Diarsenic trioxide

Administrative data

Workers - Hazard via inhalation route

Systemic effects

Long term exposure

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

5 µg/m³

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL related information

DNEL derivation method:

other: Oral to inhalation extrapolation using epidemiology data as Point of Departure (POD).

Overall assessment factor (AF):

3

Dose descriptor starting point:

NOAEL

Value:

2.2 µg/kg bw/day

Modified dose descriptor starting point:

NOAEC

Value:

15.4 µg/m³

Explanation for the modification of the dose descriptor starting point:

The POD is a NOAEL of 1.7 µg As/kg bw/day extrapolated - as presented below in the discussion section - from the HEALS drinking water exposure epidemiology study conducted by Ahsan et al. (2006) in Araihazar (Bangladesh). As the reference value is expressed as µg of As/kg bw/day, the value needs to be corrected for the molecular weight of diarsenic trioxide.

Inhalation (chronic) DNEL = 1.7 µg As/kg bw/day, corresponding to 2.2 µg As2O3/kg bw/day.

The POD is then amended for an adult bodyweight of 70 kg and a breathing volume of 10 m3 (workers, 8 h shift) to yield an uncorrected chronic inhalation DNEL, as follows:

Inhalation (chronic) DNEL (uncorrected) = 2.2 µg As2O3/kg bw/day x 70 kg / 10 m3/d = 15.4 µg As2O3/m3 (rounded to 15 µg As2O3/m3).

AF for dose response relationship:

1

Justification:

No correction required.

AF for differences in duration of exposure:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for interspecies differences (allometric scaling):

1

Justification:

No correction required; the POD is an epidemiology study.

AF for other interspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for intraspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for the quality of the whole database:

1

Justification:

No correction required; an extensive database is available.

AF for remaining uncertainties:

3

Justification:

In consideration of the oral-to-inhalation extrapolation, in order to cover any remaining uncertainties also with respect to possible local effects, an assessment factor of 3 is applied to the above derived “uncorrected” DNEL.

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

Local effects

Long term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

Workers - Hazard via dermal route

Systemic effects

Long term exposure

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

112 µg/kg bw/day

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL related information

DNEL derivation method:

other: Oral to dermal extrapolation using epidemiology data as Point of Departure (POD).

Overall assessment factor (AF):

1

Dose descriptor starting point:

NOAEL

Value:

2.2 µg/kg bw/day

Modified dose descriptor starting point:

NOAEL

Value:

15.4 µg/kg bw/day

Explanation for the modification of the dose descriptor starting point:

In consideration of the systemic nature of arsenic compounds carcinogenicity, route-to-route extrapolation from the data obtained from drinking water studies to the dermal route is considered justified. For this purpose, modification of the oral DNEL by considering a dermal absorption factor of 2% (refer to Toxicokinetics section, reference Wester et al., 1993) is appropriate. The chronic dermal DNEL can therefore be derived as follows:

Dermal (chronic) DNEL = 2.2 µg As2O3/kg bw/day (oral) / 2% = 112 µg As2O3/kg bw/day.

AF for dose response relationship:

1

Justification:

No correction required.

AF for differences in duration of exposure:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for interspecies differences (allometric scaling):

1

Justification:

No correction required; the POD is an epidemiology study.

AF for other interspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for intraspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for the quality of the whole database:

1

Justification:

No correction required; an extensive database is available.

AF for remaining uncertainties:

1

Justification:

No correction required.

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

Local effects

Long term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

Workers - Hazard for the eyes

Local effects

Hazard assessment conclusion:

high hazard (no threshold derived)

Additional information - workers

Basis of the risk assessment

The mode of action by which inorganic arsenic produces its non-carcinogenic and carcinogenic effects in various organs and systems is not entirely elucidated and still the subject of intense debate. Based on a wide body of available evidence, inorganic arsenic compounds do not appear to be mutagenic but may cause indirect DNA damage, including chromosomal aberrations, micronucleus formation and sister chromatid exchanges in vitro and in vivo. Because this indirect toxicity was observed to occur generally at high concentrations, above those that would be attained systemically in animals and humans, it does not seem to be the basis for the carcinogenicity of arsenicals, either in animals or in humans, particularly at lower doses (Tsuji et al., 2019). Instead, there is growing evidence for a mode of action involving the formation of reactive trivalent metabolites interacting with critical cell sulfhydryl groups, leading to cytotoxicity and regenerative cell proliferation. The cytotoxicity results in non-cancer toxicities and the cell proliferation enhances the development of epithelial cancers. In other tissues such as vascular endothelium, different toxicities develop, not cancer. This mode of action implies a non-linear, threshold dose-response relationship for both cancer and non-cancer endpoints, requiring sufficient concentrations of trivalent arsenic to disrupt normal cell function (Cohen et al., 2013; Tsuji et al., 2019).

In the present report, based on the fact that diarsenic trioxide appears to have a similar mode of action for cancer and non-cancer effects, carcinogenicity is taken as the lead effect for risk assessment purposes.

When evaluating the results of the animal and human studies, it is important to remember that there are considerable differences in arsenic biotransformation between species and between individuals of a same species (see section on Toxicokinetics). Also, a number of aspects raise issues regarding the usefulness of some studies for quantifying, comparing and interpreting results, especially regarding relevance to humans and risk assessment. These include definition of the arsenic species analysed, concentrations/doses, types of cells, simulation of natural exposure for example (Cohen et al., 2013). Therefore, for the hazard assessment, preference is given to human data to derive safe exposure levels.

The most extensive studies on arsenic involve exposure via drinking water. In line with the threshold concept but adopting a conservative approach, the present risk assessment has selected the NOAEL of 8 µg As/L in drinking water from the study by Ahsan et al.(2006) as a point of departure (POD) as it is the lowest NOAEL identified across the current studies available. In this publication, arsenic exposure was measured in well water at baseline and in urine samples collected at baseline and during follow-up ensuring a proper characterization of the population exposure. This information is therefore used to derive conservative safe exposure levels for all relevant exposure routes. Carcinogenicity of arsenic is of systemic nature, and not dependent on the route of entry into the body. For example, after long-term ingestion of elevated levels of arsenic in drinking water, not only is cancer of the bladder observed, but also of the lungs and the skin.

As discussed in the section on Carcinogenicity, the shape of the dose-response relationship between ingestion of arsenic via drinking water and cancer and the existence (or not) of a threshold are still subject to scientific controversy. The proposed cytotoxicity-based mode of action and a very large body of evidence support the existence of such a threshold. From the data investigated, it seems that inorganic arsenic compounds do not appear to be mutagenic but may cause indirect DNA damage. This basically provides an argument to excludes a linear, non-threshold approach. Although indirect genotoxicity was observed, it generally occurs at “high concentrations” (higher than expected human exposure levels). Therefore, indirect genotoxicity is unlikely to be the unique basis of the carcinogenicity of arsenic either in animals or in humans, particularly at low doses. Instead, a MoA involving the formation of reactive trivalent metabolites interacting with critical cell sulfhydryl groups, leading to an excess of cytotoxicity and regenerative cell proliferation was retained. This implies a “non-linear”, threshold dose-response relationship for both cancer and non-cancer endpoints, with a possible threshold around 100-150 µg As/L in drinking water (based on meta-analysis of existing relevant epidemiology studies). In line with the threshold concept and adopting a very conservative approach by selecting the lowest NOAEL observed in an epidemiology study in which the exposure to Arsenic was appropriately evaluated, a point of departure (POD) for the extrapolation of safe exposure levels (DNEL) in a REACH context was chosen.

For worker exposure, inhalation and dermal DNELs are established. Consumer exposure to arsenic substances can be ruled out since the substances are not made available to the general public as such or in any preparation or mixture. However, since members of the general population could be exposed indirectly to arsenic from industrial emissions via the environment, inhalation, dermal and oral DNELs for the general population have also been established.

Point of Departure (POD) for DNEL derivation

As discussed in the Carcinogenicity section, cancer risk assessment for diarsenic trioxide has traditionally followed a linear dose-response approach using data from drinking water studies at relatively high exposure levels. However, mode of action considerations and the outcome of several recent reviews suggest the existence of a threshold for carcinogenic and non-carcinogenic effects at around 100 µg As/L in drinking water.

In line with the threshold concept but adopting a conservative approach, the present risk assessment has selected the HEALS study by Ahsan et al. (2006) as key for determining the point of departure (POD) for the extrapolation of safe exposure levels in a REACH context for both workers and the general population.

In this study the authors evaluated dose-response relations between arsenic exposure from drinking water and premalignant skin lesions by using baseline data on 11746 participants recruited in 2000-2002. Several measures of arsenic exposure were estimated for each participant based on well-water arsenic concentration and usage pattern of the wells and on urinary arsenic concentration. In different regression models, consistent dose-response effects were observed for all arsenic exposure measures. Control group consists in people with drinking water containing <8.1 µg/L of arsenic. Drinking water containing 8.1-40.0, 40.1-91.0, 91.1-175.0, and 175.1-864.0 µg/L of arsenic was associated with adjusted prevalence odds ratios (OR) of premalignant skin lesions of 1.91 (95% confidence interval (CI): 1.26, 2,89), 3.03 (95% CI: 2.05, 4.50), 3.71 (95% Cl: 2,53, 5.44), and 5.39 (95% CI: 3.69, 7.86), respectively. To take into account water intake, the authors used also a Cumulative Arsenic Index (well concentration * daily consumption * days of use/year).

When considering this value, the median of the NOAEL group is 24 mg and the median of the LOEL group (prevalence OR: 1.83 and range: 1.25-2.69) is 137 mg. This is equivalent per day to 66 and 675 µg/day for NOAEL and LOAEL respectively.

When arsenic urinary excretion is considered, this value the median of the NOAEL group is 48.3 µg/g of creatinine and the median of the LOEL group (prevalence OR: 1.75 and range: 1.23-2.48) is 124.3 µg/g of creatinine. With the hypothesis that urinary excretion is equivalent to 60% of the intake (Buchet et al, 1981) and the creatinine excretion is 1300 mg/day for a 60 kg adult, this is equivalent to 1.7 and 4.5 µg/kg/day for NOAEL and LOAEL, respectively. The NOAEL of 1.7 µg/kg As bw/day is then corrected for molecular weight to 2.2 µg/kg As2O3 bw/day.

This study has identified a large number of people presenting lesions (between 57 and 242 in low and high dose group respectively) and is therefore statistically robust. However, the lesions identified were not malignant and can be considered as very early indicators of arsenic overload. The 2.5% prevalence of lesions in the control group tends to indicate a high sensitivity in the lesion identification. Therefore, the LOAEL in this study can be considered as the median value of the lowest category with an OR of 1.91, i.e. 24 µg/L. This value is close from the NOAEL of 0.8 µg/L derived from Tseng, 1968 (Tseng, W.P., H.M. Chu, S.W. How, J.M. Fong, C.S. Lin and S. Yeh. 1968. Prevalence of skin cancer in an endemic area of chronic arsenicism in Taiwan. J. Natl. Cancer Inst. 40: 453-463) used by EPA (*EPA. 1998d. National emissions standards for hazardous air pollutants for primary lead smelters. U.S. Environmental Protection Agency. Code of Federal Regulations. 40 CFR 63.Fed Regist 63(74)19200. ) to derive acceptable values in drinking water.

Inhalation and dermal DNEL values for workers

As described in the preceding sections, chronic inhalation and dermal DNEL values for workers were calculated using the oral NOAEL of 2.2 µg As2O3/kg bw/day as POD, amending for bodyweight and breathing volume (DNEL inhalation) or dermal penetration (DNEL dermal). The resulting values are:

Workers	Oral	Dermal	Inhalation
Long term, systemic	-	112 µg As2O3/kg/day	5 µg As2O3/m3

Alternative assessment approaches: linear extrapolation from epidemiological data

Because the threshold approach is not unanimously accepted, the results of selected other risk assessment methodologies are presented below, for comparative purposes.

German Ausschuss fur Gefahrstoffe (AGS)

The German AGS (2011) derived an “ERB” (Exposure Risk Relationship) for arsenic exposure in the workplace. They used the study by Lubin et al. (2008) as the basis, with a point of departure associating a workplace concentration of 135 µg/m3 with an added risk of cancer at a level of 6.5%. In the absence of a documented threshold or mechanistic data suggesting otherwise, a linear calculation procedure was applied, acknowledging the uncertainties and inherent conservatism of this approach. The following values were determined for workers:

Tolerance Risk (4 per 1,000) = 8.3 µg As/m³ and

Acceptance Risk (4 per 10,000) = 0.8 µg As/m³

These values are in the same range as the chronic inhalation DNEL presented above.

Committee of the Health Council of the Netherlands (DECOS)

In 2012, DECOS considers lung cancer to be the lead (most sensitive) adverse health effect for arsenic in relation to inhalation exposure. The committee did assess studies of lung cancer in the three copper smelter cohorts for suitability for risk modelling (Lubin et al., 2000; Lubin et al., 2008; Enterline et al., 1995; Järup et al., 1989). DECOS concluded that all these studies had shortcomings, with Lubin et al (2000) being the strongest study with the fewest limitations.

Based on this study, DECOS calculate an excess lung cancer mortality risks for arsenic of:

4 per 1,000 (or 4 x 10^-3) for 40-year occupational exposure to 28 μg/m³

4 per 100,000 (or 4 x 10^-5) for 40-year occupational exposure to 0.28 μg/m³.

Thus, the occupational arsenic exposure associated with a 10^-6lung cancer risk is 0.007 μg/m³. This cancer risk estimate can be extrapolated to continuous lifetime exposure scenario, assuming occupational exposure of 8 h/d, 5 d/w and a lifetime exposure of 70 y. From this extrapolation, the concentration of arsenic in air associated with a 10^-6excess lifetime risk of cancer can be estimated as 0.9 ng/m³.

ECHA Risk Assessment Committee (RAC)

Based on a previous evaluation by DECOS and AGS reported above, RAC also attempted to define a cancer dose-response relationship for arsenic substances based on linear extrapolation from the data in the epidemiological study by Lubin et al. (2008). RAC committee stated that despite mechanistic indications of a threshold mode of action, the available data do not allow the identification of a threshold. Therefore, the following excess lifetime mortality risks were defined:

Workers: Based on a 40-year working life (8 h/day, 5 days/week):

An excess lifetime lung cancer mortality risk = 1.4 x 10-4 per µg As/m³ (inhalable particulate fraction).

Finally, for the sake of completeness, it is noted that, in 2017, ECHA RAC further suggested a biological guidance value of 10 µg As/L urine (post-shift sample at end of a working week, as combined As(III), As(V), monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA).

This value is in line with the systemic effects/long-term inhalation DNEL of 5 µg/m3(workers) proposed in this dossier for diarsenic trioxide.

References

Ahsan H et al. (2006). Arsenic exposure from drinking water and risk of premalignant skin lesions in Bangladesh: baseline results from the health effects of arsenic longitudinal study. Am. J. Epidemiol. 163(12):1138-1148.

Buchet JP et al. (1981). Comparison of the urinary excretion of arsenic metabolites after a single oral dose of sodium arsenite, monomethylarsonate or dimethylarsinate in man. Int. Arch. Occup. Environ. Health 48:71–79.

Cohen SM et al. (2013). Evaluation of the carcinogenicity of inorganic arsenic. Crit. Rev. Toxicol. 43(9):711-752.

Tsuji JS et al. (2019). Dose-response for assessing the cancer risk of inorganic arsenic in drinking water: the scientific basis for use as a threshold. Critical Reviews in Toxicology.
https://doi.org/10.1080/10408444.2019.1573804.

General Population - Hazard via inhalation route

Systemic effects

Long term exposure

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

2.5 µg/m³

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL related information

DNEL derivation method:

other: Oral to inhalation extrapolation using epidemiology data as Point of Departure (POD).

Overall assessment factor (AF):

3

Dose descriptor starting point:

NOAEL

Value:

2.2 µg/kg bw/day

Modified dose descriptor starting point:

NOAEC

Value:

7.7 µg/m³

Explanation for the modification of the dose descriptor starting point:

The POD is a NOAEL of 1.7 µg As/kg bw/day extrapolated - as presented above in the worker DNEL discussion section - from the HEALS drinking water exposure epidemiology study conducted by Ahsan et al.(2006) in Araihazar (Bangladesh). As the reference value is expressed as µg of As/kg bw/day, the value need to be corrected for the molecular weight of diarsenic trioxide.

Inhalation (chronic) DNEL = 1.7 µg As/kg bw/day, corresponding to 2.2 µg As2O3/kg bw/day.

The POD is then amended for an adult bodyweight of 70 kg and a breathing volume of 20 m3 (general population) to yield an uncorrected chronic inhalation DNEL, as follows:

Inhalation (chronic) DNEL (uncorrected) = 2.2 µg As2O3/kg bw/day x 70 kg / 20 m3/d = 7.7 µg As2O3/m3.

AF for dose response relationship:

1

Justification:

No correction required.

AF for differences in duration of exposure:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for interspecies differences (allometric scaling):

1

Justification:

No correction required; the POD is an epidemiology study.

AF for other interspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for intraspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for the quality of the whole database:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for remaining uncertainties:

3

Justification:

In consideration of the oral-to-inhalation extrapolation, in order to cover any remaining uncertainties also with respect to possible local effects, an assessment factor of 3 is applied to the above derived “uncorrected” DNEL.

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

Local effects

Long term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

General Population - Hazard via dermal route

Systemic effects

Long term exposure

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

112 µg/kg bw/day

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL related information

DNEL derivation method:

other:

Overall assessment factor (AF):

1

Dose descriptor starting point:

NOAEL

Value:

2.2 µg/kg bw/day

Modified dose descriptor starting point:

NOAEL

Value:

112 µg/kg bw/day

Explanation for the modification of the dose descriptor starting point:

In consideration of the systemic nature of arsenic compounds carcinogenicity, route-to-route extrapolation from the data obtained from drinking water studies to the dermal route is considered justified. For this purpose, modification of the oral DNEL by considering a dermal absorption factor of 2% (refer to Toxicokinetics section, reference Wester et al., 1993) is appropriate. The chronic dermal DNEL can therefore be derived as follows:

Dermal (chronic) DNEL = 2.2 µg As2O3/kg bw/day (oral) / 2% = 112 µg As2O3/kg bw/day.

AF for dose response relationship:

1

Justification:

No correction required.

AF for differences in duration of exposure:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for interspecies differences (allometric scaling):

1

Justification:

No correction required; the POD is an epidemiology study.

AF for other interspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for intraspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for the quality of the whole database:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for remaining uncertainties:

1

Justification:

No correction required.

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

Local effects

Long term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

General Population - Hazard via oral route

Systemic effects

Long term exposure

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

2.2 µg/kg bw/day

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL related information

DNEL derivation method:

other: Extrapolation using epidemiology data as Point of Departure (POD)

Overall assessment factor (AF):

1

Dose descriptor starting point:

NOAEL

Value:

2.2 µg/kg bw/day

Modified dose descriptor starting point:

NOAEL

Value:

2.2 µg/kg bw/day

Explanation for the modification of the dose descriptor starting point:

The NOAEL of 2.2 µg As2O3/kg bw/day was determined - as presented in the worker DNEL discussion section - from the HEALS drinking water exposure epidemiology study conducted by Ahsan et al. (2006) in Araihazar (Bangladesh).

AF for dose response relationship:

1

Justification:

No correction required.

AF for differences in duration of exposure:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for interspecies differences (allometric scaling):

1

Justification:

No correction required; the POD is an epidemiology study.

AF for other interspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for intraspecies differences:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for the quality of the whole database:

1

Justification:

No correction required; the POD is an epidemiology study.

AF for remaining uncertainties:

1

Justification:

No correction required.

Acute/short term exposure

Hazard assessment conclusion:

high hazard (no threshold derived)

DNEL related information

General Population - Hazard for the eyes

Local effects

Hazard assessment conclusion:

medium hazard (no threshold derived)

Additional information - General Population