Secondary Logo

Journal Logo


Update of the Follow-Up of Mortality and Cancer Incidence among European Workers Employed in the Vinyl Chloride Industry

Ward, Elizabeth1,2; Boffetta, Paolo1; Andersen, Aage3; Colin, Didier1; Comba, Pietro4; Deddens, James A.2; De Santis, Marco4; Engholm, Göran5; Hagmar, Lars6; Langard, Sverre7; Lundberg, Ingvar8; McElvenny, Damien9; Pirastu, Roberta10; Sali, Davide1; Simonato, Lorenzo11

Author Information
  • Free


The current study is an update of mortality and cancer incidence among individuals included in the European Multicentric Cohort Study of Workers in the Vinyl Chloride Industry. 1,2 The first analysis of the study found a nearly threefold excess of liver cancer with 24 observed and 8.4 expected [standardized mortality ratio (SMR) = 2.86, 95% confidence interval (CI) = 1.86–4.25]. Excess liver cancer was clearly related to time since first exposure, duration of employment, and estimated ranked and quantitative exposures. No strong association was observed with any other cause of death or incident cancer. The analysis of the updated study was designed to evaluate a number of issues, including the possibility that vinyl chloride (VC) induces hepatocellular carcinoma as well as angiosarcoma; the exposure-response relation between cumulative VC exposure and liver cancer/angiosarcoma, especially at lower cumulative exposures; the potential associations of brain cancer, lung cancer, soft-tissue sarcoma, non-Hodgkin’s lymphoma and malignant melanoma with VC exposure; the potential association between exposure to VC and nonmalignant respiratory disease; and the potential association between VC exposure and deaths from nonmalignant liver disease. Detailed results are published in an IARC Internal Report, 3 which can be obtained from the corresponding author.

Subjects and Methods

This multicentric study was conducted in four countries (Italy, Norway, Sweden, and the United Kingdom); results for most of the national study cohorts have been published previously. 4–16 The study includes a total of 19 factories (Table 1); vital status has been updated for 17 factories, and incidence data are available for 13 factories in three countries (Norway, Sweden, and the United Kingdom). The years through which incidence and mortality have been updated range from 1993 through 1997.

Table 1:
Characteristics of the Cohort

In the majority of these factories there is mixed VC monomer (VCM)/polyvinyl chloride (PVC) production (11 factories), two produce VCM only, five produce PVC only, and one is a PVC-processing plant. Analyses were carried out for 12,700 male workers with at least 1 year of employment. The observation period began in 1955 (the year for which reference rates are first available) for all factories, except for three factories for which the start of the observation period was as follows: factory 1 in Italy, 1972; factory 3 in Italy, 1974; and factory 2 in Sweden, 1961.

Of the total 12,700 persons in the analysis, 9,688 (76.3%) were alive (ranging by country from 66% to 89%), 2,665 (21.0%) were deceased (ranging by country from 10% to 33%), 63 (0.5%) were lost to follow-up, and 284 (2.2%) had emigrated (these individuals were traced until date of emigration).

Age- and calendar period-specific (males only) national (that is, for each country) mortality rates from the World Health Organization database ( have been used as the reference for the SMR analysis. Similar incidence rates from cancer registries were used for the incidence analysis. 17 We searched for the best evidence for diagnosis of liver cancers by reviewing all available documentation, including death certificates, cancer registry records, medical records, and listings of angiosarcoma cases from several registries. 18,19 We used the best-evidence diagnosis to classify liver cancers for internal Poisson regression analyses only. Person-years at risk were calculated using a modified life-table approach. 20 The ratio of observed to expected deaths gives the SMR or the standardized incidence ratio (SIR); 95% CIs for the SMR or SIR are calculated using a Poisson distribution.

We examined mortality according to time since first employment (TSFE), calendar period of hire, and age at hire. Exposure variables are employment as an autoclave worker (ever/never/undetermined), duration of employment, ranked level of exposure (RLE), and cumulative exposure in parts per million (ppm)-years to VCM in air. The reason to focus on autoclave workers stems from reports on very high peaks of exposure experienced by these workers. Autoclave workers were exposed to unreacted monomer when reactors were opened for cleaning. Workers were also lowered into reaction vessels to clean them manually, and were exposed to peak VCM exposures of several thousand ppm. That practice was phased out in the late 1960s and the 1970s after the identification of acroosteolysis among autoclave cleaners. 21

With the exception of two factories, employment histories were not updated beyond the original data collected, 2 and therefore, the original work histories were used in the analysis. As a result, duration of employment will be underestimated for workers who were employed at the time of the original follow-up. Other VC exposure parameters should not be greatly affected by the censoring of work history data because exposures to VC in the study plants were estimated to be 1 ppm for all jobs from 1976 through 1988. Details of the job exposure matrix (JEM) were published previously. 1 Calendar period-specific JEMs were provided by industrial hygienists for 13 of the 19 factories. Jobs were grouped into 22 broad categories. We estimated typical exposures for each job category on the basis of a number of sources. In most factories, up until the mid-1970s, only limited air measurement data were available. These data were supplemented by knowledge of exposure conditions, processes, and changes over time. For several factories, interviews with those who worked in earlier time periods provided information on how often workers could smell VC, which is detectable at about 400 or 500 ppm. These interviews and knowledge of industrial hygienists who worked at the time revealed that use of protective equipment, such as respirators, was rare. Since the mid-1970s, air measurements have been systematically collected and provide the basis for the more recent exposure assessments. The JEMs produced for each factory were validated by two independent industrial hygienists who had several years of experience in the VC industry. There were six factories for which JEMs were not developed. For four of these, JEMs were estimated on the basis of those developed for factories in the same country. For the remaining two plants, which have together had only one death from liver cancer, no JEM was available. Quantitative estimates of exposure were available for 9,775 members (82%) of the cohort. In addition to quantitative estimates of cumulative exposure (expressed as ppm-years), study subjects were classified in an RLE index on the basis of their maximum exposure level at any job.

We examined mortality from 15 categories or causes of death, in addition to the eight causes specified to be of a priori interest. Cutpoints for TSFE, duration of employment, cumulative exposure, and calendar period at hire were based on dividing the distribution of observed deaths from all causes into approximately equal quintiles. For liver cancer and cirrhosis of the liver, the cutpoints for cumulative exposure based on all observed deaths resulted in more than two-thirds of the deaths falling into the highest two exposure categories. To better characterize exposure response for these causes of death, we redefined cutpoints such that an equal number of observed deaths fell into each of five categories. We conducted Poisson regression analysis to assess the contribution of several variables simultaneously. We included calendar time and age at risk in all of the Poisson regression models. For cancers of a priori interest, we conducted internal exposure-response analyses for incident and deceased cases combined. For liver cancers and angiosarcomas of the liver, we also stratified by dose to understand better the overall shape of the exposure response and the exposure response at lower levels of exposure. For all liver cancer this stratification was achieved by creating 13 exposure categories with four to five observed cases per stratum. For angiosarcomas we combined the lowest two dose categories because no death was observed in the lowest category. In addition to categorical analyses in which we estimated relative risks (RRs) for each exposure stratum, we created a continuous exposure variable using the midpoints of the exposure range in each stratum. For the highest dose stratum, the midpoint was determined from the distribution of cumulative doses for individuals who reached that stratum. In modeling risk, we considered three models: (1) log linear [RR = exp(B*cumexp)], (2) log linear with log transformed exposure [RR = exp(B*log cumulative exposure)], and (3) additive [RR = (1 + B cumexp)]. We evaluated the fit of each model from its deviance, and we retained model 2 because it had the best fit. We present individual RRs for categories of cumulative exposure, as well as continuous RRs and CIs based on model parameters in Figures 1 and 2. We calculated the latter continuous RRs assuming that risk in the exposed occurs above a baseline risk equal to the baseline risk in the categorical analysis (that is, risk in the lowest or reference group). By using the same reference category, we made the continuous RRs comparable with the categorical RRs.

Log relative risk and 95% confidence interval of all liver cancer (N =71), by cumulative exposure to vinyl chloride (ppm-years). Shown are results of Poisson regression analysis adjusted for age and calendar period. exp(−intercept 1 + beta 1* logdose)/exp(intercept 0) was plotted where intercept 1, beta 1 comes from the logdose model and intercept 0 from the categorical model. The reference category is 0–130 ppm-years.
Log relative risk and 95% confidence interval of liver angiosarcoma (N = 37), by cumulative exposure to vinyl chloride (ppm-years). Shown are results of Poisson regression analysis adjusted for age and calendar period. exp(−intercept 1 + beta 1* logdose)/exp(intercept 0) was plotted where intercept 1, beta 1 comes from the logdose model and intercept 0 from the categorical model. The reference category is 0–450 ppm-years.

The rate difference is another useful measure for evaluating exposure response, especially when the baseline risk is extremely low, as is the case for angiosarcoma. In calculating rate differences, we examined the same five dose categories shown in Table 2, except that a baseline category (<287 ppm) was created with no case of angiosarcoma. We calculated crude and standardized rates of liver cancer and angiosarcoma using seven age and four calendar time strata. We also calculated rate differences and 95% CIs.

Table 2:
Results of Poisson Regression Analysis of Liver Cancer, Angiosarcoma, and Hepatocellular Carcinoma by Time Since First Employment, Duration of Employment, and Cumulative Exposure


For the 12,700 subjects included in the analysis, the mean length of follow-up is 29 years, and 80% of cohort members have been traced for 15 or more years. The total number of person-years at risk accumulated by the cohort is 324,701. Mortality from all causes is decreased (2,664 deaths; SMR = 0.85, 95% CI = 0.82–0.88), whereas mortality from all malignant neoplasms is close to that expected (883 deaths; SMR = 0.99, 95% CI = 0.93–1.06) (Table 3). Fifty-three deaths from liver cancer were observed in the updated study compared with 24 in the original study. The increase in SMR for liver cancer is less pronounced [2.40 (95% CI = 1.80–3.14) in the current analysis and 2.86 in the original analysis]. Among other causes of death of a priori interest, there was no marked excess (Table 3), nor was any overall excess or exposure-response trend noted in non-a priori causes of death. In the cancer incidence analysis, there were 760 cases observed, and 896 expected, for a SIR of 0.85 (95% CI = 0.79–0.91) (Table 4). Among 20 cancer sites for which reference rates are available, the only notable result was a slight excess of thyroid cancer based on seven cases (SIR = 2.21, 95% CI = 0.89–4.55).

Table 3:
Mortality by Detailed Causes
Table 4:
Incidence of Selected Cancer Sites

Liver Cancer and Angiosarcoma of the Liver

A total of 71 cases of liver cancer were identified on the basis of best evidence and used in the internal analysis (all SMR analyses were restricted to 53 cases identified from death certificates). In addition to the 62 cases identified in the mortality or cancer incidence follow-ups, 9 cases were identified from other sources, including 5 cases identified from linkage with the angiosarcoma registry maintained by the Association of Plastics Manufacturers in Europe. 19 The 71 cases of liver cancer comprised 37 angiosarcomas, 10 hepatocellular carcinomas, 7 cases of other known histology, and 17 cases of unspecified type of liver cancer.

Liver cancer mortality was elevated in the Italian (SMR = 1.63, 95% CI = 1.01, 2.50), Swedish (SMR = 2.86, 95% CI = 1.37–5.26), and United Kingdom (SMR = 3.91, 95% CI = 2.42–5.97) cohorts, with only one liver cancer death observed in the Norwegian cohort (SMR = 2.62, 95% CI = 0.07–14.6). Liver cancer mortality was elevated in PVC production (SMR = 2.28, 95% CI = 1.09–4.18) and mixed VCM and PVC production plants (SMR = 2.85, 95% CI = 2.05–3.87), but not in the VCM production and PVC-processing plants, each of which had fewer than two expected liver cancer deaths. Among persons hired in 1964 or later, liver cancer mortality was only slightly increased (SMR = 1.22, 95% CI = 0.61–2.18); however, 4 of the 11 liver cancer deaths in this category were angiosarcomas. Among 25 liver cancer deaths among former autoclave workers, there were 16 angiosarcomas, 1 hepatocellular carcinoma, and 7 liver cancers of other or unknown histology.

In the Poisson regression analysis of liver cancer, we found strong positive trends for TSFE, duration of employment, and cumulative exposure (Table 2), and we found negative trends for later calendar period of hire and later age at hire, when these were included in models as the only predictor variables (data not shown). TSFE, calendar period of hire, and age at hire were not confirmed to be important predictors when included in models with cumulative exposure. When we further subdivided cumulative dose into 13 categories, we found a fairly monotonic increase in rate ratios, with an RR of 82 (95% CI = 21–323) in the highest-dose category of >18,300 ppm-years. When we used the midpoints of the dose categories to model dose as a continuous variable, the (natural) logarithmic transformation of cumulative exposure provided the best fit (Table 5). The RR for one logarithmic unit of cumulative exposure was 2.0 (95% CI = 1.7–2.4). Figure 1 shows the estimated categorical rate ratios for cumulative exposure and liver cancer along with the continuous curve and CIs estimated from model parameters.

Table 5:
Summary of Model Results

When exposure response was estimated for only the 37 angiosarcomas in a model including 12 dose categories (Figure 2), the RR for a logarithmic unit of cumulative exposure was 2.9 (95% CI = 2.2–3.9). It is not possible to calculate an SMR or SIR for angiosarcoma, because age- and calendar time-specific reference rates are not available. The incidence of angiosarcomas of the liver with unknown etiology in the general population is estimated to be about 0.1 per million population per year. 22,23 The reference group in the Poisson regression analysis shown in Table 2, with 4 angiosarcomas among approximately 200,000 person-years (about 2 per 100,000), therefore has an approximately 200 times greater risk of angiosarcoma than the general population. Table 6 presents crude and age-adjusted rates, as well as rate differences and CLs for angiosarcomas and liver cancers by dose category, defined so that there is no angiosarcoma case in the 0–286 dose category.

Table 6:
Rate Differences for All Liver Cancer and Angiosarcoma

There are 28 liver cancer deaths that are not known to be angiosarcomas, an excess of approximately 5 deaths over the expected number. If angiosarcoma deaths are excluded, the resulting SMR is approximately 1.27 (95% CI = 0.84–1.83). The Poisson regression analysis for hepatocellular carcinomas (N = 10) is shown in Table 2. Marked trends were found for TSFE, duration of employment, and cumulative exposure; the RR for “ever-work” as an autoclave worker was 2.97 (95% CI = 0.80–11.1).

To examine exposure-response trends at low doses, we conducted a separate analysis among individuals who had cumulative exposures <1,500 ppm-years, an arbitrary cutpoint that yielded 20 liver cancer cases on the basis of best evidence and is equivalent to approximately 30 ppm exposure for 45 years. The SMR, based on 16 liver cancer deaths, was 1.31 (95% CI = 0.75–2.13). The results of the Poisson regression analysis are shown in Table 5. The RR for one logarithmic unit of cumulative exposure was 2.0 (95% CI = 1.3–3.0), which is nearly identical to the result obtained when analyzing the full cohort. There also appeared to be a strong exposure-response trend in the categorical analyses for the seven angiosarcomas that occurred among persons with <1,500 ppm-years of cumulative exposure (data not shown), but the continuous models did not converge.

Other Malignant Neoplasms

The overall SMR for brain cancer (N = 24; SMR = 0.93, 95% CI = 0.60–1.39) has decreased slightly compared with the previous analysis (N = 14; SMR = 1.07, 95% CI = 0.59–1.80). There is no trend in SMR with respect to RLE, TSFE, duration of employment, cumulative exposure, or calendar period of hire. The SMR for brain cancer (1.02) among ever-autoclave workers was similar to that among never-autoclave workers (0.91). In the Poisson regression analyses we saw no monotonic trend, although the RR was elevated in the middle-dose category of cumulative dose (RR = 3.45, 95% CI = 0.94–12.6) (Table 7).

Table 7:
Mortality from Selected Causes of Death, by Country, Ranked Level of Exposure, Cumulative Exposure to Vinyl Chloride, and History of Work as an Autoclave Cleaner

The SMR for lung cancer in the current mortality analysis (SMR = 0.95, 95% CI = 0.84–1.07) is virtually identical to that in the previous analysis (SMR = 0.97, 95% CI = 0.82–1.14). There is little association between VC exposure and lung cancer mortality, as estimated by RLE, duration of employment, cumulative exposure, or ever/never autoclave work. The SMR for lung cancer was 1.47 (95% CI = 0.80–1.47) in VCM production and 1.43 (95% CI = 0.85–2.26) in PVC processing. In the Poisson regression analyses, negative trends were suggested with increasing TSFE, duration of employment, and cumulative exposure (Table 7). We saw no trend with increasing cumulative exposure for persons who had ever worked as packers and baggers, among whom there were 53 lung cancer deaths. There was, however, a trend with increasing cumulative exposure among persons who had only worked as packers and baggers, including 30 lung cancer deaths (data not shown).

There were a total of six deaths from tumors of the connective tissue (soft-tissue sarcoma) observed in the study, yielding an SMR of 1.89 (95% CI = 0.69–4.11). In the best-evidence analysis for liver cancer, however, it became apparent that three of the six deaths coded as tumors of the connective tissue were actually angiosarcomas of the liver, the primary site of which had been miscoded (these cases were included in the Poisson regression analysis as liver cancers).

The overall SMR for non-Hodgkin’s lymphoma of 1.19 (N = 18, 95% CI = 0.78–1.75) is decreased from that in the previous study (SMR = 1.70, 95% CI = 0.69–3.71). The SIR is 0.78 (95% CI = 0.48–1.21) based on 20 cases. Among the four countries, Italy has the highest SMR, 1.86 (95% CI = 0.84–3.54), and as in the previous study much of the excess is in the mixed production plants (SMR = 1.52, 95% CI = 0.96–2.27). There is no clear trend with TSFE, duration of employment, cumulative exposure, or calendar period at hire. In the Poisson regression analyses, there appears to be some elevation in rate ratios above the baseline category of cumulative dose (Table 7).

The present analysis found an excess in malignant melanoma mortality (N = 15; SMR = 1.60, 95% CI = 0.90–2.65), which was similar to that found in the previous study (N = 7; SMR = 1.63, 95% CI = 0.65–3.35). An excess in melanoma has previously been noted for the Norwegian cohort and found to be dose-related 7,9,16; in the present analysis, the SMR in the Norwegian cohort is 6.27 (95% CI = 2.04–14.6). An excess in melanoma mortality is also present in the Italian cohort (SMR = 1.96, 95% CI = 0.53–5.01). In both the SMR (Table 7) and Poisson regression (Table 7) analyses a trend of increasing risk with increasing VCM exposure was suggested. There is, however, no association with previous employment as an autoclave cleaner.

Nonneoplastic Diseases

There was no excess in respiratory disease overall or in any subcategory of respiratory disease. Mortality from this group of diseases, however, was increased among Norwegian workers (SMR = 2.51, 95% CI = 1.08–4.94). There were only three deaths from pneumoconiosis, with 5.07 expected. Detailed analyses conducted for “bronchitis, emphysema, and asthma” revealed no evidence of an association between VC exposure and these causes of death. There was a negative trend with duration of employment in both the SMR and Poisson regression analyses. The Poisson regression analysis suggested decreasing RRs with increasing cumulative exposure (Table 7). Respiratory disease mortality was not elevated among workers ever or only employed as packers and baggers, nor was there any trend with duration of employment for this job title or for cumulative exposure to VCM (data not shown).

Although there was a deficit in mortality from cirrhosis of the liver in the cohort overall (SMR = 0.77, 95% CI = 0.57–1.02), marked elevations in RRs were observed for all dose categories above the reference category in the Poisson regression analysis. The SMR for cumulative exposures of <524 ppm-years (the reference group in the Poisson regression analysis) is 0.31 (95% CI = 0.13–0.61), whereas the SMR for exposures of 524 or more ppm-years is 1.30 (95% CI = 0.9–1.8). Thus, the Poisson regression results are not as inconsistent with the SMR results as they might appear. The deficit in liver cirrhosis is largely due to the Italian cohort, which had 2 observed and 15.17 expected deaths in this category.


The results of the updated study are generally consistent with the original study with respect to liver cancer and angiosarcoma. A strong relation is observed between cumulative VC exposure and occurrence of liver cancer. An even sharper exposure-response relation is observed for angiosarcoma. The exposure response can be represented by a log-linear model, with an increase in the logarithm RR of approximately 2.0 per logarithmic unit of cumulative dose for liver cancer and 2.9 for angiosarcoma. A marked exposure-response trend with both duration of employment and cumulative VC exposure was present for the ten known cases of hepatocellular carcinoma, suggesting that VC exposure may be associated with this tumor as well. An association of hepatocellular carcinoma with VC exposure is biologically plausible, given that hepatocellular carcinomas have been induced by VC in rodents. 24,25 Furthermore, thorotrast, which is another major cause of angiosarcoma in humans, is also a cause of hepatocellular carcinoma. 26 In addition, cases of hepatocellular carcinoma, together with angiosarcoma, have been reported among workers highly exposed to VC. 27–29 A recent study has found mutations of the p53 gene in 11 of 18 hepatocellular carcinomas in patients with VC exposure 30; p53 mutations have also been noted in VC-related angiosarcomas. 31

The quantitative exposure-response data generated in this study have several limitations. The exposure matrix has characterized exposures for broad categories of jobs and time periods, and therefore the cumulative exposure estimates generated for individuals may be quite imprecise. Furthermore, for the majority of the cohort, we have not updated the occupational history beyond the late 1980s, when the data were collected. Nevertheless, given the strong decrease in exposure levels that occurred after 1975, it is unlikely that this omission would have any practical influence on the exposure variables, with the exception of duration of exposure. The cumulative exposures estimated for the majority of the cohort are considerably higher than those that would result from exposure levels permitted under current occupational standards, which range from 1 to 7 ppm as a time-weighted average. Risk assessments for occupational exposures base their calculations on 45 years of working life, for which an average exposure of 5 ppm would yield a cumulative dose of 225 ppm-years. The observation of six angiosarcomas among workers with estimated cumulative exposures <1,500 ppm-years and the clear exposure response observed in this study at cumulative exposures <1,500 ppm-years suggest that an increased risk of angiosarcoma may be present at cumulative VC exposures within an order of magnitude of those permitted by current standards. The apparent discrepancy in the magnitude of the SMRs for angiosarcoma (estimated 200-fold) and liver cancer (one- to twofold) is explained by the very low rate of the former neoplasm in the reference population.

Brain cancer was of a priori interest on the basis of findings in some prior studies. 32–34 Evidence for an association of brain cancer with VC exposure in the current study was generally negative. Lung cancer was also of a priori interest on the basis of a summary analysis of VC-exposed cohorts, 34 but the findings in the current study provide no evidence of a lung cancer excess in the total cohort. A possible association between VC polymer dust exposure and lung cancer has been suggested by some early studies 33–35 as well as a recent investigation of PVC baggers in a VCM-PVC manufacturing plant in Italy (SMR = 1.39, 95% CI = 0.75–2.36). 14 The present study does not provide estimates of exposure to PVC dust and therefore no firm conclusion can be drawn with respect to lung cancer risk due to inhalation of the polymer. Nevertheless, when analyses were restricted to individuals who had only been employed as packers and baggers, there was a trend for lung cancer with increasing cumulative VCM exposure. It is important that lung cancer mortality among workers potentially exposed to PVC dust be examined in cohorts where exposure measurements for PVC dust are available. 36

One study of a large U.S. cohort of workers exposed to VC reported an SMR of 2.7 for soft-tissue sarcoma (95% CI = 1.4–4.7) based on 12 deaths; this excess was associated with longer duration of employment. 37,38 The finding from the current study that three of the six observed deaths from soft-tissue sarcoma were actually angiosarcomas of the liver suggests caution in the interpretation of results on soft-tissue sarcoma among VC workers based on death certificate information only.

The data from this study do not provide strong support for an association between VC exposure and non-Hodgkin’s lymphoma, nor are the results consistently negative. The evidence for an association of VC exposure with malignant melanoma is stronger than that for non-Hodgkin’s lymphoma, but the data are also limited for this relatively rare cause of death.

A new finding of the updated study is increased mortality from cirrhosis of the liver associated with moderate to high levels of VC exposure, although in part this pattern is explained by the extremely low SMR in the reference category owing to a large deficit in cirrhosis in the lowest exposure category in the Italian cohort. The areas where the Italian factories are located have elevated rates of liver cirrhosis, and it may be that the deficit in the lowest VC exposure category is explained by a “healthy worker effect,” in particular by selecting out of the workforce heavy drinkers. This unusual pattern nonetheless detracts from a causal interpretation of the finding. Numerous case reports and cross-sectional studies have documented VCM-induced hepatic disease associated with high exposures. 21 Increased mortality from liver cirrhosis has been observed among ever-autoclave workers 14 and among VCM manufacturers. 15 Only a few studies of occupational cohorts exposed to acute hepatotoxins (carbon tetrachloride, chlorinated naphthalenes, or other solvents) have demonstrated increased cirrhosis mortality, suggesting persistent subclinical injury after high exposures. 39–41

Our study found no evidence of increased mortality from all diseases of the respiratory system; pneumoconiosis; or bronchitis, emphysema, and asthma, nor was there any indication of increased respiratory disease mortality among workers in curing, packing, and blending jobs. Prior cross-sectional studies have found pneumoconiosis, pulmonary function abnormalities, and increased respiratory symptoms among workers exposed to PVC dust. 36,42–45 Our findings do not contradict the cross-sectional studies, as it is possible that the respiratory effects of VC or PVC dust exposure are unlikely to lead to death or that the categories of respiratory disease-related death are too broad to detect a more specific outcome.


Davide Sali worked on this study under the tenure of a Special Training Award from the International Agency for Research on Cancer.


1. International Agency for Research on Cancer. Mortality and Cancer Incidence Results of the European Multi-Centric Cohort Study of Workers Employed in the Vinyl Chloride Industry. IARC Internal Report No. 89/007. Lyon: IARC, 1989.
2. Simonato L, L’Abbé KA, Anderson A, Belli S, Comba P, Engholm G, Ferro G, Hagmar L, Langard S, Lundberg I, Pirastu R, Thomas P, Winkelmann R, Saracci R. A collaborative study of cancer incidence and mortality among vinyl chloride workers. Scand J Work Environ Health 1991; 17: 159–169.
3. Ward E, Boffetta P, Andersen A, Colin D, Comba P, Deddens J, De Santis M, Engholm G, Hagmar L, Langard S, Lundberg I, McElvenny D, Pirastu R, Sali D, Simonato L. Update of the Follow-Up of Mortality and Cancer Incidence among European Workers Employed in the Vinyl Chloride Industry. IARC Internal Report No. 01/001. Lyon: International Agency for Research on Cancer, 2000.
4. Byren D, Engholm G, Englund A, Westerholm P. Mortality and cancer morbidity in a group of Swedish VCM and PCV production workers. Environ Health Perspect 1976; 17: 167–170.
5. Fox AJ, Collier PF. Mortality experience of workers exposed to vinyl chloride monomer in the manufacture of polyvinyl chloride in Great Britain. Br J Ind Med 1977; 34: 1–10.
6. Molina G, Homberg B, Elofsson S, Holmlund L, Moosing R, Westerholm P. Mortality and cancer rates among workers in the Swedish PVC processing industry. Environ Health Perspect 1981; 41: 145–151.
7. Heldaas SS, Langard SL, Andersen A. Incidence of cancer among vinyl chloride and polyvinyl chloride workers. Br J Ind Med 1984; 41: 25–30.
8. Belli S, Bertazzi PA, Comba P, Foa V, Maltoni C, Masina A, Pirastu R, Reggiani A, Vigotti MA. A cohort study on vinyl chloride manufacturers in Italy: study design and preliminary results. Cancer Lett 1987; 35: 253–261.
9. Heldaas SS, Andersen A, Langard S. Incidence of cancer among vinyl chloride and polyvinyl chloride workers: further evidence for an association with malignant melanoma. Br J Ind Med 1987; 44: 278–280.
10. Jones RD, Smith DM, Thomas PG. A mortality study of vinyl chloride monomer workers employed in the United Kingdom in 1940–1974. Scand J Work Environ Health 1988: 14: 153–160.
11. Hagmar L, Akesson B, Nielsen J, Andersson C, Linden K, Attewell R, Moller T. Mortality and cancer morbidity in workers exposed to low levels of vinyl chloride monomer at a polyvinyl chloride processing plant. Am J Ind Med 1990; 17: 553–565.
12. Pirastu R, Comba P, Reggiani A, Foa V, Masina A, Maltoni C. Mortality from liver disease among Italian vinyl chloride monomer/polyvinyl chloride manufacturers. Am J Ind Med 1990; 17: 155–161.
13. Pirastu R, Belli S, Bruno C, Comba P, De Santis M, Foa V, Maltoni C, Masina A, Reggiani A. The mortality among the makers of vinyl chloride in Italy (in Italian). Med Lav 1991; 82: 388–423.
14. Pirastu R, Chellini E, Carnevale F, De Santis M, Bracci C, Comba P. Indagine epidemiologica sui lavoratori di Porto Marghera esposti a cloruro de vinile nella fasi di produzione, polimerizzazione e insacco (in Italian). Rapporti ISTISAN 1997; 22: 1–39.
15. Pirastu R, Bruno C, De Santis M, Comba P. An epidemiological study of workers exposed to vinyl chloride in the plants of Ferrara, Rosignano and Ravenna (in Italian). Epidemiol Prev 1998; 22: 226–236.
16. Langard S, Rosenberg J, Andersen A, Heldaas SS. Incidence of cancer among workers exposed to vinyl chloride in polyvinyl chloride manufacture. Occup Environ Med 2000; 57: 65–68.
17. Parkin DM, Whelan SL, Ferlay J, Raymond L, Young J, eds. Cancer Incidence in Five Continents. IARC Scientific Pub. No. 143. Vol. 7. Lyon: International Agency for Research on Cancer, 1997.
18. Baxter PJ. The British Angiosarcoma Registry. Environ Health Perspect 1981; 11: 115–116.
19. Forman D, Bennett B, Stafford J, Doll R. Exposure to vinyl chloride and angiosarcoma of the liver: a report of the register of cases. Br J Ind Med 1985; 42: 750–753.
20. Coleman M, Douglas A, Hermon C, Peto J. Cohort study analysis with a FORTRAN computer program. Int J Epidemiol 1986; 15: 134–137.
21. Falk H, Steenland K. Vinyl chloride and polyvinyl chloride. In: Rom WN, ed. Environmental and Occupational Medicine. 3rd ed. Philadelphia: Lippincott-Raven, 1998; 1251–1259.
22. Falk H, Herbert J, Crowley S, Ishak KG, Thomas LB, Popper H, Caldwell GG. Epidemiology of hepatic angiosarcoma in the United States: 1964–1974. Environ Health Perspect 1981: 41: 107–113.
23. Elliott P, Kleinschmidt I. Angiosarcoma of the liver in Great Britain in proximity to vinyl chloride sites. Occup Environ Med 1997; 54: 14–18.
24. Drew RT, Boorman GA, Haseman JK, McConnell EE, Busey WM, Moore JA. The effect of age and exposure duration on cancer induction by a known carcinogen in rats, mice and hamsters. Toxicol Appl Pharmacol 1983; 68: 120–130.
25. Maltoni C, Cotti G. Carcinogenicity of vinyl chloride in Sprague-Dawley rats after prenatal and postnatal exposure. Ann NY Acad Sci 1988; 534: 145–159.
26. Ito Y, Kojiro M, Nakashima T, Mori T. Pathomorphologic characteristics of 102 cases of thorotrast-related hepatocellular carcinoma, cholangiocarcinoma, and hepatic angiosarcoma. Cancer 1988; 62: 1153–1162.
27. Gokel JM, Liebezeit E, Eder M. Hemangiosarcoma and hepatocellular carcinoma of the liver following vinyl chloride exposure: a report of two cases. Virchows Arch A Pathol Anat Histol 1976; 372: 195–203.
28. Lelbach WK. A 25-year follow-up study of heavily exposed vinyl chloride workers in Germany. Am J Ind Med 1996; 29: 446–458.
29. Du CL, Wang JD. Increased morbidity odds ratio of primary liver cancer and cirrhosis of the liver among vinyl chloride monomer workers. J Occup Environ Med 1998; 55: 528–532.
30. Weihrauch M, Lehnert G, Kockerling F, Wittekind C, Tannapfel A. p35 mutation pattern in hepatocellular carcinoma in workers exposed to vinyl chloride. Cancer 2000; 88: 1030–1036.
31. Marion MJ. Critical genes as early warning signs: example of vinyl chloride. Toxicol Lett 1998:102–103:603–607.
32. Monson RR, Peters JM, Johnson MN. Proportional mortality among vinyl-chloride workers. Lancet 1974; 2: 397–398.
33. Waxweiler RJ, Smith AH, Falk H, Tyroler HA. Excess lung cancer risk in a synthetic chemicals plant. Environ Health Perspect 1981; 41: 159–165.
34. Doll R. Effects of exposure to vinyl chloride: an assessment of the evidence. Scand J Work Environ Health 1988; 14: 61–78.
35. Wagoner JK. Toxicity of vinyl chloride and poly(vinyl chloride): a critical review. Environ Health Perspect 1983; 52: 61–66.
36. Soutar CA, Copland IH, Thornley PE, Hurley JF, Ottery J, Adams WG, Bennett B. Epidemiological study of respiratory disease in workers exposed to polyvinyl chloride dust. Thorax 1980; 35: 644–652.
37. Mundt KA, Dell LD, Austin RP, Luippold RS, Noess R, Bigelow C. Historical cohort study of 10,109 men in the North American vinyl chloride industry, 1942–1972: update of cancer mortality to 31 December. Occup Environ Med 2000; 57: 774–781.
38. Applied Epidemiology Inc. Epidemiological study of men employed in the vinyl chloride industry between 1942 and 1972. I. Re-Analysis of Mortality through December 31, 1982. II. Update of Mortality through December 31, 1995: Final Report Prepared for the Vinyl Chloride Panel. Amherst, MA: Chemical Manufacturers Association. AEI, 1999.
39. Teta MJ, Ott MG. A mortality study of a research, engineering and metal fabrication facility in Western New York State. Am J Epidemiol 1988; 127: 540–551.
40. Blair A, Haas T, Prosser R, Morrissette M, Blackman K, Grauman D, van Dusen P, Moran F. Mortality among United States Coast Guard Marine inspectors. Arch Environ Health 1989; 44: 150–156.
41. Ward EM, Ruder AM, Suruda A, Smith AB, Fessler-Flesch CA, Zahm SH. Acute and chronic liver toxicity resulting from exposure to chlorinated naphthalenes at a cable manufacturing plant during WW II. Am J Ind Med 1996; 30: 225–233.
42. Lilis R, Anderson H, Miller A, Selikoff IJ. Pulmonary changes among vinyl chloride polymerization workers. Chest 1976; 69 (suppl 2): 299–303.
43. Mastrangelo G, Manno M, Marcer G, Bartolucci GB, Gemignani C, Saladino G, Simonato L, Saia B. Polyvinyl chloride pneumoconiosis: epidemiological study of exposed workers. J Occup Med 1979; 21: 540–542.
44. Baser ME, Tockman MS, Kennedy TP. Pulmonary function and respiratory symptoms in polyvinylchloride fabrication workers. Am Rev Respir Dis 1985; 131: 203–208.
45. Ng TP, Lee HS, Low YN, Phoon WH, Ng YL. Pulmonary effects of polyvinyl chloride dust exposure on compounding workers. Scand J Work Environ Health 1991; 17: 53–59.

dose-response relation; epidemiology; liver angiosarcoma; liver cancer; vinyl chloride

© 2001 Lippincott Williams & Wilkins, Inc.