Registration Dossier

Toxicological information


Currently viewing:

Administrative data

Description of key information

No carcinogenic effects were found in animal studies with sulfur dioxide concentrations of 10 and 30 ppm for 21 weeks or at of 10 ppm for 794 days. However, the number of animals used in these studies was insufficient. With regard to carcinogenicity in humans, none of the available studies indicated a causative link between sulfur dioxide exposure and the formation of tumours in the respiratory tract of workers.

Key value for chemical safety assessment

Additional information

1.     Introduction

In this weight of evidence approach, the carcinogenic potential of sulfur dioxide after inhalation is re-evaluated based on (i) existing assessments prepared by various organisations/states, (ii) data already submitted under REACH and the biocidal authorisation application both for SO2and “sulfite” substances, and (iii) new information retrieved from the public domain with the aim to provide a justification for non-classification of sulfur dioxide for this endpoint and waiving of testing requirements.

The published information was collected and assessed according to standard hazard/risk assessment criteria. In addition to animal data, which are typically used for the risk assessment, human data were used as supportive information. Published data on exposure to sulfites is used as additional supportive information, particularly since sulfur dioxide will react with aqueous media leading to formation of sulfites under physiologically and environmentally relevant conditions (please refer to the read across concept in Chapter3.1).

2.     Carcinogenicity after inhalation exposure of sulfur dioxide


2.1.  ATSDR


There is no definitive evidence for an increased cancer potential from sulfur dioxide in humans. Several epidemiological studies have been conducted on copper smelter workers and pulp and paper workers who can be exposed to sulfur dioxide (IARC 1992). However, the studies conducted in copper smelters focused primarily on the association between arsenic exposure and cancer. It has also been difficult to separate the potential effects of sulfur dioxide and arsenic exposures in the studies of copper smelter workers. Studies of the potential carcinogenicity of sulfur dioxide are discussed in the following paragraphs.

A cohort study of 5,403 male copper smelter workers showed that respiratory cancer risks in workers exposed to sulfur dioxide for 212 months were not significantly increased when controls for exposure to arsenic were in place (Lubin et al. 1981). The concentration of sulfur dioxide was not specified in the study. In a follow-up mortality study in the same cohort of smelter workers, a clear dose-response relationship between arsenic exposure and respiratory cancer was demonstrated (Welch et al. 1982). Although an apparent relationship of mortality to sulfur dioxide exposure was observed, workers in the medium and high sulfur dioxide exposure categories had higher exposures to arsenic than those in the low sulfur dioxide exposure groups. The concentrations of sulfur dioxide associated with each exposure category were not specified. The studies by Lubin et al. (1981) and Welch et al. (1982) are limited because exposures to sulfur dioxide and arsenic could not be completely separated.

The results of a case-cohort study of mortality in workers from eight copper smelters in the United States revealed that arsenic was the primary cause of lung cancer in copper smelter workers after adjusting for cigarette smoking and sulfur dioxide exposure (Enterline et al. 1987).

A nested case-control study of lung cancer among 308 workers in a large chemical facility revealed significantly elevated risks for workers with moderate and high potential exposure (1 year) to sulfur dioxide (Bond et al. 1986). For workers who had been exposed to sulfur dioxide, the odds ratio for lung cancer was 1.40. Application of multivariate analyses showed a significant trend (p=0.003) of increasing lung cancer risk associated with increasing intensity of sulfur dioxide exposure for the comparison with the decedent controls. The odds ratios for lung cancer were 0.48 for low exposure, 1.69 for moderate exposure, and 1.45 for high exposure. No trend was apparent for the comparison with the living controls. Also, the risk of lung cancer did not increase with duration of exposure to sulfur dioxide. Confounding effects from exposure to numerous chemicals was also possible. Products manufactured at the facility included chlorinated solvents, plastics, chlorine, caustic soda, ethylene, styrene, epoxy, latex, magnesium metal, chlornitrogen products, and glycols. Employees were exposed to an average of 7.5 chemicals during their career, and the most common exposures included chlorine, sulfur dioxide, hydrogen chloride, and carbon tetrachloride. The study authors indicated that the findings must be interpreted with caution because such findings have not been reported among other workers with similar exposures.

The relationship between ambient air pollution (including sulfur dioxide) and lung cancer has been examined. Lung cancer cases (2,439 males, 765 females) were identified in Helsinki, Finland, and standardized incidence ratios were calculated for 33 subareas of Helsinski for 1975-1978, 1979-1982, and 1983-1986 (Ponka et al. 1993). Mean annual concentrations of sulfur dioxide were 0.005-0.008 ppm. After adjustment for age, sex, and level of education, the lung cancer risk increased slightly, but nonsignificantly, with increasing sulfur dioxide concentration. Lung cancer was 1.3% higher in the subareas with the highest sulfur dioxide concentrations (0.008 ppm) compared with the subareas with the lowest concentrations (<.005 ppm). There was no consistent relationship between the concentration of nitrogen dioxide and the incidence of lung cancer. The study authors concluded that sulfur dioxide had little, if any, effect on the risk of lung cancer.

A retrospective cohort study employing a Poisson regression model for time trends of mortality to detect subtle effects of air pollution on lung cancer mortality in Japan was conducted (Tango 1994). The trend of mortality in females, aged 40-79, was examined in 23 wards of the Tokyo metropolitan area. The size of the cohort was not specified. The ward-specific time trend of nitrogen dioxide and sulfur dioxide concentrations for the years 1972 through 1988 was estimated. The result of this study showed a statistically significant (p=0.0055) association between exposure to nitrogen dioxide and an increased trend of lung cancer mortality. The association for sulfur dioxide exposure was nonsignificant (p=0.0655).

One chronic-duration animal study investigated the potential carcinogenicity of inhaled sulfur dioxide in mice (Peacock and Spence 1967). An experimental group of 35 male and 30 female mice was exposed to 500 ppm sulfur dioxide for 5 minutes/day, 5 days/week, for 2 years. The control group consisted of 41 males and 39 females. Female mice exposed to sulfur dioxide exhibited a significant increase in the incidence of lung tumors (13/30 adenomas and carcinomas versus 5130 in controls; 4/30 primary carcinomas versus none in the controls). The incidence of lung adenomas and carcinomas was also higher in the treated males (15/28 versus 11/35 in controls), but the increase was not significant. The incidence of primary carcinomas in treated males was similar to that of the controls.


These data provide only limited evidence for the carcinogenicity of sulfur dioxide in mice. However, the determination of carcinogenic potential is complicated by study limitations. A dose-response relationship could not be established because multiple doses were not tested. Due to the small group sizes, it cannot be concluded that increased tumor incidences did not result from chance alone. Quality studies in additional species are required before the carcinogenicity status of the substance can be determined.

2.2.  IARC          Human carcinogenicity data


In four US and one Swedish cohort studies of copper smelters, increased lung cancer risks were observed in relation to exposure to arsenic, but no independent effect of sulfur dioxide was seen. One proportionate mortality study from the USA and Canada, as well as two US and one Finnish cohort studies, analysed cancer risks among sulfite pulp mill workers. Three of them suggested an increase in risk for stomach cancer; however, potential confounding factors were not adequately controlled. Lung cancer risks were not elevated in any of these studies. In case-control studies performed at a chemical facility in Texas with a complex exposure environment, increased risks for lung cancer and brain tumours were indicated in workers with high exposure to sulfur dioxide; however, the findings using two different control groups were not consistent.

A population-based case-control study from Canada suggested an increased risk for stomach cancer in men exposed to sulfur dioxide, but no excess was indicated for lung cancer.

No epidemiological study was available on cancer risks associated with exposure to sulfites, bisulfites or metabisulfites.          Carcinogenicity in experimental animals


Sulfur dioxide was tested for carcinogenicity in one study in mice by inhalation exposure. A significant increase in the incidence of lung tumours was observed in females.

Sulfur dioxide was tested for enhancement of carcinogenicity when administered with benzo[a]pyrene in two studies in rats and in one study in hamsters. One incompletely reported study found an increase in the incidence of lung tumours in rats exposed to sulfur dioxide in conjunction with benzo[a]pyrene. The other study in rats suffered from limitations owing to the high incidence of lung tumours in controls given benzo[a]pyrene. The study in hamsters was inadequately reported.          Evaluation


There is inadequate evidence for the carcinogenicity in humans of sulfur dioxide, sulfites, bisulfites and metabisulfites. There is limited evidence for the carcinogenicity in experimental animals of sulfur dioxide. There is inadequate evidence for the carcinogenicity in experimental animals of sulfites, bisulfites and metabisulfites.


Overall evaluation: Sulfur dioxide, sulfites, bisulfites and metabisulfites are not classifiable as to their carcinogenicity to humans (Group 3).

2.3.  MAK Report

2.3.1.     Supplement 1998          Carcinogenicity


In an inhalation study, groups of 35 male and 30 female LX mice were exposed for life to sulfur dioxide concentrations of 500 ml/m3 (purity unknown) for 5 minutes on 5 days a week. 41 male and 39 female animals served as controls. Only mice that survived 300 days or longer were considered in the evaluation, as lung tumours occurred only after this time. In female mice the incidence of pulmonary adenomas and carcinomas was significantly (p = 0.02) increased compared with that in the controls (13/30 compared with 5/30). The incidence of pulmonary carcinomas was not significantly increased (4/30 compared with 0/30). In male animals the incidence of pulmonary neoplasms was not significantly increased (15/28 compared with 11/35), and that of pulmonary carcinomas was not increased (2/28 compared with 2/35) (Peacock and Spence 1967). As a result of the now unusual test conditions of daily exposure for 5 minutes to the very high, strongly irritating concentration of 500 ml/m3—in other studies the LC50 for mice was determined to be 130 ml/m3, and 600 ml/m3 was lethal after exposure for 5 hours (see documentation “Schwefeldioxid” 1974, available only in German)—the relevance of these findings is questionable. Therefore, this study cannot be used for the evaluation.


In an inadequately documented study with rats exposed for a lifetime to sulfur dioxide concentrations of 10 ml/m3, the incidence of lung tumours was not increased. After daily exposure for 1 hour to benzo[a]pyrene concentrations of 10 mg/m3, the incidence of lung tumours was slightly increased. The combined exposure intensified this effect (Table 2; Laskin et al. 1976). A statistical evaluation of the tumour incidences was not carried out.


In a follow-up study with male Sprague-Dawley rats, the incidence of lung tumours was not increased after 21-week inhalation exposure to sulfur dioxide concentrations of 10 or 30 ml/m3 and life-long observation. Intratracheal instillation of 1 mg benzo[a]pyrene in 0.2 ml gelatine increased the incidence of lung tumours to 90% and, in combination with a sulfur dioxide concentration of 30 ml/m3, to 93% (Table 3; Gunnison et al. 1988). Also in this study there were no data regarding the statistical significance of tumour incidences.


In a study available only as a summary, male Syrian golden hamsters were exposed for 104 weeks via the nose to benzo[a]pyrene concentrations of 2 or 10 mg/m3 (superfine B[a]P-coated NaCl particles) alone or in combination with sulfur dioxide concentrations of 172 ml/m3. According to the authors, benzo[a]pyrene alone induced some neoplastic changes; in combination with sulfur dioxide more tumours developed earlier in the upper respiratory tract. The incidence of tumours was not quantified (Pauluhn et al. 1985). This study cannot be included in the evaluation as a result of inadequate documentation.          Manifesto (MAK value/classification)


The available epidemiological studies do not provide any evidence of carcinogenic effects of sulfur dioxide. In long-term inhalation studies with rats, the incidence of lung tumours was not increased after life-long exposure to sulfur dioxide concentrations of 10 ml/m3 and after 21-week exposure to sulfur dioxide concentrations of 30 ml/m3.

After short-term exposure of healthy volunteers to concentrations of up to 0.5 ml/m3 for a maximum of 2 hours, no effects on lung function were observed. In some studies bronchoconstriction was observed in healthy, probably hyperresponsive test persons exposed to sulfur dioxide concentrations of between 0.6 and 2 ml/m3 and in most cases simultaneous exercise. As no changes in lung function parameters were observed in workers exposed to average sulfur dioxide concentrations of 0.67 and 0.78 ml/m3, a MAK value was originally established at 0.5 ml/m3. Classification in Peak Limitation Category I was retained. In view of the studies with volunteers, bronchoconstriction cannot, however, be ruled out at this exposure peak level. The MAK value does not apply for asthmatics, because bronchoconstriction, in particular during physical activity, cannot be ruled out.


2.3.2.     TRGS 900


Recently, the maximum allowable concentration for sulfur dioxide was raised from 0.5ppm to 1 ppm, with the following justification: Justification for raising:


Considering the exposure situation in the Nowak study (1997), where hyperventilation was clearly accelerated to increase the penetration of SO2in the respiratory tract, and the low physical stress and thus less hyperventilation in two other studies (van Thriel et al. 2010 Raulf-Heimsoth et al. 2010) in healthy volunteers with a documented exposure without adverse effect at 2 ppm, the AGW was provisionally set at 1 ppm. Since SO2is a locally acting irritant substance, it is assigned to the peak limit category I. The former exceeding factor of 1 is maintained due to the currently available limited data at moderate physical exertion.



2.3.3.     Supplement 2013


In the supplement 2013, the MAK value was newly established at 2.7 mg/m³ (1 ppm) based on two studies with volunteers exposed to sulfur dioxide from 2010.


2.4.  Evaluationof further existing literature


The results of experimental studies on carcinogenicity via inhalation are summarised in Annex I.

2.4.1.    Animal data


Taking the results from two animal studies on sulfur dioxide (Gunnison, A.F. et al., 1988; and Laskin, S. et al., 1970) into account, there is no evidence that sulfur dioxide has any carcinogenic effect in a concentration range of 10 and 30 ppm for 21 weeks and of 10 ppm for 794 days. However, the number of animals used in these studies were not in line with current guideline standards, thus rendering these results more of an indicative nature.

The studies by Ohyama, K. et al. (1999), Ohyama, K. (1993) and Pauluhn, J. et al (1980) show neoplastic effects in co-exposure with other substances, but no information on sulfur dioxide exposure itself is available. Therefore, these data are not suitable for hazard characterisation.

The study by Peacock, P.R. and Spence, J.B. (1967) shows neoplastic effects but the results are disregarded since the quality of the data are questionable due to the high number of control animals with lung damage. Presumable, the animals were already ill at the time of testing.

The four new studies by Qin, G. and Meng, Z (2006), Bai, J. and Meng, Z. (2009a), Qin, G. andMeng, Z (2009b), Qin, G. and Meng, Z (2010) were also disregarded due to the fact that only the enhancement of proto-oncogenes and tumour suppressor genes were investigated. However, no distinct potential of carcinogenicity was observed since various genes were enhanced or decreased including the enhancement of apoptosis genes. Therefore, it is not possible to predict the impact to thewhole organism and the studies have to be disregarded for hazard characterisation.


2.4.2.    Human data

With regard to carcinogenicity in humans, only those studies were evaluated which investigated cancer risks in populations where sulfur dioxide exposure played a dominant role. This criterion is for example fulfilled by workers engaged in the sulfite pulping process in the paper industry with only limited engagement in other processes. The most important study is the worldwide mortality study in the pulp and paper industry coordinated by the International Agency for Research on Cancer (Lee et al., 2002; IARC), which included sulfite pulp mill workers from 12 countries with at least one year of employment and positively verified exposure to sulfur dioxide. Because of an absence of adequate evidence for carcinogenicity in humans, IARC concluded that sulfur dioxide, sulfites, bisulfites and metabisulfites are not classifiable for their carcinogenicity to humans (Group 3).

Overall, it can be concluded that increased incidences or increased mortality from certain endpoints of cancer were detected also in other studies. However, subjects of these populations were generally either already included in the cohorts of the IARC study, or were deliberately excluded because the available information was insufficient according to the demands of the worldwide study. The evidence in these studies is more limited since the exposure criteria are poorly defined, co-exposures were only partly analysed as possible confounders, and the subject populations were rather small, thus limiting the significance of effects. In summary, increased lung cancer risk (but mostly not significant) was demonstrated repeatedly, whereas results on other cancer types as e. g. stomach, pancreas and kidney were conflicting and not confirmed by the IARC study. Here, co-exposure to other known or suspected carcinogens played a major role such as wood dust, paper dust, chlorine and organochlorines. Potential confounding factors were not adequately controlled (IARC, 1992; entry adopted from the OECD SIAR on sulfur dioxide).


2.5.  Summary


No carcinogenic effects were found in animal studies with sulfur dioxide concentrations of 10 and 30 ppm for 21 weeks or at concentrations of 10 ppm for 794 days. However, the number of animals used in these studies was insufficient in comparison to current guideline standards. With regard to carcinogenicity in humans, none of the available studies indicated any causative link between sulfur dioxide exposure and the formation of tumours in the respiratory tract of workers.

Overall, the epidemiological evidence from well-conducted investigations has not shown that exposure to sulfur dioxide is correlated in any way to a carcinogenic potential for humans.

Therefore, with regard to the existing data and the existing evaluations of various international scientific bodies, no evidence for a carcinogenic potential of sulfur dioxide could be detected which would justify a classification of the substance as carcinogenic.


3.     Supportive evidence


3.1.  Read-across concept for sulfur dioxide, sulfites, hydrogensulfites, in aqueous solution:

Standard, guideline-conform carcinogenicity studies specifically for sulfur dioxide are not available. However, published studies with the hydrolysis products of sulfur dioxide in aqueous media are available, which allow for an assessment of the carcinogenic potential of systemically available sulfite dissociation products under physiological conditions.

Sulfur dioxide reacts rapidly and extensively in water to sulfites and hydrogensulfites, based on the pH-dependant equilibrium in aqueous solutions which is summarised in the following equations[1][2].

SO2+ H2OD`H2SO3´       H2SO3DH++ HSO3-D2H++SO32-              2HSO3-DH2O +S2O52-

In other words, sulfite (2-) and hydrogensulfite (1-) anions are in equimolar equilibrium at most physiological (i.e., neutral) pH values, relevant for systemic effects and local lung effects. Even in the acidic medium of the stomach, a large percentage will be present as hydrogensulfite, and only to a lesser extent as hydrated SO2; the presence of “free” SO2is not feasible under any phsiological conditions.

Since the nature of the cation (i.e., sodium, potassium, ammonium…) is not assumed to contribute substantially to differences in toxicity and solubility (all compounds are very soluble in water), only the chemical and biological properties of the "sulfite" anion are considered as relevant determinants. Based on the described equilibrium correlations, unrestricted read-across between SO2, sulfites, hydrogensulfites and metabisulfites is considered justified.

For this endpoint, read-across from a long-term study withpotassium metabisulfiteis therefore considered to be feasible considering the toxicokinetic properties of sulfur dioxide. For further information on the read-across concept please refer to Doc IIA.


3.2.  Evaluation of existing studies


The results of experimental studies on carcinogenicity via inhalation are summarised in Annex II.


3.2.1.    Animal data:

Taken together the results from three studies on either sodium or potassium metabisulfite (Tanaka et al., 1979; Til et al., 1972; Feron and Wensfoort, 1972), there was no evidence that metabisulfite had any carcinogenic effect. The drinking water study by Tanaka et al. (1979) on potassium metabisulfite in mice is considered as most suitable for the assessment of carcinogenicity, because the size of the experimental groups (50 animals of each sex) corresponded to that recommended in the OECD Guidelines for Carcinogenicity studies (451, 453). The highest concentration of 2% potassium metabisulfite, corresponding to an estimated dose of 2500 mg/kg bw/d K2S2O5, did not indicate any carcinogenic potential. However, there is merely some indication for a possible evidence for a tumour-promoting potential of metabisulfite in glandular stomach carcinogenesis (Furihata et al., 1989; Takahashi et al., 1986), which however does not raise concern in consideration of the overall weight-of-evidence data base.

3.2.2.    Human data:

Only a few studies of pulp and paper mill workers (Milham and Demers, 1984; Robinson, et al. 1986; Anderson, et al. 1998; Rix, et al. 1997) are available which analysed whether the workers might be at an increased risk for several site-specific malignancies. However, analyses of exposure-response relationships were not possible, because no exposure levels were available in any of the studies.

The retrospective cohort study in Danish sulfite pulp mill workers had a 2-fold increased risk for stomach cancer and pancreatic cancer (Rix et al., 1997). Other cancers with elevated risks were leukaemia (SIR 1.84) and soft-tissue sarcomas (SIR 2.37). The increased risk for stomach cancer found in this study was in accordance with that of other studies from sulfite pulp mills. The stomach cancer risk was increased in a retrospective cohort mortality study among workers in American sulfite mills (Robinson et al., 1986). They found 11 cases out of 523 observed deaths (SMR 149) with an increasing risk by time since the first employment. When process-specific analyses were conducted, the risk of lymphosarcoma and reticulosarcoma was increased only for men who had worked in sulfate mills. A proportionate mortality study among pulp and paper workers in the United States and Canada indicated a statistically significant excess risk of stomach cancer (Milham and Demers, 1984). Higher proportionate mortality ratios (PMRs) for lymphosarcoma (statistically significant) and kidney, pancreatic and rectal cancers were associated with jobs in the sulfite process. Hodgkin´s disease deaths occurred primarily in sulfate (Kraft) process workers.

The International Agency for Research on Cancer (IARC, 1992) has evaluated the evidence for carcinogenicity and concluded: There is limited evidence for sulfur dioxide carcinogenicity in experimental animals. There is however inadequate evidence for sulfites, bisulfites and metabisulfites for carcinogenicity in experimental animals.

3.3.  Summary

The available data on long-term oral exposure of experimental animals to sodium and potassium metabisulfite allow an evaluation of the carcinogenic potential of sulfite compounds for humans exposed via the oral route. There was no indication that metabisulfite had any carcinogenic effect itself. Taking into account the applicability of the read-across approach between the different sulfites, the carcinogenicity assessment of the sulfites and hydrogensulfites can be extrapolated also to SO2.

[1]Holleman-Wiberg, Lehrbuch der Anorganischen Chemie, 101.Auflage(refer to supportive report SR-2)

[2]Handbook of Chemistry and Physics, Ed. Lide, DR, 88thedition, CRC Press (refer to supportive report SR-3)

Justification for classification or non-classification

The available data on inhalation exposure of experimental animals to sulfur dioxide as well as on oral exposure of experimental animals to sodium and potassium metabisulfite altogether allow for a weight-of-evidence assessment of the carcinogenic hazard of sulfur dioxide for humans. There is no indications that sulfur dioxide has any carcinogenic effect.

This is supported by the conclusions of various health organisations (IARC, ATSDR, MAK) which likewise do not suggest that classification of sulfur dioxide as carcinogenic is required.