Science of the Total Environment 521–522 (2015) 359–371

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Assessment of human exposure to environmental sources of nickel in Europe: Inhalation exposure Jurgen Buekers a,⁎, Katleen De Brouwere a, Wouter Lefebvre a, Hanny Willems a, Marleen Vandenbroele b, Patrick Van Sprang b, Maxime Eliat-Eliat b, Keegan Hicks c, Christian E. Schlekat d, Adriana R. Oller d a

Flemish Institute for Technological Research (VITO), Boeretang 200, 2400 Mol, Belgium ARCHE (Assessing Risks of Chemicals) Consulting, Stapelplein 70, 9000 Gent, Belgium University of Waterloo, 200 University Avenue W, Waterloo, Ontario N2L 3G1, Canada d NiPERA, 2525 Meridian Parkway, Suite 240, Durham, NC 27713, USA b c

H I G H L I G H T S • • • • •

Local respiratory effects drive health risks of Ni via inhalation. DNELs for chronic inhalation 20 ng/m3 (EU air guidance) and 60 ng/m3 are considered. Most of the EU population is exposed to Ni air levels below the DNELs. A tiered modelling approach was used to assess Ni air levels near industrial sites. Majority of sites compliant with 60 ng Ni/m3 (Tier I) or 20 ng Ni/m3 (Tier I–IIb).

a r t i c l e

i n f o

Article history: Received 16 December 2014 Received in revised form 24 February 2015 Accepted 26 February 2015 Available online xxxx Editor: F.M. Tack Keywords: Nickel Inhalation DNEL Dosimetric Modelling REACH

a b s t r a c t The paper describes the inhalation nickel (Ni) exposure of humans via the environment for the regional scale in the EU, together with a tiered approach for assessing additional local exposure from industrial emissions. The approach was designed, in the context of REACH, for the purpose of assessing and controlling emissions and air quality in the neighbourhood of Ni producers and downstream users. Two Derived No Effect Level (DNEL) values for chronic inhalation exposure to total Ni in PM10 (20 and 60 ng Ni/m3) were considered. The value of 20 ng Ni/m3 is the current EU air quality guidance value. The value of 60 ng Ni/m3 is derived here based on recently published Ni data (Oller et al., 2014). Both values are protective for respiratory toxicity and carcinogenicity but differ in the application of toxicokinetic adjustments and cancer threshold considerations. Estimates of air Ni concentrations at the European regional scale were derived from the database of the European Environment Agency. The 50th and 90th percentile regional exposures were below both DNEL values. To assess REACH compliance at the local scale, measured ambient air data are preferred but are often unavailable. A tiered approach for the use of modelled ambient air concentrations was developed, starting with the application of the default EUSES model and progressing to more sophisticated models. As an example, the tiered approach was applied to 33 EU Ni sulphate producers' and downstream users' sites. Applying the EUSES model demonstrates compliance with a DNEL of 60 ng Ni/m3 for the majority of sites, while the value of the refined modelling is demonstrated when a DNEL of 20 ng Ni/m3 is considered. The proposed approach, applicable to metals in general, can be used in the context of REACH, for refining the risk characterisation and guiding the selection of risk management measures. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Nickel (Ni) is a natural element of the earth's crust. Natural sources like volcanic eruptions and windblown dust contribute to Ni in the atmosphere. Worldwide estimates of these natural emissions range from 8500 tons/year in early 1980s to 30,000 tons/year in the early ⁎ Corresponding author. E-mail address: [email protected] (J. Buekers).

http://dx.doi.org/10.1016/j.scitotenv.2015.02.092 0048-9697/© 2015 Elsevier B.V. All rights reserved.

1990s (ATSDR, 2005). Additionally, anthropogenic activities like mining and smelting of Ni ores, manufacturing of Ni containing articles (e.g., stainless steel), fossil fuel combustion and waste incineration lead to supplemental emissions of Ni into the air. The anthropogenic emission rate to air is estimated to be a factor 1.4 to 1.8 higher than the natural one (IARC, 2012). Fossil fuel combustion is reported to be the major contributor of atmospheric Ni in Europe and the world, accounting for around 62% of the anthropogenic emissions in the 1980s (Barbante et al., 2002; ATSDR, 2005). Rydh and Svärd (2003) estimated that in

360

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

1999 57,000 tons of Ni was released into the air from the combustion of fossil fuels worldwide. Other anthropogenic sources of Ni in the atmosphere are nickel smelting and refining processes which accounted for 17%, municipal incineration for 12%, steel production for 3%, other Ni-containing alloy production for 2% and coal combustion for 2% (ATSDR, 2005). For the EU27 total Ni emissions are estimated at about 1790 tons Ni/year of which 612 tons Ni/year is emitted to air. Traffic and industrial processes such as production (refining) and downstream use industries (metal surface treatment, production of batteries, etc.) have been identified as important sources for emissions of Ni to air (ECB, 2008; personal communication Patrick Van Sprang ARCHE). Under REACH (Registration Evaluation Authorization & Restriction of Chemicals; EC 1907/2006), producers and importers of Ni and Ni substances, that are registered at N 10 tons/year, have to demonstrate that manufacturing, selling, or using these substances in downstream applications has no adverse effects on human health and the environment. One aspect of the Chemical Safety Assessment (CSA) for REACH is the assessment and management of risks associated with indirect human exposure to each of the many environmental sources of Ni. The EU legislation requires an assessment of multimedia exposure for substances produced/used in the EU. Dietary Ni intake constitutes the main exposure pathway in terms of the total (absorbed) dose relevant for systemic health effects (De Brouwere et al., 2012). Inhalation of Ni is a minor source in terms of total daily doses relevant for systemic effects; however, there is also a concern about inhalation exposure to Ni due to possible adverse local effects on the respiratory tract. This paper focuses on exposure and risk characterisation of Ni (as total Ni in ambient air) via the inhalation pathway. REACH requires a separate CSA for every substance that is registered in tonnage bands of 10 tons/year or above. However, although different chemical forms of nickel have different toxicological properties, a generic approach was used here because the speciation of emitted Ni substances and effects of atmospheric ageing on speciation of Ni particles is poorly understood. Thus, it is not possible to precisely apportion environmental Ni to its different chemical forms (Hughes et al., 1995). In general, however, oxidic and soluble forms of nickel predominate in ambient air (Füchtjohann et al., 2001; Galbreath et al., 2003; Tirez et al., 2011). In this manuscript a generic human inhalation exposure and risk characterisation approach was performed which is considered to be applicable to typical mixtures of Ni prevailing in ambient air, both at regional and local scale (i.e., near industrial sites). The most critical adverse effects of Ni substances in rodent and/or human studies after inhalation are on the respiratory tract (e.g., lung inflammation, lung and nasal tumours) (e.g., ATSDR, 2005). A review of these data and a rationale for selecting the most appropriate points of departure to derive reference concentrations for Ni in ambient air based on respiratory tract effects are included in Oller et al. (2014). The aims of the present study are i) to select appropriate Derived No Effect Level (DNEL) values to protect the general population from adverse respiratory effects associated with Ni exposure, starting from toxicity data and epidemiological data; ii) to characterise the regional EU ambient air exposure to Ni and to perform the risk characterisation for respiratory effects of Ni at the EU regional level; iii) to develop a tiered approach to aid industries in refining their local emission assessment; and iv) to perform risk characterisation at the local exposure level. To fulfil these aims we undertook the following analyses: 1) a review of the most recently derived ambient air standards and their underlying assumptions, including the EU air guidance value which can be considered as both a DNEL (Derived No Effect Level) and a DMEL (Derived Minimal Effect Level) (see Results and discussions); 2) a derivation of an ambient air PM10 DNEL value that incorporates the concept of a practical threshold for the respiratory carcinogenicity (as well as toxicity) of nickel and includes full dosimetric adjustments; 3) a determination of the range of inhalation Ni exposure within the EU population (regional exposure assessment); 4) comparison of regional ambient air exposures to DNEL values (regional risk characterisation); 5) development of a

tiered approach for assessing Ni inhalation exposure levels at local scales around Ni producing/using industries when measured data are lacking; and 6) demonstration of the applicability of this tiered approach to the risk characterisation of exposures in the Ni producing/ using sectors (as an example of industrial Ni sites). The approaches described in this study can be used in the context of fulfilling REACH requirements, both for nickel and for other metal substances to refine the risk characterisation when the screening approaches fail to demonstrate safe use. The applied models are not Ni specific and are valid for other substances defined as non-reactive in the atmosphere. 2. Materials and methods 2.1. Review of ambient air reference or guidance values for nickel In considering the selection of a suitable DNEL to protect the general population from the lifetime effects associated with inhalation of Ni present in ambient air, a brief review of ambient air guidance or reference values derived in the last fifteen years within and outside the EU was undertaken. 2.2. Derivation of an ambient air PM10 DNEL value for nickel In the last ten years, there has been growing recognition that nickel and other compounds may cause tumours by threshold driven pathways (Bolt and Huici-Montagud, 2008; Hengstler et al., 2003; SCOEL, 2011). In addition, more information on the toxicological effects of nickel compounds in rats and humans has become available (e.g., Oller et al., 2008; Goodman et al., 2009, 2011; Kraut et al., 2010) and the use of dosimetric models to refine the calculations of human equivalent concentrations (HECs) to the animal or workplace exposures have been more widely used (e.g., Maruyama et al., 2006). Recently, Oller et al. (2014) considered these issues in the derivation of Ni HECs corresponding to the PM10 aerosol subfraction of ambient air that can be used as modified points of departure for the derivation of a DNEL for nickel. Since information on environmental exposure to different Ni species is not routinely collected, the nickel DNEL needs to cover all Ni forms present in ambient air. Ni monoxide and complex Ni–Fe oxides together with Ni sulphate predominate in ambient air, with minor contributions from Ni sulphides, and metallic nickel (e.g., Galbreath et al., 2003; Füchtjohann et al., 2001; Tirez et al., 2011; Oller et al., 2014). Both cancer and non-cancer respiratory effects are the main drivers for setting ambient air reference values for nickel. For respiratory toxicity effects, Oller et al. (2014) derived PM10 HECs based on rodent effects of Ni sulphate and Ni oxide but these HECs are protective for exposure to other Ni compounds as well. The authors applied toxicokinetic adjustments by considering equivalent retained doses in the pulmonary region of the respiratory tract in rats and in humans. For cancer effects, Oller et al. (2014) and others (e.g., SCOEL, 2011; Bal et al., 2011; Goodman et al., 2011) considered that nickel compounds are likely to be indirect genotoxic carcinogens with a practical threshold. It is possible then to derive a nickel reference value for cancer that is based on a practical threshold identified in epidemiological cancer studies (i.e., a DNEL) instead of relying on a linear dose–response extrapolation and the acceptance of 1/105 or 1/106 extra cancer risk (i.e., a DMEL). Furthermore, exposure to soluble, oxidic and sulphidic Ni with focus on low sulphidic Ni exposure (thus mirroring Ni speciation in ambient air) can be considered in the epidemiological studies and dosimetric adjustments can be made to account for differences in particle size distribution between Ni refinery exposures and exposures in ambient air. This approach was taken by Oller et al. (2014) to derive a PM10 Equivalent Concentration (EC) corresponding to the practical threshold for lung cancer effects derived from epidemiological data presented in Goodman et al. (2011).

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

The PM10 H(EC) nickel values derived by Oller et al. (2014) based on non-cancer and cancer respiratory effects, respectively, were used here as the starting values in the derivation of a DNEL for ambient air. In the final step of the DNEL derivation, informed assessment factors (AF) are applied to the H(EC) values to derive the final ambient air DNEL. A careful evaluation of the degree of conservatism included in the H(EC) calculations was undertaken to identify the extent to which remaining uncertainties need to be accounted for in the selection of assessment factors for the final DNEL derivation. For the risk characterisation of health effects associated with ambient air Ni exposures, both the newly derived DNEL as well as a DNEL corresponding to the current EU ambient air target value of 20 ng/m3 (defined as PM10) were considered. 2.3. Exposure assessment 2.3.1. Regional exposure Nickel concentrations in air serve as input for the inhalation pathway. Data from the 2012 AirBase database were retrieved for estimating the regional exposure (database accessed September 2014; most recent year for which data were available was 2012). The AirBase database is the public air quality database system of the European Environment Agency (EEA; http://www.eea.europa.eu/themes/air/airbase/). The EU countries have an obligation to report under the Council Decision 97/ 101/EC a reciprocal Exchange of Information (EoI) on ambient air quality. Aggregation of data and calculation of statistical parameters is in accordance of Annex IV of the EC Decision 2001/752/EC. Member states need to report on the analytical principle or measurement method. Monitoring stations are classified as background, traffic and industrial. Stations near industrial sites were excluded for this purpose of deriving typical and Reasonable Worst Case (RWC) regional EU concentrations as it is likely that measured Ni concentrations around these points are due to local emissions and do not reflect the regional situation. The air quality statistics in AirBase are based on hourly values, daily average values and daily 8-hour maximum values. Based on the daily average data, annual mean values are calculated, which were used in our approach to reflect long-term exposure. In the Airbase database a distinction is made between Ni measured in PM10 (particulate matter with aerodynamic diameter b 10 μm) which was given code 5015 in Airbase and Ni in aerosol (code 15). The latter means that it is unknown in which PM aerosol fraction Ni was measured (e.g., total PM, PM10, PM2.5). Therefore, code 15 measurements are further defined in this text as Ni in PM. The derivation of the regional exposure to Ni concentrations in air was done in parallel for data of Ni in PM10, Ni in PM and both datasets combined. For EU countries for which data of at least 15 monitoring stations (from traffic & background stations) were available, a log-normal distribution was fitted and the mean and 90th percentile (P90) value were derived for each country. Based on all of these fitted distributions, an average P90/mean ratio was calculated and this was applied for countries having minimal monitoring data (between 1 and 15 stations) and for which mean Ni concentrations could be calculated. With this approach a P90 was calculated for each country and the average of all countries was taken forward as the RWC. This RWC value is further reported as the PECregional. 2.3.2. Local exposure The Predicted Environmental Concentration in the air at the local scale (PEClocal) is defined as the sum of the regional air background (PECregional) and the local air concentration: PEClocal ¼ PECregional þ Clocal

ð1Þ

where Clocal refers to air concentrations resulting from air emissions from Ni producing/using industrial sites. When available, relevant air monitoring data were used to estimate the PEClocal. In case no local air monitoring data were available, Clocal was modelled (see Section 2.4).

361

2.4. Tiered approach to estimate local exposures A tiered approach using different air dispersion models was followed to predict local air Ni-concentrations (Clocal) and to assess the risks associated with Ni inhalation for the general population. The Risk Characterisation Ratio (RCR) was calculated as the ratio of the PEClocal/DNEL. The different air dispersion models used in this approach are further described below. The approach starts with the tier I model EUSES 2.0 which can be applied with a minimum of data collection and modelling efforts, assuring conservative predictions (as defined in current REACH guidance) and progresses towards higher tier, more sophisticated tools (GPM, Gaussian Plume Model and IFDM, Immission Frequency Distribution Model) to obtain more realistic predictions for Clocal. Application of the tiered approach involves in the first instance, the EUSES 2.0 model (Tier I). If this step results in a RCR higher than 1, the tiered approach prompts the application of the GPM model (Tier IIa tool). Should this still result in a RCR above 1, the IFDM model (Tier IIb tool) can be applied in a further refining step. Irrespective of the model used, predictions of Clocal include only emissions from stacks; diffuse emissions (fugitive emissions of Ni from sources within the industrial site in addition to the emissions that originate from the stack) are not considered due to the lack of available data to quantify them. 2.4.1. Tier I — EUSES model The first model (Tier I; EUSES 2.0, 2004) is a simplified OPS (Operational Model for Priority Substances; van Jaarsveld, 1990) model that estimates the Ni air concentration at 100 meter distance from a central stack (default stack height: 10 m; exhaust temperature identical with temperature in the environment) at standard Dutch weather conditions (default wind speed: 3 m/s; precipitation 700 mm/year). Input parameters are the produced/used substance tonnage, the Ni release factor per ton of Ni and the emission days/year. De Bruin et al., 2010 investigated the potential impact and importance of source parameter refinement on the emission distribution and environmental fate of chromium VI as modelled by EUSES. By increasing the heat content or the emission source height compared to the defaults of EUSES, the modelled air concentrations and depositions are reduced. Results showed that estimated air concentrations became more realistic. According to de Bruin et al. (2010), the selected default settings of EUSES (stack height: 10 m, concentration at 100 m of source, exhaust temperature identical with temperature environment) can be considered as (realistic) worst-case assumptions for the emission source. This is because the height of industrial stacks will often be taller than 10 m and emissions will rise due to some heat content. Consequently, the air concentration and atmospheric deposition will decrease with increasing heat content due to atmospheric mixing and dilution processes. When the stack height is increased, emissions enter into the air at a higher altitude, which causes ground level concentrations to decrease. 2.4.2. Tier IIa — GPM model A Gaussian Plume Model (GPM) can be used to generate more realistic predictions of Clocal than the EUSES model. The Gaussian plume models assume ideal plume geometry, ideal steady state of air pollutant emissions and meteorological conditions, a uniform flat terrain and a complete conservation of mass. These models can estimate Ni air concentrations at several distances up to 50 km from the central stack in one wind direction. One of the primary determinants in the GPM calculations is the effective stack height. As the gases are heated in the industrial facility, the hot plume will be thrust upward some distance above the top of the stack, i.e., the effective stack height. For point sources, the plume rise is calculated using the equations of Briggs (1969, 1971, 1972). Once the plume has reached its effective stack height, dispersion will begin in three dimensions depending on Gaussian plume equations, wind speed and atmospheric conditions. The model assumes that dispersion will take the form of a normal Gaussian curve, with the

362

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

Table 1 Emission characteristics of industrial sites in the Ni sulphate sector (Producers & Downstream Users, DU) used to assess the local Ni air concentration in the tiered modelling approach.

Producers

DU1: Metal surface treatmenta

DU2: Production of batteries

DU3: Production of Ni salts

DU4: Use of Ni sulphate in pigment production

DU5: Selective plating DU6 : Formulation of Ni compoundsa

DU7: Production of Ni-containing glass DU8: Metal surface treatment in the abrasives industrya

Minimum Maximum Median Minimum Maximum Median Minimum Maximum Median Minimum Maximum Median Minimum Maximum Median Median Minimum Maximum Median Minimum Maximum Minimum Maximum

Tonnage

Daily emissions to air

Release factor to air

Emission days to air per year

(T/year)

(kg Ni/emission day)

(g Ni/T)

(days/year)

257 1840 914 b1 14 2 175 796 715 177 51,113 967 100 136 118 0.4 10 800 78 5 15 0.2 3

0.01 0.21 0.14 b0.01 0.24 0.01 b0.01 0.06 0.04 b0.01 4.93 0.11 b0.01 0.02 0.01 b0.01 b0.01 0.20 0.02 0.02 0.14 b0.01 0.03

7 72 68 b0.01 10,791 1224 2 25 14 1 72 20 4 73 39 1800 50 50 50 1086 1460 2000 2000

200 365 350 211 336 220 247 360 276 200 365 355 340 362 351 215 200 270 240 120 365 200 200

a For some companies, release factors to air (g Ni/T) are based on SPERC (Specific Emission Release Category) information. With an ERC or Emission Release Category, labels are given based on use characteristics. These include: intended technical fate, life cycle stage, dispersion, indoor/outdoor use, level of containment, release promotion during service life. Specific ERCs (SPERCs) try to obtain a more realistic first tier emission estimate for substance uses in chemical industry and its downstream users (http://www.arche-consulting.be/Metal-CSAtoolbox/spercs-tool-for-metals).

maximum concentration in the centre of the plume. The “standard” algorithm used in the plume studies is the Gaussian plume model as developed by Sutton (1932). The input parameters generally required are the emission rate, wind speed, source height, gas temperature and gas velocity. The GPM model which was used in this project is freely available in Microsoft Excel format at the website1 of ARCHE (Assessing Risks of Chemicals) Consulting. 2.4.3. Tier IIb — IFDM model The more refined Immission Frequency Distribution Model (IFDM; Cosemans and Roekens, 2002; Cornille, 2009) can be used to generate more precise values for Clocal. IFDM is a bi-Gaussian model, with steady-state bi-Gaussian plume for each hour. It simulates air quality and deposition, and needs as input parameters information on point sources (x,y coordinates; emission rate; stack height & diameter; gas exhaust temperature) and dynamic meteorological data. The advantage of the IFDM model over the EUSES and GPM models is that it takes into account site-specific, measured, dynamic (hourly) meteorological conditions. IFDM uses one year of hourly meteorological data collected near the industrial sites. The IFDM model is designed to simulate non-reactive pollutant dispersion on a local scale. The Gaussian dispersion parameters are dependent on the stability of the atmosphere and the wind speed following the Bultynck and Malet formulation (Bultynck and Malet, 1972). More information on the IFDM model can be found in the European Model Database.2 Lefebvre et al. (2011a,b, 2013) have shown that IFDM is capable of accurately estimating black carbon concentrations from point sources in Flanders, which is also relevant for Ni given that both nickel and black carbon are non-reactive pollutants. 2.4.4. Selection of relevant location for air quality predictions DIRECTIVE 2004/107/EC establishes target values for arsenic, cadmium, nickel and benzo(a)pyrene in ambient air and also gives guidance on air sampling. In case air is sampled in the vicinity of point sources, 1 2

http://www.arche-consulting.be/Metal-CSA-toolbox/Local-air-modeling-(GPM-tool). http://air-climate.eionet.europa.eu/databases/MDS/index_html.

sampling points should be selected in such a way to provide data on areas with zones and agglomerations where the population is likely exposed to the highest concentrations averaged over a calendar year. EUSES estimates the air concentration at a standard distance of 100 m from the source, which in many cases does not reflect the distance of the closest population to the emission sources. For the GPM model, the maximum concentration of all combinations of distance and wind speed compared to the point source was here selected. In a more refined approach based on IFDM modelling, the maximum concentration at residences near the industrial site was selected. The distance and orientation of nearest residences to the industrial facilities of the companies investigated in this paper were retrieved from questionnaires completed by the companies, and by visual inspection using Google Maps satellite images. 2.5. Risk characterisation approach Risk characterisation is performed by comparing effect and exposure levels; under REACH, this is addressed by calculating the Risk Characterisation Ratio (RCR) as follows: RCR ¼ Exposure = DNEL

ð2Þ

with DNEL: Derived No Effect Level; and Exposure: PECregional or PEClocal. In the case of risk characterisation for local health effects on the respiratory tract following inhalation, both exposure levels and DNELs are expressed as Ni concentrations in air (ng Ni/m3). In the REACH Chemical Safety Reports (CSR), the RCRs are calculated at the regional level (using PECregional) and at each local site for the nickel producing and using industries (using PEClocal). For the local sites, demonstration of how the tiered approach can be applied will be given for one nickel substance, as an example. 2.6. Case study Data for 33 Ni sulphate producers and downstream using sites, were collected and subjected to the tiered approach described earlier. An

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371 Table 2 Stack characteristics of industrial sites in the Ni sulphate sector (Producers & Downstream Users).

Minimumb Maximumb Medianb a b

Stack height

Stack diameter

Gas exit velocitya

Exhaust temperature

(m)

(m)

(m/s)

(°K)

6 140 18

0.2 4.9 0.6

b1 31 7

285 427 298

At exhaust temperature. Information was not available for all recorded industrial sites in the Ni sulphate sector.

overview of operational conditions (tonnage produced, release factor, emission days/year) and stack characteristics of industrial sites in the Ni sulphate sector are shown respectively in Tables 1 and 2. 3. Results and discussion 3.1. Review of ambient air reference or guidance values for nickel In the EU, a target value of 20 ng Ni/m3 has been in place since 2004 and has entered into force on 31st December 2012 (EC, 2001). This value applies to the PM10 subfraction of ambient air Ni (averaging period: 1 year). The value of 20 ng Ni/m3 is based on non-cancer effects in rodents and lung cancer effects in workers. This value can be considered both as a DNEL (Derived No Effect Level) and a DMEL (Derived Minimal Effect Level). This value represents a DNEL for non-cancer respiratory effects associated with exposure to the most toxic of the Ni compounds (e.g., water soluble nickel compounds) and a DMEL for the carcinogenic effects of Ni using a cancer non-threshold unit risk for refinery dust (driven by nickel subsulphide) and an acceptable excess cancer risk of 1/106. Particle dosimetry was not taken into account. In the last 10 years, ambient air reference values were also derived by other groups (see Table 3). For example, in 2005, the U.S. Agency for Toxic Substances and Disease Registry (ATSDR) derived an inhalation chronic Minimal Risk Level (MRL) for nickel of 90 ng/m3 based on respiratory effects from rodent studies. California's Office of Environmental Health Hazard Assessment (OEHHA) derived two Reference Exposure Levels (RELs), one for nickel and nickel compounds (except nickel oxide) of 14 ng Ni/m3 based on lung and nasal epithelial and lymphatic pathological effects in rats exposed to nickel sulphate, and one for nickel oxide of 20 ng Ni/m3 based on lung pathology in mice (OEHHA, 2008). Both MRL and REL values are based on non-cancerous health effects only and do not consider carcinogenic effects. The Texas Commission on Environmental Quality (TCEQ) completed the development of ambient air Chronic Reference Values (ReVs) and Effect Screening Levels (ESLs) for nickel in 2011 (TCEQ, 2011). With a Hazard Quotient of 1, the TCEQ derived a chronic ReV of 230 ng/m3; this was based on chronic active lung inflammation and associated lesions observed in rats exposed to Ni sulphate. A chronic ESL based on a linear dose–response for cancer of 59 ng/m3 was derived from studies of lung cancer in industrial workers, and a tolerable excess cancer risk of 1/105

363

(Haney et al., 2012). In the derivation of reference values, all recent U.S. efforts considered 0.03 mg Ni/m3 to be the NOAEC for respiratory toxicity effects in the nickel sulphate rat study. In summary, for non-cancer effects a range of threshold values from 14 to 230 ng/m3 have been derived, while for respiratory cancer effects values of 20–60 ng/m3 have been derived based on linear dose–response and acceptance of extra lung cancer risk values of 1/105 or 1/ 106. Importantly, while the values derived by ATSDR, California OEHHA and Texas TCEQ applied some form of dosimetric adjustment to their calculations, no adjustment was incorporated in the derivation of the current EU ambient air guidance value (EC, 2001). 3.2. Derivation of an ambient air DNEL value for nickel As described in Section 2.2, a DNEL to protect the general public from nickel exposures in PM10 can be derived based on the H(EC) calculations reported by Oller et al. (2014). A brief summary of the H(EC) derivation approach is provided below. The most sensitive endpoints of chronic toxicity (alveolar region of the lung), in rats (most sensitive animal species) exposed to the nickel compounds most commonly found in ambient air (nickel sulphate the most toxic or nickel oxide the most retained), were selected as the starting points for the HEC derivation. Then, Oller et al. (2014) applied a full dosimetric adjustment, based on experimental nickel particle size and lung retention data, as well as PM10 particle size and human experimental or calculated lung retention information. After applying the toxicokinetic adjustment (including adjustment for exposure duration), the most conservative retention-based PM10 Human Equivalent Concentration (HEC) corresponding to the NOAEC for Ni sulphate was equal to 9.4 μg Ni/m3. For Ni oxide, the most conservative HECLOAECs were equal to 11.0 μg Ni/m3 or 27.5 μg Ni/m3 depending on the selected combination of clearance rates for rats and humans. These values can be used as adjusted points of departure for all forms of Ni present in ambient air (Oller et al., 2014) (See Table 4). Oller et al. (2014) described the sources of uncertainty associated with the assumptions made in each step of their calculations. For toxicity effects observed in animals, the calculation of a HEC based on equivalent retained human doses in the pulmonary region of the lung was more conservative than a calculation based on deposited doses, while variations in the particle size of the available PM10 data had little influence on the calculated values. The most conservative parameter values (e.g., for retention time T1/2), were selected for the HEC derivation. Additional informed (Ni-specific) uncertainty factors that could be applied to these HECs to derive a DNEL include a factor for remaining differences in toxicodynamic (TD) responses between rats and humans and a factor for intra-individual differences in toxicokinetic (TK) and susceptibility/metabolism among the general population. A rationale for the selection of assessment factors is provided in Table 4. For the DNEL derivation based on respiratory toxicity effects, a default interspecies toxicodynamic assessment factor (AF) of 3 could be applied under the assumption that humans are ~ 3-times more sensitive to the local toxicity effects of nickel sulphate than rats (REACH R.8 guidance,

Table 3 Examples of U.S. nickel ambient air references values for chronic-duration exposure. Reference or guidance value (ng Ni/m3)

Valid for

Substance tested in study

Type of value

Species

Effect considered

Derived by

Date

90 14 20 230 59

Nickel forms found in air Ni & Ni compounds excluding Ni oxide Ni oxide Ni & inorganic Ni compounds Ni & inorganic Ni compounds

Ni sulphate Ni sulphate Ni oxide Ni sulphate Nickel refinery exposure with low levels of sulphidic Ni

MRLa RELb RELb ReVc ESLlineard

Rats Rats Mice Rats Workers

Non-carcinogenic Non-carcinogenic Non-carcinogenic Non-carcinogenic Lung cancer

ATSDR California OEHHA California OEHHA TCEQ TCEQ (2011) and Haney et al. (2012)

2005 2008 2008 2011 2011-2012

a b c d

Minimum risk level. Reference exposure level. Reference value. Effect screening level.

364

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

Table 4 Summary of nickel DNEL (Derived No Effect Level) derivation for ambient air. Adjusted starting values correspond to the PM10 HEC (Human Equivalent Concentration) or EC (Equivalent Concentration) values described in Oller et al. (2014); these values are based primarily on animal respiratory toxicity and human carcinogenicity data. Respiratory toxicity

Starting value(s) from rat chronic study

Adjusted starting value(s)—Dosimetric and exposure duration adjustments

Respiratory cancer

Soluble Ni

Oxidic Ni

Total Ni

NOAEC = 0.03 mg/m3 (Ni sulphate)

LOAEC = 0.5 mg/m3 (Ni oxide)

Practical threshold value for lung cancer = 0.1 mg Ni/m3 for workers

Respirable aerosol fraction, MMAD ~ Respirable aerosol fraction, MMAD ~ 2 μm 2 μm

Inhalable aerosol fraction

HEC-NOAEC = 9.4 μg Ni/m3 (Ni sulphate)a

HEC-LOAEC = 11.0–27.5 μg Ni/m3 (oxide)b

EC-practical threshold = 0.5 μg Ni/m3 (total forms of Ni) c

PM10

PM10

PM10 Assessment factors to account for uncertaintyd regarding: Inter species differences toxicokinetic (TK)

1 Already accounted in dosimetric adjustment

Not applicable

Inter species toxicodynamic (TD) differences between rats and humans for respiratory local effects

1–3 REACH recommends a default factor of 3 when no substance-specific data are available. A factor of 1 could be justified because comparisons of rat and human data do not indicate that humans are more sensitive than rats to the local respiratory toxicity effects of Ni particulates.

Not applicable

Intraspecies:inter individual differences among members of the general public (TK–TD)

10 5 A default factor of 5 has been considered appropriate for local effects (ECETOC, 2003 and references therein), while ECHA's default factor is 10 in the absence of substance-specific data. An informed factor of 5 could be applied given that effects are local and that inorganic Ni compounds do not undergo metabolism by enzymatic isoforms (e.g., P450).

Dose–response relationship

Severity of effect

4 Differences in toxicokinetics (TK) between adult workers and the general public have already been considered in the dosimetric adjustment. Remaining TK differences in thoracic region deposition between adults and children were shown to be no higher than 2-fold based on dosimetric modelling.e Thus an additional factor of 2 could be applied to account for remaining toxicokinetic differences between workers and the general public including children. Modelling of retained doses in alveolar region of the lung using a dosimetric A factor of 2 (rather than 2.5) is considered sufficient to account for remaining toxicodynamic (TD) differences model demonstrates no more than a 2-fold difference between adults and between workers and the general public for respiratory children in TK.e This suggests that a factor of 5 may be sufficient to cover inter individual differences in TK (2-fold) and TD (2.5-fold) for nickel. cancer based on the following: For soluble Ni an additional factor of 2 could be applied to also cover possible a) Epidemiological data for Ni is based on relatively large cohort (N100,000 workers) from North-America and respiratory sensitization effects that have been associated with soluble Ni Europe with differences in ethnic background and reflecting compounds. intra-worker variability. b) Furthermore, in the REACH guidance (R8) it is mentioned that intra-species variability in cancer susceptibility is already covered when extrapolating DMEL and non-threshold genotoxic carcinogenicity data from worker studies to the general population; it is often assumed that different large population groups have similar cancer susceptibility. Higher deposited doses in children that could be associated with higher ventilation rate are already accounted for by the additional TK factor. c) A TD factor of 2 for remaining inter individual variability could be derived as the difference between the ECHA AF of 5 used for intra-worker variability and the AF of 10 used for variability for the general public. d) Ni compounds are not lipophilic and do not undergo metabolism e) local nature of cancer effects Thus a TD factor no greater than 2 could be justified based on a)–e) above, resulting in an overall AF no higher than 4 for TK and TD uncertainties. 1 3 1 The factor 3 is based on REACH The point of departure for ‘total nickel’ is a practical guidance for LOAEC → NOAEC threshold nickel exposure value calculated based on derivation and Ni-specific data. soluble Ni exposure (in presence of other insoluble Ni exposures) but applied as ‘total Ni forms’ and based on data from N 100,000 workers. 1 1 1–2 Cancer is an irreversible effect and as such application of an additional factor for severity of effect could be considered. The only tumour sites relevant to nickel inhalation exposure are respiratory. The starting practical threshold is derived from workers with equal or higher smoking rates than the general population; therefore a possible interaction between Ni exposure and smoking is accounted for. Considering all conservative assumptions made in the derivation, a factor higher than 2 for severity of effect does not seem to be justified.

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

365

Table 4 (continued) Respiratory toxicity Soluble Ni Quality of database

Overall uncertainty assessment factor Final DNEL values Overall uncertainty & dosimetric factors

Respiratory cancer Oxidic Ni

Total Ni

1 The database for respiratory toxicity effects of Ni includes 4 chronic inhalation studies with 4 different Ni substances in 2 animal species together with supporting human data. 10–30 15–45 310–940 ng Ni/m3 240–700 ng Ni/m3f 32–97 714–2083

1 The database is relatively large and includes refinery and non-refinery cohorts. 4–8 60–125 ng Ni/m3 800–1667

a

See Table 4 of Oller et al. (2014). See Table 4 of Oller et al. (2014). Two values (11.0 and 27.5 μg Ni/m3) were derived for Ni oxide using two different combinations of Ni oxide clearance rates for rats and humans. This value considers differences in frequency and length of exposure and particle size, between workers and general population. See Table 5 of Oller et al. (2014). d Table 6 of Oller et al. (2014) describes sources of uncertainty in the calculations of adjusted starting values, how the calculations were able to address these uncertainties and what uncertainties still remain. e Based on Fig. 7 of Jarabek et al., 2005. f DNELs based on most conservative HEC-LOAEC for Ni oxide. b c

2010). However, comparative studies with nickel compounds and other metal compounds and particulates suggest that rats are more sensitive to the toxicity effects of inhaled metal (and nickel) particles than mice and humans (Oberdörster, 1995; Mauderly, 1997; Nikula et al., 2001; Greim and Ziegler-Skylakakis, 2007). For this reason a factor of 1 could also be justified for interspecies toxicodynamic differences in local effects (ECETOC, 2003, 2010). This is further confirmed by comparing the predictions (based on animal data) with the existing (albeit limited) in human studies of respiratory toxicity (Oller et al., 2014). As shown in Table 4, factors of 5 or 10 to account for intraspecies differences in TK and TD among the general population have been recommended (REACH R.8 guidance, 2010; ECETOC, 2003; Hattis et al., 1987, 1999; Hattis and Silver, 1994; Renwick and Lazarus, 1998). For nickel, an informed factor of 5 could be applied for local respiratory toxicity effects based on the fact that inorganic nickel compounds do not undergo metabolism by enzymatic isoforms (e.g., P450). From the perspective of intraspecies differences in lung retained Ni doses, modelling of retained doses in alveolar region of the lung using the MPPD (Multiple Path Particle Deposition) model demonstrates no more than a 2-fold difference between adults and children (based on Fig. 7 of Jarabek et al., 2005). This suggests that a factor of 5 may be sufficient to cover inter individual differences in TK (2-fold) and TD (2.5-fold) for nickel. However, the DNEL derivation based on respiratory toxicity effects observed in rodents, as well as the reference values reviewed in Section 2.1, do not take the possibility of respiratory sensitization into account. Water soluble nickel compounds are the only Ni compounds considered to be respiratory sensitizers and even for these compounds there are only a few case reports (Bright et al., 1997; Cirla et al., 1982; SCOEL, 2011). In addition, coexposure to chromium or hard metals usually occurs making it more difficult to assign causality of these few cases to nickel. Given the relative small weight of evidence and the lack of dose–response information on which to base a DNEL derivation, an additional protection factor of 2 for possible respiratory sensitization effects in the general population was considered. Thus, a total factor of 10 can be applied to the soluble Ni DNEL derivation to cover possible intraspecies differences. An AF for the dose–response relationship needs to be applied to derive a NAEC from the LOAEC for Ni oxide. A factor of 3 seems appropriate as the resulting NAEC of 0.17 mg Ni/m3 is 6-fold higher than the NOAEC for the more toxic Ni sulphate. This difference is nevertheless much lower than the difference in maximum tolerated dose (MTD) for these two substances in the chronic studies (Ni oxide is 18-fold less toxic than Ni sulphate). For this reason, and the fact that fibrosis was not reported after exposure to Ni oxide in the rat studies, a factor of 3 is applied to the Ni oxide HEC-LOAEC values of 11 or 27.5 μg Ni/m3. A final AF for database quality of 1 can be considered (REACH R.8 guidance, 2010) as the database for respiratory toxicity effects of nickel includes 4 chronic inhalation studies with 4 different nickel substances together with supporting human data.

Depending on the combination of assessment factors used (e.g., 10– 30 for sulphate and 15–45 for Ni oxide, Table 4), DNEL values of 240– 700 ng Ni/m3 (based on Ni oxide) or 310–940 ng Ni/m3 (based on Ni sulphate) can be calculated. The lowest value from these ranges (240 ng Ni/m3) can be considered to represent a conservative PM10 nickel DNEL based on respiratory toxicity effects of the main forms of Ni in ambient air. This value is 125 to ~2000 fold lower than the original point of departure values. For cancer effects, it is possible to derive a DNEL by considering nickel compounds as indirect genotoxic carcinogens with a practical threshold, and using data from epidemiological studies of nickel workers. Exposure in Ni producing and using worksites is always mixed (different combinations of soluble, sulphidic, oxidic and metallic Ni). Nickel in ambient air consists mainly of soluble and oxidic Ni, while sulphidic Ni is considered to be the most potent carcinogen of the insoluble Ni compounds present in mixed refinery operations. Sulphidic Ni comprises at most 10% of the total Ni exposure in ambient PM10. To identify a practical threshold for exposure to soluble and oxidic compounds, dose–response curves for these two compounds were analysed separately, taking into account differences in response relative to the presence of sulphidic Ni exposure (for oxidic Ni the presence of soluble Ni was also considered) (Oller et al., 2014). In total, lung cancer data from 22 process areas arising from 13 cohorts of geographically distinct Ni producing and using operations were included, encompassing N100,000 workers. The non-linear dose–response curves showed clear inflection points. Based on all measured data (taking into account soluble, oxidic and sulphidic Ni) a soluble inhalable Ni concentration of 0.1 mg Ni/m3 was derived as practical threshold (Oller et al., 2014). Applying this to ‘total Ni’ is a conservative estimate as the total Ni in the selected studies was always as high or higher than soluble Ni alone. A similar practical threshold value was recently considered as a starting point for the derivation of an occupational reference value for Ni (SCOEL, 2011). Starting with the value of 0.1 mg/m3, and considering that the thoracic area of the respiratory tract is the relevant area for lung tumours, Oller et al. (2014) calculated the practical threshold equivalent concentration for people exposed to ambient air (PM10) particle size aerosols for 24 h/day, 52 weeks/year, for 75 years. Application of these adjustments resulted in a conservative time-adjusted, dosimetry-adjusted value of 500 ng Ni/m3 (Table 4) corresponding to the workplace practical threshold exposure value of 0.1 mg Ni/m3 (Oller et al., 2014). The dosimetric adjustment yielded an equivalent PM10 Ni concentration that is 200-fold lower than the original value of 0.1 mg Ni/m3. To derive a DNEL, application of an additional factor no greater than 4 for remaining intra-human variability is proposed. Differences in toxicokinetics (TK) between adult workers and general public were already considered in the dosimetric adjustment. Remaining TK differences in thoracic region deposition between adults and children were

366

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

Table 5 Comparison of underlying approaches for the derivation of the two selected DNELs for the risk characterisation of Ni effects in ambient air. DNEL

20 ng Ni/m3 (as PM10)

60 ng Ni/m3 (as PM10)

Year Reference

2001 EC (2001)

2014 This paper and Oller et al. (2014)

Based on animal respiratory toxicity data Exposure Starting POD study

Ni sulphate 60 μg Ni/m3 (LOAEC mice) or 30 μg Ni/m3 (NOAEC rat)

PM10 HEC dosimetric adjustment

Not applied → 60 μg Ni/m3 (LOAEC) or 30 μg Ni/m3 (NOAEC)

AF (LOAEC to NOAEC × exposure adj. × TK–TDinter × TK–TDintra) Acceptable levels

6000 (10 × 6 × 10 × 10) applied to LOAEC or 600 (1 × 6 × 10 × 10) applied to NOAEC 10–50 ng Ni/m3

Ni sulphate and Ni oxide 30 μg Ni/m3 (sulphate NOAEC rats) 500 μg Ni/m3 (oxide LOAEC rats) Applied → 9.4 μg Ni/m3 (sulphate NOAEC) 11 μg Ni/m3 (oxide LOAEC) 30 (1 × 1a × 3 × 10)-sulphate 45 (3 × 1a × 3 × 5)-oxide 310 ng Ni/m3 (sulphate) 240 ng Ni/m3 (oxide)

Based on human respiratory cancer data Exposure Cohort size Dose–response

PM10 EC dosimetric adjustment AF (TK–TD intra)

Acceptable levels

Refinery dust 5000 (1 cohort) Linear (lung cancer unit risk 2.4 or 3.8 × 10−4 extra cancer cases per μg Ni/m3 corresponds to 2.6 to 4 ng Ni/m3 refinery aerosol fraction for 10−6 acceptable risk) Not applied → 2.6–4 ng Ni/m3 Not applied Considered that cancer risk was overestimated by up to 10-fold (Ni subsulphide b10% in ambient air) 20 ng Ni/m3

Workplace dust N100,000 (13 cohorts) Non-linear (practical threshold lung cancer = 0.1 mg Ni/m3 inhalable aerosol fraction) Applied → 500 ng Ni/m3 8 (4 × 2)b

60 ng Ni/m3

a

HEC calculation already accounts for differences in length of exposure. b Includes TK–TD intraspecies (4) and severity of effect (2); DNEL: Derived No Effect Level; POD: Point of Departure; NOAEC: No Observed Adverse Effect Concentration; LOAEC: Lowest Observed Adverse Effect Concentration; TK: Toxicokinetic; TD: Toxicodynamic; HEC: Human Equivalent Concentration; EC: Equivalent concentration; AF: Assessment Factor.

shown to be no higher than 2-fold based on dosimetric modelling for children and adults (Jarabek et al., 2005). Remaining toxicodynamic (TD) differences between workers and the general public for respiratory cancer can be covered by a factor of 2 based on several lines of evidence as indicated in Table 4. This suggests that using an AF no higher than 4 (2 × 2) for remaining TK–TD differences between workers and the general public including children could be justified. A factor of 1 for uncertainties related to the dose–response relationship was applied (Table 4). The starting practical threshold value for ‘total Ni’ is a conservative value (see Table 4). For database quality a factor of 1 was also applied as the database is relatively large and includes refinery and non-refinery workers. Cancer is an irreversible effect and as such application of an additional factor for severity of effect was taken into account. The only tumour sites relevant for Ni exposure are respiratory. The starting practical threshold is derived from 13 cohorts of workers from which the percentage of ever smokers is higher than for the general public, and thus accounts for possible interactions between Ni exposure and cigarette smoke. Considering all conservative assumptions made in the derivation, a factor higher than 2 for severity of effect does not seem to be justified. Applying all AF factors, yields a final lung cancer-based DNEL in the range of 60–125 ng Ni/m3 (see Table 4). The DNEL values of 60–125 ng Ni/m3 for the general public are 800–1667 times lower than the original practical threshold value for lung cancer of 0.1 mg Ni/m3 for workers. The most conservative DNEL value of 60 ng Ni/m3 is based on N100,000 workers, making it less likely that significant residual (undetected) cancer risks will be associated with that exposure level. In summary, by applying a full dosimetric adjustment and a threshold approach to cancer and non-cancer respiratory effects, DNEL values in the range of 60–125 ng Ni/m3 and 240–310 ng Ni/m3, respectively, could be conservatively supported for total Ni (all species) in ambient air. The values at the lower end of these ranges are not very different from, the existing ambient air regulatory or guidance values summarized in Table 3 and discussed in Section 2.1 (20–60 ng Ni/m3 and 14– 230 ng Ni/m3, respectively based on cancer and non-cancer effects). Table 5 summarizes the underlying assumptions in the derivation of the EU air quality guidance value of 20 ng Ni/m3 and compares them to those underlying the derivation of the DNEL in the present study.

For the purpose of this risk assessment the ambient air exposure values were compared here to both 20 ng Ni/m3, which is the current EU PM10 ambient air guidance value (as a very conservative surrogate DNEL-DMEL), and to 60 ng Ni/m3 which corresponds to the lower value in the PM10 DNEL range calculated here based on the H(EC) values reported by Oller et al. (2014) and the application of assessment factors. 3.3. Local inhalation effects drive health risks of Ni inhalation The contribution of Ni via inhalation should also be taken into account for the total, aggregated systemic dose. Exposure via inhalation, and its contribution to aggregated exposure are very marginal for humans indirectly exposed to Ni via the environment when the Ni air concentration is below the possible DNEL values of 20 and 60 ng Ni/ m3. Continuous inhalation exposure to air concentrations up to 20 and 60 ng/m3, leads to an additional systemic dose via inhalation of 0.0057 and 0.017 μg/kg bw per day for an adult (inhalation rate of 20 m3/day and bw 70 kg), equivalent to an additional internal dose of 0.003 and 0.009 μg/kg bw per day (assuming 100% deposition and an inhalation absorption factor of 50%). The additional doses are less than 3 and 9% of the typical systemic dose via the oral pathway. The reasonable worst case total absorbed dose (including all routes of exposure except inhalation), relevant for systemic effects (regional background + local contribution) for the EU population can be calculated as 16.7 μg Ni/day or 0.24 μg Ni/kg bw/day (De Brouwere et al., 2012). The value of 0.24 μg Ni/kg bw/day is several-fold lower than the absorbed dose associated with the DNEL for oral elicitation of systemic effects of 3.6 μg/kg/day (De Brouwere et al., 2012). This implies that as long as Ni air concentrations are below 20 to 60 ng Ni/m3, the DNEL for systemic effects (aggregated dose) would not be reached. This calculation confirms that it is the local effects after inhalation and not the systemic effects that are driving the health risks associated with inhalation exposures to nickel compounds. 3.4. Regional exposure RWC background Ni air concentrations at EU scale were derived from the AirBase database for 2012, and corresponded to Ni in PM10,

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

367

Table 6 Distribution of annual Ni ambient air concentrations (ng Ni/m3) within and across EU Member States for 2012 by AirBase (data accessed September 2014). Country

Ni in PMb Stationsa

Mean

AT BE BG CH CY CZ DE DK EE ES FI FR GB HU IE IT LT LU LV NL NO PL RO SE SI Average

12 2 6

1.3 3.1 2.0

2.4 6.0 3.8

3 53

6.7 1.4

12.8 3.0

6 4 79

2.4 5.4 3.3

4.6 10.4 5.1

27 2 5

Combinedc

Ni in PM10

1.4 1.2 1.5

P90

Stations

Mean

P90

14

2.2

4.3

11

0.8

1.6d

94

2.0

3.9

4 40

1.1 2.6

2.2 5.8

76 5 2 5

4.0 0.5 6.3 1.6

7.6 1.0 12.0 3.1

1 87

2.4

4.4

3.2 2.3 2.9

6 2

1.8 0.1

3.5 0.2

24 4 4

2.3 0.5 3.6

4.6 0.9 6.9 4.5

4.6

Stations

Mean

P90

12 16 6 11 3 53 94 6 4 79 4 40 27 2 5 76 5 2 5 6 3 87 24 4 4

1.3 2.3 2.0 0.8 6.7 1.4 2.0 2.4 5.4 3.3 1.1 2.6 1.4 1.2 1.5 4.0 0.5 6.3 1.6 1.8 0.2 2.4 2.3 0.5 3.6

2.4 4.4 3.8 1.5d 12.7 3.0 3.7 4.5 10.3 5.1 2.1 5.8 3.2 2.2 2.9 7.6 1.0 11.9 3.1 3.5 0.3 4.4 4.6 0.9 6.9 4.5

a

Number of monitoring stations. b It is unknown in which PM fraction Ni was measured. c Data of Ni in PM and Ni in PM10. d P90 values derived for Ni in PM or in PM10 may differ marginally from the combined dataset as the P90/Mean ratio differs according to available data in the analysis (Ni in PM, Ni in PM10 or combined); AT: Austria; BE: Belgium; BG: Bulgaria; CH: Switzerland; CY: Cyprus; CZ: Czech Republic; DE: Germany; DK: Denmark; EE: Estonia; ES: Spain; FI: Finland; FR: France; GB: Great Britain; HU: Hungary; IE: Ireland; IT: Italy; LT: Lithuania; LU: Luxembourg; LV: Latvia; NL: Netherlands; NO: Norway; PL: Poland; RO: Romania; SE: Sweden; SI: Slovenia.

Ni in PM (unknown in which PM fraction Ni is measured) and the combination of both. Monitoring stations defined as industrial were excluded (see Table 6). The REACH guidance stipulates that data have to be properly assigned to the appropriate spatial scale (local or regional scale) by taking into account sources of exposure and the environmental fate of the substance. In practice, such an approach is in most cases not possible due to the lack of additional information in the datasets on exact location or presence of potential point sources (e.g., nearby industrial activities, roads, and effluent emission). We are aware of the presence of local metal industry point sources near measuring stations of some of the AirBase database, however, not for all of them. A pragmatic approach is to ensure that monitoring stations near local point sources are excluded from the regional scale assessment. This was done by omitting the data of AirBase monitoring stations identified as industrial type. Recent AirBase analysis shows that in 2012, data are available from 239 monitoring stations in 16 countries for Ni in PM, 339 monitoring stations in 10 countries for Ni in PM10 and 578 monitoring stations in 25 countries for the combined dataset. The 2012 P90 values were: 4.6 ng Ni/m3 for Ni in PM10, 4.5 ng Ni/m3 for Ni in PM, and 4.5 ng Ni/m3 for the combined dataset (see Table 6). These concentrations are slightly lower than values derived for the years 2007 & 2009, possibly due to the increasing number of monitoring stations or decreasing background due to emission reductions related to the EU air pollution policy. The value 4.5 ng Ni/m3 was chosen here as the background EU value instead of the PM10 value of 4.6 ng Ni/m3. The reason for this is that there are only 10 EU countries for which Ni in PM10 are reported, whereas the combined dataset is based on 25 countries and this improves geographical coverage. The information on Ni in PM does not specify in which fraction nickel is measured (e.g., PM10 or TSP) but since PM10 is a subset of TSP this is a conservative assumption. The 90th percentile exposure is well below both DNEL values.

3.5. Local exposure The maximum additional Ni air concentration at the local scale (Clocal) resulting in a RCR b1 was derived by subtracting the EU RWC PECregional of 4.5 ng Ni/m3 from the selected DNEL values of 20 and 60 ng Ni/m3. Depending on the selected DNEL value, the maximum allowable Clocal would result in values of 15.5 and 55.5 ng Ni/m3. Therefore, compliance with the DNEL values of 20 and 60 ng Ni/m3 is achieved when the modelled local air Ni concentrations are below 15.5 or 55.5 ng Ni/m3. The tiered approach assessing the Ni air concentration in the vicinity of industrial sites is presented in Fig. 1. The approach recommends the use of reliable air monitoring data at locations relevant for the general population if available. If measured data are not available, air Ni-concentrations should be estimated using modelling tools. The tiered approach recommends to first use screening air dispersion models, and to use more detailed air quality models only in cases where further refinement of the assessment is needed. For the EUSES 2.0 model a maximum allowable daily air emission of 56 or 200 g Ni/day was derived. This is based on the fact that in EUSES an emission of 1 kg/s corresponds with a local air concentration, estimated at 100 m of the point source, of 24 × 10− 6 kg/m3. If the maximum allowable Clocal is 15.5 or 55.5 ng Ni/m3 (in accordance with a RCR b1) this corresponds with an air emission of 56 or 200 g Ni/day. Assuming 365 days of emission, this would result in a yearly emission of 20 or 73 kg Ni/year. For 29 out of 33 industrial sites included in the survey, RCR was demonstrated to be below 1 following the tiered approach for the DNEL of 20 ng Ni/m3. When the DNEL of 60 ng Ni/m3 was used, a RCR b 1 was shown for all 33 industrial sites. Table 7 gives an overview of a number of industrial sites belonging to the Ni sulphate sector where, according to the different tiers, safe use/production of Ni could be demonstrated when using a DNEL of 20 ng Ni/m3. Fig. 2 illustrates the followed tiered approach for the Ni sulphate sector. In those cases where the RCR values would still exceed 1, even after applying a tier IIB model, the companies

368

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

Fig. 1. Tiered approach for assessing compliance with Ni Derived No Effect Levels (DNELs) for respiratory effects. Scenario A is based on current EU PM10 ambient air target value (20 ng/m3) while scenario B is based on the DNEL value derived in this manuscript taking into account dosimetric adjustments (60 ng Ni/m3; see Table 4).

are advised to collect relevant measured exposure data. When the RCR is N 1 but close to 1, a further assessment of the site-specific risk could be performed by considering the nickel background ambient air data relevant to the country where the site is located. When measured and/or higher tier modelled data indicates the RCR is clearly above 1, appropriate risk reduction measures need to be applied. Fig. 3 presents the total Ni air concentrations for nine different industrial sites of the Ni sulphate sector using the different tiers. Modelled local air concentrations were generally the highest for the predictions made by EUSES. For the seven sites for which both EUSES and GPM data were available, GPM

predicted higher Ni air concentrations than EUSES (by a factor of approximately 2.5) for two sites. There are several reasons that can explain these results. By changing the default stack characteristics in EUSES to more realistic values in the GPM model, the maximum air concentration as a function of distance can occur further away from the emission source than the 100 m distance from the stack (de Bruin et al., 2010). In the EUSES model a standard distance of 100 m from the point source is used. For the GPM model we used the maximum air concentration value, regardless of the distance and wind speed. A second factor explaining the observed larger predictions using the GPM model

2 2

1

369

compared to EUSES is that EUSES considers a standard wind speed of 3 m/s, while the GPM model calculates air concentrations for wind speeds ranging from 1 m/s to 19 m/s. Finally, results given by the IFDM model, in which meteorological data were taken into account, always predicted lower values than EUSES (median of all values factor of 20 lower) or the GPM model (factor of 15). The proposed tiered approach was applied to total Ni levels. In the future, when more data might become available on speciation of Ni in ambient air, one could compare DNEL information for specific Ni compounds with exposure to these compounds. Approaches exist to estimate fugitive (non-stack) emissions at industrial sites (Bleux et al., 2007); however diffuse emissions at the industrial sites were currently not considered as emission data are mainly lacking. This may result in an underestimation of the modelled Ni air concentration closer to industrial sites. However, for sites at which monitoring data were available, the measured concentrations were always substantially lower than the EUSES modelled concentrations.

1

2 DNEL = 60 ng Ni/m3 DNEL = 20 ng Ni/m3

Number of additional sites with RCR b 1 based on Tier IIb

Number of sites for which no detailed stack information was available

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

3 8

3 6

3

1 4

2

1 6

3 4

3

1 3

2

2

3

The approaches used in this study can be used in the context of fulfilling REACH requirements, both for nickel and for other metal substances where the standard approaches (e.g., EUSES model) fail to demonstrate safe use. The models mentioned in this paper to predict local air nickel concentrations (EUSES, GPM, IFDM) are not specific to nickel but are valid for other substances defined as non-reactive in the atmosphere (e.g., van Leeuwen and Vermeire, 2007; Lefebvre et al., 2011a). The tiered approach proposed here to demonstrate compliance at the site level and to guide companies on risk management options has not, to our knowledge been described and published before. These models to predict local air concentrations could be applicable to other jurisdictions outside the EU provided that the model input data are available. The approach applied to the derivation of an alternative DNEL is also not limited to the EU (e.g., U.S. EPA uses a dosimetric approach in their Reference Concentration, RfC, derivation with the subsequent application of assessment factors) and could have applicability in other jurisdictions.

1

DNEL = 20 ng Ni/m3 DNEL = 60 ng Ni/m3 DNEL = 20 ng Ni/m3

2

Number of additional sites with RCR b 1 based on Tier IIa Number of additional sites with RCR b 1 based on Tier I

DNEL = 60 ng Ni/m3

3.6. Approach's applicability

1

RCR : Risk Characterisation Ratio. a Consortium information.

1a

2

1 4

3

1

1 1 1 1 4 10

1 1 1 1 6 9

Producers DU1: Metal surface treatment DU2: Battery production DU3: Production of Ni salts from Ni sulphate DU4: Use of Ni sulphate in pigment production DU5: Selective plating DU6: Formulation of Ni compounds DU7: Production of Ni-containing glass DU8: Metal surface treatment in the abrasives industry

DNEL = 60 ng Ni/m3 DNEL = 20 ng Ni/m3

Number of sites for which monitoring showed compliance Number of recorded industrial sites

Table 7 Result of the tiered approach (see Fig. 1) for industrial sites in the Ni sulphate sector.

4. Conclusions Human health risks associated with environmental nickel exposure via inhalation are driven by local effects rather than systemic effects. Based on 2012 data, the P90 regional value for measured nickel concentrations was estimated to be 4.5 ng Ni/m3 which is below the current EU ambient air target PM10 value of 20 ng/m3 or an alternative DNEL of 60 ng Ni/m3 (derived here based on recent nickel data). By contrast, local concentrations are often unknown. A tiered modelling approach was developed to estimate the nickel air concentration in the vicinity of industrial installations. When measured data is not available the tiered approach goes from a standard model with minimal input information (Tier I, EUSES) to more complex models (Tier IIa and IIb). When this approach is applied to predict local emissions near Ni sulphate producers and downstream users, the EUSES model was able to demonstrate compliance with a DNEL of 60 ng Ni/m3 for the majority of the sites. The value of the refined models is demonstrated when the most conservative DNEL value of 20 ng/m3 (PM10) corresponding to the EU ambient air target value is considered. The tiered approach can also be applied for metals other than nickel when default predictive tools such as EUSES fail to demonstrate compliance. Acknowledgements This work was funded by the Nickel REACH Consortia and the Nickel Institute.

370

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371

Fig. 2. Presentation of tiered approach for downstream users active in the production of Ni salts from Ni sulphate. Percentage represents the cumulative percentage of sites compliant with the respective DNEL.

Fig. 3. PEClocal (ng Ni/m3) 1) assessed by modelling (EUSES, GPM, IFDM): Clocal + PECregional or 2) assessed by monitoring data for different industrial sites of the Ni sulphate sector. Industrial sites for which only data on EUSES modelling was available are not shown. The horizontal band presents the range of selected DNEL values (20–60 ng Ni/m3). For site 3, site 5 and site 9 available measured data were shown.

References ATSDR Agency for Toxic Substances and Disease Registry, 2005. Toxicological Profile for Nickel. Atlanta. US Public Health Service, GA. Bal, W., Protas, A.M., Kasprzak, K.S., 2011. Genotoxicity of metal ions: chemical insights. Met. Ions Life Sci. 8, 319–373. Barbante, C., Boutron, C., Moreau, A.L., Ferrari, C., Van de Velde, K., Cozzi, G., et al., 2002. Seasonal variations in nickel and vanadium in Mont Blanc snow and ice dated from the 1960s and 1990s. J. Environ. Monitor. 4, 960–966. Bleux, N., Berghmans, P., Liekens, I., Sleeuwaert, F., van Rompaey, H., Mensink, C., Torfs, R., 2007. Evaluation of methods to quantify fugitive dust sources in Flanders — fugitive process emissions. DustConf 2007, How to Improve Air Quality, The Netherlands (Available at http://www.dustconf.com [Accessed September 2014]). Bolt, H.M., Huici-Montagud, A., 2008. Strategy of the scientific committee on occupational exposure limits (SCOEL) in the derivation of occupational exposure limits for carcinogens and mutagens. Arch. Toxicol. 82, 61–64. Briggs, G.A., 1969. Plume rise. USAEC Critical Review Series, TID-25075. National Technical Information Service, Virginia: Springfield. Briggs, G.A., 1971. Some recent analyses of plume rise observation. In: Englund, H.M., Beery, W.T. (Eds.), Proceedings of the Second International Clean Air Congress. Academic Press, New York, pp. 1029–1032. Briggs, G.A., 1972. Discussion on chimney plumes in neutral and stable surroundings. Atmos. Environ. 6, 507–510.

Bright, P., Burge, P.S., O'Hickey, S.P., Gannon, P.F., Robertson, A.S., Boran, A., 1997. Occupational asthma due to chrome and nickel electroplating. Thorax 52, 28–32. Bultynck, H., Malet, L.M., 1972. Evaluation of atmospheric dilution factors for effluents diffused from an evaluated continuous point source. Tellus 24, 453–472. Cirla, A.M., Baruffini, A., Pisati, G., Zedda, S., 1982. Allergic bronchial reactions due to stainless steel welding fumes. Lav. Um. 30, 17–20. Cornille, P., 2009. The Flemish atmospheric dispersion model IFDM. Potential and limitations: point of view of typical users. Int. J. Environ. Pollut. 8, 778–781. Cosemans, G., Roekens, E., 2002. Assessing uncontrolled emissions near a lead works. In: Borrego, C., Schayes, G. (Eds.), Air Pollution Modeling and Its Applications XV. Kluwer Academic/Plenum Publishers, New York, pp. 507–508. De Brouwere, K., Buekers, J., Cornelis, C., Schlekat, C.E., Oller, A.R., 2012. Assessment of indirect human exposure to environmental sources of nickel: oral exposure and risk characterization for systemic effects. Sci. Total Environ. 419, 25–36. De Bruin, Y.B., de Knecht, J., Hollander, A., Bakker, J., van Jaarsveld, H., Hogendoorn, E., 2010. Risk assessment using EUSES. Refinement options to estimate atmospheric transportation by its Operational Priority Substances model (OPS). Hum. Ecol. Risk. Assess. 16, 945–961. DIRECTIVE 2004/107/EC, 2004. Arsenic, Cadmium, Mercury, Nickel and Polycyclic Aromatic Hydrocarbons in Ambient Air. EC European Communities, 2001. Ambient Air Pollution by As, Cd and Ni Compounds. Position Paper. Office for Official Publications of the European Communities, Luxembourg (Available at http://ec.europa.eu/environment/air/pdf/pp_as_cd_ni.pdf. [Accessed September 2014]). ECB European Chemicals Bureau, 2008. European Union Riks Assessment Reports on Nickel and Nickel Compounds. Office for Official Publications of the European Communities, Luxembourg. ECETOC European Centre for Ecotoxicology and Toxicology of Chemicals, 2010. Guidance on assessment factors to derive DNELs. Technical Report No. 110. ECETOC, Brussels. ECETOC European Centre for Ecotoxicology and Toxicology of Chemicals, 2003. Derivation of assessment factors for human health risk assessment. Technical Report No. 86. ECETOC, Brussels. EUSES European Union System for the Evaluation of Substances 2.0, 2004. Prepared for the European Chemicals Bureau by the National Institute of Public Health and the Environment (RIVM), Bilthoven, The Netherlands. (RIVM Report no. 601900005; 2004). Available via the European Chemicals Bureau. Available at http://ihcp.jrc.ec. europa.eu/our_activities/public-health/risk_assessment_of_Biocides/euses [Accessed September 2014]). Füchtjohann, L., Jakubowski, N., Gladtke, D., Klockow, D., Broekaert, J.A., 2001. Speciation of nickel in airborne particulate matter by means of sequential extraction in a micro flow system and determination by graphite furnace atomic absorption spectrometry and inductively coupled plasma mass spectrometry. J. Environ. Monitor. 3, 681–687. Galbreath, K.C., Crocker, C.R., Nyberg, C.M., Huggins, F.E., Huffman, G.P., 2003. Ni speciation of urban particulate matter from Broward county, Florida. Abstr. Pap. Am. Chem. Soc. 48, 779–781. Goodman, J.E., Prueitt, R.L., Dodge, D.G., Thakali, S., 2009. Carcinogenicity assessment of watersoluble nickel compounds. Crit. Rev. Toxicol. 39, 365–417. Goodman, J.E., Prueitt, R.L., Thakali, S., Oller, A.R., 2011. The nickel ion bioavailability model of the carcinogenic potential of nickel-containing substances in the lung. Crit. Rev. Toxicol. 41, 142–174. Greim, H., Ziegler-Skylakakis, K., 2007. Risk assessment for biopersistent granular particles. Inhal. Toxicol. 19 (Suppl. 1), 199–204. Haney Jr., J.T., McCant, D.D., RLJr, Sielken, Valdez-Flores, C., Grant, R.L., 2012. Development of a unit risk factor for nickel and inorganic nickel compounds based on an updated carcinogenic toxicity assessment. Regul. Toxicol. Pharmacol. 62, 191–201.

J. Buekers et al. / Science of the Total Environment 521–522 (2015) 359–371 Hattis, D., Silver, K., 1994. Human interindividual variability — a major source of uncertainty in assessing risks for noncancer health-effects. Risk Anal. 14, 421–433. Hattis, D., Erdreich, L., Ballew, M., 1987. Human variability in susceptibility to toxicchemicals. A preliminary-analysis of pharmacokinetic data from normal volunteers. Risk Anal. 7, 415–426. Hattis, D., Banati, P., Goble, R., Burmaster, D.E., 1999. Human interindividual variability in parameters related to health risks. Risk Anal. 19, 711–726. Hengstler, J.G., Bogdanffy, M.S., Bolt, H.M., Oesch, F., 2003. Challenging dogma-thresholds for genotoxic carcinogens? The case of vinyl acetate. Annu. Rev. Pharmacol. 43, 485–520. Hughes, K., Meek, M.E., Newhook, R., Chan, P.K.L., 1995. Speciation in health risk assessments of metals: evaluation of effects associated with forms present in the environment. Regul. Toxicol. Pharmacol. 22, 213–220. IARC International Agency for Research on Cancer, 2012. Monographs on the evaluation of carcinogenic risks to humans. A Review of Human Carcinogens: Arsenic, Metals, Fibres, and dusts. IARC, Lyon. Jarabek, A.M., Asgharian, B., Miller, F.J., 2005. Dosimetric adjustments for interspecies extrapolation of inhaled poorly soluble particles (PSP). Inhal. Toxicol. 17, 317–334. Kraut, A., Molino, L., MacKenzie, C., 2010. Dermatologic and respiratory symptoms in a nickel processing operation. Poster Presented at the AOHC (American Occupational Health Conference), Miami, Florida, USA, May 2010. Lefebvre, W., Fierens, F., Trimpeneers, E., Janssen, S., Van de Vel, K., Deutsch, F., et al., 2011a. Modeling the effects of a speed limit reduction on traffic-related elemental carbon (EC) concentrations and population exposure to EC. Atmos. Environ. 45, 197–207. Lefebvre, W., Vercauteren, J., Schrooten, L., Janssen, S., Degraeuwe, B., Maenhaut, W., et al., 2011b. Validation of the MIMOSA–AURORA–IFDM model chain for policy support: modeling concentrations of elemental carbon in Flanders. Atmos. Environ. 45, 6705–6713. Lefebvre, W., Degraeuwe, B., Beckx, C., Vanhulsel, M., Kochan, B., Bellemans, T., et al., 2013. Presentation and evaluation of an integrated model chain to respond to traffic- and health-related policy questions. Environ. Model Softw. 40, 160–170. Maruyama, W., Hirano, S., Kobayashi, T., Aoki, Y., 2006. Quantitative risk analysis of particulate matter in the air: interspecies extrapolation with bioassay and mathematical models. Inhal. Toxicol. 18, 1013–1023. Mauderly, J.L., 1997. Relevance of particle-induced rat lung tumors for assessing lung carcinogenic hazard and human lung cancer risk. Environ. Health Perspect. 105, 1337–1346. Nikula, K.J., Vallyathan, V., Green, F.H., Hahn, F.F., 2001. Influence of exposure concentration or dose on the distribution of particulate material in rat and human lungs. Environ. Health Perspect. 109, 311–318.

371

Oberdörster, G., 1995. Lung particle overload: implications for occupational exposures to particles. Regul. Toxicol. Pharmacol. 27, 123–135. OEHHA Office for Environmental Health Hazard Assessment, 2008. TSD for non-cancer RELs, nickel and nickel compounds, including nickel oxide. Reference Exposure Levels. California Environmental Protection Agency, California (Available at: http:// www.oehha.ca.gov/air/Allrels.html [accessed February 2015]). Oller, A.R., Kirkpatrick, D.T., Radovsky, A., Bates, H.K., 2008. Inhalation carcinogenicity study with Ni metal powder in Wistar rats. Toxicol. Appl. Pharmacol. 223, 262–275. Oller, A.R., Oberdörster, G., Seilkop, S., 2014. Derivation of PM10 size-selected human equivalent concentrations of inhaled nickel based on cancer and non-cancer effects in the respiratory tract. Inhal. Toxicol. 26, 559–578. REACH Registration Evaluation Authorisation and Restriction of Chemicals, 2010. Guidance on information requirements and chemical safety assessment. Chapter R.8: Characterisation of Dose [Concentration]–Response for Human Health. European Chemicals Agency, Helsinki (Available at http://echa.europa.eu/web/guest/guidance-documents/ guidance-on-reach [accessed September 2014]). Renwick, A.G., Lazarus, N.R., 1998. Human variability and noncancer risk assessment — an analysis of the default uncertainty factor. Regul. Toxicol. Pharmacol. 27, 3–20. Rydh, C.J., Svärd, B., 2003. Impact on global metal flows arising from the use of portable rechargeable batteries. Sci. Total Environ. 302, 167–184. SCOEL Scientific Committee on Occupational Exposure Limits, 2011. Recommendation From the Scientific Committee on Occupational Exposure Limits for Nickel and Inorganic Nickel Compounds. SCOEL/SUM/85. Sutton, O.G., 1932. A theory of Eddy diffusion in the atmosphere. Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 135, pp. 143–165. TCEQ Texas Commission on Environmental Quality, 2011. Nickel and Inorganic Nickel Compounds Development Support Document. Chief Engineer's Office, Texas (Available at http://www.tceq.state.tx.us/assets/public/implementation/tox/dsd/final/ june11/nickel_&_compounds.pdf [Accessed September 2014]). Tirez, K., Silversmit, G., Vincze, L., Servaes, K., Vanhoof, C., Mertens, M., et al., 2011. Speciation and fractionation of nickel in airborne particulate matter: comparison between selective leaching and XAS spectroscopy. J. Anal. At. Spectrom. 26, 517–527. van Jaarsveld, H.J.A., 1990. A quantitative model analysis of year to year changes in concentration and deposition. Presented at the NATO CCMS Meeting, Vancouver, Canada. van Leeuwen, C.J., Vermeire, T.G., 2007. Risk Assessment of Chemicals: An introduction. 2nd ed. Springer, New York, p. 167.