Carcinogenicity bioassays of vinyl chloride monomer: a model of risk assessment on an experimental basis.

Data are presented regarding the final results of the Bentivoglio (Bologna) project on long-term carcinogenicity bioassays of vinyl chloride (VC). The experimental project studied the effects of the monomer, administered by different routes, concentrations and schedules of treatment, to animals (near 7000) of different species, strains, sex and age. To our knowledge this is the largest experimental carcinogenicity study performed on a single compound by a single institution. The results indicate that VC is a multipotential carcinogen, affecting a variety of organs and tissues. In the experimental conditions studied, the neoplastic effects of the monomer were also detected at low doses. The experimental and biological factors greatly affect the neoplastic response to VC. Long-term carcinogenicity bioassays are, at present, a unique tool for the identification and quantification of environmental and occupational risks. Precise and highly standardized experimental procedures are needed to obtain data for risk assessment.


Introduction
The present report deals with the presentation of the final results of our project on the long-term carcinogenicity bioassays of vinyl chloride (VC) (BT project).
To our knowledge this project is the most extensive experimental carcinogenesis study ever performed on one industrial compound by a single institution.

Planning, Materials, Methods and Performance of the Experiment Planning
The experiments of the project were planned (a) to test the carcinogenicity of the compound; *Institute of Oncology and Tumor Center, Bologna, Italy.
(b) to obtain information on the site and type of tumors; (c) to evaluate the possible effects of the routes of administration, with particular regard to the ones reproducing potential human exposure; (d) to assess, in quantitative terms, the level of risk. The planning of the experiments was aimed at achieving these goals.
The compound was tested on animals of different species, strain, sex and age (Table 1), since it is known that these factors may modify the neoplastic response qualitatively and quantitatively. The choice of the animals was made with the intention of having an integrated system of complementary biological models which could express a range, as wide as possible, of neoplastic responses. VC was administered by different routes: intraperitoneal (IP) injection, subcutaneous (SC) injection, inhalation and ingestion (by stomach tube), the latter two being the major routes of potential human exposure.
The monomer was administered at different concentrations: 14 by inhalation levels and 6 ingestion levels for various periods of time, by continuous or intermittent treatment (Table 2).
October 1981 The plan of the project is presented in Tables 3-9.
Material VC was supplied from the same source in all cases, and it contained very low amounts of impuri-ties (Table 10). The oil employed as a vehicle in the ingestion and injection experiments was pure virgin olive oil from Tuscany.
The animals (except for the golden hamsters) were breeds which have been routinely employed in our laboratory for many years. It should be pointed out that, whatever their use, all the animals of our colony undergo periodic examination and complete autopsy, giving us extensive information concerning their pathology.
The chambers for inhalation exposure were built basically of stainless steel and glass.
For the ingestion treatment glass syringes and stainless steel needles with round tips were used.
To control the level of exposure in the inhalation experiments, an automatic gas chromatography system was used.     10 animals of each of the three exposed and control groups the treatment was planned to last 104 weeks, but it had to be stopped because of animal intolerance.

Methods and Procedures
For the experiment on VC, as well as for any other long-term experimental bioassays performed in our laboratory, the procedure has been always the same highly standardized and controlled one. In particular, the following points in our laboratory standard procedures, should be emphasized.
Compounds. All shipments of VC used were examined in order to determine whether they meet the required standards.
Concentrations. The concentrations, particularly when VC was given by inhalation, were controlled by continuous gas chromatographic monitoring . 6 Modalities of Treatment. Treatment was always performed by the same people. This is particularly important for gavage, since the animals become accustomed to the same operator.
Control of the Animals. The conditions of the animals was checked three times daily. Every two weeks the animals were examined to detect any gross changes.
Weight of the Animals. The animals were weighed every two weeks during treatment and every eight weeks after the end of treatment.
Duration of the Experiments. In the VC project, as in any other long-term bioassays performed in our laboratory, the animals were kept alive until spontaneous death.
Autopsy. Full autopsy was performed on each animal. All parts of the body were explored, including the central nervous system. Specimens for histology included the brain, Zymbal glands, interscapular brown fat, salivary glands, tongue, lungs, liver, kidneys, adrenals, spleen, pancreas, stomach, intestine, bladder, uterus, gonads and any other organ with pathological lesions.
Histology. Specimens were trimmed in the standard way. Sections were routinely stained with Haematoxylin-Eosin and, when necessary, with special techniques.
Histopathological Examination. All sides were screened by a junior pathologist and then reviewed Environmental Health Perspectives by a senior pathologist . The same classification of and the gross and microscopic observations were  the lesions were used by all pathologists. classified and coded following our laboratory codes Classification ofData. All the anatomical sites (Tables 11-13).  Presentation of Pathological Data. The results of all VC experiments, as well as those of any other experiment performed in our laboratory, will be presented in the final report (now in press) with the same types of tables, in the same sequence.
This type of presentation has been made possible by the knowledge of the basic pathology of the animal used, which enabled us to make an approximated census of the expected lesions.
Such a procedure permits a quick comparison among the results of different experiments of the same project and possibly of the results of projects studying different compounds.
Interpretation of the Data. The data were subjected to statistical analysis. Although statistical analysis provides an extremely important tool for interpreting the meaning of the results of long-term bioassays, it should be stressed that there may be smaller differences between exposed and control groups which do not reach statistical significance, while these differences could still have meaning from an oncological point of view (particularly in the case of tumors which are infrequent in the animal colony). October 1981 Therefore, the most important data should be commented on both in the light of the statistical analysis performed and from a biological point of view.
The methodological protocol adopted meets the requirements of the recent Good Laboratory Practice Act.

Results
Part of these results, namely those dealing with seven basic experiments on the effects of long-term exposure to a range of 14 doses by inhalation (from 30,000 to 1 ppm) and of six doses by ingestion (from 50 mg to 0.03 mg/kg bw), on Sprague-Dawley rates, were presented previously (1,2). A report, for limited circulation, dealing with part of the results has also appeared (3).
The results of the whole project, with detailed tables, will appear in a monograph which will encompass data on survival rate, body weight, regressive and inflammatory changes, benign and malignant tumors, neoplastic precursors, and the most important proliferative changes.
With this report we are presenting only tables summarizing the most outstanding results and information, and what we do believe to be the integrative documentation and strictly necessary comments. 9 Tables 15-31 presented data on the incidence of the tumors which have been considered as dependent or possibly correlated to VC exposure, in 17 different experiments on the effects of VC in different animal systems, by different routes, at different doses and with various schedules of treatment. Explanations of abbreviations used in the tables are given in Table 14.
The possible leukemogenic effect of VC in golden hamsters is expressed both by the slight increase in incidence but more by the decrease in latency time (from 16 weeks in animals treated at 10,000 ppm to 36 weeks in control animals).
Examples of the most characteristic microscopic features of these tumors were given in a previous publication (4).
The data on dose-response relationship in longterm treatment experiments, by inhalation and by ingestion, in Sprague-Dawley rats, Wistar rats and Swiss mice, with reference to the incidence of total malignant and benign tumors, and the most impor-10 tant neoplasias observed, are shown in TIables 32-63.
The striking effect of the influence of scheduled treatment is pointed out by the results shown in Table 64.
Examples of the marked influence of the animals used in determining the neoplastic response are shown in Tables 65-67, which point out the effects of species, strain and age.

Conclusions
VC-dependent tumors are identified on the basis of one or more of the following parameters: (a) sharply enhanced incidence; (b) rare or exceptional occurrence in the colony of the animal used; (c) dose-response relationship; (d) association of precursor lesions.
From the presented data the following conclusions may be drawn.
(1) VC causes tumors in all the different animal systems tested.
(2) VC is a multipotential carcinogen, since it causes tumors of different types in different sites (Table 68).
(3) Some types of tumors are observed in all the animals studied, i.e., liver angiosarcoma, whereas others are observed in only one animal system.
(4) The degree of evidence of correlation between VC treatment and the tumors considered as VC-dependent varies from tumor to tumor.
(5) VC shows carinogenic effects both when given by inhalation and ingestion and possibly by injection. (6) Both through inhalation and ingestion experiments there is a clear-cut dose-response relationship.
(7) The duration of treatment and schedule of treatment greatly affects the neoplastic response. (8) The neoplastic response, in qualitative and quantitative terms, is greatly affected by the species, the strain and the sex of the animals studied.
(9) Newborn animals appear to be extremely responsive and easily develop liver tumors, both hepatocarcinomas and angiosarcomas.
(10) VC produces carcinogenic effects on embryos via the placenta.
(11) With the above criteria for identifying VCdependent tumors, VC shows carcinogenic effects even at low doses, namely down to 50 ppm and less. (12) The results of the seven basic experiments studying the effects of doses of VC as given by inhalation (BT1, 2, 6, 9, 15), and ingestion (BTi1, 27), have been subject to statistical analysis following the Fisher exact probability test (p -0.05). The total cancer-bearing animals and the tumors Environmental Health Perspectives  Liver angioma Extra-liver angiosarcoma Extra-liver angioma Nephroblastoma Neuroblastoma Angioblastic hyperplasia in liver Angioblastic dysplasia in liver Neoplastic nodules of liver Nodular hyperplasia of liver Diffused hyperplasia of liver Marked Very marked aThe incidence of total malignant and benign tumours is given as the total number of tumors per 100 animals (one animal may bear more than one malignant or benign tumor) on the basis of the tumors observed among the animals alive, when the first tumor was observed in the experiment.
The incidence of specific tumour is given, as percent of the animals bearing the tumor considered, referred to the animals alive when the first tumor was observed (in parentheses).
October 1981 significantly in excess in these experiments, in relation to dose, are given in Tables 69 and 70.
The Fisher exact probability test at 95% confidence is, in relation to the above, not "sensitive" enough, in our experimental conditions. Biologically, in our opinion, the following results, although not statistically significant according to the test used, should be given proper attention.
Extrahepatic angiosarcomas of different sites are observed at a very low incidence dose in untreated Sprague-Dawley rats of our colony. Results of experiments BT1 and particularly BT9, however, strongly suggest a relationship between these tumors and VC exposure. This relationship is supported by the excessive incidence of extrahepatic vascular tumors in mice treated with VC (BT4).
Few cases of hepatomas have been observed in treated groups, particularly in BT1. This tumor is exceptionally rare in our colony of animals, and none have been observed in the control group of the 17 experiments. Moreover the relationship with treatment is supported by the fact that a high incidence of hepatomas has been observed in Sprague-Dawley rats, following neonatal exposure to a high dose for a short period (BT14).
In view of their rareness or nonobservation in the colony of animal used, for the following tumors it should be stressed that attention should be paid to                       Liver angiosarcomas are extremely rare in the colony used (4 cases over several thousand untreated animals). Therefore, one must consider the onset of these tumors as important even at doses not shown by statistical analysis, and particularly below 50 ppm (5 liver angiosarcomas out of 120 animals at 25 ppm, and 1 liver angiosarcoma out of 20 120 animals at 10 ppm), and at 1 mg/kg (3 liver angiosarcomas out of 150 animals), and at 0.3 mg/kg (1 liver angiosarcoma out of 150 animals).
The onset of a few nephroblastomas observed after inhalation treatment at doses below 100 ppm and in groups treated by ingestion with 50 and 16.65 mg/kg, is not casual in our opinion, given the extreme rarity of these tumors in rats.
Neuroblastomas have never been observed by us, up to the present, in the Sprague-Dawley rats used in our laboratory as control or otherwise Environmental Health Perspectives  treated. Therefore we consider as dependent on treatment the onset of these tumors, even at doses below 10,000 ppm, i.e., 6000 and 2500 ppm.
The meaning in oncological terms of the results at the lowest doses may be better evaluated in considering, not singly, but together, the tumors found to be VC-dependent (Table 71).
None (or no increase) of the specifically VC related tumors shown in Table 71, observed in the seven basic experiments, was found at doses of 5 and 1 ppm (by inhalation) and 0.03 mg/kg (by ingestion).

General Comments
VC long-term experimental study led to the discovery of VC carcinogenicity, and as a direct consequence, to what probably has been the greatest effort ever made at controlling the exposure to an industrial carcinogen in the workplace (Table  72).
Moreover, long-term carcinogenicity bioassays on VC are a crucial step in the field of environmental and occupational carcinogenesis which, in turn,  possible target organs and, in general terms, on the quality of neoplastic response; may represent a tool for obtaining information on the relative risk represented by different compounds, provided that they are tested under the same standard conditions (Table 73); have revealed the need to identify Environmental Health Perspectives   animal systems more equivalent to humans in neoplastic response, which in turn depends on partly-known factors, such as basic "spontaneous" tumorigram and enzymatic profiles.

Prospects
At present the most important goal of research on environmental and occupational carcinogenesis is, in our own view, the extrapolation of results     from animal to human, both in qualitative and in quantitative terms. VC carcinogenicity may again provide an important tool towards solving this problem.
We now know a great deal about the effects of VC in experimental animal systems, both in qualitative and quantitative terms.
On the other hand, epidemiological investigations October 1981  sure (so as to define homogeneous exposed groups), and collect all possible available data on pathology, we shall have an opportunity, unique at present, to compare animal and human data, both in qualitative and quantitative terms, and to help find a possible key for extrapolating from animals to humans.

The Cost
With the presentation made in Paris last November (2) and with today's report, ten years of work on our VC experimental project seem to be nearly concluded. After having presented the results, we also wish to present the data of the cost of the project, which cannot be expressed only in financial terms.
The cost of the BT project of long-term carcinogenicity bioassays on vinyl chloride includes the cost of (1) the planning and setting-up of experimental apparatus, including inhalation facilities, of of LAS and Zymbal gland CA in Sprague-Dawley rats in relation to schedule of treatment with VC administered by inhalation.  (2) the study of nearly 7000 animals up to the point of their natural death, equivalent to more than 3,000,000 rodent days; (3) ten years of work; (4) the routine examination of some 200,000 histological slides; (5) a financial commitment equivalent to more than $2,000,000 U.S. at present prices (the average cost of a rat throughout the world in this type of experiment is $300 U.S.); (6) the availability of the same team of scientists throughout the entire 10 years of the project, a prerequisite which may be difficult or even impossible to ensure in many countries at the present time; (7) the highly motivated commitment of those scientists to a type of work which is long-lasting, onerous and often tedious; (8) the effort involved in maintaining the   VC was found to produce liver enlargement and microscopic hepatic degenerative changes (5) 1970 Zymbal gland carcinomas were reported in rats exposed to 30,000 ppm of VC, by inhalation (6) 1970 An increase in atypias in respiratory cells was observed among workers heavily exposed to VC (7) July 1971 A vast project of long-term carcinogenicity bioassays on VC was started in Bentivoglio, near Bologna, Italy (BT project) August 1972 Zymbal gland carcinomas, nephroblastomas and liver angiosarcomas were observed in rats exposed to VC by inhalation (Maltoni, BT project) April 1973 The first data of the BT project were released to the scientific community: the oncogenic effect was observed up to 250 ppm (4) 1973 Splenomegalic liver disease was found among poly(vinyl chloride) production workers (8) December 1973 For the first time a case of liver angiosarcoma in a poly(vinyl chloride) production worker was correlated to VC exposure (9) February 1974 On the basis of the BT project data indicating a carcinogenic effect at 250 ppm, OSHA proposed a TLV of 50 ppm February 1974 The BT project data showed that VC is a multipotential carcinogen, producing a variety of tumors, in different animal species 1974 The BT project data indicated a carcinogenic effect at 50 ppm (10); OSHA proposed new stricter rules

1974
Early epidemiological observations (paralleling the experimental information) indicated an increase in tumors other than liver angiosarcomas (of brain, lung, liver, hemolymphoreticular tissues) among workers of VC-PVC industries (11)

1974-75
BT project data showed that VC had carcinogenic effects in rats also when given by ingestion (12) 1976 In rats of the BT project exposed to VC by inhalation, angiosarcomas were observed down to the level of 25 ppm, and Zymbal gland carcinomas down to the level of 10 ppm (13) consistency of the methodology, which has as its reverse side the limits placed on the exercise of imagination-the most positive element in scientific life; (9) the effort involved in establishing and preserving objectivity and balance in the evaluation and interpretation of data; (10) and finally, the strength required to withstand the sense of loneliness arising from the lack of co-operation of many of those bodies which should properly be concerned with the progress of science in this field, not excluding part of the scientific community whose Environmental Health Perspectives indifference sometimes degenerates into frank hostility.
The high costs probably represent the reason why, in the field of experimental and environmental carcinogenesis, words overlap facts, opinions overlap data, and meetings and commissions reports submerge good laboratory work.