Human risk assessment of heavy metals: principles and applications

Abstract

Humans are exposed to a number of "heavy metals" such as cadmium, mercury and its organic form methylmercury, uranium, lead, and other metals as wel as metalloids, such as arsenic, in the environment, workplace, food, and water supply. Exposure to these metals may result in adverse health effects, and national and international health agencies have methodologies to set health-based guidance values with the aim to protect the human population. This chapter introduces the general principles of chemical risk assessment, the common four steps of chemical risk assessment: hazard identification, hazard characterization, exposure assessment, risk characterization, and toxicokinetic and toxicity aspects. Finally, the risk assessments performed by international health agencies such as the World Health Organisation, the Environmental Protection Agency of the United States, and the European Food Safety Authority are reviewed for cadmium, lead, mercury, uranium, and arsenic.

Full text

2
Human Risk Assessment of Heavy Metals:
Principles and Applications
Jean-Lou C. M. Dorne,
1*
George E. N. Kass,
1
Luisa R. Bordajandi,
1
Billy Amzal,
1
Ulla Bertelsen,
1
Anna F. Castoldi,
1
Claudia Heppner,
1
Mari Eskola,
1
Stefan Fabiansson,
1
Pietro Ferrari,
1
Elena Scaravelli,
1
Eugenia Dogliotti,
2
Peter Furst,
3
Alan R. Boobis
4
and
Philippe Verger
5
1
European Food Safety Authority, Largo N. Palli 5, I-43100 Parma, Italy
2
Istituto Superiore di Sanita, Viale Regina Elena 299, I-00161 Rome, Italy
3
Chemisches Landes- und Staatliches Veterina
¨runtersuchungsamt, Joseph-Ko
¨nigstrasse 40,
D-48147 Mu
¨nster, Germany
4
Imperial College, Department of Experimental Medicine and Toxicology, Burlington Danes,
Hamersmith Campus, Du Cane Road, London, W12 ONN, UK
5
World Health Organisation, Department of Food Safety and Zoonoses, 20 Avenue Appia,
CH-1211 Geneva, Switzerland
ABSTRACT 28
1. INTRODUCTION 29
2. PRINCIPLES OF CHEMICAL RISK ASSESSMENT 29
2.1. Risk Assessment of Non-Genotoxic and Genotoxic
Carcinogens 30
2.2. The Four Pillars of Risk Assessment 33
3. TOXICOLOGY OF HEAVY METALS 34
3.1. General Principles 34
3.2. Toxicokinetics 35
Metal Ions in Life Sciences, Volume 8 Edited by Astrid Sigel, Helmut Sigel, and Roland K. O. Sigel
rRoyal Society of Chemistry 2011
Published by the Royal Society of Chemistry, www.rsc.org
*
Correspondence to ojean-lou.dorne@efsa.europa.eu4
Met. Ions Life Sci. 2011,8, 27–60

3.2.1. Absorption, Distribution, Metabolism, and Excretion
of Heavy Metals and Metalloids 35
3.2.2. Physiologically-Based and Population-Based
Toxicokinetic Models 36
3.3. Toxicodynamics 37
3.4. Selected Molecular Mechanisms of Action: Epigenetic
Mechanisms of Carcinogenicity 38
4. ANALYTICAL TECHNIQUES AND EXPOSURE
ASSESSMENT OF HEAVY METALS 39
4.1. Analytical Techniques for the Detection of Heavy Metals
and Metalloids in Biological Samples 39
4.2. Data Sources for the Estimation of Human Dietary Exposure 40
4.3. Combining Occurrence and Consumption Data in Humans
for Exposure Assessment 42
5. APPLICATIONS TO THE HUMAN RISK ASSESSMENT OF
HEAVY METALS AND METALLOIDS 43
5.1. Hazard Identification and Characterization 44
5.1.1. Cadmium 44
5.1.2. Lead 44
5.1.3. Methylmercury 45
5.1.4. Uranium 45
5.1.5. Arsenic 46
5.2. Exposure Assessment of Heavy Metals and Metalloids 47
5.2.1. Cadmium 47
5.2.2. Lead 47
5.2.3. Methylmercury 48
5.2.4. Uranium 48
5.2.5. Arsenic 49
5.3. Risk Characterization of Heavy Metals and Metalloids 50
5.3.1. Cadmium 50
5.3.2. Lead 51
5.3.3. Mercury 51
5.3.4. Uranium 52
5.3.5. Arsenic 52
6. CONCLUSIONS AND FUTURE PERSPECTIVES 53
ACKNOWLEDGMENTS 54
ABBREVIATIONS AND DEFINITIONS 54
REFERENCES 55
ABSTRACT: Humans are exposed to a number of ‘‘heavy metals’’ such as cadmium,
mercury and its organic form methylmercury, uranium, lead, and other metals as well
as metalloids, such as arsenic, in the environment, workplace, food, and water supply.
Exposure to these metals may result in adverse health effects, and national and
28 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

international health agencies have methodologies to set health-based guidance values
with the aim to protect the human population. This chapter introduces the general
principles of chemical risk assessment, the common four steps of chemical risk assess-
ment: hazard identification, hazard characterization, exposure assessment, risk char-
acterization, and toxicokinetic and toxicity aspects. Finally, the risk assessments
performed by international health agencies such as the World Health Organisation, the
Environmental Protection Agency of the United States and the European Food Safety
Authority are reviewed for cadmium, lead, mercury, uranium, and arsenic.
KEYWORDS: arsenic .cadmium .lead .mercury .risk assessment .toxicokinetics .
toxicity .uranium
1. INTRODUCTION
Humans are exposed to a range of ‘‘heavy metals’’ such as cadmium, mer-
cury and its organic form methylmercury (CH
3
-Hg), uranium, lead, and
other metals as well as metalloids, such as arsenic, in the environment,
workplace, food and water supply. In history, a plethora of epidemiological,
toxicological and molecular evidence from all around the globe has shown a
variety of health risks to human populations associated with environmental,
occupational, and dietary exposure to such metals. Consequently, health
agencies have been setting health-based guidance values to prevent the
occurrence of adverse health effects in humans.
The aim of this chapter is first to introduce the four steps of chemical risk
assessment for non-genotoxic and genotoxic carcinogens, namely hazard
identification, hazard characterization, exposure assessment and risk char-
acterization. The toxicology and risk assessment performed by international
health agencies on cadmium, lead, mercury, uranium and arsenic are
reviewed together with potential future developments.
2. PRINCIPLES OF CHEMICAL RISK ASSESSMENT
Risk has been defined as a function of hazard and exposure. The Interna-
tional Program on Chemical Safety (IPCS) of the World Health Organisa-
tion (WHO) has defined hazard as ‘‘the inherent property of an agent or
situation having the potential to cause adverse effects when an organism,
system or (sub) population is exposed to that agent’’ and risk as ‘‘the
probability of an adverse effect in an organism, system or (sub)population
caused under specified circumstances by exposure to an agent’’ [1]. The
qualification and quantification of hazard and risk are the corner stones of
29HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

the risk assessment paradigm. In terms of food safety, the European Union
has defined ‘‘hazard’’ as a biological, chemical or physical agent in, or
condition of, food and ‘‘risk’’ as a function of the probability of an adverse
health effect and the severity of that effect, consequential to a hazard [2].
2.1. Risk Assessment of Non-Genotoxic and Genotoxic
Carcinogens
Risk assessment of chemicals in humans relies on a mechanistic assumption
that such chemicals may either be genotoxic or non-genotoxic. Genotoxic
carcinogens and their metabolites are assumed to act via a mode of action
that involves a direct and potentially irreversible DNA-covalent binding
whereas non-genotoxic carcinogens or their metabolites are assumed to act
via an epigenetic mode of action without covalent binding to DNA. In terms
of risk assessment, a linear low dose-response relationship for life time
exposure with no threshold or a dose without a potential effect is usually
assumed for genotoxic carcinogens whereas a threshold level of exposure,
below which no significant effects are induced, is assumed for non-genotoxic
carcinogens (and for almost all non-cancer endpoints). For the latter, this
implies that homeostatic mechanisms are able to balance biological pertur-
bations produced by low levels of intake, and that structural or functional
changes leading to adverse effects, which may include cancer, would be
observed only at higher intakes [3,4].
Worldwide, the risk assessment of genotoxic carcinogens is performed
using one of the three major methods namely linear extrapolation from high
dose animal studies to low exposures in humans, the threshold of tox-
icological concern and the margin of exposure approach.
The linear extrapolation (LE) approach has been used by the US Envir-
onmental Protection Agency (US-EPA), Norway and in the European
Union for industrial chemicals, non-threshold carcinogens and for carci-
nogens for which the mode of action is unclear. LE often involves modelling
of dose-response data from high dose carcinogenicity studies in animals
using the lower end of the observed range of tumor incidences. Hence, a risk
estimate of cancer for low dose life time exposure in humans (1 in 10
5
or 10
6
)
can be derived and often LE has involved the lower 95% confidence interval
of the bench mark dose (BMD) producing a 10%, 5%, 1% increase in tumor
incidence compared to background incidences (BMDL10, BMDL05,
BMDL01) from mostly animal data or on rare occasions human epide-
miological data when available. Overall, LE provides estimates of the pos-
sible range of cancer risk associated with lifetime exposure to a particular
concentration of a genotoxic carcinogen in food, air or from other exposure
30 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

routes (e.g., a risk of 0–1 in a million). LE has limitations in the fact that the
potency of the carcinogen in animals is assumed to relate directly to the
potency in humans and such assumptions are still not supported by sub-
stantive data [5]. In addition, considerable uncertainty is introduced by the
extent to which it is often necessary to extrapolate to human exposure levels.
The threshold of toxicological concern (TTC) was originally proposed by
Cramer et al. [6] to establish exposure thresholds predicted to be without
adverse effects based on the distribution of potencies of a large number of
compounds. One of the main advantages of the TTC approach is that low
exposure risk can be evaluated without the need for chemical-specific data
from animal toxicity studies as proposed in a TTC decision tree by Kroes
et al. [7]. From this analysis, threshold values for three groups of non-gen-
otoxic chemicals were proposed according to their toxicity in relation to
human exposure and expressed in mg/kg b.w./day for a 60 kg adult with
group I (30) (low), group II (9) (intermediate) and group III (1.5) (high)
[5,7,8]. However, this approach is not relevant to heavy metals since metals
were excluded when the TTCs were derived [3,8].
The margin of exposure (MOE) approach was introduced after an inter-
national conference organized by the International Life Sciences Institute
(ILSI), the Joint Food and Agricultural Organization of the United Nations/
WHO (FAO/WHO) Expert Committee on Food Additives (JECFA),
and the scientific committee of the European Food Safety Authority (EFSA)
[9–11]. The MOE is defined as the ratio of a specified point on a dose-
response curve for adverse effects obtained in animal experiments (in the
absence of human epidemiological data) and human intake data. Like for
the LE approach, the preferred reference points describing the dose-response
relationship are the BMD and BMDL. Overall, the Scientific Committee of
EFSA considered that an MOE of 10,000 or more, based on a BMDL10
derived from animal cancer bioassay data and taking into account the
uncertainties in the interpretation, ‘‘would be of low concern from a public
health point of view and might reasonably be considered as a low priority for
risk management actions’’ [9]. EFSA has recently conducted a risk assess-
ment for the metalloid arsenic using this approach [12] (see Sections 5.1.5
and 5.2.5).
For non-genotoxic carcinogens, threshold levels of toxicity are defined as
‘‘without appreciable health risk’’ when consumed every day or weekly for a
lifetime such as the acceptable/tolerable daily intake (ADI/TDI) or provi-
sional tolerable weekly intake (PTWI) used in Europe and by the WHO, the
tolerable daily intake or tolerable concentration in Canada or the ‘reference
dose’ (RfD) in the United States by the US Environmental Protection
Agency (EPA) and the Agency for Toxic Substances and Disease Registry
(ATSDR) [13,14]. Despite the nomenclature differences, these health-based
guidance values are all determined by dividing a surrogate for the threshold
31HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

determined from chronic/subchronic animal studies using the most sensitive
species (usually mouse, rat, rabbit or dog), such as the no observed adverse
effect level (NOAEL) or the BMDL 95% lower confidence limit, by a default
uncertainty factor (UF) of a 100-fold [15]. The BMD is defined as a dose
level, derived from the estimated dose-response curve, associated with a
specified change in response, the benchmark response (BMR) (e.g., 0.1%,
1%, 5% or 10% incidence). The BMD limit (BMDL) is the lower confidence
bound, and is often used as the reference point. e.g., for a BMR of 5%, the
BMDL05 can be interpreted as a dose for which the response is likely to be
smaller than 5% and for which the term ‘‘likely’’ is defined by the statistical
confidence level, usually 95% confidence [16].
The 100-fold uncertainty factor has been further split to allow for dif-
ferences in toxicokinetics (TK), relating the external dose to the internal
dose: i.e., absorption, distribution, metabolism, and excretion, and in tox-
icodynamics (TD), relating the concentration of the proximate toxicant
(parent compound, metabolite or both) in the target organ(s) and the sen-
sitivity of the target organ(s) itself [17,18]. Renwick [17] proposed TK and
TD values of 4 and 2.5 for interspecies differences and even values of 3.16 for
human variability. These were derived from the analysis of a small database
describing interspecies differences, expressed as the ratio between the animal
species and humans for TK processes and parameters (e.g., liver weight, liver
blood flow, renal blood flow, absorption, elimination) as well as for TD
sensitivity to a chemical (e.g., sedation, pain relief) [19]. The subdivision was
subsequently adopted by the IPCS workshop on the derivation of guidance
values [20]. The main aim of this subdivision was to allow for chemical-
specific TK and ideally to derive chemical-specific adjustment factors
(CSAFs) [21,22]. Further refinements have been developed using the ther-
apeutic drug database and include pathway-related uncertainty factors (PR-
UFs) as an intermediate option between CSAFs and the UF when the
pathway of metabolism is known but compound-specific TK data are not
available. These have been derived for human variability in TK for phase I,
phase II, and renal excretion in subgroups of the human population and
interspecies variability [18,22–25] for test species for CYP1A2, glucur-
onidation, and renal excretion [15,26–29].
Ideally, CSAF or a physiologically-based toxicokinetic model (PB-TK)
when compound-specific data are available as recommended by the WHO
[22] and this approach has been recently explored by the panel on con-
taminants in the food chain (CONTAM) of the EFSA for the risk assess-
ment of cadmium in food for which a PB-PK model together with human
BMDL was used to set a PTWI for humans (see Section 5).
Beyond the mechanistic assumptions of genotoxicity and thresholded
toxicity, the application of the four pillars of risk assessment is common and
summarized below.
32 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

2.2. The Four Pillars of Risk Assessment
The four steps of the risk assessment, namely (1) hazard identification,
(2) hazard characterization, (3) exposure assessment, and (4) risk char-
acterization have enabled scientists and public health agencies to protect
consumers and the environment from adverse health effects that may result
from acute and chronic chemical exposure [30].
1. Hazard identification has been defined as ‘‘the identification of biolo-
gical chemical and physical agents capable of causing adverse health
effects and which may be present in a particular food or group of
foods’’. The main purpose of hazard identification applied to metals is
to evaluate the weight of evidence for adverse health effects, based on an
assessment of all the available data regarding toxicity and mode of
action (non-genotoxic/genotoxic) of the particular metal. In practice, a
review of studies regarding the mode of action (evidence for muta-
genicity, genotoxicity), the TK of the metal (absorption, distribution,
metabolism, and excretion), the nature of any toxicity or adverse health
effect occurring, and the affected (target) cell(s)/organ(s)/tissue(s) site
(TD) is performed. Toxicological studies in animals (mainly mouse, rat,
rabbit, and dog) play a critical role in hazard identification and ideally
use international guidelines and good laboratory practices (GLPs) and
include acute (single dose studies), sub-chronic (repeated dose studies:
28–90 days) and chronic studies (up to 2-year study) and/or more
specific endpoints (reproductive and developmental toxicity, neuro-
toxicity, immunotoxicity...) [31]. However, in the case of most heavy
metals (cadmium, mercury, methylmercury, lead) and metalloids
(arsenic), epidemiological human data were available and these have
been used to select critical studies for the setting of health-based gui-
dance values. For uranium, the results of chronic/sub-chronic (28–90
days) studies from the most sensitive species were selected.
2. Hazard characterization (also known as dose–response assessment)
constitutes ‘‘the qualitative and/or quantitative evaluation of the nature
of the adverse health effects associated with biological, chemical and
physical agents which may be present in food’’ [32]. Currently, the
BMD approach is preferred to the NOAEL/LOAEL approach because
it makes extended use of the dose-response data from studies in the
most sensitive species of experimental animals or from observational
epidemiological studies to estimate the shape of the overall dose-
response relationship for a particular endpoint so that both genotoxic
and non-genotoxic carcinogens can be assessed. In practice, the iden-
tification of the reference point (NOAEL/LOAEL or BMD/BMDL)
constitutes a basis for the risk characterization of a particular chemical.
33HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

An important distinction between thresholded toxicants and genotoxic
carcinogens is that the ADI/TDI for the former is derived in this step
usually by applying UFs whereas the MOE is derived in the risk
characterization part taking into account the human exposure.
3. Exposure assessment is ‘‘the qualitative and/or quantitative evaluation of
the likely intake of biological, chemical and physical agents via food as
well as exposure from other sources if relevant’’. For chemical con-
taminants in the food and the feed chain, exposure assessment integrates
the occurrence and the concentrations of the compound in the human
diet measured using validated analytical techniques and the human
consumption patterns for the different food categories available. Addi-
tionally, a range of intake/exposure scenarios are taken into account so
that special subgroups of the population that may be at either high
dietary exposure or high consumers are taken into account [4,32].
4. Risk characterization is the final step and represents ‘‘the qualitative
and/or quantitative estimation, including attendant uncertainties, of
the probability of occurrence and severity of known or potential
adverse health effects in a given population based on hazard identi-
fication, hazard characterization and exposure assessment’’ [32]. In
practice, risk characterization integrates the hazard identification and
characterization, leading to a health-based guidance value PTWI/TDI
and the human exposure, estimated from either a deterministic or
probabilistic method to conclude on the likelihood of adverse effects
for public health. In contrast, for genotoxic carcinogens, the MOE is
calculated in this step by dividing the point of departure, often a
BMDL, with the human exposure. Currently, an MOE of 10,000 is
considered of low public health concern but the interpretation has also
to be taken on a case by case basis. In summary, from the identifi-
cation and characterization of the toxicological effects (dose-response)
of a chemical a health-based guidance value is derived. Using vali-
dated analytical techniques, the amount of the chemical is measured in
a biological matrix (water, food, air, etc.) and combined with the
human consumption (via oral route or inhalation) of the biological
matrix to estimate human exposure. Exposure is then related to the
health-based guidance value to characterize the potential risk of
adverse health effects in humans after acute or chronic exposure [4].
3. TOXICOLOGY OF HEAVY METALS
3.1. General Principles
The old adage by Paracelsus stipulates ‘‘Sola dosis fecit venenum – it is only
the dose which makes a chemical a poison’’ and applies to heavy metals and
34 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

metalloids since such substances are undesirable in food and the environ-
ment. In principle, the toxicity of chemicals including metals arises from two
basic processes: what the body does to the chemical (toxicokinetics, TK) and
what the chemical does to the body (toxicodynamics, TD) [4].
3.2. Toxicokinetics
TK involves the translation of the external dose of a chemical to an internal
dose leading to overall elimination from the body, i.e., absorption from the
site of exposure, often the gastro-intestinal tract, distribution in body fluids/
tissues, metabolism to biologically inactive/active metabolites and ultimately
excretion in the urine/feces. Potential bioaccumulation in tissues of either the
parent compound or metabolites is an important aspect for TK and depends
on the absorption, distribution, metabolism, and excretion of the com-
pound. The biological half-life of the compound and its lipophilicity provide
good descriptors as to whether it will bioaccumulate or not. Although some
adjustment factors can be used to translate TK parameters from animals to
humans, only human toxicokinetics is addressed in this section.
3.2.1. Absorption, Distribution, Metabolism, and Excretion of
Heavy Metals and Metalloids
The main absorption routes of heavy metals are usually oral and inhalation.
Absorption from dermal exposure can still exist but at very limited level
(e.g., about 0.1% for uranium). The solubility of the metal forms is highly
influencing the absorption fraction, in either oral or pulmonary routes. Non-
soluble forms have generally a very limited absorption (below 1%) range of
values for various heavy metals absorbed via both routes. Oral absorption is
very variable ranging from 1–10% for cadmium, 10–50% for lead, 1–30%
for methylmercury, 1–6% for uranium, 40–100% for soluble forms of
arsenic [12,33,34].
Transport and distribution models are not always clear-cut for heavy
metals. For most of them, heavy metals get rapidly attached to blood cells
once absorbed. Blood (via erythrocyte binding) and plasma are typically the
main transport routes. Metabolic pathways for most heavy metals and
metalloids are generally complex and multiple and not always identified. For
example, accumulation of uranium in tissues may not be constant over time
during chronic exposure and can significantly accumulate in non-target
organism such as brain and teeth [33]. Furthermore, high inter-individual
variability is observed in human susceptibility and has been attributed to
genetic polymorphism in the enzymes associated with the metal metabolism,
especially in the case of arsenic [34]. Most of the studies dealing with this
35HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

genetic basis of variability in the human metabolism of arsenic concentrate
on the polymorphisms of arsenic-methyltransferases and glutathione-S-
transferases (mainly omega 1 and omega 2 isoforms) [34]. For most metals,
long-term accumulation occurs to a large proportion in the kidney for As,
Cd, and mercury, and in blood for lead, whereas uranium accumulates in
most organs and is released via the urinary route.
Most heavy metals are excreted via the kidney in the urine, and to a much
lesser extent by the gastrointestinal tract. The half-life, which characterizes
the elimination of heavy metals from the body, varies widely between metals.
It can be larger than 10–12 years for cadmium and lead, with inter-individual
variability of about 30% [35], 4 days for arsenic, 60 days for mercury and 0.5
to 1 year for uranium.
3.2.2. Physiologically-Based and Population-Based Toxicokinetic
Models
For most chemical compounds, the TK of heavy metals can be assessed
using compartment models such as the physiologically-based TK model
(PB-TK) or the population TK models.
The PB-TK models describe in more details the metabolic pathways and
allow the calculation of heavy metal concentrations in the main organs in the
body. On the other hand, the numerous parameters require substantial
parameter information and make any statistical evaluation and fitting more
difficult. It generally requires thorough sensitivity analysis and model vali-
dations. They usually provide estimates of the main TK parameters for a
typical individual, for a given body weight. Conversely, they are usually not
suitable to assess inter-individual variability of those parameters because
models become computationally too intricate. PB-TK models have been
built and used in humans for most heavy metals, such as arsenic [36–38],
cadmium (e.g., 8 compartment-model in [39,40]), lead [41], methylmercury
[42], chromium and uranium [43].
An alternative to PB-TK models is a population approach such as
population TK models, which are usually simpler (one or two compart-
ments), focused on the main elimination routes of the compound, and
making rough and global assumptions on other pathways of elimination. In
case of poor prior knowledge, this approach allows a simplified and parsi-
monious description of the compound’s elimination, hence enabling more
sophisticated statistical evaluations (such as the estimation of population
variability). Population models are therefore often an interesting option in
the area of human risk assessment, as they can provide a more precise and
reliable estimate of chemical-specific UFs [24]. However, in some cases
where the simple toxicokinetic assumptions are not met (like zero-order
36 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

absorption or linearity), such an approach could lead to models with poor
fits and high residuals. Moreover, such an approach does not allow for the
evaluation of a compound’s concentration in all organs. Recent population
models for heavy metals have been developed, e.g., for cadmium a
1-compartment model [35] and for arsenic a TK-TD model [44]. The choice
between PB-TK and population-based TK models depends on the precise
aim of the model and on the available data.
3.3. Toxicodynamics
Toxicological effects may occur when the toxic species, which either is the
parent compound or one or more of its metabolites, reaches a critical target
within the body. The cells in our body are equipped with a range of powerful
defence and repair mechanisms, and toxicity is only observed once this
protective barrier has been overwhelmed. The key defence mechanisms
comprise among others, small antioxidant molecules such as ascorbic acid
and a-tocopherol, the tripeptide glutathione (GSH) and a range of anti-
oxidant enzymes such as superoxide dismutases, catalase, GSH transferases
and GSH peroxidases [45]. Our cells are therefore well equipped to deal with
toxic compounds that induce conditions of oxidative stress. Indeed, the
majority of toxic drugs and environmental compounds and the effects
caused by ionizing and non-ionizing radiation, through the direct generation
of oxygen-based (ROS) (e.g., superoxide anion radical or hydroxyl radical)
or nitrogen-based free radicals (RNS) (e.g., peroxynitrite) or through the
depletion of cellular thiols via oxidation or conjugation, lead to conditions
of oxidative stress. These result in direct or indirect damage to cellular
proteins, phospholipids, and nucleic acids and in turn to a spectrum of
cellular effects ranging from cancer to cell death.
Toxic (non-essential) metals have been shown to induce conditions of
oxidative stress either through their ability to undergo redox-cycling and
generate ROS such as superoxide or as a consequence of enhanced pro-
duction of ROS by damaged mitochondria. For example, lead is able to
generate ROS [46] and similarly, enhanced formation of ROS from mito-
chondria occurs in cells exposed to arsenic [47], probably as a result of the
ability of the metal ion to bind to protein thiol groups and induce mito-
chondrial damage through opening of the mitochondrial permeability
transition pore [48].
A unique feature of toxic metals is the ability of the complexes, formed
between the metal ion and the nucleophilic sites on cellular proteins, to
mimic endogenous substrates or conformations. This property is responsible
for the selective transport of metal ions into or across cells and to interfere
with the functioning of target enzymes [49]. An example of this ionic
37HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

mimicry is the ability of lead to activate protein kinase C by acting as a
surrogate for the enzyme’s normal activator, Ca
21
[50]. Perturbations in cell
signaling in response to oxidative stress can lead to changes in cell pro-
liferation and cell differentiation, but from a toxicological point of view,
changes in cell survival signals, such as the growth factor-dependent phos-
phoinositide 3-kinase pathway, play a critical role in the development of a
number of diseases such as cancer [51]. When damage to the cell becomes
excessive, generally as a consequence of damage to mitochondria, cell death
pathways are activated, and these typically take the form of apoptosis,
autophagic cell death or necrosis [52,53].
3.4. Selected Molecular Mechanisms of Action: Epigenetic
Mechanisms of Carcinogenicity
A growing body of evidence indicates that epigenetic alterations, including
DNA methylation and histone modification, contribute to the toxicity of
heavy metals. For instance, cadmium can affect both gene transcription and
translation through the induction of ROS in mitochondria. This causes a
perturbation of cellular redox homeostasis thereby affecting a large set of
transcription factors characterized by reactive cysteines. The comprehensive
analysis of gene expression of human cell lines exposed to non-toxic doses of
cadmium confirmed the induction of cell protection and damage control
genes, such as metallothionein (MT), antioxidant and heat shock proteins,
and revealed several other alterations in genes involved in signaling and
metabolism (reviewed in [54]). Moreover, by inducing oxidative modifica-
tion of proteins cadmium can also target these proteins to degradation.
The key role of epigenetic events in toxicity is similarly well documented for
arsenic (reviewed in [55]). Inorganic arsenic induces hypermethylation of
DNA gene promoters, as shown for the tumor suppressor gene p53, both in
cells in vitro and in subjects exposed to arsenic-contaminated drinking water.
Chronic exposure to arsenic may also lead to loss of global DNA methylation
due to S-adenosylmethionine (SAM) depletion as well as to alteration of
global histone H3 methylation. The alteration of specific histone methylations
represents both gene silencing and activation marks. Arsenic is a carcinogen
with transplacental activity and several studies report alteration of genetic
programming following prenatal exposure that could impact tumor formation
much later in adulthood (reviewed in [56]). Arsenic exposure in utero exa-
cerbated skin cancer response in adulthood in association with distortion of
tumor stem cell dynamics [57]. Finally, in newborns from mothers exposed to
inorganic arsenic through contaminated water in Thailand, altered transcript
profiles in cord blood were reported including changes of stress-related genes
and breast cancer/estrogen-signature genes [58].
38 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

By modulation of gene expression and signal transduction heavy metals
may affect cell proliferation, differentiation, apoptosis, and other cellular
activities, thus contributing to carcinogenicity.
4. ANALYTICAL TECHNIQUES AND EXPOSURE
ASSESSMENT OF HEAVY METALS
4.1. Analytical Techniques for the Detection of Heavy
Metals and Metalloids in Biological Samples
The occurrence data for heavy metals in food are usually obtained from
routine monitoring programs conducted at the level of a specific country to
check the compliance for which maximum levels are laid down in legislation.
In Europe, the implementation of the Rapid Alert System (RASFF) for
Food and Feed in Europe has provided a helpful tool to perform systematic
monitoring of specific notifications regarding heavy metals that may be
above maximum levels in food and feed.
To obtain reliable occurrence data on heavy metals in food, the avail-
ability of suitable analytical methods for their determination is of utmost
importance. The complexity of food samples, together with the low con-
centrations at which heavy metals occur, requires sensitive, selective, and
reliable analytical techniques, which can also be applied to biological and
environmental samples. Usually, the analytical methods comprise a sample
preparation step involving the digestion (mineralization) or dry ashing of the
sample, followed by the instrumental determination. Atomic absorption
spectrometry (AAS), either flame AAS (F-AAS) or graphite furnace AAS
(GF-AAS), as well as inductively coupled plasma atomic emission spectro-
metry (ICP-AES), inductively coupled plasma-optical emission spectroscopy
(ICP-OES) and inductively coupled plasma mass spectrometry (ICP-MS)
are techniques commonly used for measuring trace metals in food samples,
and vary widely in cost, ease of operation, and analytical performance such
as LODs, linear range, and robustness.
The instrumental techniques applied for the determination of trace con-
centrations of natural uranium include radiometric methods (g-spectrometry,
a-spectrometry, and b-counting) and mass spectrometric (MS) methods
(secondary ion MS, thermal ionization MS, and especially ICP-MS), which
are more sensitive for long-lived radionuclides such as uranium [59,60].
For metals such as arsenic and mercury, speciation is an important
characteristic that provides information on the chemical form present in the
samples, crucial to accurately assess the toxicity. In those cases, additional
steps to separate the different species before detection are needed, such as
39HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

pre-concentration, extraction, and separation. The latter is usually
performed using well established separation techniques such as gas chro-
matography (GC), liquid chromatography (LC), and lately capillary elec-
trophoresis (CE), coupled to selective elemental detection systems. For
arsenic speciation, the most commonly used methods involve LC separation
followed by ICP-MS or AAS [61,62]. In the case of mercury speciation, a
number of analytical methods have been proposed for the determination of
methylmercury, including GC coupled with atomic fluorescence spectro-
metry (GC-AFS) and LC coupled to ICP-MS [63]. Sample preparation still
remains in many cases the bottleneck of the whole analytical procedure. The
selection of the sample preparation methods depends on the matrix and the
analyte. Currently, sample preparation methods tend to move towards more
environmental friendly approaches (less consumption of organic solvents),
to miniaturization, automatization, and ideally to on-line coupling with the
final instrumental determination. This will lead to extracts that are less
manipulated by the analyst, decreasing the probability of experimental
errors. Solid phase extraction (SPE), pressurized liquid extraction (PLE),
microwave assisted extraction (MAE), and solid phase micro-extraction
(SPME), are some of the extraction techniques that fulfil some of the above
mentioned requirements and offer high throughput and the possibility of
on-line coupling with the separation/detection instrumental techniques [64].
The sampling of food for the analysis of metals requires specific precautions
in order to avoid contamination or losses during handling, storage, and
transport to the laboratory. Sampling methods and detailed performance
criteria to be fulfilled by the methods of analysis for cadmium, lead, and
mercury used by the laboratories are laid down in Regulation (EC) No 333/
2007. These performance criteria include recovery ranges, limits of detection
(LOD), limits of quantification (LOQ), and precision requirements. The need
for contamination control together with technological advances will lead to
the development and implementation of effective and efficient analytical
methods, including both sample preparation and final instrumental determi-
nation. The implementation of quality assurance and quality control (QA/
QC) measures are also of utmost importance to ensure reliable occurrence
data on contaminants and decrease the uncertainty of the measurements.
4.2. Data Sources for the Estimation of Human Dietary
Exposure
The estimation of human dietary exposure from food and water corresponds
to the third pillar of risk assessment. This step combines dietary consump-
tion data with occurrence data of heavy metals, i.e., the concentration of a
40 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

heavy metal obtained through analytical methods. Ideally, such concentra-
tions are available for a comprehensive and consistent list of food categories
but in practice, these conditions are rarely met.
The most commonly used information on consumption data is derived
from dietary surveys, usually conducted at the national level on a repre-
sentative sample of individuals. In general, these surveys provide estimates
of consumption over a limited time frame, and not on lifetime consumption.
The various dietary assessment instruments can focus on a ‘short term’
diet, usually covering a period that ranges from one day to a few days, in
the case of one administration versus replicate administrations of 24-hour
dietary recalls, food records or weighed records [65]. Such data should be
harmonized to be used for international risk assessment and the easiest
way for such an objective is grouping the food consumed at national level
into broad categories at regional level. The Concise European Food Con-
sumption Database established by EFSA to support exposure assessments
in the EU [66] is compiling data from European countries based on this
principle. Currently, 20 countries provided national food consumption
data in the adult population and to optimize the degree of comparability
between these dietary estimates, consumption data have been aggregated in
15 broad food groups and 29 subcategories. Other surveys exist based on
food frequency questionnaires, dietary history questionnaires or household
purchases which cover a longer period of time in terms of dietary habits.
They are often defined as providing information on habitual diet [67]
making it difficult to quantify individual consumptions [22]. Ultimately,
regional food consumption surveys performed with similar methodologies
would allow a better picture of the dietary habits all around the world.
In parallel, local food consumption surveys, also performed using inter-
nationally recognized methodologies would aim in describing dietary
patterns of local populations in view of the protection of particular groups
at risk.
For heavy metals, most of the analytical data available for risk assessment
are customarily produced to check for regulatory compliance to specific
norm values. Other data exist which are specifically generated for risk
assessment purpose and are particularly useful for estimating the dietary
exposure to heavy metals. They are generated using the so called ‘‘Total Diet
Study’’ approach [20]. Total diet studies consist in the analysis of the con-
centration of various chemicals in food sampled on the market and prepared
to account for the potential increase or decrease in centration during the
home cooking process. The samples of a considered food category are
pooled in order to be representative of an average contamination and to
increase the cost effectiveness. These data provide risk assessors with a
realistic picture of the distribution and trend for chemicals under
consideration.
41HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

Occurrence data should ideally be consistent in terms of the analytical
methods employed, and provide information on the average and/or the
extreme occurrence of heavy metals in an exhaustive and consistent list of
food categories representative of the diet of the population. In practice, these
conditions are rarely met. Departures from these requirements are likely to
raise concerns on the accuracy of exposure assessment calculations. In
international investigations, careful evaluations on the comparability of
figures produced at the country level need to be performed. In addition,
special efforts are needed to handle non-detect values, i.e., samples for which
the concentration is below the limit of detection/quantification. Data of this
nature are typically left-censored (see Abbreviations and Definitions) [68].
The approach applied can have a great impact on the dietary estimates of the
heavy metal under assessment. Deterministic and probabilistic approaches
have been introduced to deal with the statistical handling of laboratory data
[69]. The comparative performance of these methods varies depending on the
pre-defined scenarios and the variables (sample size, frequency of non-
detects, and departure of empirical values from known statistical distribu-
tions) [70]. In food safety, the most commonly used method is currently the
substitution of results below the LOD/LOQ by half of the value of LOD or
LOQ or to estimate upper (setting all values at the LOD/LOQ at that value)
and lower (setting all values at the LOD/LOQ to zero) boundaries [20].
4.3. Combining Occurrence and Consumption Data in
Humans for Exposure Assessment
Dietary exposure assessment is generally recognized as a tiered approach.
The first steps should be based on conservative and cost-effective methods
and only when necessary, refinements should be performed. Data on food
consumption and chemical occurrence are usually combined using either a
deterministic approach, also called ‘‘point estimate’’, or a probabilistic
approach [65].
The ‘‘point estimate’’ approach is based on the selection of a fixed level in
the distribution of consumption multiplied by a fixed value chosen from the
distribution of concentration. The value of contamination could be the 95th
percentile or the maximum authorized levels in the regulation (food addi-
tives) or an average summary value (mean or median) of the occurrence data
(contaminants such as heavy metals, pesticide residues). Values of the same
nature are used from consumption distributions, so that often combinations
of different average/high values from the consumption and concentration
sides are used to evaluate various risk scenarios. This method does not
42 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

reflect the exposure of the overall population, but it is often considered the
most appropriate for screening purposes [71]. In practice, the fixed levels
utilized to calculate a ‘‘point estimate’’ are generally chosen assuming a
conservative scenario, thus being on the safe side when determining the
absence of safety concern. For example, the combination of highest levels of
residues with highest percentiles of food consumption is usually referred to
as ‘‘worst case scenario’’.
Conversely, probabilistic approaches use the full distributions of occur-
rence and consumption data, thus exploiting the variability in both quan-
tities. These probabilistic methods result in more realistic pictures, often
expressed in terms of a range of possible exposure values, thus incorporating
an estimation of the uncertainty associated with exposure estimates, pro-
vided reliable data is available together with the relevant modelling tools.
A variety of empirical, semi-parametric and parametric models have been
described, depending on whether the actual data set is used (non-parametric
approach), or parameters of a theoretical statistical distribution (log-
normal, Weibull, exponential) are estimated before data use (parametric
approach).
When assessing the potential health impact of the consumption of food
containing heavy metals, two main aspects have to be taken into account:
The external dose which can be expressed as the amount of chemical ingested
and the internal dose corresponding to the TK of the compound. Applying
key parameters such as the biological half-life, bioavailability, clearance, and
tissue concentrations to the ingested amounts of a heavy metal, a PB-TK or
a population-based TK model can reduce the uncertainty in the exposure
estimates since the variability in internal dose and its time-dependency are
taken into account (see Section 3) [72].
5. APPLICATIONS TO THE HUMAN RISK
ASSESSMENT OF HEAVY METALS AND
METALLOIDS
This section aims to summarize the hazard identification and characteriza-
tion, exposure assessment and risk characterization steps for cadmium, lead,
methylmercury, uranium, and the metalloid arsenic. For readability and
conciseness, each step considers only the most recent risk assessments per-
formed by international agencies such as the JECFA (FAO/WHO), EPA,
ASDTR, the European Commission’s Scientific Committee for Food (SCF),
and the panel on contaminants in the food chain of the European Food
Safety Authority.
43HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

5.1. Hazard Identification and Characterization
5.1.1. Cadmium
Historically, the most relevant and sensitive endpoint for cadmium toxicity
is an increased risk for potential renal damage and biomarkers of the renal
function from human studies that are excreted in the urine, i.e., s-2
microglobulin, have been used to set its PTWI. The JECFA evaluated
cadmium in 1988 and set a PTWI of 7 mg/kg b.w. per day using a 10%
prevalence rate of s-2 microglobulinemia in humans, assuming an
absorption rate of 5%, a daily excretion of 0.005% of the body load con-
centration (reflecting its long half-life) corresponding to 50 mg/g renal cortex
over a 50-year period [73]. This value was confirmed by the SCF in 1995 and
the following JECFA assessments [74]. In 2008, the ATSDR established a
minimal risk level for chronic oral exposure of 0.1 mg/kg/day based on
multiple approach namely NOAEL/LOAEL values and BMD modelling for
increased prevalence of s-2 microglobulinemia [75]. For the recent EFSA
assessment, the CONTAM panel developed a PB-TK model from human
PB-TK data together with a human BMD/BMDL derived from a meta-
analysis of published studies relating urinary cadmium and s-2 micro-
globulin (TD). The PTWI of 2.5 mg/kg b.w. per week for cadmium was
derived from the human BMDL, a CSAF for human variability in TD and a
back-translation using the human PB-TK model [33,35].
5.1.2. Lead
Historically, international agencies have used adverse neurodevelopmental
effects of lead in children using intelligence quotients (IQ) as the critical
endpoint to derive a PTWI. In 1992, the SCF endorsed the JECFA PTWI
derived in 1986 of 25 mg/kg b.w. per week which was based on an analysis
relating lead blood concentrations and children’s IQ scores [76]. Recent
studies have shown that children with lifetime average lead concentrations
between 50 and 99 mg/L scored 4.9 points lower on full-scale IQ tests com-
pared with children who had lifetime average blood lead concentrations
o50 mg/L [77]. In adults, lead exposure has been shown to be linked to
neuro-motor disturbances [78], elevated blood pressure [79], and chronic
renal disease (decrease in glomerular filtration rate) [80]. The most recent
assessment by EFSA was based on a dose-response modelling of the meta-
analysis relating lead blood concentrations and its effects on children’s full
scale IQ) by Lanphear et al. [81]. A BMDL01 (for a decrease in IQ of 1
point) of 12 mg B-Pb/L was derived as a reference point concentration when
assessing the risk of intellectual deficits in children. In adults, EFSA also
identified a BMD-01 (for the mean annual increase of SBP by 1%) for
44 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

systolic blood pressure of 36 mg/L and a BMDL10 for chronic kidney disease
of 15 mg/L [82].
5.1.3. Methylmercury
Historically, human developmental neurotoxicity has provided the basis for
setting the health-based guidance values for methylmercury by different
regulatory agencies from 1950s and 1970s. The critical data sets relate to
poisoning episodes in Japan and Iraq, or to more recent large scale epide-
miological studies relating childhood development and neurotoxicity in
relation to in utero exposure (reviewed in [83]). In 1972, the WHO estab-
lished a TWI of 3.3 mg methylmercury/kg b.w. based on the data from Japan
[84] which was then lowered to a PTWI of 1.6 mg/kg b.w. from the growing
epidemiological evidence of neurodevelopmental risks to fetuses and chil-
dren from longitudinal studies in the Faroe and Seychelle islands. The latter
studies used methylmercury in maternal hair as the critical biomarker dose.
Hair concentrations of 14 mg/kg were first related to a maternal blood
concentration of 0.056 mg/L and to a daily intake of methylmercury of
1.5 mg/kg b.w that would be expected to have no appreciable adverse effects
on children. A total UF of 6.4 was applied to give a PTWI of 1.6 mg/kg b.w.
per week.
This PTWI of 1.6 mg/kg b.w. per week was also considered by EFSA in its
2004 risk assessment of methylmercury [85]. In 1995, the US-EPA set a RfD
of 0.1 mg methylmercury/kg b.w. per day based on a study in Iraqi children
who were exposed to methylmercury in utero. In a later evaluation [86], the
BMDL
05
from the Faroes study was used to set a maternal daily intake of
about 1 mg/kg b.w. per day and a composite UF of 10 (intra-human varia-
bility and data gaps) to derive an identical RfD of 0.1 mg/kg b.w. per day.
5.1.4. Uranium
Nephrotoxicity is the most sensitive endpoint for uranium chemical toxicity
both in experimental animals and humans. In 1989, the US-EPA established
an RfD of 3 mg/kg b.w. per day for uranium (soluble salts) based on a 30-day
oral study in rabbits using a LOAEL of 2.8 mg uranium/kg b.w. per day for
initial body weight loss and moderate nephrotoxicity [87]. A LOAEL of
0.06 mg uranium/kg b.w. per day for nephrotoxicity based on a 91-day oral
study in male rats was taken as the key study by the WHO and a UF of 100
was applied to derive a TDI of 0.6 mg/kg b.w. per day [88,89]. The ATSDR
set a minimal risk level of 2 mg/kg b.w. per day for intermediate duration
(15–364 days) of uranium ingestion by applying an UF of 30 (3 for using the
LOAEL and 10 for human variability) to a LOAEL from a 91-day oral
study in rabbits of 0.05 mg uranium/kg b.w. per day for nephrotoxicity [90].
45HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

Most recently, EFSA endorsed the 1998 WHO TDI for soluble uranium of
0.6 mg/kg b.w. per day [34] after a thorough examination of the recent tox-
icokinetic and toxicological database which did not provide evidence for a
new TDI.
5.1.5. Arsenic
Toxicity of arsenic is complex because of the presence of inorganic and
organic species and the large number of toxic endpoints. Inorganic arsenic is
recognized to be much more toxic than its organic forms. In 1989, a JECFA
evaluation [91,92] confirmed their provisional maximum TDI (PMTDI)
derived in 1983 for inorganic arsenic of 2 mg/kg b.w. and converted to a
PTWI of 15 mg/kg b.w. The PTWI was based on human dose-response data
from Nova Scotians relating skin lesions and arsenic concentrations in
contaminated well water.
The US-EPA derived a RfD of 0.3 mg/kg b.w. per day based on a human
NOAEL of 0.8 mg/kg b.w. per day relating skin lesions in Taiwan and
inorganic arsenic concentrations by applying a UF of 3 to account for
sensitive subjects and the lack of data on reproductive toxicity [93,94]. In
2005, the US-EPA used lung and bladder cancer as endpoints with ED01
values for inorganic arsenic in drinking water estimated at 79–96 mg/L for
lung cancer risk, and at 304–474 mg/L for bladder cancer risk [95]. The
National Research Council (NRC) [96,97] has estimated ED01 (i.e., 1%
effective dose, which according to the NRC is the concentration of arsenic in
drinking water that is associated with a 1% increase in the excess risk) for
various studies using different statistical models. Under different modelling
approaches, the ED01 values for lung cancer estimated for the southwestern
Taiwanese population ranged from 33 to 94 mg/L and for the Chilean
population from 5 to 27 mg/L. For bladder cancer, the ED01 values for the
southwestern Taiwanese population ranged from 102 to 443 mg/L based on a
1% increase relative to the background cancer mortality in the US [97],
whilst the previous estimations, in which the reference was the background
cancer mortality in Taiwan, were 404 to 450 mg/L [96].
Studies presented in [98] established a chronic oral minimal risk to
humans (MRL) of 0.3 mg/kg b.w. per day, applying a similar approach to
that of the US-EPA RfD, based on the NOAEL for skin lesions of
0.8 mg/kg b.w. per day. The recent EFSA risk assessment has used the MOE
approach using dose-response data from key epidemiological studies
(skin, lung, and bladder cancers, skin lesions) and selected a benchmark
response of 1% extra risk together with a range of benchmark dose
lower confidence limit (BMDL01) values between 0.3 and 8 mg/kg b.w. per
day [12].
46 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

5.2. Exposure Assessment of Heavy Metals and Metalloids
5.2.1. Cadmium
Only deterministic exposure assessments were performed for cadmium at
regional or international level. In 2006, the JECFA [99] used the GEMS/
Food regional diets and the regional average concentrations of cadmium to
conclude in a mean dietary exposure ranging from 2.8 to 4.2 mg/kg b.w./week
with a value of 3.8 mg/kg b.w./week for the European region. The earlier
European Community SCOOP study [100] showed a mean dietary exposure
ranging from 0.7 to 2.9 mg/kg b.w./day. More recently the EFSA assessed
cadmium dietary exposure based on the occurrence data and the con-
sumption data as reported in the EFSA’s Concise European Food Con-
sumption Database. The mean dietary exposure across European countries
was estimated to be 2.3 mg/kg b.w. per week (range from 1.9 to 3.0 mg/kg b.w.
per week). This difference between JECFA and EFSA might indicate that a
refined assessment based on more disaggregated and representative samples
can result in lower estimates of cadmium exposure from food. EFSA also
estimated the high exposure to cadmium which resulted in a value of 3.0 mg/
kg b.w. per week (range from 2.5 to 3.9 mg/kg b.w. per week). Due to their
high consumption of cereals, nuts, oilseeds and pulses, vegetarians have a
higher dietary exposure of up to 5.4 mg/kg b.w. per week. Regular consumers
of bivalve molluscs and wild mushrooms were also found to have higher
dietary exposures of 4.6 and 4.3 mg/kg b.w. per week, respectively.
In the US, based on the data from a Total Diet Study carried out by the
US Federal Drug Administration (FDA) in 2003 [101], the US FDA con-
cluded in a dietary of 1.5 mg/kg b.w./week. This difference emphasizes the
interest of TDS for estimating the mean dietary exposure based on more
accurate occurrence data.
5.2.2. Lead
The situation for lead exposure is complicated by the fact that key measures
aimed at reducing the release of lead from anthropogenic sources, including
the phasing out of leaded petrol, have led to a major reduction in lead levels
in the environment over the past 50 years. Consequently, blood lead levels in
the general population have decreased from 150–330 mg/L in the 1960’s to
around 15 mg/L [102]. The JECFA evaluated lead at its 53th meeting (WHO,
1999, http://www.inchem.org/documents/jecfa/jeceval/jec_1260.htm). The
exposure assessment focused on the contribution from the diet based on the
WHO GEMS Food regional diets and on levels of occurrence for lead in
food. The JECFA proposed a simple Monte-Carlo simulation to estimate
the dietary exposure in various regions related to frequently consumed
47HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

foods. This simulation was based on estimates of mean intakes in the United
States converted to distributions by assuming that they follow a log normal
distribution with a geometric mean equal to 0.76 times the arithmetic mean
and a geometric standard deviation of 0.76. It resulted in an overall dietary
exposure ranging from about 7 to 30 mg/day/person (about 1 to 4 mg/kg b.w.
per week assuming 60 kg b.w.). The Committee noted that since the model
was based on data for one country, results do not reflect any geographic
difference in lead concentrations. Moreover, summing distributions does not
account for correlations in the consumption of particular foods, in that high
consumption of one food may tend to be accompanied by high consumption
of another. Such correlations would require access to raw data on con-
sumption, which are not usually published.
EFSA performed a deterministic assessment of lead dietary exposure for
adults. In the case of average adult consumers, lead dietary exposure ranges
from 0.36 to 1.24 mg/kg b.w. per day, with major contribution from the
consumption of cereal products, potatoes, leafy vegetables, and tap water.
For children aged 1–3 years mean lead dietary exposure range from 1.10 to
3.10 mg/kg b.w. per day. Compared to dietary exposure, non-dietary expo-
sure to lead is likely to be of minor importance for the general population in
the EU. However, house dust, soil and lead in paints on toys can be an
important source of exposure to lead for children due to their tendency to
ingest soil and mouth toys [82].
5.2.3. Methylmercury
The JECFA assessed the dietary exposure to methylmercury by combining the
mean level of occurrence with the mean consumption for fish and other
seafood from the GEMS Food regional diets. This deterministic assessment
resulted in exposure values ranging from 0.3 to 1.5 mg/kg b.w./week [74,103].
Similarly for the EU, the EFSA reported the mean weekly estimated dietary
exposure would be between 0.1 to 1.0 mg/kg b.w. of mercury from fish and
seafood products. Consequently, the exposure of a fraction of the population
is likely to be above the health based guidance value of 1.6 mg/kg b.w./week,
and in its opinion, the EFSA CONTAM panel performed a probabilistic
analysis of the likelihood of exceeding the PTWIs using the French con-
tamination data as reported to SCOOP in combination with the distribution
of fish and seafood product consumption in France. The probability for a
population to reach an exposure over the available health based guidance
value was calculated to be 1.2% for adults and 11.3% for children [85].
5.2.4. Uranium
EFSA recently estimated the total uranium exposure by multiplying
occurrence values (g/L for water and g/kg for foods) by consumption values
48 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

(g/day). Values of individual body weight of participants in the Concise
European Food Consumption Database were used to express uranium
exposure in g/kg b.w. per day. In order to provide summary figures of
uranium exposure in Europe, the median of 19 country-specific uranium
exposure values calculated for all water-based products and food were
reported according to four different exposure scenarios. These scenarios
were determined using combinations of average and 95th percentile values of
occurrence and consumption figures. Notably, scenario 1 used mean values
for dietary consumption in conjunction with water and food mean occur-
rence values, scenario 2 used 95th percentile consumption and mean
occurrence values, scenario 3 used mean consumption and 95th percentile
occurrence values, and scenario 4 used 95th percentile consumption and
occurrence values. The median overall lower- and upper-bound dietary
exposure to uranium across European countries is between 0.050 and
0.085 g/kg b.w. per day. This figure comprises around 0.04 g/kg b.w. per day
from water (tap and bottled) and water-based products (tea, coffee, beer,
and soft drinks). For high consumers the median country-specific overall
dietary exposure to uranium was estimated to be between 0.09 and 0.14 g/
kg b.w. per day, 0.082 g/kg b.w. per day coming from water and water-based
products [34].
Two specific subgroups of the population were looked at in more detail.
As a very conservative scenario, it can be assumed that the population of
some local communities with high uranium concentrations in their water
supply can be exposed at the 95th percentile concentration level for life-time.
At the same time there might be high consumers of water among these
subpopulations at the 95th percentile consumption level. In such a situation,
water could contribute 0.36 g/kg b.w. per day as a median across the coun-
tries studied, and a country maximum of 0.51 mg/kg b.w. per day. Con-
tribution from food is not considered likely at the 95th percentile
concentration level of uranium at the same time, but more likely at the mean
concentration level of 0.040 g/kg b.w. per day and possibly 0.066 g/kg b.w.
per day in a high consumption scenario. Thus, also in such a situation the
TDI would not be exceeded even if the estimated exposure would be in that
case more than 10 times higher than the median value. This example shows
that in certain situations a worst case scenario could be useful in reinsuring
the risk managers about the absence of safety concern [34].
5.2.5. Arsenic
The European Commission Scientific Cooperation project calculated a mean
dietary exposure to total arsenic in the adult population in three European
countries with complete dietary studies of between 37 and 66 mg/day with an
estimated seafood contribution in excess of 50% [100]. In the United States,
49HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

dietary exposure ranged from 2 mg/day in infants to 92 mg/day in 60–65-year-
old men [104]. From a toxicological point of view the amount of inorganic
arsenic is considered the most important. Tao and Bolger [104] assumed that
10% of the total arsenic in seafood was inorganic and that 100% of the
arsenic in all other foods was inorganic and average daily exposure to inor-
ganic arsenic ranged from 1.3 mg in infants to 12.5 mg in 60–65-year-old men.
The worldwide JECFA assessment estimated total arsenic dietary expo-
sure to range from below 10 mg/day to 200 mg/day and emphasized that these
values are not only reflective of different dietary habits but mirror important
variations in assumptions used to calculate them [73]. The recent EFSA
assessment, estimated dietary exposure to inorganic arsenic using a deter-
ministic approach and several assumptions. The amount of inorganic arsenic
was assumed to be 0.03 mg/kg in fish, 0.1 mg/kg in other seafood, and to
represent 70% of the total arsenic measured in other food categories. For
food consumption, the EFSA concise database was used and several sce-
narios were elaborated and resulted in a mean exposure ranging from 0.13 to
0.56 mg/kg b.w./day and in a dietary exposure at the 95th percentile ranging
from 0.37 to 1.22 mg/kg b.w./day. Consumer groups with higher inorganic
arsenic exposure levels such as high consumers of algae-based products
could be exposed up to 4.03 mg/kg b.w. per day. Infants fed only on cow’s
milk formula reconstituted with water containing arsenic at the average
European concentration level have intakes of inorganic arsenic that are
about 3-fold higher than those of breast-fed infants.
5.3. Risk Characterization of Heavy Metals and Metalloids
5.3.1. Cadmium
Risk characterization was performed by JECFA which concluded that an
excess prevalence of renal tubular dysfunction would not be expected to
occur if urinary cadmium concentration remains o2.5 mg/g creatinine since
the PTWI of 7 mg/kg b.w. per week would not be exceeded. This was based
on an estimation of cadmium intake of ranging from 2.8 to 4.2 mg/kg b.w.
per week, which equates to 40–60% of the PTWI of 7 mg/kg b.w. per week
[74,105]. Back in 1981, the US Environmental Agency published an assess-
ment on health effects of cadmium. The dietary exposure for most Amer-
icans was estimated to be 10–50 mg/day and the threshold level was set at
200 mg cadmium/g wet human renal cortex, it was estimated that an intake of
200 mg/day would result in reaching the threshold after 50 years exposure.
Later, the EPA [106] set an RfD of 0.5 mg Cd/kg b.w. per day for water and a
RfD for cadmium in food of 1 mg/kg b.w. per day. The assessment of the
Joint Research Centre [107] identified the LOAEL for Cd-U to be Z2mg/g
50 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

creatinine based on low molecular weight proteinuria and bone changes.
They found the margins of safety (MOS) between the LOAEL and the
predicted exposure to be Z3 for more than 50% of the population and o1.0
for 5% of the population (both smokers and non-smokers).
A MOS of 3 or more is considered as sufficiently protective for the general
population. In the most recent risk assessment of cadmium, the EFSA
opinion, the mean exposure for adults across Europe was found to be close
to, or slightly exceeding the new TWI of 2.5 mg/kg b.w. (as established in the
scientific opinion) and vegetarians, children, smokers, and people living in
highly contaminated areas may exceed the TWI by about twofold [33].
Although the risk for adverse effects on kidney function at an individual
level is very low, it was recommended that the current exposure to cadmium
at the population level should be reduced [33].
5.3.2. Lead
EFSA recently re-investigated the PTWI of 25 mg/kg b.w. set by the SCF and
JECFA. As a basis for its risk assessment procedure, EFSA performed a
BMD analysis on the three key toxicological endpoints for lead, namely
developmental neurotoxicity in young children, cardiovascular toxicity and
nephrotoxicity in adults. The following benchmark dose (lower limits)
(BMDLs) were derived from blood lead levels (B-Pb) for developmental
neurotoxicity: BMDL
01
,12mg/L; cardiovascular toxicity: BMDL
01
,34mg/L;
nephrotoxicity: BMDL
10
,15mg/L (B-Pb). The dietary lead intakes predicted
from toxicokinetic models to yield the BMDL
01
for developmental neuro-
toxicity, cardiovascular toxicity and BMDL
10
for nephrotoxicity were 0.50,
1.50 and 0.63 mg/kg b.w./day, respectively. Based on these results, EFSA
concluded that the PTWI of 25 mg/kg b.w. set by the SCF is no longer
appropriate [82].
5.3.3. Mercury
The EFSA’s CONTAM Panel used the intake estimates from the SCOOP
data and the JECFA PTWI of 1.6 mg/kg b.w. per week [85]. The panel
concluded that mercury intake in Europe was very variable between coun-
tries depending on fish consumption but in most cases mean intakes were
below the PTWI. There were indications, however, that proportions of
young children might exceed the PTWI and that adults with high fish con-
sumption would have intakes above the PTWI. Data quality at the Eur-
opean level was not sufficient to assess the size of these population groups
and it was recommended to perform specific intake studies on methylmer-
cury, especially for women of childbearing age and children and that such
exposure should be minimized [85].
51HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

On request of the Codex Commission on Food Additives and Con-
taminants, the JECFA examined the dietary impact of the current guideline
levels of methylmercury in fish (1.0 and 0.5 mg/kg in predatory and non-
predatory fish, respectively on exposure and risk) [103]. The Committee
compared the current situation with a scenario where no guideline levels
were in effect or were enforced. More complete exposure data in Europe
including internal dose levels would allow direct comparison of exposure
with the dose-effect relationships, which are the basis for the hazard char-
acterization [108]. JECFA concluded that, for the general population, the
setting of maximum levels for methylmercury in fish was not an effective
mean to reduce exposure. Because most marketed seafood contains mercury
concentrations below the maximum levels, excluding the food items con-
taining this contaminant at the high end of a log-normal distribution of
concentrations would not significantly diminish average exposure. The
impact of reducing exposure to predatory fish would be greater for women
of childbearing age because predatory fish make up a larger proportion of
their diets than in the case of children, and is a larger vector of exposure for
those that would exceed the PTWI (23% for children versus 70% for
women). Hence, JEFCA concluded that advice targeted at population
subgroups that might be at methylmercury exposure greater than the PTWI
and potentially at risk could effectively lower such exposure. Another
recommendation was to weight the risks and benefits in any advice aimed at
different subpopulations.
5.3.4. Uranium
In 2009, EFSA evaluated whether dietary exposure to uranium in foodstuffs
and water (tap and bottled) and water-based drinks would pose a health risk
to consumers in Europe. For most of the population, including the worst case
scenario (i.e., high consumption of highly contaminated food and water), the
estimated exposure to uranium was below the TDI, and considered not to
pose any significant health risk. Nevertheless, for infants fed with infant
formula reconstituted with water containing uranium, the exposure (expres-
sed on a body weight basis) was estimated to exceed that of adults by up to
threefold. Such exposure was recommended to be avoided [34].
5.3.5. Arsenic
The EFSA estimates of human dietary exposure to inorganic arsenic
(see Section 5.2.5) in Europe were within the range of the BMDL
01
values identified with little or no MOE and thus concluding that the pos-
sibility of a risk to high consumers cannot be excluded. Consumer groups
52 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

with the highest exposure levels included high consumers of rice,
algae-based products and children whereas breast-fed and formula-fed
infants below 6 months of age had the lowest estimated dietary exposure.
Organic sources of arsenic such as arsenobetaine from fish and most
seafood, was widely assumed to be of no toxicological concern whereas
arsenosugars and arsenolipids could not be considered because of lack of
toxicological data. Overall, EFSA recommended a reduction of such dietary
exposure to inorganic arsenic and the need to produce speciation data
for different food commodities to refine risk assessment, support dietary
exposure assessment and dose-response data for the possible health
effects [12].
6. CONCLUSIONS AND FUTURE PERSPECTIVES
This chapter has highlighted the principles and applications of chemical risk
assessment in humans for heavy metals and metalloids. Further research is
needed regarding the complex and multiple metabolic routes and accumu-
lation organs and the multiple long-term health effects of these metals. More
complete exposure assessments allow to describe the dietary exposure as a
dynamic process determined by the accumulation phenomenon due to suc-
cessive dietary intakes and by the toxicokinetics ruling the elimination
process in between intakes. This has been expressed in a recent study on
methylmercury [108].
Toxicological as well as modeling tools are also of critical interest to
estimate the inter-individual variability in susceptibility to toxicity. This
includes further research on genetic polymorphism, TK and TD modelling,
the use of OMICs (genomics, proteomics, metabolomics) to depict mole-
cular mode of actions and develop biomarkers, as well as statistical meth-
odologies to model such complex dynamic systems (e.g., the use of Bayesian
methods, non-linear mixed effects models, etc.). A relevant example for such
models is the EFSA risk assessment for cadmium for which a human BMD/
BMDL was derived from a meta-analysis of published studies relating
urinary cadmium and biomarkers of renal effects (s-2 microglobulin)
without the need to extrapolate from animals to humans. The PTWI was
then derived using a PB-TK model and a chemical specific adjustment
factor for cadmium variability in TD without the need of the 100-fold
uncertainty factor [33,35,109]. Depending on data availability, these
approaches will prove useful to risk assessors to provide more transparent
science-based risk assessment that integrate quantitative descriptors
regarding variability and uncertainty in TK and TD of single toxicants and
chemical mixtures [3,69].
53HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

ACKNOWLEDGMENTS
The authors would like to thank the members of the working groups on
mercury (EFSA, 2004), cadmium (EFSA, 2009), uranium (EFSA, 2009),
arsenic (EFSA, 2009), and lead (EFSA, 2010). The views presented in this
review are those of the authors’ only; they do not reflect the views of the
European Food Safety Authority, the Istituto Superiore de Sanita, the
Chemisches Landes- und Staatliches Veterina
¨runtersuchungsamt, the
Imperial College London, or the World Health Organisation.
ABBREVIATIONS AND DEFINITIONS
Left-censored
data
Conducting dietary exposure assessment consists in
combining deterministically or probabilistically food
consumption figures with occurrence of a given chemical
substance in a number of food categories. The occur-
rence data reported to be below the limit of detection
(LOD) of the analytical method are commonly called
left-censored data. The statistical treatment of those
values is likely to have a critical influence on the results
of the assessment.
AAS atomic absorption spectrometry
AFS atomic fluorescence spectrometry
ATSDR Agency for Toxic Substances and Disease Registry
B-Pb Pb in blood
b.w. body weight
Cd-U Cd in urine
CE capillary electrophoresis
EC European Commission
ED effective dose
EU European Union
F-AAS flame atomic absorption spectrometry
FDA Food and Drug Administration
GC gas chromatography
GEMS Groundwater Environmental Monitoring System
GF-AAS graphite furnace atomic absorption spectrometry
GSH glutathione (reduced)
ICP-AES inductively coupled plasma atomic emission
spectrometry
ICP-MS inductively coupled plasma mass spectrometry
ICP-OES inductively coupled plasma optical emission
spectrometry
54 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

IQ intelligence quotient
LC liquid chromatography
LOD limit of detection
LOQ limit of quantification
MAE microwave-assisted extraction
MOS margin of safety
MRL minimal risk limit
MS mass spectrometry
MT metallothionein
NRC National Research Council
PLE pressurized liquid extraction
PMTDI provisional maximum tolerable daily intake
QA quality assurance
QC quality control
RASFF Rapid Alert System for Food and Feed
RNS reactive nitrogen species/nitrogen-based radicals
ROS reactive oxygen species/oxygen-based radicals
SAM S-adenosylmethionine
SBP systolic blood pressure
SCF Scientific Committee for Food
SCOOP Scientific Cooperation on Questions Related to Food
SPE solid phase extraction
SPME solid phase micro-extraction
TDS Total Diet Study
TWI tolerable weekly intake
US-EPA United States-Environmental Protection Agency
REFERENCES
1. WHO, Project to update the principles and methods for the assessment of che-
micals in food, 2008, http://www.who.int/ipcs/food/principles/en/.
2. EC (European Commission), REGULATION (EC) No 178/2002 OF THE
EUROPEAN PARLIAMENT AND OF THE COUNCIL laying down the
general principles and requirements of food law, establishing the European Food
Safety Authority and laying down procedures in matters of food safety., 2002,
http://eurlex.europa.eu/pri/en/oj/dat/2002/l031/l03120020201en00010024.pdf.
3. J. L. Dorne, Toxicology, 2010, (in press).
4. J. L. C. M. Dorne, L. R. Bordajandi, B. Amzal, P. Ferrari and P. Verger, Trac-
Trends in Analytical Chemistry, 2009, 28, 695.
5. I. Pratt, S. Barlow, J. Kleiner and J. C. Larsen, Mutat. Res., 2009, 678, 113.
6. G. M. Cramer, R. A. Ford and R. L. Hall, Food Cosmetics Toxicol., 1978, 16,
255.
55HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

7. R. Kroes, A. G. Renwick, M. Cheeseman, J. Kleiner, I. Mangelsdorf,
A. Piersma, B. Schilter, J. Schlatter, F. van Schothorst, J. G. Vos and G.
Wurtzen, Food Chem. Toxicol., 2004, 42, 65.
8. A. G. Renwick, Toxicol. Appl. Pharmacol., 2005, 207, 585.
9. EFSA (European Food Safety Authority), The EFSA Journal, 2005, 282,1.
10. S. Barlow, A. G. Renwick, J. Kleiner, J. W. Bridges, L. Busk, E. Dybing,
L. Edler, G. Eisenbrand, J. Fink-Gremmels, A. Knaap, R. Kroes, D. Liem, D. J.
G. Muller, S. Page, V. Rolland, J. Schlatter, A. Tritscher, W. Tueting and
G. Wurtzen, Food Chem. Toxicol., 2006, 44, 1636.
11. J. O’Brien, A. G. Renwick, A. Constable, E. Dybing, D. J. G. Muller,
J. Schlatter, W. Slob, W. Tueting, J. van Benthem, G. M. Williams and A.
Wolfreys, Food Chem. Toxicol., 2006, 44, 1613.
12. EFSA (European Food Safety Authority), The EFSA Journal, 2009, 7,1.
13. M. L. Dourson, S. F. Felter and D. Robinson, Human and Ecological Risk
Assessment, 1997, 3, 579.
14. R. Truhaut, Food Addit. Contam, 1991, 8, 151.
15. J. L. C. M. Dorne and A. G. Renwick, Toxicol. Sci., 2005, 86, 20.
16. EFSA (European Food Safety Authority), The EFSA Journal, 2009, 1150,1.
17. A. G. Renwick, Food Additives and Contaminants, 1993, 10, 275.
18. J. L. C. M. Dorne, K. Walton and A. G. Renwick, Food Chem. Toxicol., 2001,
39, 681.
19. J. L. C. M. Dorne, J. Appl. Toxicol., 2007, 27, 411.
20. WHO, International Programme on Chemical Safety: Chemical-specific Adjust-
ment Factors for Interspecies Differences and Human Variability: Guidance
Document for Use of Data in Dose/concentration Response Assessment, 2005.
http://www.who.int/ipcs/methods/harmonization/areas/uncertainty/en/
index.html
21. J. L. C. M. Dorne, K. Walton and A. G. Renwick, Food Chem. Toxicol., 2005,
43, 203.
22. WHO, Exposure Assessment for Chemicals in Food, Report of the FAO/WHO
Workshop, Annapolis, Maryland, USA, 2005.
23. J. L. C. M. Dorne, K. Walton and A. G. Renwick, Food Chem. Toxicol., 2001,
39, 1153.
24. J. L. C. M. Dorne, K. Walton, W. Slob and A. G. Renwick, Food Chem.
Toxicol., 2002, 40, 1633.
25. A. G. Renwick, J. L. C. M. Dorne and K. Walton, Human Ecol. Risk Assess.,
2001, 7, 165.
26. K. Walton, J. L. Dorne and A. G. Renwick, Food Chem. Toxicol., 2001, 39,
1175.
27. K. Walton, J. L. Dorne and A. G. Renwick, Food Chem. Toxicol., 2001, 39, 667.
28. K. Walton, J. L. C. M. Dorne and A. G. Renwick, Human Ecol. Risk Assess.,
2001, 7, 181.
29. K. Walton, J. L. C. M. Dorne and A. G. Renwick, Food Chem. Toxicol., 2004,
42, 261.
30. C. Svendsen, A. M. Ragas and J. L. C. M. Dorne, in Health Benefits of Organic
Foods: Effects of the Environment, CABI Publisher, 2008, Vol. 119, Chapter 6.
56 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

31. OECD, OECD Guidelines for testing of chemicals: Full list of test guidelines,
2009.
32. CAC, Codex Alimentarius Commission Procedural Manual, 16th edn., Joint
FAO/WHO Food Standards Programme, 2006..
33. EFSA (European Food Safety Authority), The EFSA Journal, 2009, 980,1.
34. EFSA (European Food Safety Authority), The EFSA Journal, 2009, 1018,1.
35. B. Amzal, B. Julin, M. Vahter, A. Wolk, G. Johanson and A. Akesson, Environ.
Health Persp., 2009, 117, 1293.
36. C. M. Liao, H. H. Shen, C. L. Chen, L. I. Hsu, T. L. Lin, S. C. Chen and
C. J. Chen, J. Hazardous Materials, 2009, 165, 652.
37. S. Mann, P. O. Droz and M. Vahter, Toxicol. Appl. Pharmacol., 1996, 140, 471.
38. D. H. Yu, Chemosphere, 1999, 39, 2737.
39. G. L. Diamond, W. C. Thayer and H. Choudhury, J. Toxicol. Environ. Health-
Part A, 2003, 66, 2141.
40. B. Engstrom and G. Nordberg, Toxicology, 1978, 9, 195.
41. J. G. Pounds and R. W. Leggett, Environ. Health Persp., 1998, 106, 1505.
42. J. F. Young, W. D. Wosilait and R. H. Luecke, J. Toxicol. Environ. Health-Part A,
2001, 63, 19.
43. E. J. O’Flaherty, Crit. Rev. Toxicol., 1998, 28, 271.
44. M. Begum and P. K. Sen, Environmetrics, 2007, 18, 515.
45. B. Halliwell and J. M. Gutteridge, in Free Radicals in Biology and Medicine,
Oxford Biosciences, 4th edn., 2007, pp. 1.
46. M. F. Casado, A. L. Cecchini, A. N. C. Simao, R. D. Oliveira and R. Cecehini,
Food Chem. Toxicol., 2007, 45, 945.
47. P. Watcharasit, A. Thiantanawat and J. Satayavivad, J. Appl. Toxicol., 2008,
28, 466.
48. N. Larochette, D. Decaudin, E. Jacotot, C. Brenner, I. Marzo, S. A. Susin,
N. Zamzami, Z. H. Xie, J. Reed and G. Kroemer, Exper. Cell Res., 1999, 249, 413.
49. C. C. Bridges and R. K. Zalups, Toxicol. Appl. Pharmacol., 2005, 204, 274.
50. J. Markovac and G. W. Goldstein, Nature, 1988, 334, 71.
51. P. X. Liu, H. L. Cheng, T. M. Roberts and J. J. Zhao, Nature Reviews Drug
Discovery, 2009, 8, 627.
52. S. Gupta, G. E. N. Kass, E. Szegezdi and B. Joseph, J. Cell. Molec. Med., 2009,
13, 1004.
53. G. E. N. Kass, Chemico-Biological Interactions, 2006, 163, 145.
54. P. Joseph, Toxicol. Appl.Pharmacol., 2009, 238, 272.
55. K. Salnikow and A. Zhitkovich, Chem. Res. Toxicol., 2008, 21, 28.
56. M. Vahter, Ann. Rev. Nutrition, 2009, 29, 381.
57. M. P. Waalkes, J. Liu, D. R. Germolee, C. S. Trempus, R. E. Cannon,
E. J. Tokar, R. W. Tennant, J. M. Ward and B. A. Diwan, Cancer Research,
2008, 68, 8278.
58. R. C. Fry, P. Navasumrit, C. Valiathan, J. P. Svensson, B. J. Hogan, M. Luo,
S. Bhattacharya, K. Kandjanapa, S. Soontararuks, S. Nookabkaew, C. Mahi-
dol, M. Ruchirawat and L. D. Samson, PLOS Genetics, 2007, 3, 2180.
59. C. C. May, P. J. Worsfold and M. J. Keith-Roach, Trends Anal. Chem., 2008,
27, 160.
57HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

60. X. Hou and P. Roos, Anal. Chim. Acta, 2008, 608, 105.
61. S. G. Capar, W. R. Mindak and J. Cheng, Anal. Bioanal. Chem., 2007, 389, 159.
62. P. Kuban, R. Guchardi and P. C. Hauser, Trends Anal. Chem., 2005, 24, 192.
63. D.-Y. Yang, H.-Y. T. Truong, Y. W. Chen and N. Belzile, Anal. Chim. Acta,
2009, 633, 157.
64. K. Ridgway, S. P. D. Lalljie and R. M. Smith, J. Chromatography A, 2007,
1153, 36.
65. R. Kroes, D. Muller, J. Lambe, M. R. Lowik, K. J. van, J. Kleiner, R. Massey,
S. Mayer, I. Urieta, P. Verger and A. Visconti, Food Chem.Toxicol., 2002,
40, 327.
66. EFSA (European Food Safety Authority), Food Consumption Database, 2008.
http://www.efsa.europa.eu/EFSA/ScientificPanels/efsa_locale-1178620753812_
DATEX.htm
67. W. C. Willett, in Nutritional Epidemiology, 2nd edn. Oxford University Press,
Oxford, 1998.
68. R. W. Hornung and L. D. Reed, Appl. Occup. Environ. Hyg., 1990, 5, 46.
69. D. R. Helsel, in Nondetects and Data Analysis: Statistics for Censored Envir-
onmental Data, John Wiley & Sons, Chichester, 2005.
70. P. Hewett and G. H. Ganser, Ann. Occupat. Hyg., 2007, 51, 611.
71. B. Parmar, P. F. Miller and R. Burt, Regulat. Toxicol. Pharmacol., 1997, 26, 44.
72. J. Tressou, A. Crepet, P. Bertail, M. H. Feinberg and J. C. Leblanc, Food Chem.
Toxicol., 2004, 42, 1349.
73. FAO-WHO, Evaluation of certain food additives and contaminants (Thirty-third
report of the Joint FAO/WHO Expert Committee on Food Additives), TRS 776-
JECFA 33, 1988.
74. FAO-WHO, Evaluation of certain food additives (Sixty-first report of the Joint
FAO/WHO Expert Committee on Food Additives), WHO Technical Report
Series, No. 922, 2004.[2003, TRS 922-JECFA 61], 2004.
75. ATSDR (Agency for Toxic Substances and Disease Registry), Toxicological
Profile for Cadmium (Final Report), NTIS Accession No. PB99-166621, 1999..
76. EC (European Commission), Opinion on the potential risk to health presented by
lead in food and drink, 1992. http://ec.europa.eu/food/fs/sc/scf/reports/scf_
reports_32.pdf
77. T. A. Jusko, C. R. Henderson, B. P. Lanphear, D. A. Cory-Slechta,
P. J. Parsons and R. L. Canfield, Environ. Health Persp., 2008, 116, 243.
78. K. Murata, T. Iwata, M. Dakeishi and K. Karita, J. Occupat. Health,2009,51,1.
79. B. S. Glenn, K. Bandeen-Roche, B. K. Lee, V. M. Weaver, A. C. Todd and
B. S. Schwartz, Epidemiology, 2006, 17, 538.
80. J. J. Fadrowski, A. Navas-Acien, M. Tellez-Plaza, E. Guallar, V. M. Weaver
and S. L. Furth, Arch. Intern. Med., 2010, 170, 75.
81. B. P. Lanphear, R. W. Hornung, J. Khoury, K. Yolton, P. Baghurstl,
R. L. Bellinger, K. N. Dietrich, R. Bornschein, T. Greene, S. J. Rothenberg,
H. L. Needleman, L. Schnaas, J. Graziano and R. Roberts, Environ. Health
Persp.s, 2005, 113, 894.
82. EFSA (European Food Safety Authority), The EFSA Journal, 2010, in
preparation.
58 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60

83. A. F. Castoldi, C. Johansson, N. Onishchenko, T. Coccini, E. Roda, M. Vahter,
S. Ceccatelli and L. Manzo, Regulatory Toxicology and Pharmacology, 2008, 51,
201.
84. FAO-WHO, Evaluation of mercury, lead, cadmium and the food additives
amaranth, diethylpyrocarbonate, and octyl gallate. FAO Nutrition Meetings
Report Series, No. 51A, 1972, WHO Food Additives Series, No. 4, 1972; FAS 4/
NMRS 51A-JECFA 16. 249-253, 1972..
85. EFSA (European Food Safety Authority), The EFSA Journal, 2004, 34,1.
86. US-EPA (United States-Environmental Protection Agency), Water Quality
Criterion for the Protection of Human Health: Methylmercury, EPA-823-R-01-
001, 2001.
87. US-EPA (United States-Environmental Protection Agency), Integrated Risk
Information System (IRIS) on Uranium, Soluble Salts. National Center for
Environmental Assessment, 1989. http://www.epa.gov/iris/subst/0421.htm
88. WHO, Guidelines for Drinking-Water Quality – Second Edition – Volume 2 –
Health Criteria and Other Supporting Information – Addendum, 1998.
89. WHO, Guidelines for drinking-water quality – Third Edition, 2004.
90. ATSDR (Agency for Toxic Substances and Disease Registry), Agency for Toxic
Substances and Disease Registry. Toxicological profile for uranium, 1999.
91. FAO-WHO, Evaluation of certain food additives and contaminants, WHO Food
Additive Report Series, No. 18, 1983.
92. FAO-WHO, Evaluation of certain food additives and contaminants, WHO Food
Additive Report Series, No. 24, 1989.
93. US-EPA (United States-Environmental Protection Agency), Integrated Risk
Information System (IRIS) on Arsenic, 1998.
94. US-EPA (United States-Environmental Protection Agency), Integrated Risk
Information System (IRIS) on Arsenic, 2001.
95. US-EPA (United States-Environmental Protection Agency), Issue Paper: Inor-
ganic Arsenic Cancer Slope Factor, Final Draft. July 23, 2005 report of the EPA
Intra-Agency Arsenic Cancer Slope Factor Workgroup, 2005.
96. NRC (National Research Council), Arsenic in drinking water, 1999. http://
www.nap.edu/openbook/0309063337/html/R1.html
97. NRC (National Research Council), Arsenic in drinking water 2001 Update, 2001.
http://www.nap.edu/openbook/0309076293/html/R1.html
98. ATSDR (Agency for Toxic Substances and Disease Registry), Toxicological
profile for arsenic, 2007.
99. FAO-WHO, Evaluation of certain food contaminants: 64th report of the Joint
FAO/WHO Expert Committee on Food Additives, WHO Technical Report
Series 930, 2006.
100. EC (European Commission), SCOOP Report of task 3.2.11: ‘‘Assessment of the
dietary exposure to arsenic, cadmium, lead and mercury of the population of the
EU Member States’’, 2004.
101. S. K. Egan, P. M. Bolger and C. D. Carrington, J. Exposure Science Environ-
mental Epidemiology, 2007, 17, 573.
102. CDC, Fourth National Report on Human Exposure to Environmental Chemicals,
2009.
59HUMAN RISK ASSESSMENT OF HEAVY METALS: PRINCIPLES
Met. Ions Life Sci. 2011,8, 27–60

103. FAO-WHO, Evaluation of certain food additives and contaminants. Sixty-seventh
report of the Joint FAO/WHO Expert Committee on Food Additives, WHO
technical report series 940, 2007.
104. S. S. H. Tao and P. M. Bolger, Food Additives and Contaminants, 1999, 16, 465.
105. FAO-WHO, Evaluation of certain food additives and contaminants (Fifty-fifth
report of the Joint FAO/WHO Expert Committee on Food Additives)., TRS 901-
JECFA 55, 2000.
106. US-EPA (United States-Environmental Protection Agency), Drinking Water
Criteria Document on Cadmium. Final Draft, 1985.
107. EC (European Commission), Cadmium metal and cadmium oxide, Summary
Risk Assessment Report, CAS No: 7440-43-9 and CAS No: 1306-19-0 Joint
Research Centre. EINECS No: 231-152-8 and EINECS No: 215-146-2, EUR
23424 EN, 2008.
108. P. Verger, S. Houdart, S. Marette, J. Roosen and S. Blanchemanche, Regulat.
Toxicol. Pharmacol., 2007, 48, 259.
109. EFSA (European Food Safety Authority), Technical report of EFSA prepared
by Assessment Methodology Unit on Meta-analysis of Dose-Effect Relationship
of Cadmium for Benchmark Dose Evaluation, EFSA Scientific Report, 2009, 254,
1–62.
60 DORNE et al.
Met. Ions Life Sci. 2011,8, 27–60