1 Introduction

The Ministry for the Environment has confirmed a comprehensive policy framework for managing contaminated land in New Zealand. As part of this, a Technical Advisory Group (TecAG) has been set up to develop a national methodology for deriving and applying national soil guideline values (SGVs) designed to protect the health of New Zealanders. A critical part of the derivation of SGVs is the use of toxicological criteria. A Toxicology Advisory Group (ToxAG) has been set up to provide recommendations to TecAG regarding toxicological criteria appropriate for use in the development of national SGVs for New Zealand.

This document presents a review of the toxicology literature, and recommendations for toxicological criteria, for priority contaminants in soil.

This review briefly describes the toxicological status of the contaminants (ie, their mode of action and effects) and summarises reference health standards already considered by international agencies, especially those that have developed soil guideline values. “Reference health standards” (RHS) is used in this report to refer to any value set by a regulatory or advisory body that provides an estimated daily (sometimes weekly or monthly) amount of a substance that can be taken into the body either without any, or an unacceptable additional, risk of detrimental health effects occurring (based on available scientific information). “Toxicological intake value” (TIV) is used specifically for values recommended for use in New Zealand. Dermal absorption and background exposure to contaminants are also considered.

The information and recommendations presented in this document have been endorsed by the Toxicology Advisory Group.

1.1 Background to toxicological criteria used in the derivation of soil guideline values

1.1.1 Mode of action

When their effects on human health are being considered, contaminants are often referred to as either threshold or non-threshold contaminants. Threshold contaminants are those considered to manifest toxic effects only if exposure exceeds a threshold dose level, and include (by convention) non-genotoxic carcinogens and non-carcinogens. A variety of toxicological criteria have been derived by organisations worldwide for chemicals displaying threshold critical toxicity. The most well established of these, and most universally adopted in chemical risk assessment programmes, is the tolerable daily intake (TDI) that was originally used for contaminants within foodstuffs by the Joint Food and Agriculture Organization (FAO)/World Health Organization (WHO) Expert Committee on Food Additives (JECFA). The TDI is defined as an estimate of the amount of a contaminant – expressed on a bodyweight basis, eg, mg/kg bw/day – that can be ingested daily over a lifetime without appreciable health risk (based on the available scientific information). The term “Acceptable Daily Intake” (ADI) is also used by JECFA and is also an estimate of the amount of a substance – expressed on a bodyweight basis, eg, milligrams per kilogram bodyweight per day (mg/kg bw/day) – that can be ingested daily over a lifetime without appreciable health risk (based on the available scientific information). The ADI is applied to food additives and veterinary drug residues, while the TDI is used for contaminants and naturally occurring toxicants. The United States Environmental Protection Agency (US EPA) uses largely the same methodology as JECFA/WHO but has adopted the terms “reference dose” (RfD, for oral and dermal exposure) and “reference concentrations” (RfC, for inhalation exposures instead of TDI or ADI), though using a very similar definition. US EPA (2009) defines RfD as an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily oral exposure to the human population (including sensitive groups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. The US Agency for Toxic Substances and Disease Registry (ATSDR), which is responsible for preparing toxicological profiles for priority hazardous substances commonly found at contaminated sites in the Unites States, derives minimal risk levels (MRLs) using a similar methodology to TDIs, RfDs and RfCs. Minimum risk levels are defined as an estimate of daily human exposure to a hazardous substance that is likely to be without an appreciable risk of adverse non-cancer health effects over a specified route and duration of exposure, and are typically derived for each of chronic, intermediate (up to one year), and acute exposure (eg, ATSDR, 2007).

Non-threshold contaminants conventionally include genotoxic carcinogens, and are considered to have effects at all levels of exposure. Different organisations have used different approaches to determine the potency of non-threshold contaminants and potency is typically expressed either as (1) a slope factor (US) or maximum likelihood estimate, both of which are the increased risk per daily dose, or (2) a risk-specific dose (Canada) or index dose (UK), which is an estimate of the amount of a contaminant, expressed on a bodyweight basis, eg, mg/kg bw/day, that can be ingested daily over a lifetime with a minimal or negligible increase in risk. These values are typically obtained by dividing the acceptable increased risk level by the slope factor, although they may also be obtained by dividing a specified dose that produces a certain response by a given factor (eg, NHMRC, 1999; EA, 2008). Various approaches to estimation of carcinogenic potency have been adopted by different international agencies, although recent guidance has converged on the use of linear extrapolation from a point of departure from the dose-response curve with the BMDL₁₀ (the lower bound of a 95th confidence interval on a benchmark dose (BMD) corresponding to a 10% tumour incidence), the favoured point of departure (eg, EFSA, 2005; US EPA, 2005; EA, 2008). However, a significant difference between US and Canadian agencies and other international agencies appears to be that the US EPA and Canadian agencies typically apply cross-species scaling to cancer potency estimates derived from animal studies (US EPA, 2005), while a number of European agencies, WHO, and Australian agencies do not (eg, NHMRC, 1999; Kroese et al, 2001 citing Health Council of Netherlands, 1994; EFSA, 2005; FAO/WHO 2006; EA, 2008).

The classification of carcinogens as genotoxic or non-genotoxic refers to their mode of action. Genotoxic carcinogens are those that act by causing damage to genetic materials and generally have effects at all levels of exposure (ie, non-threshold contaminants). In contrast, non-genotoxic carcinogens do not act on genetic material and are considered to have a threshold above which toxic effects are manifested (ie, threshold contaminants). This is the approach adopted in several current New Zealand government publications (eg, MfE and MoH, 1997; MoH, 2005). However, more recent understanding of the mechanisms of carcinogenesis leads to a blurring of the boundaries between genotoxicity and non-genotoxicity. For example, a carcinogen may elicit genotoxicity due to indirect action on DNA, and have a dose-response that is non-linear and most similar to a threshold response.

Recent guidance from the US EPA (2005) addresses this difficulty in empirically distinguishing between a true threshold and a non-linear low-dose relationship by using a slightly different definition to distinguish between different modes of action of carcinogens. Specifically, the US EPA (2005) defines three scenarios: (1) a linear dose-response is assumed if the carcinogen is DNA-reactive and directly mutagenic, or activity displays linearity at low doses, or there is insufficient evidence to define an alternative mode of action; (2) a non-linear dose-response is appropriate when there is sufficient data to ascertain the mode of action and conclude it is not linear at low doses and the substance is not mutagenic or displays other activity that would suggest linearity of response; (3) both linear and non-linear approaches may be used when there are multiple modes of action.

For the purposes of this report, the terms “non-threshold” and “threshold” are used. Non-threshold contaminants refer to substances for which the dose-response is demonstrated to be linear or there is insufficient evidence to indicate non-linearity – including genotoxic carcinogens that act both directly and indirectly with DNA. Threshold contaminants include those substances for which dose-response is demonstrated to be non-linear at low doses or which exhibit a threshold for response – including non-genotoxic carcinogens.

Further, given the tendency of WHO (FAO/WHO, 2006), a number of European agencies (eg, Health Council of Netherlands, 1994 cited in Kroese et al, 2001; EA, 2008), and Australian agencies (NHMRC, 1999) to not apply cross-species scaling in determining cancer potency, cross-species scaling has not been used in estimating cancer potency in this report.

Finally, contaminants may elicit both carcinogenic (typically non-threshold) and non-carcinogenic (typically threshold) effects. Typically the most sensitive toxicological endpoint is used to set the final value (based on comparison of the dose associated with the acceptable excess lifetime risk level (risk-specific dose) with the TDI or equivalent).

1.1.2 Acceptable risk level

An acceptable risk level is often used to define the acceptable risk associated with exposure to non-threshold contaminants. In New Zealand, an acceptable increased risk level of 1 in 100,000 was first used in the national drinking water standards (MoH, 1995) and this has since been adopted in a number of government publications (eg, MfE and MoH, 1997; MoH, 2005). This falls in the “mid-range” of acceptable risk levels used by international agencies, which range from one in a million (eg, US, Canada) to 1 in 10,000 (The Netherlands).

There are numerous approaches and models that have been used to estimate carcinogenic potency, yielding markedly different estimates (EFSA, 2005). More recently there has been a tendency to move towards simple linear extrapolation from a point of departure on the dose-response curve to the origin (eg, Kroese et al, 2001; US EPA, 2005; EA, 2008). Typically, the benchmark dose approach is used and a BMD₁₀ (the dose that gives rise to a 10% response) or BMDL₁₀ (the lower 95% confidence limit of the BMD₁₀) is used as the point of departure in an appropriate animal carcinogenicity study, although other doses, eg, BMD₀₅, BMD₂₅, may be used (NHMRC, 1999; EFSA, 2005). A variety of models may be used to estimate the BMD/L10 of interest (NHMRC, 1999; IPCS, 2004; US EPA, 2005). This approach is less well developed for use with human cancer data, and traditional quantitative risk models may need to be used.

More recent guidance from European agencies and WHO for non-threshold contaminants often do not specify an acceptable risk level per se, but focus either on the margin of exposure (MOE, the ratio of the BMDL₁₀ to the estimated intake in humans) (EFSA, 2005; FAO/WHO, 2006) or application of a large default factor to the BMDL₁₀ (EA, 2008). The magnitude of the MOE is then subject to consideration as to what constitutes an acceptable level of risk, with the EFSA (2005) stating that an MOE of 10,000 or more is of low concern from a public health point of view. This factor (10,000) is the default factor applied to a critical BMDL₁₀ derived from animal studies in UK guidance (EA, 2008), and is equivalent to calculating a risk-specific dose for an excess lifetime cancer risk of 1 in 100,000 from the BMDL₁₀ using low-dose linear extrapolation. Where dose-response modelling of human data is used, estimates of the dose corresponding to an excess lifetime cancer risk of 1 in 100,000 is used (EA, 2008).

The consensus of the Toxicology Advisory Group was that the acceptable increased risk level should remain at 1 in 100,000 (=10^–5) or that, where appropriate, a default factor of 10,000 could be applied to BMDL₁₀ values to derive toxicological intake values for non-threshold contaminants. It is recognised that in most cases in New Zealand, selection of appropriate carcinogenic potency estimates for a given contaminant will be based on the available literature as opposed to the derivation of values per se. Thus, these recommendations provide some guidance as to the preferential selection of carcinogenic potency estimates.

Finally, to facilitate comparison of different estimates of the potency of non-threshold substances in this document, where slope factors are used, a toxicological intake value (risk-specific dose) has been calculated assuming an acceptable risk level of 10^–5.

1.1.3 Background exposure

People may be exposed to contaminants from sources such as food, air and water; collectively this exposure from other sources is termed background exposure. For the majority of contaminants of concern from land contamination, the background exposure will primarily be from food and water.

Exposure to non-threshold contaminants is based on an agreed acceptable increase in risk, and therefore exposure should be limited as much as reasonably practicable. It is assumed that exposure from other sources (food, air, water) is also similarly controlled by the same principle. Therefore, background exposure is not taken into account for non-threshold contaminants.

For threshold contaminants, different countries have taken different approaches, which can be grouped into three main approaches:

Background exposure is ignored – this is a non-conservative approach.
Background exposure may be taken into account by subtracting the estimated background exposure from the tolerable daily intake, and the residual amount used to derive the soil guideline value.
A proportion of the TDI may be allocated to exposure from soil.

Canadian agencies combine the latter two approaches by subtracting the background exposure from the TDI, and then allocating 20% of this residual TDI to exposure from soil (with air, water, food and consumer products also assigned 20%) (CCME, 2006). The UK protocol also combines these approaches, but where the estimated background exposure is less than 50% of the TDI, all the residual amount is allocated to exposure from soil. Where the estimated background exposure is more than 50% of the TDI, 50% of the TDI is allocated to exposure from soil (EA, 2008). In New Zealand, three contaminant-focussed publications about contaminated land (MfE, 1997; 2010; MfE and MoH, 1997) assign 100% of the TDI as being from soil sources and do not consider background or dietary intakes, with an exception made in the case of copper (MfE and MoH, 1997), where only 10% of the TDI is assigned to soil sources. The allocation of 10% is nominally based on the approach adopted in the relevant Drinking Water Guidelines (MoH, 1995) and appears to be overly conservative in relation to exposure from soil sources. A fourth publication (MfE 2006) subtracted the estimated background exposure from the TDI and used the residual to calculate soil guideline values.

The Toxicology Advisory Group recommends that New Zealand adopts a variation on the UK approach (EA, 2008). Specifically, background exposure is subtracted from the TDI with the residual allocated to exposure from soil. However, in contrast to the UK approach, the Toxicology Advisory Group recommends that where background exposure comprises greater than 50% of the TDI, the proportion allocated to exposure from soil is considered on a case-by-case basis. Further, in cases where background exposure is negligible or no data on background exposure exists, the Toxicology Advisory Group recommends that a maximum of 95% of the TDI should be allocated to exposure from soil. This is expected to provide a slight degree of precaution for substances for which determining the background exposure may be problematic.

Dietary intake

Where applicable, estimates of the dietary intake of various substances are primarily based on the 2003/04 New Zealand Total Diet Survey (NZTDS) (Vannoort and Thomson, 2005) or national nutrition studies (copper only: Russell et al, 1999; MoH, 2003). Mean dietary intakes are used and are considered to represent a long-term average. Where data from the NZTDS is used, dietary intakes are provided for a child 1–3 years old and the mean of an adult male and female aged 25+. Dietary intakes provided in the national nutrition surveys are expressed as total intake, and the bodyweights provided in those reports were used to derive intakes expressed as mg/kg bw/day.

Dietary intake from water is based on data provided in a survey of the chemical quality of drinking water supplies (Davies et al, 2001), which provides details of chemical analysis of water collected from consumers’ taps across New Zealand. It is noted that reviews of drinking water quality have been carried out subsequent to this survey (eg, MoH, 2006; 2007); however, the Davies et al (2001) report is the most recent that provides data in a usable manner.

1.1.4 Primary routes of exposure

At a contaminated site the oral, inhalation and dermal routes of exposure are of primary interest in deriving soil guideline values. Ingestion is generally considered to be the primary route of exposure for most contaminants of concern at contaminated sites, although inhalation and dermal absorption may also contribute to toxic effects. Where exposure via inhalation or dermal exposure contributes to a systemic response, the intakes from each of these exposure pathways can be added to the intake arising from ingestion to estimate the total intake used in derivation of soil guideline values. Where toxic effects are dependent on the route of exposure, separate soil guideline values should be derived for each route of exposure. Discussion on some general aspects of oral, inhalation and dermal exposures relevant to determining toxicological criteria are discussed below. Further details on the derivation of soil guideline values are discussed in MfE (2009).

Oral bioavailability

Ingestion is generally considered to be the primary route of exposure for most contaminants of concern at contaminated sites. While it is generally acknowledged that not all of the contaminants present in soil are absorbed into the human body (ie, are bioavailable), there is generally insufficient data to assume anything less than 100% bioavailability. Cadmium and lead are exceptions. The tolerable intakes established for cadmium are primarily based on toxicokinetic modelling of dietary cadmium intake, in which gastrointestinal absorption of cadmium from food in humans is considered to be in the range of 1 to10%. Various approaches have been used for lead, including the application of physiologically based toxicokinetic or other models (US EPA, 1994; DEFRA and EA, 2002) or application of a single factor to account for the reduced bioavailability of lead (Baars et al, 2001). Other agencies have assumed the lead is 100% bioavailable (NCSRP, 1996).

Further there may be differences in the bioavailability of substances used in studies to establish the tolerable intake or cancer potency factor, and soil – particularly for exposure via ingestion. For example, substances in drinking water will be more bioavailable than that substance in soil; substances in food may be more bioavailable than that same substance in soil. Further, substances administered in animal laboratory tests are likely to be more bioavailable than those substances in soil. Fasting and nutritional status may also influence oral absorption. Practically, these differences make little difference to the recommended intake as there is typically insufficient data to be able to take this factor into account quantitatively; it is generally assumed that the bioavailability of the substance is the same. However, such information enables a better assessment of the degree of conservativeness associated with the intake value when applied to the derivation of soil guideline values. This is discussed for individual contaminants where relevant in this report.

Inhalation exposure

Inhalation will be a negligible route of exposure for contaminants of limited or no volatility (semi-volatile and inorganic substances) as the amount of dust considered to be inhaled typically represents a very small fraction of exposure. For example, based on inhalation parameters for New Zealand residential sites (MfE, 1997; 2010; MfE and MoH, 1997) (inhalation rates of 3.8 m³/day for a child and 20 m³/day for an adult, and dust concentration of 0.026 mg/m³ exposure), a child will inhale 0.098 mg of dust and an adult 0.52 mg. This is <0.1% of the amount of soil ingested by a child (100 mg/day) and about 2% of the soil ingested by an adult resident (25 mg/day). Similarly for industrial sites, using the parameters in MfE and MoH (1997: inhalation rate of 9.6 m³ during a working day, and a respirable dust concentration of 0.071 mg/m³ exposure) an adult worker would inhale 0.68 mg, which is about 3% of the soil ingested. Furthermore, the majority of soil dust particles that are likely to be inhaled will be captured in the nose or throat, thus would actually contribute to oral exposure. Therefore, the inhalation route is only considered for volatile contaminants.

Dermal absorption

Dermal absorption of a substance may contribute to a systemic response associated with the ingestion of that substance. The skin absorption factor is the only contaminant-specific parameter required for the dermal absorption pathway. Dermal absorption of semi-volatile and inorganic substances is considered on a case-by-case basis.

Dermal absorption of volatile organics is especially difficult to assess, because most studies have involved occluding (covering) the skin: this may give artificially high skin absorption values, since these compounds would also be expected to volatilise from the skin. The US EPA Region III recommends using a dermal absorption value of 0.05% for substances with a vapour pressure similar to that of benzene (vapour pressure approximately 95.2 mm Hg) (US EPA, 1995). This would include chemicals such as 1,1-dichloroethane 1,1,1-trichloroethane, and other volatiles with vapour pressure similar to or greater than that of benzene. For volatiles such as ethylbenzene, tetrachloroethene, toluene, and xylenes – which have vapour pressures lower than that of benzene (and less volatilisation from the skin may occur) – a default skin absorption value of 3% is recommended. These numbers are considered to only apply to non-occluded skin, which would be the scenario expected for most environmental exposures. However, if the skin is occluded for any reason, higher dermal absorption values (up to 100%) should be used.

previous next

Draft Toxicological Intake Values for Priority Contaminants in Soil