probably started work involving exposure to VC between 1939 and 1949; - those exposed to botv/een 11 and 115 ppm VC, and probably starting work between 1950 and 1959; SPI-00726 -45- - those exposed to less than 11 ppm VC and starting work after 1959. The authors presented no data or total deaths, deaths from cancer or the number of workers in the groups. SMR's were estimated from the vital statistics of the urban and total population of (probably) the USSR for 1959, 1969 and 1975, but the values in the two papers are not identical. However, the following information was drawn from these publications : - of the total deaths from cancer, 31.7% were due to cancer of the stomach, 26.S?o were due to cancer ol the lung and 14.3* to cancer of the lymphatic and haematopoietic tissues; - ir. mer, there was an inc rease in death from cancer of the lung and the lymphatic and haematcpciet c tissues (SMR's 1.5 ana 2.4 respectively); - in women there was an incre ase in deaths from cancer of the stomach, colon and the lymphatic and haematopoietic tissues. The increases in deaths from cancer seemed to be related to tie level but not duration of exposure, this being clear in the first of the ajbove groups with the highest exposure. In the third group, with the lowest ex osure the mortality from cancer was comparable to that of the normal populati on. The authors stated that "liver angiosarcomas were not observed. However, a retrospective analysis of disease cases revealed 71 c< ses of chronic hepatitis. After these workers had been removed from exposure o VC, the majority was found to be free from changes in the liver" (Fila ;ova et al, 1982). It is uncertain how this opinion was obtained. 1.2 Studies of Workers in Poly1 <_inyl Chloride Processing Studies of workers employed n PVC-processing may provide information on the effects on health of exposure to low levels of VC. During the conversion of PVC into fabricated articles some of the residual VC is released from the polymer but the resulting exposure levels wiill clearly be much lower than those at VC/PVC production plants. SPI-00727 -46- 1.2.1 USA a) Chiazze et al (1977) carried out a proportional mortality study on 4,341 workers who had died during the peri cd 1964-1973 among current and former employees at 17 PVC-processing companies w ith a total of 55 plants. Data on all deaths were obtained because it was not possible to identify those workers who had been exposed to VC only. The to'; al number of employees could not be precisely determined but was estimated to have been between 65,000 and 70,000 by the end of 1973. The deceased employee s only included those in the following categories : - those who died during empl oyme n t; - those who died after reti ning from the company, with retirement benefits; - those who died after te iininating employment but who were in a company life-insurance plan. Proportional mortality rates PMRs) for various causes of death were estimated from the vital statistics of 11 e US and adjusted for age, race, sex and year. In the calculation of PMRs the e xpected number of deaths from specific causes in the study population was obtai ned from the relative frequencies of these causes by age and by calendar year i n a comparison population. Among the 3,248 white males and 601 white female emf 1 oyees who had died there was an excess of deaths from cancer, ie. for men, 666 (562 expected) and for women, 181 (138 expected), The causes of cancer death are presented below : Cause of death Among Men All cancer Cancer of buccal cavity and pharynx Cancer of digestive system Stomach cancer Colon cancer Rectum cancer Liver cancer Respiratory cancer Cancer of bone. skin and connective tissue Cancer of genital organs Cancer of urinary organs Brain, CNS cancer Lymphomas Leukaemias Other cancers Observed 666 15 209 41 73 25 6 205 24 44 36 16 42 19 56 Expected 56? 16.9 162.1 31.5 52.5 19.6 4.2 176.7 15.1 52.8 33.1 13.9 32.4 '23.4 34.6 Statistical Significance (P) 0.001 n.s. 0.001 n.s. 0.01 n.s. n.s. 0.05 0.05 n.s. n.s. n.s. n.s. n.s. 0.001 SPI -47- Cause of death Among Women Observed All cancer Cancer of buccal cavity ar.d pharynx Cancer of digestive system Stomach cancer Colon cancer Rectum cancer Liver cancer Respiratory cancer Cancer of bone, skin connective tissue Breast cancer Cancer of genital organs Cancer of urinary organs Brain, CNS cancer Lymphomas Leukaemias Other cancers 181 3 53 8 24 8 0 12 7 44 19 11 4 9 1 18 No case of death from ASL was ound. Expected 137.7 1.9 35.3 5.0 15.3 3.9 0.63 11.9 3.3 32.4 23.3 4.5 3.5 7.6 5.0 9.6 Statistical Significance (P) 0.001 n. s. 0.01 n. s. 0.05 0.05 n. s. n. s. 0.05 0.05 n. s. 0.01 n. s. n.s. 0.05 0.01 This PMP study must be inter freted with caution because the "healthy worker" effect which produces a low Mortality rate for non-malignant diseases in the population studied could accoji nt completely for the increases in the PMRs of death from cancers. b) Chiazze et al (1980)' made a case-control analysis of the above-mentioned 44 deaths from breast cancer, in comparison with 134 control subjects matched for age and selected from 5 PVC-processing companies. When such variables as extent of exposure, length of employment, continuous versus intermittent employment, marriage status and child-bearing history were compared for the cases and the controls, no statistically-significant increase in relative risk was found. However, given the number of cases and controls, the smallest relative risk which could have been detectedl was 3. SPI-00729 -48- 1.2.2 Sweden Molina et al (1981) studied a :ohort of 2,073 workers with at least 3 months employment between 1S45 and 31 Dt cember 1974 in 4 PVC-processing plants. 103 could not be traced to follow-up, the period of which was from 1 January 1961 to 31 December 1976. Standard morta ity rates were estimated from the national statistics. The presentation of the data is rather confusing. Somewhat ambiguous data are presented on a sub-group of 1,771 workers with at least 6 months exposure between 1945 and 1974, excluding those whose exposure ceased before 1961. Of these 1,771 persons, 73 died between 1969 a id 1976 (87.7 expected). In a longer follow-up period between 1961 and 1976, i.here were 51 deaths from cancer in the group (expected, 44.6), of which 11 were from cancers of the digestive organs including 1 1iver cancer. 1.2.3 Federal Republic of Germany Greiser et al (1982) made a cohort mortality study of 4,007 workers at 2 PVC-processing units, employment c|t which was the only criterion for inclusion in the cohort. 92.1* were traced for follow-up, the follow-up period being from 1944 to 31 December 1974 and representi ig 52,896 person-years. The mortality findings are given b<:low and compared with those from Greiser's study of workers in VC/PVC production pi ,mts - see section 1.1.3 b), above. SPI-00730 -49- Cohorts of Greiscr 0982) PVC-?rocessiny VC/PVC-Product ion Number of workers 4,007 7,021 Persons years of follow-up 52,896 73,734 % follow-up 92.1 93.2 Cause of death Obs. Exp. Statistical Significance Obs. Exp. Statistical Significance (P) All causes All cancer 360 379.6 62 81.9 Cancer of the digestive organs 15 30.3 Stomach cancer 7 13.5 Colon cancer 1 5.3 Liver cancer 3 0.8 Respiratory cancer 25 24.4 Bladder cancer 2 2.8 Brain cancer 5 1.1 Cancer of the lymphatic and haematopoietic 2 6.3 tissues n.s. 0.01 0.01 0.05 0.05 0.05 n.s. n.s. 0.01 0.05 414 434.7 94 90.6 45 32.7 18 14.4 6 5.8 12 0.9 24 26.6 1 2.9 2 1.3 15 7.7 n.s. n.s. 0.01 n.s. n.s. 0.001 n.s. n.s. n. s. 0.05 The significant deviations from expected in the PVC -processing cohort are mainly based on less deaths than expect id. except the deaths of liver and brain cancer. In contrast, all significant f indi lgs i n the VC/PVC production cohort are based on more cancer deaths than expected 1.3 Evaluation of the Epidemiol oqical Studies In these studies excess mortali ty from liver cancer was frequently found and an excess of deaths from cancer o : the respiratory tract, brain, thyroid and skin (malignant melanoma), and of the lymphatic and haematopoietic tissues was sometimes reported. The relationship betwe ;n deaths from cancer in the various target organs and exposure to VC is discussed ielow. iSsSNe*!* SPI-00731 W -50- 1.3.1 Liver cancer Angiosarcoma of the liver is s jch a rare disease that it proved much easier to establish that there was an exc ess incidence of it in a study than was the case with the more common cancers. In 12 of the 20 studies review(d there was a statistically significant excess of deaths due to liver cancer in W 3 rkers exposed to VC and the relationship of this form of cancer to exposure to \C is thus well-established. In the updated cohort mortality study (E.H.A., 1986) )f all VC- and PVC-workers in the USA VC-exposure appeared to increase cancer mo Iity from liver angiosarcoma, from other liver cancers and from biliary tract c ancer. This last finding is the first and only one relating biliary tract cancer mo tality to VC-exposure. It is of interest to consider v hy 8 of the studies failed to show an excess of deaths from primary liver ca ncers. The average latency period for ASL in occupationally-exposed workers i between 20 and 29 years (Forman et al, 1985), and thus the average observation ti rfe was clearly too short in the Masuda (1979) and Buffler et al (1979) studies. The integrated dose of exposure no peak exposure were probably less in the studies by Ott et__ aj_ (1975), Buffle et__al_ ( 1979), Chiazze et__al_ (1977, 1980), Frentzel-Beyme et al (1978) and Greiser et al (1982, cohort of PVC-processors). There is no evident explanatior for the absence of an excess mortality from liver cancer? in the studies by Filatov a et al (1982) and Fedotova (1983). fhe study by Ott et al is compTjtely-comparable in terms of cohort size, average exposure period, average observ; tion period and type cf plant (PVC production), with that of Nicolson et al (1975 , 1984) and yet Ott found no excess mortality from liver cancers. This is probably because peak exposure and integrated exposure of the Ott population were clearly less than those of the Nichclscn population. This ndicatcs that the likelihood that humans develop ASL falls significantly with decreasing peak and integrated ex posure to VC. 1.3.2 Respiratory cancer The mortality from this was signi ficantly increased in the studies of Buffler et al (1979), Heldaas et al (1984), W. xweiler et al ( 1976), Filatova et al ( 1982) and SPI-00732 -51- Fedotova (1983). The level of exposure, as deduced from the rates of mortality from ASL, was moderate for the populations of Buffler et al and Heldaas et al, and high for that of Waxweiler et al . Of 'hree cohorts which had experienced high exposure to VC, those cf Nicolson et al (1975, 1984) and Theriault and Allard (1981) showed a decrease in mortality from cance of the respiratory tract, while in the study cf Greiser et al (1982) it was norn al. Beaumont and Ereslow (1981) considered the statistical power of 8 for the epidemiological studies to detect an increased relative risk of respiratory canc ?r of 1.5 in workers exposed to VC. The only 2 studies of high statistical power (EEH, 1978 and Fox and Collier, 1977) failed to show an increase in relative risk. In addition, in the updated stud ies of VC-workers in the United States and the United Kingdom (Environrrertel Heal th Associates 1986, and Jones 1986) the power to detect a relative risk of respect- vely 1.24 and 1.27 at a significance level of 5% was 80S. However, in the US-VC-workers there were 115 respiratory cancer deaths (122.5 expected) a relative risk 0.94, and in the UK-VC-workers 85 respiratory cancer deaths (94.3 expected) a r lative risk 0.90. Taken overall, these findings do not suggest that there is a cauls al relationship between exposure to VC and death from respiratory cancer. Some of V) e excess mortality found may be due to smoking. 1.3.3 Brain cancer SPI-00733 An increase in death from brain cancer was found in the cohorts of EEH (1978), Byren et__fil_ (1976) and Waxweil ;r etal (1976). Beaumont and Breslow (1981) calculated the statistical power to detect an increase in the relative risk of brain cancer for 5 of the epidemi ological studies. Those of Waxweiler (1976) and EEH (1978) had sufficient power detect a relative risk of 2-3, and an increase in brain cancer was found in thes studies and also in that of Byren et al (1976). Beaumont and Breslow concluded t!i at therefore the most reasonable interpretation is, that the data are consistent ith an aetiological hypothesis for vinyl chloride and cancer of the brain. However in the EEH study the increase in this type of cancer was net related to the le /el or duration of exposure and, in addition, no such increase was found by Nicho son et al (1975, 1984), Greiser et al (1982) or Theriault et al (1981) in whose s tudies the exposure levels were very high. In the update of the EEH study ( Environmental Health Associates, 1986) 23 deaths from brain cancers were found aga inst 12.76 expected. Although this is considered as highly significant by the autli ors, this task force could not establish by the method of Breslow et al (1983) a significant trend of the SKR with length of exposure, with the length cf the fe riod of follow-up or with age of first exposure, as was found for deaths from live r cancer. In the updated study of VC-workers in the UK (Jones, 1986) the power to detect a relative risk of brain cancer of 2.2 at the 5% level was 80%. Only 4 bra i n cancer deaths were found against 6.18 expected (relative risk 0.65). It should also be noted that stati sties on brain cancer for the general population are of doubtful reliability becau se of the limited number of brain post-mortem examinations. Therefore comparisc n between observed and expected cases is of limited validity. There is thus a serious doubt about a possible relationship between exposure to VC and death r$ suiting from brain cancer. 1.3.4 Cancer of the lymphatic and haematopoietic tissues Increased deaths from these type of cancer were reported by Waxweiler et al (1976), Greiser et al (1982), Filatova et al (1982)/Fedotova (1983; same study) and Nicolson et al (1984). The most significant increase was in the VC/PVC-production cohort of Greiser et al (15 found against 7.7 expected) but there may be some doubt about the expected figure as the German statistics for the single year 1968 had to be used for the period 1944-1968. It is possible that reliable figures were not available prior to 1968 because of the influence cf the Second World War on German demographic parameters. Greiser et__a^ (1562) stratified the VC/PVC production cohort on the basis of duration of exposure. ECETOC tested the trend of the SMR's in the subgroups in relation to durition of exposure by the method of Breslow et al (1983). A significant trend could n ;t be established. In the EEH population, 2G deaths rom cancer of the lymphatic cr haematopoietic tissues were found against 17.0 ecpected and there was no relationship between mortality and the level of exposu re or integrated ' dose of VC. In the update (Environmental Health Associates, 1)86) 37 deaths from lymphatic or haematopoietic cancer were observed against 36.28 t xpected. The power to detect a relative risk of 1.45 at 5% significance was 80% significant trend of the SMR with length of exposure, length of period of follow-up or with age of first exposure could not be established by ECETOC by the method of Breslow et al (1983). In the updated study of VC-workers in the UK the power to detect a relative risk of lymphatic and haematopoietic cancer of 1.8 was 0%. 16 deaths from this type of cancer were observed against 12.36 expected (r ilative risk 1.3). Information about exposure SPI-00734 1! -53- conditions was not sufficient to trace a relation between VC-exposure with this type of cancer mortality. Nicholson et al (1980) and Waxy eiler et al (1976) found, respectively, 3 deaths (1.14 expected) and 4 deaths (2 5 expected) from such cancers but did not relate them to exposure. Filatova et al (1982) observed a significant increase in the SMR (=12) for this type of cancer i i female workers, a finding which is difficult to assess because the authors give r o absolute cr expected values of cancer deaths. In the remaininc studies no increase was noted. Overall, there is equivocal evi d|ence, that a relationship exists between exposure to VC and death from cancer of th e lymphatic and haematopoietic tissues. 1.3.5 Malignant melanoma The cohort, cancer incidence stu Jy by Heldaas et al (1984), in which an increased incidence of this type of skin cfe ncer was found, is not directly comparable to the other studies which were of cancer mortality. In the updated EEH-study (Environmental Health Associatefc , 1986) 6 skin melanoma deaths were observed against 7.36 expected. The power of this study to detect a relative risk of skin melanoma deaths of 2.1 at a 5% si orificance level was 80%. There are numerous examples in t he literature of the finding of an excess incidence of malignant melanomas, in stu dies of occupational cohorts in a variety of industries and countries, as ex enplified below for workers not exposed to VC. An increase in melanoma morbidity or mortality has beer, observed : in Norwegian workers exposed to bestos, 3 melanoma cases (0.6 expected) (Hilt et al,1983); in Swedish rubber workers, 21 mel anoma cases (9.1 expected) (Holmberg et al,1983); in Swedish workers in the telecom tunications industry, 12 melanoma cases (4.6 expected) (Vagero et al ,19651; in Swedish electrical engineers, 3 melanoma deaths (0.9 expected) (Olin et_al,1985); in English semiconductor workers, 3 melanoma cases (0.68 expected)(Sorahan et al,1985); in workers at 8 UK oil refineries 14 melanoma deaths (6.48 expected) (Alderson and Rushton, 1982 V, In employees of the Lawrence Live rmore National Laboratory, a high energy physics research facility in California, where the re were 7 melanoma cases in females (1.43 expected), and 22 in males (6.80 expected) (Reynolds e t'al.1985). SPI-00735 -54- In all these cohort studies, no s flecific chemical is implicated as the causal agent of malignant melanoma of the skin and it seems that this can arise in a study as a purely coincidental finding. There seems to be no doubt thn t solar radiation is a causal factor in the development of malignant melanoma (Magnus, 1981). Scandinavian people have a higher risk of developing melanoma (Amst "onq, 1984); it is suggested that infrequent but intense exposure to sunlight co r|tributes more to the risk of melanoma of the fair-skinned Scandinavians that dloes continuous exposure of people in the more southern countries. This could ex plain the results of Heldaas et al if PVC-workers had relatively more exposure to s inshine than did the general population. However, the 4 workers with malignant me anoma did not have any excessive exposure to sunshine nor were other skin tumours knowr to be related to sunshine found (Heldaas, personal communication), Therefore, the finding of Heldaas et al deserves more attention in future studies. Up until now the total evidence from all the studies dees rot indicate a causal relationship between exposure to VC and the incidence of, or mortality from, malignant melanoma. 1.3.6 Cancer of the thyroid The increased incidence of this t,ype of cancer found by Heldaas et al (1984) was not paralleled by an increase in nrortality in the studies by Nicholson et al (1976) and Greiser et al (1982) in which there were high exposures to VC and long periods of observation and follow-up. Thus the evidence for a causal relationship between exposure to VC and cancer of the thyroid is weak, especially since Heldaas et al found only 2 cancers of that type. 1.4 The Register of Angiosarcoma (f the Liver (ASL) Cases Since 1974, ICI Pic has maintained, on behalf of the Association of Plastics Manufacturers in Europe, a worldwice register of histologically confirmed ASL cases resulting from exposure to VC (F Drman et al, 1985). The register is doubtless incomplete, although the collabora ting organisations believe that they have been informed about nearly all of the cases. By January 1986, 120 men, of whom three were still alive on 1 January, 1986 , had been recorded as having VC-related ASL. Of the 120 cases, 119 were reported tc be hepatoangiosarcomas and one a SPI-00736 -55- cholangiosarcoma. No cases were recorded in women but few women have been employed in manufacturing PVC. Table 4 shows the breakdown of the cases by date of death, It should be noted that the regis ter was begun in 1974 when exposure to VC first became firmly established as a c ause of ASL. Under-reporting is therefore less likely tc be a problem after this date, when special attention began to be paid to the disease. In Table 5 the age-distribution of the 120 men at the time of diagnosis is given. The mean age at diagnosis was 52. Tt seems that the incidence of ASL reached a peak 20 - 29 years after first exposure , but 18 cases occurred after more than 30 years. The average latency period for the ASL cases was 22.5 years, and the average length of exposure 18.3 years. The major ty of cases occurred 15 to 29 years after first exposure, as is common with occup altionally-induced cancers. There has been no case of ASL recorded in any individual who has been exposed since the levels of VC in the atmosphere were drastically re i uced in 1974. In Table 6 ere shown the princi pal occupations of the 120 affected men. At least 43? of the men (53/120) had beer employed as autoclave cleaners, an occupation which was almost certainly associ ited with the highest exposure to VC. Only 82 new cases have been diagnosed since 1975 despite the large number of workers in the industry. Forman et al (1985) have estimated the number of future cases of VC-related ASL by analysing the 110 cases diagnosed at the time, in terms of latency, and assuming that the hazard was eliminated in :he late 196C's. They estimate that a further 155 cases might occur for latencies o- less than 35 years, and in the absence of data for latencies greater than 35 yec rs, they suggest that there might be around 50 cases for these extended latency deriods. Thus the total estimate for future cases is in the range 200-250, or about :wice as many as have already been reported. 1.5 Angiosarcoma of the Liver in ihe General Population The incidence of angiosarcoma of 11 e liver is difficult to estimate as it is a rare condition. Apart form VC, thorium iioxide (in the x-ray contrast medium Thorotrast) and arsenic can induce ASL (Roth, 1955; Tesluk et al, 1955). Edmonson (1958) found only one case in 52.CC0 autopsies (0.002?), Rein and Huth (1975) report on 6 cases (0.021c') of primary vascular liver tumours in 2C,C7S autopsies, while Brady et cl SPI-00737 -56- (1977) give 0.25 per mill ion as the annual morbidity incidence rate for histologically confirmed angiosarcoma of the liver among residents of New York State (excluding New York City) for the years 1970 to 1975. In a later and more extensive study of the same geographical area Vianna et al (1981) conclude to an average annual incidence of 0.25 per million. Baxter et al (1977) report that between 1963-1973 an average of f)ur cases per year of ASL were recorded in the UK, while a panel agreed with these d agnoses only in a third of the reported cases. In their case-control study over the period 1958 to 1975, Brady et al (1977) report on 27 cases with confirmed ASL. Ore case was excluded because histologic assessment could not confirm ASL. Seven of the patients had documented exposure to VC (all chemical operators). Thorium dioxide (Thorotrast) or arsenical pesticides (Bradey et al, 1977). Of the other 19 patients, 5 lived in highly industrialised areas no more than 1350 m from VC fabrication or polymerisation factories. Neither residential history nor ambient air concentration data were obtained for the remaining 14 patients and statist cal analysis for a geopgraphical factor was not carried out. The authors conclude that the geographical factor might be important in the etiology of this disorder )ut it must be considered unresolved at present. Vianna et al studied the same area and expanded the study by 4 more years (1958 to 1979) including the New York Cit, cases from 1973 through 1979. A total of 43 patients diagnosed with ASL were reported. For five females residing within one mile of plastic producing factories for a period ranging from 8-62 years (medium 26.8 years', VC is mentioned as pesssible cause. Furthermore, it is noted that 19 of the 43 patients resided in four urban counties which contain the greatest number of VC polymerisation and fabricatiipn factories. Dalderup et al (1976) investigated all cases diagnosed as ASL ir. Th< Netherlands in the period 1950-1975. In that period the population increased from about 10 to 14 millions. From the 27 cases reported bj-' the pathologists only cases were proven angiosarcoma. None of the 27 cases had any traceable contact wi th vinyl chloride and a geographical factor v:as not found. One of the cases was prabably induced by chronic use of arseric drugs. Thorotrast induced cases were not ncluded in this series. Saric et al (1976) studied the inc dence of malignant primary tumours of the lung, bronchus and liver in 1968-1971 in a city in Yugoslavia with several factories including a PVC industry. There vas no causal relationship between the liver tumours recorded and the area of reiidence and there were no haemangiosarcomas. SPI-00738 -57- Of the 14 cases of ASL in the per ods 1963-1973 in England and Wales, and 1965-1973 in Scotland, which were confirmed by a panel, one case had occupational exposure to VCM. One other case was a man whc had lived for 6-7 years before his death in 1970 within half a mile of the PVC pla nt in which the occupationally exposed person had worked. For the other 12 cases there was no geographical factor (Baxter et al , 1977). 1.6 Conclusions Occupational exposure to high evels of VC, (probably hundreds of ppm during several years) causes mortality due to angiosarcoma and possibly other primary liver cancers. There is insufficient evidence to establish any relationship between exposure to VC and an increased incidence of cancer of the brain, lung, thyroid, lymphatic or haematopoietic tissies, end skin (malignant melanoma). The annual incidence of angiosa rcoma of the liver in the general population hasbeen reported to be in the orde of 0.25 per million, but may even be lower when stringent pathological criteria ire used. In addition to VC, arsenic and Thorotrast may cause ASL. but for many of the ASL cases studied no plausible cause could be found. Theoretically exposure f om the ambient air, in particular near VC or PVC plants, could present a cancer -isk, but the results from the few studies done do not indicate such a possibility or are inconclusive. CLASTOGENICITY AND MUTAGENIC1 TY 2.1 Studies The finding of carcinogenic and clastogenic effects in workers exposed to VC led to studies on its possible mutagen c effects. A number of cytogenetic studies have demonstrated that exposure tc VC is associated with an increased frequency of chromosomal aberrations in the peripheral lymphocytes of exposed workers (Funes-Cravioto et al (1975). Ducatman et al (1975), Szentezi et al (1976), Purchase et al (1976), Fomenko et al (1976), Purchase et al (1978), Hansteen et al (1978) Kucerova et al (1979), Anderson et al (1980) and Anderson et al (1981). Chromcsomal aberrations induced by VC were described as dicentrics, fragments, rings ind translocations. Leonard et al (1977) described such aberrations amongst VC e>posed workers but underlined that they had received SPI-00739 -58- frequent X-ray exposures and th< t the positive results could be attributed to X-rays rather than VC. Both Hansteen et al (1978) and Anderson et al (1980) found a decrease in chromosomal aberrations of circulating lymphocytes to the levels found in control groups (people on-site or off-site not exposed to vinyl chloride) with the improvement of plant hygiene. Fleig and Thiess (1978) failed to demonstrate an increased frequency of chromosomal aberrations in 10 persons classified as "exposed persons showing no symptoms of VC llness". Picciano et al (1977) failed to show an increased prevalence of chromosome abberations in 209 workers employed for up to 28 years in a VC plant. The average exposure had been to 50 ppm or less since 1959. Two studies reported that there was a small increase in sister chromatid exchange rate (Kucerova et al, 1979; Georgena and Tsoneva, 1981), whereas two other studies failed to demonstrate such an ircrease (Hansteen et_a1_, 1978; Anderson et_aj_, 1981). In the bone marrow cells of VC exposed workers the frequency of chromosomal breakage was higher than that repo i'ted for unexposed people (Hansteen et al, 1978). The urine of workers exposed to C did not induced mutations in the Salmonella typhimurium strain TA100, even in |he presence of rat liver homogenates or 6-glucuronidase (Matlern et al ,197 ). Some other studies on the possibl = effects of VC on reproduction are also noted here for completeness. Infante et 1_ (1976) compared the outcome of pregnancies in the wives of workers before and af :er they had been exposed to VC with the outcome in the wives of workers in the PVC processing and rubber industries. They found an excess of foetal loss in the group of wives whose husbands had been exposed to VC. However, Paddle (1976) and Clemmeji sen ( 1982) pointed out some fallacies in the study, and Downs et al (1977) have questioned some of the reported findings on the clastogenic, mutagenic and genotok ic effects of VC, concluding that there was evidence of an association between occupational exposure to VC and chromosomal abberations in the lymphocytes, but none for ar association between foetal loss in wives and exposure of the husbands , or between exposure to VC and excess birth defects in communities near PVC-prc Auction plants. Infante et al (1976) reported higher incidences of birth defects (cleft lip and palate, abnormal genital organs, clubfoot, and abnormalities in the CNS) in communities in Ohio where PVC polymerisation plants were located. Edmonds et al (1976) reinvestigated the data ind found no correlation between congenital abnormalities of the CNS and exposure to VC. In Canada, Theriault et al (1983) SPI-00740 -59- found higher incidences of birth defects, among them abnormalities of the Ch.'S, in a community where a PVC polymerisa :ion plant was located. The occupational and residential histories cf parents fho gave birth to malformed infants and normal infants (controls) did not indicate an association with exposure. 2.2 Conclusions Exposure to VC causes chromosomal abberations in human beings, but only at the levels existing before the markt d reduction in occupational exposure in the mid-1970's. It is uncertain whetier it causes chances in the rate of sister chromatid exchange. From the scant evidence available it seems unlikely that VC has caused any adverse effects on human health via its mutagenic potentia 3. OTHER EFFECTS RELEVANT TO CARCII OGENICITY OR MUTAGENICITY In this section, long-term effect; of exposure to VC which are relevant to its carcinogenicity are reviewed. Fortwengler et al_ (1981) have shown that ASL originates from effects in the endothelial cells and this type of cell may also be involved in the development of acro-osteolysis (AOL) and non-malignant liver disease? related to liver cancers. 3.1 Non-malignant Liver Diseases Juhe et al (1973) investigated W. German PVC-production workers with symptoms resembling progressive sclerodermja Three of a group of 13 had a history of bleeding from oesophageal varices due to portal hypertension. Tribukh et al (1949) studied a grn up of 73 workers at a plant in the Soviet Union at which PVC resins were manufac tured and compounded. Twenty-one subjects were feurd tc have "anicteric hepatiti " (hepatitis without jaundice) and the authors reported non-tender hepatomegaly b i t gave no details. Suciu et al (1963, 1975) re ported that 30? of a group of 168 workers at two Rumanian PVC-production plants had hepatomegaly, and this was associated with splenomegaly ir. 6?. The VC conce rtration in the working environment was 880 ppiP* Four years later, following a r iduction in this concentration to 38 ppm, the prevalence of hepatogegaly was on ly 11?. SPI-00741 -60- Marsteller et al (1973) found vary ng degrees of non-cirrhotic portal fibrosis, perisinusoids 1 fibrosis and subcapsu ar fibrosis of the liver in all of a group of 0 male PVC-production workers. The drevalence of findings related to liver disease was : increased bromsulphthaleln (BSP) reteitlon thrombocytopenia hepatic-spleen surface alterations on periconeoscopy splenomegaly hepatomegaly acroosteolysis oesophageal varices 19/20 16/20 14/20 7/20 6/20 4/20 3/20 The subjects were aged 30-56 years a diagnosis and their exposure time was 1.5-21 years. Nc information on exposure 1 evels was given but all 20 workers had beer engaged in autoclave cleaning, some cf them full-time. Barnes (1976) suggested that in the early 1950s autoclave cleane -s had been exposed to levels of VC of up to 3000 ppm. After the first report that exposure to VC was related to ASL (Creech and Johnson, 1974) the performance of liver-functi on tests on workers exposed to VC was extended and it later appeared that non-c rrhotic portal fibrosis leading to portal hypertension was a more common lesi O'l than was ASL (Lelbach and Marsteller, 1981; Jones and Smith, 1982). Non-cirrhot i c portal hypertension and ASL are both rare. Iber (1969) estimated that of all su b|jects with portal hypertension similar to that encountered in cirrhosis or portap vein blockage, 3 to 5% had normal liver morphology. Falk et al (1974) reviewed the patho ogy of liver biopsy specimens from VC exposed workers and with non-malignant liver disease or ASL. They suggested that there was a close histological similarity betwe en these two diseases, both leading to portal fibrosis and sinusoidal changes. The progression of VC-induced hepatic fibrosis to ASL has been described by Smith et al_ ( 1976), Lelbach and Marsteller (1981) and Jones and Smith (1982) who sugges ed that periportal fibrosis and sinusoidal changes were precursors to ASL. Hepatlic sinusoidal cells include endothelial cells, and Fortwengler et al (1981) have giv en evidence that VC-induced ASL has its crigin in endothelial cells. They demonstra ;ed an increase in factor VIII, a known marker for endothelial cells, in the tumou r cells of ASL cases, suggesting that the endothelial cells were the target, Hepatocyt.es may not be the target cells for toxic action, which would explain why the standard biochemical liver-function tests SPI-00742 -61- were of limited value for the ea rly detection of non-cirrhotic portal hypertension and ASL in populations at risk ( Creek and Makk, 1975 ; Williams et a 1 , 1976 ; Lee et al, 1977). 3.2 Acro-osteolysis (AOL), Raynajd's Phenomenon and Effects on the Skin Filatova and Gronsberg (1957) published one of the earliest papers to record adverse effects from exposure to VC, finding effects similar to Raynaud's phenomenon ("toxic angioneurosis" ) in exposed workers. A few years later, Smirnova (1961) reported finding destruct ve bone lesions of the terminal phalanges of the hand in PVC-production workers >xposed to VC for between 3 and 9 years. Those affected also had symptoms simil ir to those of Raynaud's phenomenon, and Smirnova believed that the bone lesions >'ere caused by the exposure to VC. Suciu et al (1963, 1975) found Raynaud's ph snomenen in 6% and scleroderma in 3.6%, of the work-force of ICC at two Rumania PVC-producirg plants where the VC concentration in the working atmosphere was 88(j> ppm. After this had been lowered to 37 ppm, the. prevalence of Raynaud's phenomeno n fell to 2.9* and that of scleroderma to zero, The authors did not mention findi lg any bone lesions such as AOL. Cordier et al (1966) described AO of the hands in two male autoclave cleaners at a Belgian PVC-production plant, Several months after starting this work they developed the symptoms of Raynaud s phenomenon, eruptions on the skin of the hands, asthma and drowsiness. The most striking sign, however, was radiological evidence of lysis of some terminal phalang|e s of the hand and, in one case, the feet. In the succeeding years additional ce es were reported from France (Chatelian and Motillon, 1967), Great Britain (|Harris and Adams, 1967) and the USA (Wilson et [,1967). These last authors foufi d 31 cases among 3,000 employees engaged in VCand PVC-production. Dinman et al (1971) observed a strong association of AOL with PVC autoclave cleaning (1 case pe r 72 workers at risk) end the bagging and packing of PVC (1 case per 86 at risk) but found no cases among the 1,257 employees working on the compounding of PVC resins. The first cases of AOL in W. Gerrrja ny were observed by Jlihe and Lange (1972) among workers at PVC plants. Initially she clinical findings resembled systemic sclerosis and it emerged that some worker; mainly occupied as autoclave cleaners, had a history of bleeding from oesc phagal varices due to portal hypertension. Arteriographic evaluation of the Ivacculer tree of the hands showed the presence of vascular lesions confined to the small digital arteries, although there was some SPI-00743 -62- narrowing of the superficial and deep palmar arterial arch (Lange et al, 1974). Arterial irregularities and segme^i tal stenosis were accompanied by retardation of the flow of the contrast medium and peculiar tortuosities of the patent arteries, The results cf arteriography cor -elated well with those of non-invasive methods such as photoplethysmography, neog|r'aphy and thermography. Ward et al (1976) demonstrated tha t circulating immune complexes were present in 19 out of 28 patients with "VC disease", ie. Raynaud's phenomenon, AOL, thrombocytopenia, portal fibrosi and hepatic and pulmonary dysfunction. They suggested that the disease was an immune complex disorder, a suggestion also made by Lenrauer-Lewowicka et al (1976 Lange et al (1974) had already concluded from their results that there was np convincing evidence that VC disease was an auto-immune disease. 3.3 Conclusions Vinyl chloride-induced chronic tox|iicity affects the endothelial lining of the liver sinusoids from which ASL origina .es. Exposure to VC may also lead to periportal fibrosis and to changes in the d` stal vasculature of the hands manifested by the development cf Raynaud's phenomen Dn, scleroderma or AOL. It is not clear whether these lesions are caused by VC it self or its metabolites. These effects were seen in workers exposed in the 1950s and 1960s. There have been no reports of such effects in workers whose exposure to VC started in the early 1970s when exposure levels in most countries were lowe red to a few ppm. SPI-00744 -63- D. E X P E R I M ENTAL TOXICOLOGY 1. CARCINOGENICITY 1.1 Exposure by Inhalation 1.1.1 Viola's initial experiment The first investigation of the long-term effects of VC on animals was carried out in the Institute of General Pathc logy, Perugia, and was reported by Viola (1970, 1971). It was undertaken in an a tempt to produce AOL in experimental animals. A group of 26 3-month old male Wist<r rats of the IRE strain were exposed to VC at a concentration of 3 v/v in the irhaled air for 4 h/d, 5 d/wk for 12 months. Twenty-five rats of the same strain served as controls. The animals were slightly sleepy but otherwise appeared to tolerate these high exposures well. Tumours started to appear after the 10th month of treatment and the experiment was terminated after 12 months. Sevr nteen rats survived beyond the 10th month and epidermcid tumours appeared in thu para-auricular region in all of them. Fourteen of these tumours were reported to be epidermoid carcinomas, ? muco-epidermoid carcinomas and one a papilloma, five of the 17 animals had osteochondromas and 7 had lung tumours, but no haemangiosarconss were reported. The epidermoid tumours were reviewed by other pathologists and re-classified as Zymbal gland carcinomas and the lung tumours were metasttses from the Zymbal gland carcinomas (Maltoni et al, 1974, 1975). No tumours of these types were observed in the control rats. 1.1.2 Investigations by Maltoni BT series Because the Viola study indicated a risk of cancer from occupational and environmental exposure to VC, its carcinogenicity was then thoroughly investigated in laboratory rodents by Maltoni Maltoni numbered his experiments BT 1, BT 2, etc. and the detailed results cf the experiments are tabulated in a book, Maltoni et al (1984). Three earlier papers conlaining details of the design of the BT series of experiments as well as some important interim and final results are cited in Table 7. As the tumourigenic response in experiments BT 3 and BT 15 was low due to inadequate dosage discussion of these experiments have not been included in this report. SPI-00745 In an extensive series of studies, Maltoni administered VC in air to adult rats, mice and hamsters, and VC disso lved in olive oil, to rats by the oral and parenteral routes. He also exposed pregnant and neonatal rats to VC by inhalation to examine the sensitivity of foeta and newly-born animals. 1*1.2.a Pose-response in Sprague-Dawlay rats exposed by inhalation In one of these studies (BT 6, Tab! a 8) the Viola experiment was repeated. A group of 30 male and 30 female Sprague-Dawley rats was exposed to 30,000 ppm of VC for 52 weeks and killed at 68 weeks. Zymb; 1 gland carcinomas were observed as in Viola's experiment. In addition a high inc dence of angiosarcomas and forestomach tumours but no bone or cartilaginous tumours were induced. It is difficult to account for this discrepancy. The longer pericd of observation in Maltoni's study may have allowed mere time for the development of the hepatic tumours but there appears to bone lesions. The increase in the incidence of forestomach tumours was not observed in later experiments and appears to be an isolated observation. The carcinogenicity cf VC was convi ucingly demonstrated in a dose-response study on Sprague-Dawley rats (BT 1, Table 9) Animals of both sexes were exposed to a wide range of concentrations of VC from 30 ppm at the lowest level to 10,000 ppm at the highest for 4 h/d, 5 d/wk over 5 1 weeks. The experiment was terminated at 135 weeks. There was a clear dose-resp )nse relationship for Zymbel gland tumours and hepatic angiosarcomas, but it v.'as less clear for hepatomas, nephroblastomas and brain tumours although the inc eased incidence of these appeared to be treatment-related. In particular, haemar.giosarcoma, a rare tumour in rats, was observed in a female rat at 50 p f m Other tumours (mammary gland, forestomach, papillomas and leukaemias) occurrid in both test and control groups and were randomly distributed. There was al ;o an indication of a reduction in the latent period of the treatment-related ti mours compared with that of the same type of tumours occurring in the control g ro up. 1.1.2.b Other, dose-response inhala ion experiments in Sprague-Dawley rats The occurrence of haemangiosarcom is at 50 ppm prompted further dose-response experiments (BT 2 and BT 15, Tables 10 and 11) to explore further the carcinogericity of VC at relative ly low levels of exposure. In one of these experiments (BT 2) a comparatively narrow range of concentrations was investigated. SPI-00746