WHO FOOD ADDITIVES SERIES: 52

First draft prepared by

Dr S. BrookeTaylor
BrookeTaylor & Co. Pty Ltd, Woonona, Australia

and

Dr P. Verger
National Institute for Agricultural Research, SAFE Consortium on Food Safety, Brussels, Belgium

Polyvinyl alcohol is a synthetic resin that is prepared by polymerization of vinyl acetate, followed by partial hydrolysis of the resulting ester in the presence of an alkaline catalyst. The number of acetate groups in polyvinyl alcohol is determined by the degree of hydrolysis (86.5–89.0% hydrolysis for this food additive specification). Polyvinyl alcohol is used as a coating, binder, sealing and surface finishing agent in food products such as dairy-based desserts, confectionery and cereal products and dietary supplement tablets, in the range of 0.2–1.8% by weight.

Polyvinyl alcohol has not been evaluated previously by the Committee.

Polyvinyl alcohol (synonyms, vinyl alcohol polymer, PVA, ethenol homopolymer) is a water-soluble synthetic resin, prepared by the polymerization of vinyl acetate, followed by partial or complete catalysed hydrolysis. The primary raw material used in the manufacture of polyvinyl alcohol is vinyl acetate monomer dissolved in methanol. The polymerization involves the presence of two proprietary cataytic agents. After polymerization, the material undergoes controlled hydrolysis with aqueous sodium hydroxide, during which the ester groups in the vinyl acetate are replaced with hydroxyl groups. Polyvinyl alcohol is precipitated, washed and dried to form an odourless, tasteless, white or cream-coloured granular powder.

The physical characteristics of polyvinyl alcohol vary depending on the degree of polymerization and hydrolysis. Polyvinyl alcohol is classified into grades of partially and fully hydrolysed polymers with different intended functional uses. Polyvinyl alcohol has a history of use in cosmetic, food packaging materials, pharmaceutical and medical applications.

2.1.1 Absorption, distribution and excretion

(a) Oral administration

Three male Fischer 344 rats received water containing 0.01 mg of ¹⁴C-labelled polyvinyl alcohol, by oral gavage. More than 98% of the administered dose was excreted in the faeces within 48 h, with trace amounts (0.18%) detected in the urine. No radioactivity derived from polyvinyl alcohol was detected as expired carbon dioxide (CO₂).

In a second study, three male Fischer 344 rats received 0.1 mg of ¹⁴C-labelled polyvinyl alcohol by oral gavage, daily for 10 consecutive days. Twenty-four hours after the final dose, the quantity of radioactivity recovered from the blood, liver, kidneys, skin, muscle and adipose tissues accounted for 0.05% of the total administered dose. The principle route of excretion was in the faeces, with <0.2% of the administered dose being detected in the urine (Sanders & Mathews, 1990).

The 36th Report of the Cosmetic Ingredient Review Expert Panel reported a study attributed to the Haskell Laboratory (1960; original source documents not provided) in which groups of six Charles River rats received 100 mg (equivalent to a dose of 250 mg/kg bw) of polyvinyl alcohol (commercial grade, non-radioactive) daily for 7 days, followed by 100 mg per day of: (a) high viscosity, unhydrolysed polyvinyl alcohol labelled with ¹⁴C, average relative molecular mass, 148 000; or (b) high viscosity, polyvinyl alcohol, average relative molecular mass, 150 000; or (c) low viscosity, polyvinyl alcohol, average relative molecular mass, 32 000. Three rats per group were killed after receiving five doses of the test material, and three rats were killed after receiving 10 doses. Urine and faeces were collected each day for analysis. A similar study was carried out using groups of four male mongrel dogs that received 2 g of polyvinyl alcohol per day (equivalent to 200 mg/kg bw per day). No radioactivity was detected in the urine of any animals. For the high viscosity preparations (a) and (b), trace amounts (<1 ppm) of radiolabel were detected in the brain, kidney and liver of both rats and dogs. For the low viscosity solution of polyvinyl alcohol, (c), marginally larger quantities of radiolabel were detected in these organs (brain, 0.60–1.29 ppm; kidney, 0.52–1.35 ppm; liver, 1.21–6.91 ppm) (Cosmetic Ingredient Review Expert Panel, 1998).

(b) Intravenous administration

Groups of female BALB/cCrS1c mice aged 8–12 weeks received ¹²⁵I-labelled polyvinyl alcohol with relative molecular mass ranging from 14 800 to 434 000, by injection into the tail vein. The half-life for excretion varied from 90 min for the polyvinyl alcohol of low relative molecular mass, to 23 h for polyvinyl alcohol of highest molecular weight. The principal route of excretion was in urine. Almost 80% of polyvinyl alcohol of low relative molecular mass (14 800) was excreted by this route within 30 min of administration. The slower rate of excretion of the high-molecular-weight polyvinyl alcohol appeared to be caused by lower permeability at the renal glomeruli that was related to high relative molecular mass (Yamaoka et al., 1995).

Female Fischer 344 rats received a single intravenous injection of ¹⁴C-labelled polyvinyl alcohol (0.1 mg/kg bw) in the tail vein. Sixty-four percent of the dose was excreted in the urine and 3% in the faeces within 24 h. Seventeen percent of the dose was retained in the liver at 24 h, decreasing to 12% by day 3 and 4% by day 10. A further 2% was excreted in the urine between days 1 and 10. Faecal excretion accounted for 5% of the administered dose by day 3 and 13% by day 10. There was no evidence for metabolism of polyvinyl alcohol (Sanders & Mathews, 1990).

(c) Intravaginal administration

Five female Fischer 344 rats received a single dose of 0.5 mg of ¹⁴C-labelled polyvinyl alcohol (equivalent to 3 mg/kg bw) intravaginally. The animals were killed after 4 days and tissues were examined. There were marked variations in the amount of residual radioactivity detected in the vagina (from trace amounts to 19% of the administered dose). Most of the radioactivity derived from polyvinyl alcohol was excreted in the urine or faeces. Although low (0.1–0.8% of the dose), absorption of polyvinyl alcohol administered by this route was greater than that of polyvinyl alcohol administered orally. The greatest quantity of radiolabel (0.8%) was reported in the liver. In follow-up studies, groups of rats received 1, 3 or 10 daily doses of ¹⁴C-labelled polyvinyl alcohol (3 mg/kg bw). Blood was collected from animals in each group for up to 30 days after the last dose. Radioactivity derived from polyvinyl alcohol was very low but increasingly concentrated in the liver over time and, to a lesser extent, in the kidneys, muscle, skin, adipose tissue and spleen. The peak concentration in the liver, 24 h after the administration of 10 consecutive doses of 3 mg/kg, was <2 ppm. Concentrations of the polyvinyl alcohol only decreased slowly after the cessation of dosing (Sanders & Mathews, 1990).

2.1.2 Biotransformation

No information was available.

2.1.3 Effects on enzymes and other biochemical parameters

No information was available.

2.2.1 Acute toxicity

The results of studies on the acute toxicity of polyvinyl alcohol are summarized in Table 1.

Table 1. Acute toxicity of polyvinyl alcohol

2.2.2 Short-term studies of toxicity

Mice

In a study undertaken by the Pharmacology Department of Osaka University and reported in the Japanese Standards of Cosmetic Ingredients (JSCI), 10 mice received polyvinyl alcohol at a dose of 500 mg/kg bw per day for 20 days. The route of administration was not specified. Other than one death, no treatment-related adverse effects were observed. The report contains no original data suitable for evaluation purposes and the relative molecular mass and percentage hydrolysis of the polyvinyl alcohol used were not identified (Japanese Standards of Cosmetic Ingredients, 1968).

In another study by the same researchers, 30 mice received polyvinyl alcohol at a dose of 100, 500 or 1000 mg/kg bw per day for 26 weeks. No signs of toxicity were seen and no differences were observed in either growth rate or body-weight gain, compared with controls. No toxicologically significant histopathological changes in stomach, intestines, heart, lungs, liver, kidneys, pancreas, thyroid and spleen were reported. The report contains no original data suitable for evaluation purposes and the relative molecular mass and percentage hydrolysis of the polyvinyl alcohol were not identified.

Two groups of 50 female B6C3F₁ mice received 20 µl of a 25% aqueous solution of polyvinyl alcohol by intravaginal administration, daily for 30 days. Animals in one group were returned to their cages immediately after dosing. Animals in the other group were restrained in a vertical nose-down position for several minutes after dosing. A control group of 50 mice received 20 µml of deionized water only. No lesions related to treatment with the test material were observed (National Toxicology Program, 1998).

Polyvinyl alcohol sponges were inserted subcutaneously into male Bar Harbor C57 mice. Biopsies were taken at 1, 3, 5, 6, 9 and 18 weeks. The sponges became more resistant to cutting at each biopsy. There was evidence that fibroblast capsules were extending into the sponges, accompanied by considerable collagenous material. Vessels were observed extending into the sponges at 6 weeks. No evidence of toxicity around the implant site was reported. This study was a published report with no original data suitable for evaluation. The relative molecular mass and percentage hydrolysis of the polyvinyl alcohol were not identified (Moore & Brown, 1952).

None of the available short-term studies in mice are relevant to the current evaluation of polyvinyl alcohol.

Rats

Groups of 10 rats received one of two preparations of 0.5% polyvinyl alcohol in drinking-water for 8 weeks (equivalent to 116 or 151 mg per day or an average of 500 or 700 mg/kg bw per day). The relative molecular mass and degree of hydrolysis of the polyvinyl alcohol used were not specified. Two animals at one dose died in the seventh and eight weeks respectively. All animals in the other dose group lost condition and body weight during the study. No macroscopic pathology related to treatment with the test material was observed in the surviving animals. The absence of adequate controls prevented interpretation of sporadic lesions reported to occur in the brains, lungs and kidney of individual animals. The study was provided as results only and was not certified for GLP or quality assurance (QA). A NOEL was not identified in this study (Haskell Laboratory, 1936).

Forty-six female Holtzman rats, divided into four groups received one of three grades of polyvinyl alcohol (relative molecular mass, 37 000, 133 000 or 185 000) as a 5% solution in physiological saline or saline by subcutaneous injection, daily for 4 weeks. Animals were given distilled water containing 1% sodium chloride throughout the study. Animals were killed on days 29 or 30 and tissues were taken for weighing and for histological examination. All rats in the group given polyvinyl alcohol of relative molecular mass 133 000 had polydipsia, progressive severe hypertension and body-weight loss. By comparison, animals in the groups receiving polyvinyl alcohol of relative molecular mass 37 000 or 185 000 showed a small increase in blood pressure compared to the control group. At sacrifice, accumulation of ascitic fluid in the abdominal cavity, enlarged, pale and granular kidneys with evidence of punctuate surface haemorrhage, glomerulonephritis, and enlargement of the heart, liver and spleen were observed in animals given polyvinyl alcohol of intermediate relative molecular mass. Less marked hepatosplenomegaly and renal and cardiac enlargement were also reported in the group given high-molecular-weight polyvinyl alcohol, but tissues from this group were not otherwise reported to be grossly abnormal. No abnormalities were reported in the group given low-molecular-weight polyvinyl alcohol. Histological examination indicated severe renal, cardiac and splenic arterial lesions in the group given intermediate-molecular-weight polyvinyl alcohol, consistent with hypertension. The authors suggested that the observed pathological effects were dependent on the relative molecular mass of the polyvinyl alcohol rather than on its chemical structure. The study was a published report with no original data suitable for evaluation (Hall & Hall, 1963).

Four rats received 2 g of polyvinyl alcohol in 45 g diet (equivalent to 4400 mg/kg bw per day) for 2 weeks, followed by 4 g of polyvinyl alcohol in 45 g diet (equivalent to 8800 mg/kg bw per day) for 2 weeks. Two animals were killed after 4 weeks. The remaining two animals received 10 g of polyvinyl alcohol in 25 g diet (equivalent to 28 600 mg/kg bw per day) for a further 2 weeks and were killed after week 6. No toxicity related to treatment with polyvinyl alcohol was reported in the two animals killed at week 4. Hepatic hydrophobic degeneration was noted in the two animals killed at week 6. The study was a published report with no data suitable for evaluation. A NOEL was not identified in this study (Hueper, 1939).

Groups of 20 male and 20 female outbred Sprague-Dawley rats were given diets containing polyvinyl alcohol at a dose of 0, 2000, 3500 or 5000 mg/kg bw per day for at least 90 days. The study was conducted in compliance with USDA regulations for GLP (Part 58 of 21 CFR), EEC GLP regulation 99/11/EEC and OECD GLP principles. Ten animals of each sex per group were examined for neurobiological effects before the start of treatment and between days 88–91. Blood was obtained from up to 10 animals of each sex per group on days 28, 56 and at termination (days 95 and 96). Unformed stools accompanied by anogenital staining, were observed in males at 3500 and 5000 mg/kg bw per day. These effects were considered to be related to the high content of unabsorbed test material in the diet and not to adverse toxicological effects. No deaths related to treatment with the test material occurred during the study. All animals were killed between days 92 and 99. No significant differences in body weight, neurobehaviour, haematology, blood coagulation, clinical chemistry, urine analysis, organ weight or macroscopic and microscopic pathology were observed. Food consumption in the treated groups was comparable or slightly higher than that of controls throughout the study. The NOEL was 5000 mg/kg bw per day (Huntingdon Life Sciences, 2000a).

Two groups of 10 rats received one of two preparations of polyvinyl alcohol at a dose of 500 mg/kg bw every second day for six doses (average dose, 250 mg/kg bw per day). The preparations were stated to differ in relative molecular mass and degree of hydrolysis, but details were not provided. One animal from one group died 24 h after the first injection and one animal from the second group after the sixth injection, apparently from pre-existing conditions not related to treatment. The remaining animals had oedema and necrosis at the injection site. A number of animals showed abnormal lung pathology and a majority of animals exhibited abnormal renal pathology. However, the absence of adequate controls for this study did not allow a thorough evaluation of these findings to be made. The study was a research report that did not appear to be certified for GLP or QA, and no raw data were suitable for evaluation (Haskell Laboratory, 1936).

In a subsequent study, groups of 10 rats were given one of two preparations of polyvinyl alcohol, or a mixture of equal concentration, at a dose of 500 mg/kg bw per day for up to 16 days. The preparations were stated to differ in relative molecular mass and degree of hydrolysis, but details were not provided. No deaths occurred during the study. Animals exhibited swelling of the subcutaneous tissue at the injection site. Abnormal lung pathology and renal pathology was observed in several animals (number not stated). Degenerative testicular changes were also seen in a considerable number of animals. The absence of adequate controls for this study prevented a thorough evaluation of these findings from being made. A NOEL was not identified in these studies.

Groups of 10 rats, dehaired on the back, received one of two preparations of polyvinyl alcohol at a dose of 120 mg per day, 5 days per week, for 8 weeks. The preparations were stated to differ in relative molecular mass and degree of hydrolysis, but details were not provided. Animals lost hair at the application site during treatment, possibly owing to the application of the test material with adhesive tape. The lack of adequate controls, however, did not permit evaluation of this effect. The report states that all animals were killed at the end of the treatment period, but also states that 1 week after the last treatment, all animals showed even hair growth. No skin lesions or inflammatory reactions were seen at the application site. Animals killed at an unspecified time after completion of treatment showed no indications of abnormal pathology or histology. The study was provided as results only and was not certified for GLP or QA. A NOEL was not identified in this study (Haskell Laboratory, 1936).

Twelve albino rats aged 2 months received 1 ml of a 5% aqueous solution of polyvinyl alcohol by subcutaneous injection, five times a week for 4 weeks. One rat died at the end of week 3. Six rats were killed after the final injection and the remaining five were killed 2 weeks later. Organs and tissues were removed for macroscopic and histological examination. Necrosis and granulomatous inflammatory tissue was found to be associated with a large amount of polyvinyl alcohol retained at the injection site. Polyvinyl alcohol was also found in the lumens of blood vessels associated with a number of organs, occluding smaller vessels, in particular the capillaries of the lungs associated with endothelial degeneration. Aggregates of polyvinyl alcohol were found in the renal glomeruli. Spleens were moderately enlarged, dark red and firm. All animals showed increased numbers and swelling of Kupffer cells. The absence of adequate controls for this study prevented a thorough evaluation of these findings. The study was a published report with no original data suitable for evaluation (Hueper, 1936).

Groups of 20 male and 20 female outbred Sprague-Dawley rats were given diets containing polyvinyl alcohol at a dose of 0, 2000, 3500 or 5000 mg/kg bw per day for at least 90 days. The study was conducted in accordance with USDA GLP regulations (Part 58 of 21 CFR), EEC GLP regulation 99/11/EEC and OECD GLP principles. Ten animals of each sex per group were examined for neurobiological effects before the test and between days 88 and 91. Blood was taken from up to 10 animals of each sex per group on days 28 and 56 and at termination (days 95 and 96). Unformed stools accompanied by anogenital staining were observed in males at 3500 and 5000 mg/kg bw per day. These effects were considered to be related to the high content of unabsorbed test material in the diet and not to adverse toxicological effects. No deaths related to treatment with the test material occurred during the study. All animals were killed between days 92 and 99. No significant differences in body weight, neurobehaviour, haematology, blood coagulation, clinical chemistry, urine analysis, organ weight or macroscopic pathology were observed. Microscopic examination of an extensive range of organs and tissues from animals in the control group and the group treated with a high dose, including the gastrointestinal tract and Peyer’s patches, and all gross lesions indicated no evidence of treatment-related abnormal pathology. Food consumption in the test groups was comparable to or slightly higher than that of controls throughout the study. This was the only short-term study provided, in any species, that is directly relevant to the safety evaluation of oral exposure to polyvinyl alcohol. The NOEL was 5000 mg/kg bw per day (Huntingdon Life Sciences, 2000a).

2.2.3 Long-term studies of toxicity and carcinogenicity

Groups of 100 female B6C3F₁ mice received 20 µl of deionized water containing 0 or 25% polyvinyl alcohol by intravaginal administration, 5 days per week for 104–105 weeks. A second control group of 100 animals were not treated. The test material was reported to have a relative molecular mass of 24 000 and to be 88% hydrolysed. No evidence of carcinogenic activity related to treatment with the test material was observed during the study. The study was subject to peer review by the National Toxicology Program Technical Reports Review Sub-Committee. The relevance of this study to the current evaluation of oral exposure to polyvinyl alcohol is very limited (National Toxicology Program, 1998).

2.2.4 Genotoxicity

Polyvinyl alcohol and preparations containing polyvinyl alcohol were not genotoxic in a range of studies in vitro and in vivo (see Table 2).

Table 2. Studies of genotoxicity with polyvinyl alcohol

2.2.5 Reproductive toxicity

(a) Multigeneration study of reproductive toxicity

In a two-generation study of reproductive toxicity, groups of 26 male and 26 female rats of both the P₀ and F₁ generations received polyvinyl alcohol in the diet at concentrations of 0, 2000, 3500 and 5000 mg/kg bw per day. Mating started in the P₀ generation at age 85 days and in the F₁ generation at age 99 days. Animals were treated from 70 days before mating, throughout the 14-day mating period, until they were killed. Females were killed on day 14 of lactation. The P₀ and F₁ males were killed after the P₀ and F₁ females.

F₁ and F₂ pups were given a macroscopic physical examination and body weights were recorded on days 4, 7, 14 and 21. On day 21, F₁ pups were randomly selected for reproductive assessment (one pup of each sex per litter).

A macroscopic postmortem examination was carried out on all parental animals. The brain, liver and kidneys were collected from five males and five females selected at random from each group. The right epididymes from 10 males selected at random from each group for each generation were collected and processed for assessment of sperm. Microscopic examination was carried out on selected tissues and organs (reproductive tissues, liver, kidney and brain) from five animals of each sex per group.

F₁ and F₂ pups underwent macroscopic postmortem examination and the brain and thymus were taken from one pup of each sex per litter for determination of organ-weight ratios. Testes and epididymides from one F₂ male pup per litter from each group, and ovaries with oviducts and uterus with vagina from one F₂ female pup per litter from each group were weighed.

No adverse macroscopic effects related to treatment with polyvinyl alcohol were noted in the parental animals. Unformed stools and anogenital staining were observed in males at 3500, 5000 and occasionally at 2000 mg/kg bw per day in both the P₀ and F₁ parental animals. These effects were considered to be related to the high content of unabsorbed test material in the diet rather than adverse toxicological effects and were consistent with the findings from a 90-day study undertaken by the same researchers. Statistically significant decreases in body-weight gain were observed in males of the P₀ generation at a dose of 2000 or 5000 mg/kg bw per day, but not in males of the P₀ generation at 3500 mg/kg bw per day, nor in any of the females of the P₀ or F₁ generations. Except during lactation, food intake was increased in both generations at 3500 and 5000 mg/kg bw per day. The authors considered that this was consistent with maintenance of normal calorific intake. No adverse effects on reproductive performance that were related to treatment with the test material were noted in males (as assessed by mating and fertility indices and sperm assessment) or females (as assessed by mating, fertility and pregnancy indices and data on estrous cycling) of either the P₀ or F₁ generations. There were no treatment-related effects on litter parameters or macroscopic and microscopic observations in either the F₁ or F₂ generations. This study was certified as having been conducted in accordance with USDA GLP regulations (Part 58 of 21 CFR), EEC GLP regulation 99/11/EEC and OECD GLP principles. The NOEL for this study was 5000 mg/kg bw per day (Huntingdon Life Sciences, 2000e).

No information on oral exposure in humans was available.

Polyvinyl alcohol is intended to be used as a moisture-barrier coating in food supplement tablets or in traditional food products, for example, dried fruits to be included in breakfast cereals or nuts to be included in yoghurt.

The quantity of polyvinyl alcohol in the final product depends on the surface area of the product to be coated rather than on its weight. Maximum estimates of levels of use of polyvinyl alcohol were derived by selecting products within each food category with the greatest proportion of moisture-sensitive components, estimating the surface area of those components and assuming that the entire surface area would be coated. In this way, the manufacturer established a common denominator for use of polyvinyl alcohol of 2.3 mg/cm² for all food products.

This additive is typically applied such that the overall weight of a product increases by a maximum of 4%. In the dossier, the manufacturer assumes that 45% of this weight gain represents polyvinyl alcohol, and thus a maximum of 1.8% of the weight of the final product comprises polyvinyl alcohol.

For this monograph, calculations of dietary intake have been performed on the basis of quantities of polyvinyl alcohol added for each food category to be considered, as described in Table 3. The results will therefore be related to finished food products.

Table 3. Proposed use of polyvinyl alcohol in foods

The budget method is used to estimate the theoretical maximum level of a food additive in the proportion of the food and/or beverage supply likely to contain it that would not result in exceedance of the acceptable daily intake (ADI) by the population (Hansen, 1979).

Maximum level of use of polyvinyl alcohol: 1.5 g/100 g or 15 000 mg/kg, intended to be used in a limited number of food categories. Solid food only: 15 000/160 = 93.75 mg/kg bw per day, corresponding to an intake of 5.6 g per day

Calculations were made using data obtained from the French food safety agency (Agence Française de Sécurité Sanitaire des Aliments, AFSSA) and using household data (SECODIP, 1993). The mean individual self-production consump

tion computed from the INSEE food consumption survey in 1991 and the mean individual eatingout consumption (Centre de Recherche pour l’Etude et l’Observation des Conditions de Vie, CREDOC) are added "by translation" to the home consumption (Table 4). The total mean intake corresponds to the sum of the average consumptions. The "high consumer" intake corresponds to the highest intake of polyvinyl alcohol at the 97.5th percentile (confectionery) plus the average intake for other categories.

Table 4. Estimated intake of polyvinyl alcohol based on household surveys in France

a Total population

The following results on the estimated daily intake of polyvinyl alcohol partially hydrolysed from all proposed food categories in the United States by population group were provided by the manufacturer and are based on data from the 1994–1996, 1998 United States Department of Agriculture (USDA) continuing survey of food intakes by individuals (CSFII) (Table 5). The manufacturer’s calculations are confirmed by the use of Market Research Corporation of America (MRCA) mean frequency of eating and USDA data on mean portion size (Table 6).

Table 5. Estimated mean daily intakes of polyvinyl alcohol by the population of the United States

a Consumers only

Table 6. Mean intake of polyvinyl alcohol based on model diets in the United States

To try to assess high levels of consumption, the approach to the monitoring of food additives used by the United Kingdom was employed (Rees & Tennant, 1994). This approach consists of the use of high levels of consumption for broad food categories based on the 97.5th percentile of consumption observed in national dietary surveys in the United Kingdom (Table 7).

Table 7. Estimating high levels of consumption of polyvinyl alcohol, in the United Kingdom

In addition, estimation of dietary intake was done using data on food consumption from Food Standards Australia New Zealand (FSANZ). These results are based on consumption by consumers only and assume that the additive is used at its maximum level for all food products in each food category considered (Table 8).

Table 8. Estimated dietary intake of polyvinyl alcohol based on individual data in Australia and New Zealand

The total consumptions are based on the sum of the category representing the highest intake of polyvinyl alcohol for consumers only, plus the average intake for all respondents for the other food categories.

These calculations overestimate exposure to polyvinyl alcohol by considering that, for the food categories considered, all the food ingested contains the additive. Nevertheless, all the results appear to be consistent with a mean consumption of around 0.5 g/person per day. For high-level consumers, it is not possible to exclude some intakes of >1 g/person per day.

It should be noted that the previous calculations do not include the potential intake of polyvinyl alcohol from food supplements. Assuming a daily ingestion of 10 tablets each weighing 1 g, corresponding to 10 g of food supplement, the additional intake of polyvinyl alcohol would be 180 mg/day.

The Committee examined a large database of studies of the toxicity of polyvinyl alcohol after administration by various routes to a number of species. Much of the information was found to be dated, not relevant to oral administration, or from studies that were not conducted in compliance with GLP or that were conducted with material that did not comply with the specification for polyvinyl alcohol as prepared at the current meeting. Nonetheless, the Committee was able to conclude that polyvinyl alcohol was very poorly absorbed after oral administration, that the acute oral toxicity was generally very low and that, taken as a whole, the results were consistent with very low toxicity and showed no evidence for carcinogenicity.

The Committee also considered a number of recent studies, which had been performed with preparations of polyvinyl alcohol complying with the food additive specification, and which met appropriate standards for GLP. A 90-day study of toxicity in rats treated orally revealed no toxicity with polyvinyl alcohol at doses of up to 5000 mg/kg bw per day. No significant differences in body weight, neurobehaviour, haematology, blood coagulation, clinical chemistry, urine analysis, organ weight or macroscopic pathology were observed. Microscopic examination of an extensive range of organs and tissues, including the gastrointestinal tract, from the control group and the groups given high doses showed no evidence for treatment-related pathology. This was the only short-term study provided, in any species, that was considered to be directly relevant to the safety evaluation of oral exposure to polyvinyl alcohol.

No toxicity was observed in a two-generation study of reproduction in rats in which the parental, first and second generations received a maximum dose of polyvinyl alcohol of 5000 mg/kg bw per day.

In both the 90-day and the two-generation studies, the most notable observations were loose or unformed stools in the groups given higher doses of polyvinyl alcohol, this being attributed to the high intestinal concentration of unabsorbed test material and increased food consumption in these groups. The Committee considered that these observations did not represent adverse effects.

There was no evidence for genotoxicity in a battery of tests undertaken with preparations of polyvinyl alcohol complying with the food additive specification.

The Committee also noted a report that more polyvinyl alcohol was absorbed after intravaginal than oral administration, and reviewed a study involving the intravaginal administration of polyvinyl alcohol to mice, 5 days per week for 104–105 weeks, that provided no evidence for local or systemic carcinogenic activity.

The Committee identified a NOEL of 5000 mg/kg bw per day for polyvinyl alcohol on the basis of the maximum dose tested from both the 90-day and the two-generation studies in rats. The Committee noted the lack of reports of any apparent toxic or carcinogenic effects in studies concerning polyvinyl alcohol as a whole, the very poor absorption of preparations of polyvinyl alcohol complying with the specification after oral administration and the absence of any effects on the gastrointestinal tract in the 90-day study in rats. Despite the absence of long-term studies or studies in a second species, the Committee considered the data adequate for the establishment of an ADI. The Committee therefore established an ADI for polyvinyl alcohol of 50 mg/kg bw per day, on the basis of the NOEL of 5000 mg/kg bw per day from the 90-day and two-generation studies in rats, with a safety factor of 100.

The intake estimate based on use levels provided by the sponsor and national food consumption data (from the USA) shows a mean ingestion of around 0.5 g per day, equivalent to 8.3 mg/kg bw per day for a 60 kg adult. Extreme intakes based on Australian and New Zealand consumption data during one day were shown to reach 2–2.5 g per day at the 97.5th percentile, corresponding to 33 and 42 mg/kg bw per day respectively.

Burford, R.G. & Chappel, C. (1968) Range-finding acute toxicity studies of polyvinyl alcohol, phthalic acid and cellulose acetate phthalate in the mouse. Unpublished report 6713/2 from BioResearch Laboratories Ltd, Point Claire, Quebec. Submitted to WHO by Colorcon, West Point, PA, USA.

Cosmetic Ingredient Review Expert Panel (1998) Final report on the safety assessment of polyvinyl alcohol. Int. J. Toxicol., 17(Suppl. 5), 67–92.

Hall, C.E. & Hall, O. (1963) Polyvinyl alcohol nephrosis: Relationship of degree of polymerization to pathophysiological effects. Proc. Soc. Biol. Med., 112, 86–91.

Haskell Laboratory (1936) The toxicity and potential dangers of polyvinyl alcohols RH-276 and RH-394 and polyvinyl acetate RH-201-U150. Unpublished report by Haskell Laboratory, USA. Submitted to WHO by Colorcon, West Point, PA, USA.

Hueper, W.C. (1936) Organic lesions produced by polyvinyl alcohol in rats and rabbits. Arch. Pathol., 28, 510–531.

Huntingdon Life Sciences (2000a) Polyvinyl alcohol (PVA): A 3-month dietary toxicity study in rats. Unpublished report No. 99-2615 from Huntingdon Life Sciences, East Millstone, NJ, USA. Submitted to WHO by Colorcon, West Point, PA, USA.

Huntingdon Life Sciences (2000b) Polyvinyl alcohol (PVA): bacterial mutation assay. Unpublished report No. 001/994757 from Huntingdon Life Sciences, Huntingdon, Cambridgeshire, England. Submitted to WHO by Colorcon, West Point, PA, USA.

Huntingdon Life Sciences (2000c) Polyvinyl alcohol (PVA): mammalian cell mutation assay. Unpublished report No. 002/002105 from Huntingdon Life Sciences, Eye, Suffolk, England. Submitted to WHO by Colorcon, West Point, PA, USA.

Huntingdon Life Sciences (2000d) Polyvinyl alcohol (PVA): mouse micronucleus test. Unpublished report No. 003/994931 from Huntingdon Life Sciences, Eye, Suffolk, England. Submitted to WHO by Colorcon, West Point, PA, USA.

Huntingdon Life Sciences (2000e) Study of the effects of polyvinyl alcohol (PVA) on fertility, early embryonic development to weaning, and the growth and development when administered in the diet to rats. Unpublished report No. 99-4171 from Huntingdon Life Sciences, East Millstone, NJ, USA. Submitted to WHO by Colorcon, West Point, PA, USA.

JSCI (1968) Bulletin of Osaka Medical College 14 March 1954, 1st Ed., reported by the Japanese Standards of Cosmetic Ingredients.

Moore, A.M. & Brown, J.B. (1952) Investigation of polyvinyl compounds for use as subcutaneous prostheses. Plast. Reconst. Surg., 10, 453–459.

National Toxicology Program (1998) National Toxicology Program technical report on the toxicology and carcinogenesis studies of polyvinyl alcohol (molecular weight 24 000) (CAS No. 9002-89-5) in female B6C3F₁ mice (intravaginal studies). Report No. NTP TR 474 of the National Toxicology Program, Research Triangle Park, NC, USA.

Sanders, J.M. & Mathews, H.B. (1990) Vaginal absorption of polyvinyl alcohol in Fischer 344 rats. Human Exp. Toxicol., 9, 71–77.

Schweikl, H., Schmalz, G. & Gottke, C. (1986) Mutagenic activity of various dentine bonding agents. Biomaterials 17, 1451–1456.

Shibuya, T., Tanaka, N., Katoh, M., Matsuda, Y.T. & Moriata, K. (1985) Mutagenicity testing of ST-film with the Ames test, chromosome test in vitro and micronucleus test in female mice. J. Toxicol. Sci., 10, 135–141.

Yamaoka, T., Tabata, Y. & Ikada, Y. (1995) Comparison of body distribution or poly(vinyl alcohol) with other water-soluble polymers after intravenous administration. J. Pharm. Pharmacol., 47, 479–486.

Zaitsev, N.A. & Sechenov, I.M. (1986) Substantiation of hygienic standards for some polymeric compounds in water with the use of gradual standardization. Gig. Sanit., 10, 75–76.

    See Also:
       Toxicological Abbreviations
       Polyvinyl alcohol (ICSC)