Review of case studies on the human
and environmental risk assessment
of chemical mixtures
Identification of priorities,
methodologies, data gaps,
future needs
Stephanie K. Bopp, Aude Kienzler, Sander
van der Linden, Lara Lamon, Alicia Paini,
Nikos Parissis, Andrea-Nicole Richarz, Jutta
Triebe, Andrew Worth
2016
EUR 27968 EN
This publication is a Technical report by the Joint Research Centre (JRC), the European Commission’s science and
knowledge service. It aims to provide evidence-based scientific support to the European policy-making process.
The scientific output expressed does not imply a policy position of the European Commission. Neither the
European Commission nor any person acting on behalf of the Commission is responsible for the use which might
be made of this publication.
Contact information
E-mail: [email address]
JRC Science Hub
https://ec.europa.eu/jrc
JRC102111
EUR 27968 EN
PDF
ISBN 978-92-79-59146-4
ISSN 1831-9424
doi:10.2788/272583
Luxembourg: Publications Office of the European Union, 2016
© European Union, 2016
Reproduction is authorised provided the source is acknowledged.
How to cite: : S. K. Bopp, A. Kienzler, S. van der Linden, L. Lamon, A. Paini, N. Parissis, A.-N. Richarz, J. Triebe,
A. Worth (2016); Review of case studies on the human and environmental risk assessment of chemical mixtures;
EUR 27968 EN; doi:10.2788/272583
All images © European Union 2016
1
Table of contents
Abstract ............................................................................................................... 3
1 Introduction .................................................................................................... 4
2 Methodology ................................................................................................... 5
3 Overview of the case study characteristics .......................................................... 7
4 Discussion of case study characteristics and findings .......................................... 14
4.1 Case study summaries in the context of other relevant literature ................... 14
4.1.1
Pesticides......................................................................................... 14
4.1.2
Phthalates ........................................................................................ 15
4.1.3
Polybrominated diphenyl esters .......................................................... 15
4.1.4
Parabens ......................................................................................... 16
4.1.5
Pharmaceuticals ............................................................................... 16
4.1.6
Food contact materials ...................................................................... 17
4.1.7
Dioxin-like compounds ...................................................................... 17
4.1.8
Cross-sectorial and environmental mixtures ......................................... 17
4.1.9
Need to address mixtures across regulatory sectors .............................. 19
4.2 Conclusions related to chemical classes considered ...................................... 19
4.3 Potential for over- and underestimating risks from combined exposure to
chemical mixtures ............................................................................................ 20
4.4 Mixture assessment approaches ................................................................. 21
4.4.1
Prediction models ............................................................................. 21
4.4.2
Grouping of chemicals ....................................................................... 22
4.5 Current limitations in performing mixture risk assessments ........................... 23
5 Conclusions .................................................................................................. 26
References ......................................................................................................... 28
List of abbreviations and definitions ....................................................................... 32
Annex 1 – Overview of individual case studies ........................................................ 35
A.1 Pesticides................................................................................................... 35
A.2 Phthalates .................................................................................................. 51
A.3 PBDEs ....................................................................................................... 55
A.4 Parabens ................................................................................................... 58
A.5 Pharmaceuticals ......................................................................................... 61
A.6 Food Contact Materials ................................................................................ 66
A.7 Dioxin-like compounds (DLCs) including dioxins, furans and PCBs ..................... 68
A.8 Cross-sectorial mixtures from consumer product and environmental exposure .... 70
2
Abstract
Humans and wildlife can be exposed to an infinite number of different combinations of
chemicals in mixtures via food, consumer products and the environment, which might
impact health. The number of chemicals and composition of chemical mixtures one might
be exposed to is often unknown and changing over time. To gain further insight into the
current practices and limitations, published peer reviewed literature was searched for
case studies showing risk assessments for chemical mixtures. The aim was to find
examples of mixture assessments in order to identify chemical mixtures of potential
concern, methodologies used, factors hampering mixture risk assessments, data gaps,
and future perspectives.
Twenty-one case studies were identified, which included human and environmental risk
assessments. Several compound classes and environmental media were covered, i.e.
pesticides,
phthalates,
parabens,
polybrominated
diphenyl
esters
(PBDEs),
pharmaceuticals, food contact materials, dioxin-like compounds, anti-androgenic
chemicals, contaminants in breast milk, mixtures of contaminants in surface water,
ground water and drinking water, and indoor air. However, the selection of chemical
classes is not necessarily representative as many compound groups have not been
covered. The selection of these chemical classes is often based on data availability,
recent concerns about certain chemical classes or legislative requirements. Several of
the case studies revealed a concern due to combined exposure for certain chemical
classes especially when considering specific vulnerable population groups. This is very
relevant information, but needs to be interpreted with caution, considering the related
assumptions, model parameters and related uncertainties. Several parameters that could
lead to an over- or underestimation of risks were identified. However, there is clear
evidence that chemicals need to be further addressed not only in single substance risk
assessment and that mixtures should be considered also across chemical classes and
legislative sectors.
Furthermore, several issues hampering mixture risk assessments were identified. In
order to perform a mixture risk assessment, the composition of the mixture in terms of
chemical components and their concentrations need to be known, and relevant
information on their uptake and toxicity are required. Exposure data are often lacking
and need to be estimated based on production and use/consumption information.
Moreover, relevant toxicity data are not always available. Toxicity data gaps can be filled
e.g. using the Threshold of Toxicological Concern (TTC) approach. Reference values used
in single substance risk assessments can be found for several chemical classes, however,
they are usually derived based on the lowest endpoint. If a refined toxicity assessment
of a mixture for a specific effect/cumulative assessment group is envisaged, this is often
hampered by a lack of specific toxicity and mode of action information.
In all case studies, concentration addition based assessments were made, mainly
applying the Hazard Index. To further characterise the drivers of the mixture risk, the
maximum cumulative ratio was calculated in several case studies. This showed that the
scientific methodologies to address mixtures are mostly agreed and lead to reasonable
predictions. However, especially for some groups of compounds that are designed as
active substances, it cannot be excluded that interactions occur and they should
therefore be addressed on a case-by-case basis.
Most of the mixtures addressed in the identified case studies examined specific chemical
groups. Only few of them looked at mixtures comprising chemicals regulated under
different legislative frameworks. The examples indicated that there is evidence for
combined exposure to chemicals regulated under different legislation as well as evidence
that such chemicals can elicit similar effects or have a similar mode of action. A mixture
risk assessment across regulatory sectors should therefore be further investigated.
3
1 Introduction
Humans and wildlife can be exposed to chemicals via food, consumer products and the
environment, which might impact their health. The number of chemicals and composition
of chemical mixtures one might be exposed to is often unknown and changing over time,
resulting in an infinite number of different combinations. In 2012, the European
Commission published a communication on the combined effect of chemicals (EC, 2012)
in which the Commission proposed actions to ensure that risks associated with chemical
mixtures are properly understood and assessed. The Communication acknowledges that
the current EU legislative framework sets strict limits for the amounts of particular
chemicals allowed in food, water, air and manufactured products, but that the potential
risks of exposure to these chemicals in combination are rarely examined.
Individual regulations already focus on the assessment of mixtures to some extent, e.g.
Regulation (EC) No. 1107/2009, indicates that "interaction between active substance,
safeners, synergists, and co-formulants shall be taken into account", and product
authorisation for plant protection products and biocidal products requires the
assessment of "cumulative and synergistic effects" of the formulations containing more
than one active substance and/or "substance of concern". But even though
methodologies for assessing/estimating the combination effects of chemicals are being
developed and used by scientists and regulators in specific circumstances, so far there
is no systematic, comprehensive and integrated approach across different pieces of
legislation (Kienzler et al., 2014; Kienzler et al., 2016; EC, 2012).
To gain further insight into the current practices and experiences in performing
chemical mixture assessments, the JRC conducted an expert survey to gather
information on the current approaches, scientific methodologies, priorities and gaps
(Bopp et al., 2015). The survey showed that the main sectors where most experience is
already gained in assessing mixtures are in the area of plant protection products and
chemicals under REACH. These were also rated highest regarding the priority for
performing mixture assessments. However, mixture assessments are also performed in
many other areas. Experts had experience with assessing mixtures, both in the context
of human health and environmental risk assessment. Mostly concentration addition
(CA) based methods are used for predicting mixture effects. In contrast, several
experts did not recommend the further use of independent action (IA) based
approaches, mainly because of the higher need for input data for IA and considering
the small differences in predictions by IA compared to CA. Novel tools (such as
in vitro methods, omics, Quantitative Structure Activity Relationships (QSARs), read-across,
Physiologically Based ToxicoKinetic (PBTK) modelling, Threshold of Toxicological
Concern (TTC) approaches, Adverse Outcome Pathways (AOPs), Dynamic Energy
Budget (DEB) models, Integrated Approaches to Testing and Assessment (IATA)), are
being increasingly used in the hazard assessment of mixtures, but mainly in a research
context for the time being, because of a lack of guidance, data and expertise. A general
need for clear and harmonised guidance – also between different legislations - for
combined exposure assessments can be identified from the survey.
As a next step, published peer reviewed literature was searched for case studies
showing risk assessments for chemical mixtures. The aim was to find examples of
mixture assessments in order to identify chemical mixtures of potential concern,
methodologies used, factors hampering mixture risk assessments, data gaps, and
future perspectives. All case studies identified were screened and relevant information
extracted. In the following parts of the report, the approach used for the case studies
review (section 2), the findings from the case studies (section 3), the discussion of the
findings (section 4) and the overall conclusions (section 5) can be found along with the
detailed presentation of the individual case studies in Annex I.
4
2 Methodology
To identify potential case studies, a literature search was performed, taking into account
all kinds of chemicals (i.e. not limited to a specific class of chemicals) and all kinds of
exposure scenarios (to intentional and unintentional mixtures through all routes and
pathways), including human health and environmental risk assessments. Assuming that
more recent case studies would include more up to date knowledge and methodology,
the review was restricted to the years 2014 until May 2016. Several additional case
studies cited from earlier years were included for completion.
The literature was continuously screened using SCOPUS since the beginning of 2014, to
find all articles related to search terms such as "combined exposure", "aggregate
exposure", "cumulative risk assessment", "cumulative toxicity", "combined toxicity",
"mixture toxicity" and "mixture effect". Criteria for the selection of relevant publications
were to include only papers showing a mixture risk assessment or testing real samples
or realistic artificial samples, in order to assess potential risks. Papers that were
exclusively looking at exposure or exclusively reporting testing of combined effects
without making the relation between combined exposure and combined effects were not
considered as a relevant case study in this review. Several of those excluded papers are
however included in the overall discussion. This approach resulted in a more detailed
review of 21 relevant case studies.
In order to extract the information in a standardised way and present the case studies in
an easily comparable format, a reporting template was developed and compiled for each
case study (see Annex 1). For each case study the following was reported:
1. Reference (title, journal, authors and year of publication)
2. Substances assessed
3. Exposure scenario
4. Background and objectives of the case study
5. Problem formulation according to the WHO/IPCS mixture assessment framework
6. Information/data sources used
7. Mixture assessment methodology
8. Overall summary of the outcome and the future perspectives (as provided by the
authors in the publications).
WHO/IPCS has published a framework for the risk assessment of combined exposure to
multiple chemicals, which describes a general approach for risk assessment of combined
exposure to multiple chemicals that can be adapted to the needs of specific users. The
framework provides a tiered approach for both the exposure and the hazard assessment.
Tier 0 is a screening level assessment using simple semi-quantitative estimates of
exposure and default dose addition for all components together. The hazard assessment
is based on available toxicity values, TTC values or the value for the most potent
component is used for all components as conservative approach. Tier 1 assessments use
generic exposure scenarios based on conservative point estimates, while for the hazard
assessment individual points of departure (POD) are used. For Tier 2 the exposure
assessment is further refined using increasingly measured data, while the hazard
assessment is based on refined potency information and groupings based on mode of
action is performed. Tier 3 uses probabilistic exposure estimates and ideally leads to
probabilistic information on the risk. The Tier 3 hazard assessment includes toxicokinetic
aspects in order to consider internal exposure.
An important step in the framework before conceiving of a mixture risk assessment is
the related problem formulation. It was therefore decided to include for each case study
5
the four related questions shown here below from the WHO/IPCS framework (Meek et
al., 2011):
1-
What is the nature of exposure? Are the key components known? Are there data
available on the hazard of the mixture itself (i.e., not extrapolated from the hazards
presented by the components of the mixture)?
2-
Is exposure likely, taking into account the context? i.e. for substances that would
be used only as industrial intermediates and are not expected to be released in the
general environment, the answer would be no.
3-
Is there a likelihood of co-exposure within a relevant timeframe? To answer this
question, the temporal aspects of external exposure, toxicokinetics and toxicodynamics
should be taken into consideration. If, based on consideration of those aspects, the
likelihood of co-exposure is low, a framework analysis of an assessment group is
unnecessary. Biomonitoring data should also be considered as they may indicate co-
occurrence of substances in the human body or elimination products.
4-
What is the rationale for considering compounds in an assessment group? Grouping
of chemicals is commonly based on chemical structures, using predictive tools such as
(Q)SARs, but may also be based on biological information, e.g. on downstream
mechanistic events using
in vitro methods, based on same outcome, same target organ.
6
3 Overview of the case study characteristics
Twenty-one case studies were finally selected within the scope of this review and
examined further. Relevant information was extracted from the identified papers and
was summarised in tables as it was reported by the case study authors. All detailed
tables for each individual case study can be found in Annex 1. Table 1 here below shows
an overview of the case studies indicating types of mixtures and methodologies included
in the assessments.
The case studies were examined regarding the substances that were assessed, the
exposure scenario that was considered, the methodology used for assessing hazard,
exposure and risk of the mixtures, with the aim to identify potential issues regarding
data availability and available methodologies.
Human vs environmental risk assessment (RA): Fourteen of the case studies
performed a human health risk assessment (HRA), 6 an environmental risk assessment
(ERA), and 1 study included both types of assessment for the same samples (exposure
to contaminants in surface waters).
Compound classes: Fourteen of the 21 case studies were looking at specific chemical
groups (i.e. 6 at pesticides, 2 at phthalates, 1 at parabens, 1 at PBDEs, 1 at
pharmaceuticals in general, 1 at antibiotics, 1 at food contact materials including 3
different cases, 1 at dioxin-like compounds). Furthermore, there were 7 case studies
focusing on cross-sectorial mixtures, i.e. including chemicals regulated under different
regulatory silos, such as human exposure to anti-androgenic chemicals, contaminants in
breast milk, 4 studies on mixtures of contaminants in water samples (including surface
water, ground water and drinking water) and 1 on indoor air.
Retrospective vs prospective RA: Most of the case studies use (bio)monitoring data
for the exposure assessment and perform a retrospective RA. However, the same
methodology could also be applied for prospective RA. The example of Junghans et al.
(2006) shows that predicted environmental concentrations (here for pesticides using
FOCUS1 scenarios), can be used as well.
Types of toxicity and exposure data used: Most of the case studies were using
toxicity data extracted from peer reviewed literature or from regulatory authority reports
and databases. In some case studies mixture toxicity was also measured. Toxicity test
endpoint values such as No observed adverse effect levels (NOAELs) or effective
concentrations (EC50) were used as well as reference values including the relevant safety
factors used in single substance risk assessments, like acceptable daily intake (ADI)
values. Exposure data were mostly based on environmental or human biomonitoring
data, but also estimated based on e.g. sales and consumption data or predicted by more
sophisticated modelling, e.g. based on food consumption and food residue data.
Mathematical approaches used: All case studies applied concepts based on the
Concentration Addition (CA) approach (for general information on the commonly used
approaches please refer to e.g. Kienzler et al. (2014). In one case study (Junghans et
al., 2006) the CA approach was compared to Independent Action (IA) based predictions.
Most of the studies used the Hazard Index (HI) approach. The Sum of Toxic Units (Σ TU)
concept was also used frequently mainly in ERA. In one case study (Han and Price,
2013) the Toxic Equivalency (TEQ) concept was applied for dioxin-like compounds. Also
the Maximum Cumulative Ratio (MCR) (Price and Han, 2011), which can help to identify
the drivers of combined risks, was used in six cases.
1 FOCUS: FOrum for Co-ordination of pesticide fate models and their USe.
http://esdac.jrc.ec.europa.eu/projects/focus-dg-sante
7
Table 1: Overview of main characteristics of the reviewed case studies
Study Reference
Chemical
HH
ERA Exposure data
Hazard data
Assessment
RA method
ID
sector
groups
1
Junghans et al Pesticides
X
Modelled exposure
Measured algal
No grouping
CA vs IA, Σ TU
2006
(FOCUS scenario R1)
toxicity of
individual
compounds and
mixture
2
Nowell et al
Pesticides
X
Literature data
Toxicity data
No grouping
Pesticide toxicity
2014
from databases
index (based on CA)
such as USEPA
ecotox DB
3
Kennedy et al
Pesticides
X
Modelled dietary and
From literature
Based on
RPF within group
2015
non-dietary aggregate
or dossiers
similar toxic
exposure (MCRA tool),
effect
using optimistic and
pessimistic scenario
4
Boon et al
Pesticides
X
Modelled dietary
From literature
Based on
RPF within group
2015
(triazoles)
exposure (MCRA tool)
or dossiers
similar toxic
according to EFSA
effect
guidance, optimistic
(hepatotoxicity)
and pessimistic
scenario
5
Evans et al
Pesticides
X
International Estimated ADIs from JMPR No classical
WHO/IPCS
2015
Daily Intakes (IEDI)
groups, but
framework, HI, Tier
grouping
0, Tier 1, Tier 2
according to
health impact
categories in
Tier 2
8
Study Reference
Chemical
HH
ERA Exposure data
Hazard data
Assessment
RA method
ID
sector
groups
6
Ccanccapa et
Pesticides
X
Sediment
Collected from
No, all together
Σ TU
al 2016
concentrations used to
pesticide
based on acute
predict pore water
property
toxicity
concentrations
database
7
Dewalque et al Phthalates
X
Human biomonitoring
EFSA TDI and
Structurally
HI
2014
data; Exposure data
RV from
similar group
from food consumption literature
with similar
etc. to estimate
toxic profile
contribution of dietary
and non-dietary
sources
8
Hartmann et
Phthalates
X
14 metabolites
Reference dose
Dose-addition
Cumulative risk
al 2015
from
detected in urine
for anti-
concept is
assessment
consumer
sample from 10 parent
androgenicity,
considered for
products
phthalates
tolerable daily
anti-androgenic
intake
phthalates
9
Meek et al
PBDEs,
X
Semi-quantitative
No TDI data,
One group of 7
Tier 0 and Tier 1 of
2011 Case
based on
estimates based on
instead LOEL
isomers with
WHO/IPCS
study A
commercial
volume of production,
for most toxic
similar use and
framework; sum of
mixtures
producing and using
congener (Tier
common target
risk quotient like
companies (Tier 0),
0), critical
organ
approach using the
use of Canadian intake
effect level
data outlined
data (Tier 1);
derived from all
comparison to
data on
biomonitoring data
neurobehavioral
effects (Tier 1)
9
Study Reference
Chemical
HH
ERA Exposure data
Hazard data
Assessment
RA method
ID
sector
groups
10
Gosens et al
Parabens
X
Aggregate exposure of
NOAEL data
n.a., only
n.a.
2013
from
children of 0-3 years
aggregate
personal
exposure
4 most frequent
care
considered
parabens, not
products
considering combined
exposure
Tier 1 WC deterministic
approach; Tier 2:
probabilistic based on
product use
11
Backhaus and
Pharma-
X
Literature data on
From literature
No grouping,
comparison of sum of
Karlsson 2014
ceuticals
sewage treatment
and databases
similarly and
PEC/PNEC ratios to
plant effluents
dissimilarly
ΣTU, use of MCR
acting
compounds
considered
together
12
Marx et al
Antibiotics
X
Predicted PEC from
PNECs from
All in one group HI and HIint
2015
hospital and ambulant
literature on
and division of
antibiotic prescription
bacteria, algae,
contribution of
information
daphnia
different
categories of
antibiotics to HI
13
Price et al
food contact
X
Literature data
Use of existing
No grouping
HI and MCR, CEFIC
2014
materials
ADI data or TTC
decision tree
14
Han and Price
Dioxin-like
X
Human biomonitoring
WHO TEF
All in one group TEQ and MCR based
2013
compounds
data for general
values
since known for on TEQs
population and two
common
10
Study Reference
Chemical
HH
ERA Exposure data
Hazard data
Assessment
RA method
ID
sector
groups
(DLCs)
groups of workers with
effects/MoA
relevant occupational
exposure
15
Kortenkamp
15 anti-
X
Data from literature
Reference
All in one group HI
and Faust
androgens
and public databases;
values (NOAEL,
selected for
2010
(phthalates,
using median human
BMD) for anti-
similar action
pesticides,
intake and highly
androgenicity
(anti-
cosmetic
exposed population
androgenicity)
ingredients)
groups
16
Price et al
Organic and
X
X
Use of large dataset of
RV from
No grouping
HI and MCR, CEFIC
2012a
inorganic
surface water (sw) and literature and
decision tree
compounds
effluent monitoring; for databases, if
detected in
sw direct consumption
not available
surface
assumed as WC, for
TTC
water and
effluents a 10x
effluent
dilution; 2 options for
samples
non-detects compared
17
Han and Price
Ground-
X
USGS groundwater
Permitted doses No grouping
HI and MCR
2011
water
monitoring data, direct
from USEPA
(VOCs, PPPs,
groundwater
and ATSDR
metals,
consumption as WC
databases
inorganics)
assumption
18
Meek et al
10
X
Based on surface water Use of TTC
All in one group HI
2011 Case
substances
monitoring data to
(pretending the as WC for Tier
study B
found in
create hypothetical
case that no
0
surface
case study on human
toxicity data
water
exposure via water
are available)
monitoring
consumption
to create
11
Study Reference
Chemical
HH
ERA Exposure data
Hazard data
Assessment
RA method
ID
sector
groups
hypothetical
mixture
(fragrances,
PPPs,
surfactants,
personal
care
products,
solvents,
petro-
chemicals
19
Malaj et al
Organic
X
Monitoring data of the
From toxicity
No grouping
Calculation of a
2014
chemicals
waterbase; measured
databases.
Chemical Risk Index
detected in
concentrations for 223
Using
considering 3
freshwater
chemicals from 4000
experimental,
organism groups
sites from 91 rivers.
predicted or
(fish, daphnia,
Use of Cmean and Cmax.
baseline toxicity
algae), number of
data to derive
sites a chemical
the acute and
occurs and its
chronic toxicity
concentrations
threshold using
related to a river
assessment
basin, derivation of 5
factors.
risk classes, other
approaches used to
compare sites
20
Evans et al
Cross-
X
From literature data of
Reference
No grouping
HI
2016
sectorial
contaminants in human doses collected
human milk
milk
from literature
contami-
and authorities
nants
12
Study Reference
Chemical
HH
ERA Exposure data
Hazard data
Assessment
RA method
ID
sector
groups
21
De Brouwere
VOCs and
X
From indoor air
Reference
No grouping
HI and MCR
et al 2014
NO2 in
monitoring and
values from
indoor air
personal monitoring
different
studies
sources
13
4 Discussion of case study characteristics and findings
4.1 Case study summaries in the context of other relevant
literature
4.1.1 Pesticides
Pesticides were among the most often investigated mixtures, with three studies
regarding human health and three environmental assessments. This might have several
reasons, like e.g. the advanced work of the European Food Safety Authority (EFSA) to
develop a methodology for cumulative risk assessment of pesticides, but also based on
the fact that pesticides are generally data rich chemicals, with extensive data
requirements for their approval.
Junghans et al. (2006) showed that pesticides in mixtures are clearly more toxic to algae
than any individual component, and that CA showed a good predictive quality over the
complete range of effects considered, irrespective of the similarity or dissimilarity of
their mechanisms of actions.
Nowell et al. (2014) developed the Pesticide Toxicity Index (PTI) as a relative ranking
tool regarding pesticide mixture impacts on freshwater organisms. Two approaches were
compared: the median-PTI based on median toxicity values within each taxonomic
group, and the sensitive-PTI based on the most sensitive endpoint within a taxonomic
group. Both were correlated well, with the median-PTI being more robust to outliers. The
median-PTI was proposed for sample and site comparison, while the sensitive-PTI was
proposed more for screening level as a conservative tool. Also the PTI is based on CA
and does not account for potential interactions.
Ccanccapa et al. (2016) looked at pesticides detected in the river Ebro (Spain), and
looked at the potential environmental risks from the 42 detected pesticides and some of
their degradation products to aquatic organisms, exposed via water or sediment pore
water. The sum of toxic units (Σ TU) based on acute EC50 values showed no risk in all
sites. However, when using risk quotients (RQ=PEC/PNEC) based on chronic NOEC
values for the individual pesticides, some exceeded 1 and could be of concern.
Kennedy et al. (2015) assessed aggregate exposure to conazole pesticides, integrating
dietary and non-dietary human exposure. Dietary exposure models used food
consumption and pesticide residue data. Non-dietary exposure was modelled for
operators and bystanders. Optimistic and pessimistic model scenarios were applied. The
outcome was presented as relative contributions from various sources to the overall
risks, e.g. showing in the hypothetical cases investigated that for a UK arable spray
operator inhalation and dermal exposure may be main routes of exposure, while non-
dietary exposure played a minor role for child bystanders. Model specifications had a
large impact on the outcome and should be duly justified.
Boon et al. (2015) used the EFSA guidance and applied an optimistic and a pessimistic
scenario for the cumulative dietary exposure to triazole pesticides using the Monte Carlo
Risk Assessment (MCRA) tool. Regarding acute exposures, in the optimistic model run
none of the scenarios exceeded the ARfD, while in one of the pessimistic scenarios an
exceedance was predicted. In the chronic assessment, again no exceedance of the ADI
were found using the optimistic model run, while 6% and 4.3% of the population had a
simulated chronic exposure exceeding the ADI in the pessimistic model run in Denmark
and Italy, respectively. The model provides also probabilities for exceedances and
related uncertainties. Results for the different countries varied substantially and when
using MRLs for the exposure calculations, animal commodities (milk and meat)
contributed most to the exposures. The authors concluded that the optimistic model runs
are rather easy to perform but probably underestimating exposure, while the pessimistic
model run is laborious and might produce unrealistically high exposures.
14
Evans et al. (2015) applied the WHO/IPCS scheme for the risk assessment of combined
exposures to the assessment of 67 pesticides, for which individual assessments by the
FAO/WHO JMPR were available. A Tier 0 screening level assessment was performed
assuming that no hazard data would be available and TTC values for each compound
were used. HI values obtained based on TTCs (Tier 0) were in the range 37.5-146, and
were up to 16 times greater than HIs based on ADIs. The Tier 1 assessment based on
ADI values and 13 WHO food cluster regions showed an HI>1 in all 13 regions, with one
region exceeding an HI of 10. The HI was never driven by just one compound and 80%
of the HI were contributed by each 9-18 chemicals in the mixtures. The Tier 2 refined
assessment was not possible due to a lack of relevant input data. However, an
assessment was presented applying a subgrouping based on the Pesticide Property
Database (PPDB2) nine human health issues as surrogate information for the mode of
action. That way in some cases the resulting HI would remain <1.
4.1.2 Phthalates
Phthalates were assessed in two case studies regarding combined human health effects.
Many more studies in the literature address combined effects of or combined exposure to
phthalates, but these were the only two recent case studies identified performing a
complete risk assessment.
Dewalque et al. (2014) investigated the daily intake of 5 phthalate diesters in the
Belgian general population, the contribution of dietary exposure to the overall exposure,
and the related risks using general reference values (HITDI) as well as anti-androgenic
specific endpoints (HIRfDAA). Exposure was assessed based on human biomonitoring
(HBM) data. Few participants of the study showed a HQ>1 for individual phthalates, but
6.2% of the adults and 25% of the children showed a HITDI>1. HITDI was 3-4 fold higher
than HIRfDAA, showing the influence of the values used for the hazard assessment. For
DEHP the main exposure pathway was via food while for all other congeners dietary
intake seemed to play a minor role.
Hartmann et al. (2015) assessed the combined risk from exposure to 11 phthalates and
14 metabolites in children in Austria using HBM data as well. Median HIs based on all
acceptable levels of exposure are far below 1. HI>1 was however found using TDIs,
whereas no exceedances were found using anti-androgenic RfD.
Dewalque et al. (2014) suggest that larger biomonitoring studies including pertinent
biomarkers of exposure to other anti-androgenic compounds should be performed. This
wide exposure was also shown earlier by Becker et al. (2009), who detected 12
phthalate metabolites in urine samples from 600 German children from 3 to 14 years
old. The phthalate concentrations decreased with increasing age and children
contamination were 3-5 times higher than in adults. This might be a situation of concern,
as anti-androgenic effects are known for phthalates and exposure could occur at all life
stages. Furthermore, phthalates are not the only anti-androgenic chemicals to which
humans are exposed. This is confirmed by another case study assessing cumulative anti-
androgenic effects of 15 chemicals including phthalates, pesticides, and cosmetics,
(Kortenkamp and Faust, 2010), which concludes that although the cumulative risk
obtained for median exposures can be judged tolerable, it exceeds acceptable levels for
people on the upper end of exposure levels.
4.1.3 Polybrominated diphenyl esters
Polybrominated diphenyl esters (PBDEs) were assessed by Meek et al. (2011) applying
the WHO/IPCS framework Tier 0 and Tier 1. Assessments were based on three
commercial mixtures. The Tier 0 RA used semiquantitative exposure estimates based on
production volume, number of manufacturing plants etc. and the LOEL for the most toxic
2 http://sitem.herts.ac.uk/aeru/ppdb/en/
15
congener in the mixture. Thus the HI exceeded 1 and a refinement was performed. The
resulting Tier 1 assessment used conservative intake estimates based on maximum
levels in air, water, dust, food and human breast milk for six age groups in the Canadian
population, versus critical effect levels based on neurobehavioural effects. The margin of
exposure (MoE) was ca. 300, while MoEs considering HBM data were 10 fold less with
however higher uncertainty. Food was considered the major source of exposure, with
highest percentages of dietary contribution for breast-fed infants.
4.1.4 Parabens
Parabens used in personal care products were investigated by Gosens et al. (2013). This
study considered no real combined exposure to multiple chemicals, but examined the
exposure from various sources to individual substances. Aggregate exposure to 4
parabens for children of 0-3 years of age was assessed in two ways: Tier 1 used a worst-
case deterministic approach based on the maximum amounts of parabens used in
products and default use amounts to predict exposure. Tier 2 used a person-oriented
probabilistic approach. Exposure was then compared to NOAEL values based on male
reproductive effects. In Tier 1, for methyl and ethyl paraben, the MoE was above the
"safe" MoE of 100, while for propyl- and butyl-paraben the MoE was only 8 and 10,
respectively. This might be due to the worst case assumptions, like use of all relevant
personal care products in parallel. The more realistic probabilistic approach in Tier 2
including an uncertainty analysis allows deriving fractions of the population that might
be exposed above a critical level. This refinement results in a much more realistic
assessment but is also very data demanding. Further refinement is difficult as detailed
data on the use of personal care products is scarce, especially for children, and it is
unknown whether extrapolation from adult use by scaling the amount of product used to
body surface is appropriate. Furthermore, those chemicals are also used in other types
of products such as pharmaceuticals and as food additives and those uses have never
been assessed together, as they are regulated under different legal frameworks. More
exposure data via these products would therefore be needed to obtain an even more
accurate aggregate estimated exposure to parabens.
4.1.5 Pharmaceuticals
Pharmaceuticals entering the aquatic environment via waste water treatment plant
(WWTP) effluents were investigated in Backhaus and Karlsson (2014). Exposure to
aquatic organisms was estimated from published monitoring data. For the hazard
assessment only limited chronic effect data were available so that the assessment was
based on acute toxicity data for algae, daphnia and fish, following the REACH guidance
for estimating PNECs with an assessment factor (AF) of 1000. Two different approaches
based on CA were used, i.e. a risk quotient based on measured concentrations (MEC)
and PNEC values (RQ=Σ(MEC/PNEC)i) and a Σ TU (with TU= MEC/EC50) for each trophic
level, multiplied in the end with the appropriate AF. The maximum cumulative ratio
(MCR) was also calculated and showed ratios between 1.2-4.2, meaning that up to 4-5
components were the main drivers of mixture toxicity. The RQ regularly exceeded 1,
indicating a potential risk depending on the dilution in the recipient stream. The top ten
mixture components contributed more than 95% of the overall mixture risk. Algae were
found as most sensitive group, fish as least sensitive. The differences between the two
approaches remained within a factor of 1.3.
Marx et al. (2015) looked specifically at mixtures of antibiotics and potential synergistic
effects to aquatic organisms exposed via WWTP effluents. Exposure for 15 antibiotics
was calculated based on known prescriptions for humans, not including veterinary uses
that were considered of lower relevance in the study area. PNEC values were derived
from the literature as well as information on binary interactions. Common HI as well as
HIint considering interactions were compared. HI over a 7 year period had a mean value
of 0.37, with 20% of all weeks exceeding 0.5, and one week exceeding HI>1. The share
of contribution to the mixture risk changed between different antibiotic classes over
16
time, but not the overall HI. HIint showed a 50% increase in risk in a worst-case
scenario, so that HI>1 would be found in 25 weeks over 7 years. It needs to be kept in
mind that most underlying data on interactions were gained from much higher
concentrations compared to realistic environmental exposures.
4.1.6 Food contact materials
Food contact materials were investigated by Price et al. (2014). The Cefic MIAT decision
tree for the assessment of mixtures (Price et al., 2012b) was applied to investigate three
examples. ADI values (and where not available TTC values) were used for calculating the
HI and MCR. All examples were classified into Group II, indicating low concern. MCR
values were in the range of 1.3-2.4, indicating that few compounds drive the overall risk.
MCR values declined with increasing HI, which means that if the risk increases, fewer
compounds drive this risk.
4.1.7 Dioxin-like compounds
Dioxin-like compounds (DLCs) were assessed by Han and Price (2013) for three datasets
comprising general population exposure (NHANES 3 ), and two worker groups with
occupational exposure (MI, Michigan dataset, and NZ New Zealand dataset). For all
individuals in the datasets the total TEQs were calculated as well as the MCR values. The
top five major contributors to total TEQs in the NHANES dataset were 12378-PeCDD,
123678-HxCDD, PCB 126, TCDD, and 23478-PeCDF. On average they accounted for
76% of the total TEQ. Total TEQs were higher in the MI and NZ datasets than in the
NHANES dataset (58.96 fg/g for MI, 25.5 fg/g for NZ, 19.72 fg/g for NHANES) explained
by the occupational exposure. Part of the difference is however also due to the different
age distributions: i.e. for persons >45 years of age, NHANES total TEQs were lower than
in the MI dataset but higher than for the NZ dataset. Average MCR values (including
2.5th percentile and 97.5th percentile) were: for NHANES 3.5 (2.2/5.7), for MI 3.6
(1.6/5.1), and for NZ 3.2 (1.4/4.6). This indicates that for all 3 groups a small number
of DLCs drives the total TEQ. As also in other studies analysing the MCR, it showed a
decreasing trend with increasing total TEQ values. Overall more highly exposed people
tend to have lower MCR values for the MI and NZ dataset, but not for the NHANES
dataset. The person age and total TEQs are positively correlated. In the NHANES dataset
two groups of age > or < 45 years can be distinguished with persons < 45 years
showing generally lower DLC levels and higher MCR values. For all three groups, the
MCR values were larger than in other investigations of MCR of different mixtures (Han
and Price, 2011; Price et al., 2012a), indicating a greater need for CRA for DLCs. A
single substance RA based only on the largest contributor would underestimate the total
TEQ by a factor of 2-6. In the case of occupational or local sources of exposure, the
impact of performing a CRA compared to single substance RA decreases.
4.1.8 Cross-sectorial and environmental mixtures
Apart from case studies addressing specific chemical groups, several case studies
address mixtures of compounds that are regulated under different pieces of legislation,
i.e. so-called
cross-sectorial mixtures. These are sometimes related to known co-
exposure via certain consumer products or known co-exposures from HBM or
environmental monitoring, but some are also based on known common effects (e.g.
anti-androgenic effects).
Kortenkamp and Faust (2010) looked at anti-androgenic effects of 15 compounds
including phthalates and other chemicals. Risks were compared using median human
intake and highly exposed population groups. Acceptable levels were derived using
NOAELs or BMDs applying appropriate uncertainty factors, or ADIs derived on anti-
androgenic effects were used. For median exposure an HI of 0.38 was calculated,
3 http://www.cdc.gov/nchs/nhanes/
17
whereas for highly exposed populations HI of 2.01 was reached. The authors suggest
that risk reductions can be achieved by limiting exposures to the plasticiser diethyl hexyl
phthalate, the cosmetic ingredients butyl- and propyl paraben, the pesticides vinclozolin,
prochloraz and procymidone and bisphenol A. All those results (considering also
Dewalque et al., 2014; Hartmann et al., 2015) indicate that combined exposures to anti-
androgens might have reached levels of concern. Moreover, it has to be highlighted that
these case studies do not take into account synergistic effects, although synergisms
have been observed with a mixture of anti-androgens with diverse mode of action (MoA)
for particular endpoints (Christiansen et al., 2009). For all these reasons, further work
should be performed to know whether this is a phenomenon of concern that should be
taken into account in anti-androgenic RA.
Price et al. (2012a) applied the Cefic MIAT decision tree for assessing combined risks to
humans and the environment from exposure via surface water and WWTP effluents.
Surface water was assumed to be directly consumed without dilution by humans. A 10-
fold dilution was assumed for effluents. All detected chemicals (up to 49 were detected
from 222 measured substances) were considered to contribute in an additive way
without grouping. The human RA followed the WHO/IPCS Tier 0 using available reference
values (RVs) or TTCs, for environmental risk assessment (ERA) RVs were available to
determine WHO/IPCS Tier 1. Non-detected chemicals were either assumed to be absent
or considered at a concentration of the limit of detection divided by 20.5. For human
health, 98% of the mixtures showed HI<1. For ERA, 68% of the mixtures had HI>1 with
one or more substances individually showing an HQ>1, 19% had HI<1 and about 12%
had an HI>1 which was not due to individual substances exceedances. The tree enables
to identify the MoA of chemicals where refinements would be most useful. Han and Price
(2011) applied the same methodology also to groundwater used for drinking water
supplies, with the aim to mainly look at the usefulness and applicability of the MCR
approach. They found that MCR is negatively correlated with HI, i.e. the risk of mixtures
with higher HI is usually driven by fewer compounds. The way how chemicals below the
limit of detection are considered can greatly influence the related MCR for mixtures with
small HI, but has little impact on MCR for mixtures with HI>1. MCR is positively
correlated with the number of chemicals detected, i.e. the higher the number of
chemicals the higher also the MCR. The average MCR in all samples was 2.2-3.1, so
mixtures were mainly dominated by few chemicals.
Boobis et al. in Meek et al. (2011) took a similar approach assessing a theoretical
mixture of potentially detectable surface water contaminants using the WHO/IPCS Tier 0.
For the cross-sectorial mixture of 10 substances, using exposure monitoring data and
TTC values, the resulting HI was 0.2. In this case a screening assessment would have
been sufficient and no further refinement would have been triggered.
Malaj et al. (2014) investigated the risk for freshwater organisms from organic chemicals
on a continental scale, using median and maximum annual exposure concentrations from
4000 European monitoring sites covering 91 European rivers. A chemical risk index for
each organism group per river was calculated. The mean concentrations were compared
to chronic toxicity thresholds, while the maximum annual concentrations were compared
to the acute toxicity threshold. For 14% of sites the organic chemicals were likely to
exert acute effects and for 42% chronic effects on sensitive fish, invertebrate or algae
species. Major contributors to the risk were pesticides, tributyltin, Polycyclic Aromatic
Hydrocarbon (PAHs), brominated flame retardants. Pesticides were responsible for 81,
87, and 96% of observed exceedances of the acute risk threshold for fish, invertebrate
and algae, respectively. The risk increased with the number of chemicals analysed per
site.
Evans et al. (2016) include a case study on breast-fed children exposed via human milk.
It was assumed that all detected chemicals would act in an additive way, but also a
subdivision in different chemical categories was performed. HI>1 was identified for
several chemical groups, i.e. organochlor pesticides and PCBs. The overall HI for the
18
whole mixture was 66, indicating a potential risk. Mapping the chemicals to the different
regulatory framework shows that some of them fall simultaneously under different
legislation and all together several regulatory silos become relevant. This underlines that
mixtures need to be addressed across regulatory silos.
In the same context of exposure to unintentional mixture, exposure via indoor air
becomes relevant. People in modern society spend approximately 90% of their time
indoors, in which 2/3 would be spent at home (Le Cann et al., 2011). De Brouwere et al.
(2014) applied the MCR tool to evaluate mixtures in residential indoor air. Several data
sets across Europe were compared. The average MCR value was 1.8, with a range from
1 to 5.8. MCR was found to be small compared to the number of chemicals in the
mixtures, indicating that generally the overall effect was driven by only a few chemicals.
The MCR was significantly declining with increasing HI. The large majority of samples
from the Flemish school survey were categorised in the "low concern group II", while
Flemish home samples were mostly falling into the "concern for combined effects group
III", and to the "single substance concern group I". Most of the OQAI dataset were
assigned to the "single substance concern group I". Substances identified as biggest
contributors were NO2, trichloroethylene, acrolein, and xylenes. These were however,
not consistently measured in all the studies, so comparison of datasets and overall
drivers is difficult. The study showed a significant number of cases where combined
effects should be considered further and a chemical-by-chemical approach would be
insufficient. However, the mixtures showing concern for combined effects were not those
with the highest HIs. Highest HI values were observed for samples where single
substances were dominating the overall risk. Personal measurements had generally a
higher HI than indoor air measurements. The average ratio for HI from personal
measurement vs. indoor air monitoring was 1.5 (range 0.15-19), thus the use of indoor
air could lead to some underestimation. A problem was the availability of reliable chronic
inhalation toxicity data for non-cancer effects. The choice of the RVs had a large impact
on the overall results. Using minimum RVs instead of the basic RVs moved most samples
to the group of "single substance of concern I".
4.1.9 Need to address mixtures across regulatory sectors
Different chemical classes are often regulated under different pieces of legislation, for
instance pesticides, biocides, cosmetic ingredients, industrial chemicals etc. Several
studies in the recent literature show that chemicals from different classes are able to
elicit similar effects (e.g. Evans et al., 2015; Kortenkamp and Faust, 2010; Maffini and
Neltner, 2014) and numerous monitoring studies provide evidence for relevant co-
exposure. Several of the case studies presented in section A.8 (De Brouwere et al.,
2014; Evans, et al., 2015; Kortenkamp and Faust, 2010; Malaj et al., 2014) show that
mixtures of chemicals spanning different regulatory silos can be of concern particularly
for vulnerable / highly exposed subgroups. An additional example is shown in Carvalho
et al. (2014), who tested two artificial mixtures, designed of 14 and 19 substances
selected to cover different classes and modes of action, each present at their individual
safety limit concentration imposed by the European legislation for surface water (Water
Framework Directive). The toxic effects of the two mixtures were investigated in 35
bioassays based on 11 organisms representing different trophic levels. The mixtures
elicited quantifiable and significant toxic effects on some of the test systems, showing
the need of precautionary actions on the assessment of chemical mixtures even in cases
where individual toxicants are present at seemingly harmless concentrations. As
concerns were identified in several cases, further case studies should address also
mixture risk assessments across regulatory silos.
4.2 Conclusions related to chemical classes considered
Among the considered case studies, several chemical classes have been addressed, but
this does not mean that the overview is representative as many groups of compounds
19
have not been covered. The selection of these chemical classes is often based on data
availability, recent concerns about certain chemical classes or legislative requirements.
However, many more chemicals, including emerging substances, could be considered in
the future. Wang et al. (2016) for example investigated perfluoroalkyl phosphonic and
phosphinic acids (PFPAs and PFPiAs) used as defoamers in pesticide formulations and
wetteners in consumer applications, which individually have a low risk. However, they
conclude that combined exposure to them could be of concern due to similar MoA with
other chemicals and their potential for long-range transport and potential for
bioaccumulation in aquatic and mammalian organisms.
Several of the case studies revealed a concern due to combined exposure for certain
chemical classes especially when considering specific vulnerable population groups. This
is very relevant information, but needs to be interpreted with caution, considering
carefully the assumptions, model parameters and related uncertainties. However, there
is clear evidence that chemicals need to be further addressed not only in single
substance risk assessment and that mixtures should be considered also across chemical
classes and legislative sectors.
4.3 Potential for over- and underestimating risks from combined
exposure to chemical mixtures
For some case studies described above, a concern for the environment or human health
was identified. Each case study however, needs to be interpreted carefully taking into
account the related assumptions and uncertainties.
Several factors in these studies might lead to an
overestimation of risks, such as e.g.
conservative assumptions that:
• all chemicals in a mixture contribute to a combined effect,
• exposure takes place to all chemicals simultaneously,
• high-end exposures for all chemicals in parallel, e.g. in lower tiers of the WHO/IPCS
scheme (Meek et al., 2011).
The idea in a tiered scheme is however to start on purpose with conservative
assumptions to reduce data requirements and allow a simpler screening assessment that
is protective. This can be refined based on more realistic assumptions and additional
data where a potential concern is identified.
However, many factors could also result in an
underestimation of the overall risk,
such as:
• the limited knowledge of the number and identity of chemicals humans and the
environment are really co-exposed to. This is usually based on known exposure data
from chemical monitoring, which can only detect the limited set of chemicals we are
specifically looking for (e.g. De Brouwere et al., 2014; Kortenkamp & Faust, 2010;
Malaj et al., 2014). This could be partly overcome by non-targeted monitoring or
effect-based monitoring, potentially followed by effect-directed analysis to identify
chemicals driving the risk;
• mixture assessments that consider only a certain compound class, which do not take
into account co-exposure to other compounds that might contribute to a combined
risk (e.g. leading to the same adverse outcome) (Dewalque et al., 2014);
• neglecting potential bioaccumulation in the organism using only external exposure
concentrations (Kortenkamp & Faust, 2010);
• not addressing possible synergistic effects by assuming only concentration addition
(Kortenkamp & Faust, 2010; Marx et al., 2015);
• neglecting chemical metabolites that are more toxic than the parent compound,
might contribute to the overall risk, since most often only the parent compounds are
considered (Malaj et al., 2014);
20
• in the case of exposure via environmental samples or effluents, the underlying
sampling and extraction method can influence the chemicals that can be detected in
the chemical (or biological) analyses. In the case of effluents, the assumptions made
on their dilution in receiving waters can underestimate concentrations for small
streams receiving several discharges (Malaj et al., 2014; Price et al., 2012a);
• in the use of monitoring data, the decision how non-detected chemicals are treated
can make a big difference. If they are considered as absent, an underestimation is
probable, while if they are assumed to be present at a certain fraction of the limit of
detection or quantification, this can lead to overestimations (e.g. De Brouwere et al.,
2014; Han and Price, 2011, 2013; Price et al., 2012a).
4.4 Mixture assessment approaches
4.4.1 Prediction models
All case studies performed a mixture risk assessment based on
Concentration
Addition (CA)
except for Gosens et al. (2013) which considered only aggregate
exposure to individual substances from different sources. One case study (Junghans et
al., 2006) applied CA and
Independent Action (IA) to the same dataset, which slightly
underestimated the mixture toxicity. However this underestimation is significant only
with increasing effect levels. At the 50% effect level the confidence interval of the EC50
predicted according to IA still overlaps with the confidence interval of the EC50 derived
from the measured concentration-response data. Moreover, the EC50 values that can be
derived from each prediction only differed by a factor of 1.3. Those results suggest that
CA provides a precautious but not overprotective approach to the predictive hazard
assessment of pesticide mixtures under realistic exposure scenarios, irrespective of the
similarity or dissimilarity of their mechanisms of action. Junghans et al. (2006) identified
two circumstances that can challenge the precautionary character of the CA approach,
one of them being rather flat concentration-response curves so that IA could predict a
higher toxicity than CA. The other could be caused by potential synergistic effects that
are not covered by CA or IA approaches. Interactions need to be assessed on a case-by-
case basis.
CA based approaches were also the ones most used by the experts participating in the
JRC expert survey (Bopp et al., 2015). Experts used mostly the HI, TEQ, and Σ TU, as
reflected also in the case studies discussed in this report. Some experts in the survey
specifically mentioned IA based calculations as an approach they have abandoned, since
the prediction outcome is usually similar to CA based predictions, while IA calculations
require a lot more input information (full dose-response curves) to enable such
calculations, which is often not available.
There is an ongoing debate about the relevance of
interactions. In the reviewed case
studies, only one included an assessment of synergistic interactions applying the HIint
approach to antibiotic mixtures (Marx et al., 2015). Toxicological interactions modulate
toxicokinetic and/or toxicodynamic mechanisms of individual chemicals. Toxicokinetic
interactions could be e.g. induction of metabolising enzymes, alterations in uptake
mechanisms, all processes linked to the influence of individual chemicals on ADME of
others. Toxicodynamic interactions can be based on e.g. modulation of homeostasis or
repair mechanisms. Boobis et al. (2011) performed a literature review, identifying 90
studies demonstrating synergisms in mammalian test systems performed at low doses
(i.e. close to the point of departure, POD) for individual chemicals. Only in 6 of the 90
studies useful quantitative information on the magnitude of synergy was reported. In
those six studies, the difference between observed synergisms and predictions by CA did
not deviate by more than a factor of 4. Cedergreen (2014) performed a systematic
literature review for binary mixtures within three groups of environmentally relevant
chemicals (pesticides, metals, antifouling agents). Synergy was defined as a minimum
two-fold deviation from CA predictions. Synergy was found in 7%, 3% and 26 % of the
21
pesticide, metal and antifoulant mixtures, respectively. The extent of synergy was rarely
more than a factor of 10. Based on some more in depth analysis, Cedergreen concluded
that true synergistic interactions between chemicals are rare and often occur at high
concentrations. Using standard models as CA is regarded as the most important step in
the RA of chemical mixtures. In the JRC expert survey (Bopp et al., 2015), most of the
experts were in favour of addressing interactions on a case-by case basis, considering
whether available information (e.g. regarding the chemical structures, MoA) can be used
to anticipate possible interactions.
Two international frameworks developed for the assessment of combined exposure to
multiple chemicals were applied in some of the case studies: the
WHO/IPCS
framework (Meek et al., 2011) and the
Cefic MIAT decision tree (Price et al.,
2012b). The latter combines the frameworks of the WHO/IPCS with the decision tree
developed by the Scientific Committees (SCHER, SCENIHR, SCCS, 2011) and
incorporates the Maximum Cumulative Ratio (MCR) (Price and Han, 2011). The MCR can
be used as a tool for prioritising mixtures, prioritising relevant refinements in the
mixture RA and identifying where a single substance RA might be sufficient. The main
characteristic of the WHO/IPCS framework is its tiered approach for both the exposure
and the hazard assessment. Screening level assessments (Tier 0) using e.g. simple
exposure estimates based on production or sales volumes and TTC values for missing
hazard information and without subgrouping of chemicals were presented in several case
studies, as well as more refined Tier 1 case studies. However, it was also shown that
further refinement (which includes grouping of chemicals based on e.g. common effect
or common MoA) is often hampered by a lack of data (e.g. Evans et al., 2015). The Cefic
MIAT decision tree was shown to be applicable to several cases and the
MCR was
demonstrated as a valuable tool in six examples. The MCR was in the range of 1.2-4.2
for pharmaceuticals in WWTP effluents (Backhaus and Karlsson, 2014), 1.3-2.4 for the
FCM cases (Price et al., 2014), 2.2-5.7 for DLC in the generally exposed population, 1-
5.8 for indoor air contaminants (De Brouwere et al., 2014), and 1-2 for groundwater
samples with 5-10 detects, and 1-5 for groundwater samples with 15-25 detected
contaminants (Han and Price, 2011). This shows that the number of chemicals that drive
the mixture risk is usually low. In all the examples, the MCR decreased with increasing
HI. This indicates that the higher the predicted risk, the lower the number of chemicals
that are substantial contributors. It was also shown to be useful to present the mixture
effects by ranking the chemicals according to their individual RQ from highest to lowest,
to identify those chemicals that are contributing most and to see which number of
chemicals reaches in the sum a certain percentage of the overall risk. It was shown for
example in Backhaus and Karlsson (2014) that the top ranking 10 chemicals contributed
95% of the combined risk. In Price et al. (2012a) it was shown that only 2-5 compounds
were significant contributors to the overall risk. This can help in further characterising a
mixture, deciding when further refined mixture assessment is needed and developing
strategies for such targeted refinements. Another important task is also to identify major
sources of exposure, which is relevant for the risk management of combined exposures.
Dewalque et al. (2014) and Kennedy et al. (2015) showed ways of comparing e.g.
dietary
vs non-dietary contributions for exposure to phthalates and pesticides,
respectively.
4.4.2 Grouping of chemicals
In the case studies reviewed here, mostly lower tier assessments were applied based on
conservative assumptions. This implies that in most cases no specific grouping was
performed, but all mixture components were supposed to act together leading to
combined effects. These assessments are usually based on agreed reference values
which were derived based on the most sensitive endpoints, i.e. not necessarily based on
the same type of effect. This is a valid conservative approach for lower tiers. If no
concern is identified considering all mixture components together, no further refinement
and grouping will be needed.
22
There is some consensus in the current frameworks that in refined assessments a
grouping is performed based on a common effect, common mode of action, or common
target organ. Depending on the choice, the groups will be larger (resulting in more
conservative assessments) or smaller (less conservative). One widely acknowledged
advanced framework is the development of cumulative assessment groups (CAGs) for
pesticides by EFSA (EFSA PPR Panel, 2014). At CAG level 1, chemicals are grouped
based on their toxicological target organ. At CAG level 2 grouping is further refined
based on common specific phenomenological effects, at level 3 based on common mode
of action and at level 4 based on common mechanism of action. With the usually
available chemical hazard information, grouping can mostly be performed until level 2.
As nowadays more and more mechanistic information is becoming available, further
refinement will be possible. Another EFSA opinion (EFSA PPR Panel, 2013) concluded
that the best approach for addressing pesticides eliciting a common adverse effect in the
same organ by dissimilar MoA is also CA.
Thus the question remains how far further refinement of groupings should go to remain
protective. Most scientific publications and international activities on the risk assessment
of mixtures conclude that risks from combined effects are relevant for mixtures of
substances with similar mode of action or effect (e.g. SCHER, SCENHIR, SCCS, 2011).
However, based on the relatively well studied adverse effects of mixtures of
pharmaceuticals, Hadrup (2014) suggested that chemicals with dissimilar mechanisms of
action could be of bigger concern in the toxicological risk assessment of chemical
mixtures than chemicals with a similar mechanism of action. Examples obtained from
cancer and HIV treatment studies, show that pharmacological combination therapy
targeting different mechanisms of action is more effective than a strategy where only
one mechanism is targeted. Another argument is that also in many diseases several
organ systems concomitantly contribute to the pathophysiology, implying that a
grouping based on common target organs may be inadequate. In further considerations,
it should be however considered that in pharmacology usually higher doses are applied,
whereas at lower concentrations some specific effects might not occur. Goodson et al.
(2015) reviewed actions on key pathways and mechanisms related to carcinogenesis for
85 chemicals ubiquitously occurring in the environment. The aim was to explore the
hypothesis that low-dose exposures to mixtures of chemicals in the environment may be
combining to contribute to environmental carcinogenesis. The results of their analysis
suggest that the combined effects of "individual (non-carcinogenic) chemicals acting on
different pathways, and a variety of related systems, organs, tissues and cells could
plausibly conspire to produce carcinogenic synergies". Additional research on
carcinogenesis and low-dose effects of chemical mixtures needs to be performed to
further investigate the hypothesis. The concept of assessing combined effects strictly
based on grouping chemicals according to their MoA or common effect(s) might need to
be revisited in order not to underestimate cancer-related risks and systemic diseases.
Further investigations on the risks from combined exposure to multiple chemicals should
consider synergies of chemicals acting via dissimilar processes, acting on different
targets and tissues, and consider synergies between certain pathways.
4.5 Current limitations in performing mixture risk assessments
In order to perform mixture risk assessments using component based approaches, it is a
prerequisite to have detailed information on the
mixture composition regarding the
chemical identity and concentrations. This is usually known for formulated
products/intentional mixtures, but is sometimes problematic for unintentional and
environmental mixtures. For example Tang et al. (2014) have shown that although a
total of 299 chemicals were screened in wastewater and recycled water samples, all
present below the individual regulatory safety limit, the known chemicals in designed
mixture toxicity testing explained less than 3% of the observed cytotoxicity and less
23
than 1% of the oxidative stress response, and were not related to the observed
genotoxicity. Neale et al. (2015) examined Danube river samples using large volume
water extracts testing in an
in vitro test battery and tried to match observed effects with
the detected organic micropollutants. Most samples showed rather low response in the
bioassays, however, depending on the endpoint, the contribution of the detected
chemicals explained in the worst case only 0.2 % of the observed effect (for PXR
activation), while five chemicals explained 80% of the observed effect for ER activation.
The sometimes large proportion of unknown toxicity in environmental samples could be
addressed by widening the range of chemicals analysed and complementing the chemical
monitoring with biological effect monitoring. De Brouwere et al. (2014) used 4 different
indoor air monitoring data sets in their case study and faced difficulties in comparing the
results for the different studies. The substances identified as biggest contributors to the
potential risk were NO2, trichloroethylene, acrolein, xylenes. These were however not
consistently measured in all the studies, so comparison of datasets and overall drivers is
difficult. The combined assessments from the different studies might lead to an
underestimation of risk due to the presence of some major contributors that were not
included in the chemical analysis.
Another major problem is the availability of relevant exposure and toxicity data, as well
as lack of information on the MoA of mixture components (e.g. Evans et al., 2016;
2015). A major
gap was identified in the information on
human and environmental
exposure and a new platform for monitoring data was therefore created. IPCheM4, the
Information Platform for Chemical Monitoring data, was developed over the last years as
one of the follow up actions to the Commission Communication (EC, 2012). It comprises
monitoring data in four modules, i.e. human biomonitoring, environmental monitoring,
indoor air and consumer products, and food / feed related monitoring data. Thus it offers
great potential for the assessment of mixtures.
Exposure is often predicted from e.g. production volumes, but also assessed using small
surveys, e.g. for parabens in personal care products (Gosens et al., 2013). The usually
limited number of individuals can lead to high uncertainties on the representativeness.
Also the limited spectrum of chemicals analysed in monitoring studies can be a problem
(e.g. De Brouwere et al. (2014) indicating problems in comparability between data sets),
as well as the reporting of monitoring results. For example, in the case of human
biomonitoring data, if only aggregated results are made available, it is impossible to
trace back the real co-occurrence of chemicals in individual humans. If monitoring data
are used, obviously the higher the number of analysed chemicals, the higher will also
become the predicted potential risk. Another major uncertainty is related to the impact
of chemicals that are analysed but below the limit of detection. Several approaches can
then be used to address these chemicals, which can greatly influence the final outcome
(Han and Price, 2011). When external exposure data are used (like environmental
monitoring data or exposure predicted from emissions or product uses), there is always
the question about internal co-exposure at a target site. Tier 3 of the WHO/IPCS scheme
therefore includes toxicokinetic modelling for the prediction of internal exposure
concentrations. Such Tier 3 refinements of the hazard assessment are however not
reported in any case study and hampered by the availability of relevant input
parameters required for the modelling (see also EURL ECVAM Toxicokinetic Strategy,
Bessems et al. (2015)).
Furthermore, the
lack of toxicity data is highlighted in many of the reviewed case
studies. For screening assessments, the TTC can be used in several cases to replace
specific reference values. If however a refinement and specific values are needed,
limited availabilities are encountered, e.g. for pharmaceuticals (Backhaus and Karlsson,
2014), pesticides (Junghans et al., 2006; Kennedy et al., 2015; Nowell et al., 2014)
cosmetics, etc. Moreover, if relevant reference values are found from the various sources
4 https://ipchem.jrc.ec.europa.eu/
24
used (i.e. literature, public databases, authorities' assessment reports etc.), it is
sometimes difficult to select the most reliable one. De Brouwere et al. (2014) for
example used chronic inhalation toxicity values and identified up to 300 fold differences
in the retrieved values. They developed a decision scheme to select the most reliable RV.
Apart from the general RVs, it is often even more difficult to find information on specific
effects, which is important when chemicals need to be grouped based on a common
effect. So what is often missing is information on the detailed MoA and also on the
toxicity values related to such a specific effect. Several databases may be useful for
obtaining tissue and organ level information as well as reference values. An example is
the publicly accessible COSMOS database5, which currently hosts toxicological data on
cosmetic and food relevant chemicals. Missing toxicity data often imply that
extrapolations have to be used, such as acute to chronic extrapolations or read-across
from similar compounds, which leads to overall increased uncertainties of the
predictions.
For both hazard and exposure assessments, additional assumptions have to be made
due to limited data availability and parameters sometimes need to be predicted by
modelling. The related model specifications can greatly influence the results (Boon et al.,
2015; Kennedy et al., 2015). Therefore it is of utmost importance that the scenario
parameters and hypotheses underlying each mixture risk assessment are clearly justified
and transparently documented to allow a proper interpretation of the results.
5 https://cosmosdb.eu/cosmosdb.v2
25
5 Conclusions
As humans and wildlife can be exposed to a virtually infinite number of different
combinations of chemicals in mixtures via food, consumer products and the
environment, it is impossible to test or assess all these possible combinations. In this
review, 21 examples of case studies on specific mixtures have been identified, but to
address the risk assessment in a wider range of mixtures, priorities need to be set. The
Scientific Committees (SCHER, SCENIHR, SCCS, 2011) have set out relevant criteria to
prioritise mixtures for the assessment, e.g. based on relevant exposure close to health
based guidance values for several components, chemicals of higher production volumes
or produced as commercial mixtures, likelihood of frequent large scale exposure etc.
Once a decision is taken to perform a mixture RA, it can then be performed as screening
level RA or higher tier RA, depending on the needs and to be resource efficient. Apart
from the tiered approach as outlined in the WHO/IPCS framewrok, the MCR can help to
further characterise the main issues around a certain mixture and to decide on the next
steps for refinement where needed, e.g. concentrate on few substances driving the risk
or identify a need to tackle further the whole mixture composition.
Monitoring data are essential in mixture risk assessment as they can give information on
identity, magnitude, duration, frequency and/or timing of real exposure, depending on
the monitoring scheme, and allow to assess the co-exposure patterns to chemicals (Qian
et al., 2015), both for human and environmental risk assessment. Numerous
retrospective ERA have been performed with monitoring data; however so far there is no
prospective RA concerning chemical substances related to various regulatory sectors
and/or uses, and although numerous chemicals fall under several regulatory frameworks
(biocides, pesticides, REACH...), the potential for co-exposure is hardly assessed or
taken into account in their risk assessment.
The data sources used are various (exposure data from modelling, monitoring or
published data from surveys; toxicological data from published databases, TTC
approach, data gap filling, etc.) and the data sets are frequently incomplete, which has a
direct impact on the possibility to perform a mixture risk assessment, as well as the
reliability and accuracy of the risk assessment. Data gaps seem to be the major hurdle
when it comes to deal with risk assessment of chemical mixtures, especially when
focusing on particular uses or population subgroups (e.g. amateur uses of pesticides,
frequency of use of personal care products for children).
It has to be taken into account that the list of compounds covered in the reviewed case
studies is unlikely to be fully representative of human and environmental exposures.
However, based on the identified cases, pesticides followed by pharmaceuticals and
personal care products dominate the observed mixture effects in cross-sectorial
mixtures. Tributyltin, polycyclic aromatic hydrocarbons, and brominated flame retardants
are also major contributors to the environmental chemical risks of the monitored
chemicals. Human exposure to parabens, phthalates and more generally anti-androgenic
chemicals seems to be of concern, particularly for highly exposed or more sensitive
population subgroups. It is important to be aware of the influence that the choice of
model specifications, the parameters and the toxic reference values considered have on
the outcome of a mixture RA. There is a need to clearly specify and justify the choices
that have been made. Thus, the results should be interpreted carefully in the light of the
models used, the underlying hypothesis and the degree of conservatism that has been
chosen.
It has to be highlighted that for both environmental and human exposure, there are
several factors that might lead to an underestimation or overestimation of the potential
risk, e.g. uncertainty in reference values used, incompleteness of monitoring data, etc. A
clear potential for underestimation results from neglecting potential synergistic effects,
bioaccumulation potential and metabolites. Another potential for underestimation results
from the assessment of specific chemical classes or regulated under a specific legislative
26
framework. Several studies reviewed here indicated a potential concern for mixtures
across several regulatory silos.
In order to facilitate mixture risk assessment in the future, it will be relevant to improve
data sharing regarding toxicity and exposure information. Relevant platforms such as
e.g. IPCheM should be further populated (e.g. by monitoring programmes such as the
European Human Biomonitoring Initiative) and made interoperable with other tools.
Future case studies on mixture RA could help to fill the knowledge gaps identified
through this review, by:
• Comparison of different populations including vulnerable subgroups;
• Inclusion of substance groups that have not been addressed in mixtures so far,
including emerging chemicals;
• Further investigations on the relevance of interactions (particularly synergisms) at
relevant low exposure concentrations;
• Developing further criteria to decide under which circumstances and for which
mixtures interactions need to be addressed;
• Investigating the impact of different approaches for grouping (based on common
effects, common MoA etc.) and related to that, investigating further combined effects
of independently acting chemicals considering interactions between pathways, as e.g.
for carcinogenesis and systemic diseases;
• Examining further the relevance to address mixtures across different regulatory
sectors.
27
References
Backhaus, T., & Karlsson, M. (2014). Screening level mixture risk assessment of
pharmaceuticals in STP effluents.
Water Research,
49(0), 157–65.
Becker, K., Göen, T., Seiwert, M., Conrad, A., Pick-Fuss, H., Müller, J., … Kolossa-
Gehring, M. (2009). GerES IV: phthalate metabolites and bisphenol A in urine of
German children.
International Journal of Hygiene and Environmental Health,
212(6), 685–92.
Bessems, J., COECKE, S., GOULIARMOU, V., WHELAN, M., & Worth, A. (2015). EURL
ECVAM strategy for achieving 3Rs impact in the assessment of toxicokinetics and
systemic toxicity.
EUR 27315 EN. http://doi.org/10.2788/197633
Boobis, A., Budinsky, R., Collie, S., Crofton, K., Embry, M., Felter, S., … Zaleski, R.
(2011). Critical analysis of literature on low-dose synergy for use in screening
chemical mixtures for risk assessment.
Critical Reviews in Toxicology,
41(5), 369–
383.
Boon, P. E., van Donkersgoed, G., Christodoulou, D., Crépet, A., D’Addezio, L.,
Desvignes, V., … van Klaveren, J. D. (2015). Cumulative dietary exposure to a
selected group of pesticides of the triazole group in different European countries
according to the EFSA guidance on probabilistic modelling.
Food and Chemical
Toxicology : An International Journal Published for the British Industrial Biological
Research Association,
79(August), 13–31.
Bopp, S., Berggren, E., Kienzler, A., van der Linden, S., & Worth, A. (2015).
Scientific
methodologies for the assessment of combined effects of chemicals – a survey and
literature review. EUR 27471 EN
Carvalho, R. N., Arukwe, A., Ait-Aissa, S., Bado-Nilles, A., Balzamo, S., Baun, A., …
Lettieri, T. (2014). Mixtures of chemical pollutants at European legislation safety
concentrations: How safe are they?
Toxicological Sciences,
141(1), 218–233.
Ccanccapa, A., Masiá, A., Navarro-Ortega, A., Picó, Y., & Barceló, D. (2016). Pesticides
in the Ebro River basin: Occurrence and risk assessment.
Environmental Pollution
(Barking, Essex : 1987),
211, 414–424.
Cedergreen, N. (2014). Quantifying synergy: A systematic review of mixture toxicity
studies within environmental toxicology.
PLoS ONE,
9(5).
Christiansen, S., Scholze, M., Dalgaard, M., Vinggaard, A. M., Axelstad, M., Kortenkamp,
A., & Hass, U. (2009). Synergistic disruption of external male sex organ
development by a mixture of four antiandrogens.
Environmental Health
Perspectives,
117(12), 1839–1846.
De Brouwere, K., Cornelis, C., Arvanitis, A., Brown, T., Crump, D., Harrison, P., … Torfs,
R. (2014). Application of the maximum cumulative ratio (MCR) as a screening tool
for the evaluation of mixtures in residential indoor air.
Science of the Total
Environment,
479-480(1), 267–276.
Dewalque, L., Charlier, C., & Pirard, C. (2014). Estimated daily intake and cumulative
risk assessment of phthalate diesters in a Belgian general population.
Toxicology
Letters,
231(2), 161–8.
EC (2012). Communication from the Commission to the Council - The combination
effects of chemicals. Chemical mixtures,
COM(2012) , 10.
ECHA (2008).
Guidance on information requirements and chemical safety assessment
Chapter R . 10 : Characterisation of dose [ concentration ] -response for
environment.
European Chemical Agency.
EFSA Panel on Plant Protection Products and their Residues (PPR). (2014). Scientific
Opinion on the identification of pesticides to be included in cumulative assessment
28
groups on the basis of their toxicological profile.
EFSA Journal,
11(7), 3293.
http://doi.org/10.2903/j.efsa.2013.3293
EFSA Panel on Plant Protection Products and their residues (PPR). (2013). Scientific
Opinion on the relevance of dissimilar mode of action and its appropriate application
for
cumulative
risk
assessment
of
pesticides.
EFSA
Journal,
11(12).
http://doi.org/10.2903/j.efsa.2013.3472
Evans, R. M., Martin, O. V., Faust, M., & Kortenkamp, A. (2016). Should the scope of
human mixture risk assessment span legislative/regulatory silos for chemicals?
Science of the Total Environment,
543(Pt A), 757–764.
Evans, R. M., Scholze, M., & Kortenkamp, A. (2015). Examining the feasibility of mixture
risk assessment: A case study using a tiered approach with data of 67 pesticides
from the Joint FAO/WHO Meeting on Pesticide Residues (JMPR).
Food and Chemical
Toxicology : An International Journal Published for the British Industrial Biological
Research Association,
84, 260–9.
Goodson, W. H., Lowe, L., Carpenter, D. O., Gilbertson, M., Manaf Ali, A., Lopez de
Cerain Salsamendi, A., … Hu, Z. (2015). Assessing the carcinogenic potential of low-
dose exposures to chemical mixtures in the environment: the challenge ahead.
Carcinogenesis,
36(Suppl 1), S254–S296.
Gosens, I., Delmaar, C. J. E., Burg, W., Heer, C. De, & Schuur, A. G. (2013). Aggregate
exposure approaches for parabens in personal care products : a case assessment for
children between 0 and 3 years old,
24(2), 208–214.
Hadrup, N. (2014). Evidence from pharmacology and pathophysiology suggests that
chemicals with dissimilar mechanisms of action could be of bigger concern in the
toxicological risk assessment of chemical mixtures than chemicals with a similar
mechanism of action.
Regulatory Toxicology and Pharmacology,
69(3), 281–283.
Han, X., & Price, P. S. (2011). Determining the maximum cumulative ratios for mixtures
observed in ground water wells used as drinking water supplies in the United
States.
International Journal of Environmental Research and Public Health,
8(12),
4729–4745.
Han, X., & Price, P. S. (2013). Applying the maximum cumulative ratio methodology to
biomonitoring data on dioxin-like compounds in the general public and two
occupationally exposed populations.
Journal of Exposure Science & Environmental
Epidemiology,
23(4), 343–349.
Hartmann, C., Uhl, M., Weiss, S., Koch, H. M., Scharf, S., & König, J. (2015). Human
biomonitoring of phthalate exposure in Austrian children and adults and cumulative
risk assessment.
International Journal of Hygiene and Environmental Health,
218(5), 489–499.
Junghans, M., Backhaus, T., Faust, M., Scholze, M., & Grimme, L. H. (2006). Application
and validation of approaches for the predictive hazard assessment of realistic
pesticide mixtures.
Aquatic Toxicology (Amsterdam, Netherlands),
76(2), 93–110.
Kennedy, M. C., Glass, C. R., Bokkers, B., Hart, A. D. M., Hamey, P. Y., Kruisselbrink, J.
W., … van Klaveren, J. D. (2015). A European model and case studies for aggregate
exposure assessment of pesticides.
Food and Chemical Toxicology,
79, 32–44.
Kienzler, A., Berggren, E., Bessems, J., Bopp, S., Linden, S. Van Der, & Worth, A.
(2014). Assessment of Mixtures - Review of Regulatory Requirements and
Guidance, 136. http://doi.org/10.2788/84264
Kienzler, A., Bopp, S. K., van der Linden, S., Berggren, E., & Worth, A. (2016).
Regulatory assessment of chemical mixtures: Requirements, current approaches
and future perspectives.
Regulatory Toxicology and Pharmacology.
29
Kortenkamp, A., & Faust, M. (2010). Combined exposures to anti-androgenic chemicals:
Steps towards cumulative risk assessment.
International Journal of Andrology,
33(2), 463–472.
Le Cann, P., Bonvallot, N., Glorennec, P., Deguen, S., Goeury, C., & Le Bot, B. (2011).
Indoor environment and children’s health: recent developments in chemical,
biological, physical and social aspects.
International Journal of Hygiene and
Environmental Health,
215(1), 1–18.
Maffini, M. V, & Neltner, T. G. (2014). Brain drain: the cost of neglected responsibilities
in evaluating cumulative effects of environmental chemicals.
Journal of
Epidemiology and Community Health, 1–4.
Malaj, E., von der Ohe, P. C., Grote, M., Kühne, R., Mondy, C. P., Usseglio-Polatera, P.,
… Schäfer, R. B. (2014). Organic chemicals jeopardize the health of freshwater
ecosystems on the continental scale.
Proceedings of the National Academy of
Sciences
of
the
United
States
of
America,
111(26),
9549–54.
http://doi.org/10.1073/pnas.1321082111
Marx, C., Mühlbauer, V., Krebs, P., & Kuehn, V. (2015). Environmental risk assessment
of antibiotics including synergistic and antagonistic combination effects.
Science of
the Total Environment,
524-525, 269–279.
Meek, M. E., Boobis, A. R., KM, C., Heinemeyer, G., M, V. R., & Vickers, C. (2011). Risk
assessment of combined exposure to multiple chemicals: A WHO/IPCS framework.
Regulatory Toxicology and Pharmacology,
60(2), S1–S14.
Neale, P. A., Ait-Aissa, S., Brack, W., Creusot, N., Denison, M. S., Deutschmann, B., …
Escher, B. I. (2015). Linking in vitro effects and detected organic micropollutants in
surface water using mixture toxicity modeling.
Environmental Science &
Technology, acs.est.5b04083.
Nowell, L. H., Norman, J. E., Moran, P. W., Martin, J. D., & Stone, W. W. (2014).
Pesticide Toxicity Index - a tool for assessing potential toxicity of pesticide mixtures
to freshwater aquatic organisms.
The Science of the Total Environment,
476-477,
144–57.
Price, P., Han, X., Junghans, M., Kunz, P., Watts, C., & Leverett, D. (2012a). An
application of a decision tree for assessing effects from exposures to multiple
substances to the assessment of human and ecological effects from combined
exposures to chemicals observed in surface waters and waste water effluents.
Environmental Sciences Europe,
24(1).
Price, P., Dhein, E., Hamer, M., Han, X., Heneweer, M., Junghans, M., … Rodriguez, C.
(2012b). A decision tree for assessing effects from exposures to multiple
substances.
Environmental Sciences Europe,
24(1).
Price, P. S., & Han, X. (2011). Maximum cumulative ratio (MCR) as a tool for assessing
the value of performing a cumulative risk assessment.
International Journal of
Environmental Research and Public Health,
8(6), 2212–2225.
Price, P., Zaleski, R., Hollnagel, H., Ketelslegers, H., & Han, X. (2014). Assessing the
safety of co-exposure to food packaging migrants in food and water using the
maximum cumulative ratio and an established decision tree.
Food Additives &
Contaminants. Part A, Chemistry, Analysis, Control, Exposure & Risk Assessment,
31(3), 414–21.
Qian, H., Chen, M., Kransler, K. M., & Zaleski, R. T. (2015). Assessment of chemical
coexposure patterns based upon phthalate biomonitoring data within the 2007/2008
National Health and Nutrition Examination Survey.
Journal of Exposure Science &
Environmental Epidemiology,
25(3), 249–55.
SCHER, SCENIHR, & SCCS. (2011).
Toxicity and Assessment of Chemical Mixtures.
30
Tang, J. Y. M., Busetti, F., Charrois, J. W. a, & Escher, B. I. (2014). Which chemicals
drive biological effects in wastewater and recycled water?
Water Research,
60, 289–
99.
Wang, Z., Cousins, I. T., Berger, U., Hungerbühler, K., & Scheringer, M. (2016).
Comparative assessment of the environmental hazards of and exposure to
perfluoroalkyl phosphonic and phosphinic acids (PFPAs and PFPiAs): Current
knowledge, gaps, challenges and research needs.
Environment International,
89-90,
235–247.
31
List of abbreviations and definitions
ADI
Acceptable Daily Intake
ADME
Absorption, Distribution, Metabolism, Excretion
AF
Assessment Factor
Aggregate
exposure to a single substance originating from different sources
exposure
AL
Acceptable Level
AOP
Adverse Outcome Pathway
ARfD
Acute Reference Dose
BMD
Benchmark Dose
CA
Concentration Addition
The effects of exposure to a mixture of compounds with a similar mode of
action are assumed to be the sum of the potency-corrected effects of
each component.
CAG
Cumulative Assessment Group
CEA
Cumulative Exposure Assessment
Cefic MIAT
CEFIC (European Chemical Industry Council) Mixture Industry Ad hoc
Team
Combined
multiple substances from one or different sources
exposure
CR
Concentration-Response
CRA
Cumulative Risk Assessment
DEB
Dynamic Energy Budget modelling
DEHP
diethylhexyl phthalate
DI
Daily Intake
DLC
Dioxin-like compound
EC50
Concentration where 50 % effect was observed/calculated
ERA
Environmental Risk Assessment
HBM
Human Biomonitoring
HH
Human Health
32
HI
Hazard Index
Sum of Hazard Quotients, i.e. ratio between exposure and the reference
value for the common toxic effect of each component in a mixture or a
CAG.
HIint
Hazard Index considering interactions
HQ
Hazard Quotient
HRA
Human Risk Assessment
IA
Independent Action
Occurs where the mode of action and possibly, but not necessarily, the
nature and sites of toxic effects differ between the chemicals in a
mixture, and one chemicals does not influence the toxicity of another.
The effects of exposure to such a mixture are the combination of the
effects of each component compounds (also referred to as response-
addition).
IATA
Integrated Approach to Testing and Assessment
Intentional
manufactured products or formulations, including commercial mixtures of
mixtures
industrial substances
JMPR
Joint FAO/WHO Meeting on Pesticide Residues
LO(A)EL
Lowest Oserved (Adverse) Effect Level
LOD
Limit Of Detection
MCR
Maximum Cumulative Ratio
MCRA
Monte Carlo Risk Assessment tool
MEC
Measured Environmental Concentration
MoA
Mode of Action
MoE
Margin of Exposure
MRL
Maximum Residue Level
NHANES
National Health and Nutrition Examination Survey
NIAS
Non-Intentionally Added Substances
NOAEL
No Observed Adverse Effect Level
PBDE
PolyBrominated Diphenyl Ether
PBTK
Physiologically Based ToxicoKinetic modelling
PCBs
PolyChlorinated Biphenyls
PEC
Predicted Environmental Concentration
33
PNEC
Predicted No Effect Concentration
POD
Point Of Departure
PPDB
Pesticide Property Database
http://sitem.herts.ac.uk/aeru/ppdb/en/
PPPs
Plant Protection Products
PTI
Pesticide Toxicity Index
QSARs
Quantitative Structure–Activity Relationship
RA
Risk Assessment
RfD
Reference Dose
RPF
Relative Potency Factor
RQ
Risk Quotient
RV
Reference Value
STU
Sum of Toxic Unit (Σ TU)
TDI
Tolerable Daily Intake
TEF
Toxic Equivalency Factors
TEQ
Toxic Equivalency
TTC
Threshold of Toxicological Concern
TU
Toxic Unit
Unintentional substances from different sources, deposited separately at a particular
mixtures
site (e.g. in surface water)
US EPA
United States Environmental Protection Agency
VOCs
Volatile Organic Compounds
WC
Worst Case
WHO/IPCS
World Health Organisation / International Programme on Chemical Safety
WWTP
Waste Water Treatment Plant
34
Annex 1 – Overview of individual case studies
Twenty-two case studies from the literature were selected as described in Section 2.
Relevant information from the case studies was extracted and is reported in the tables
below. No judgement on the quality/validity of the case studies is included here.
Reported findings and conclusions are those of the case study publications'
authors and do not necessarily represent the views of the authors of this
report.
A.1 Pesticides
ID
1
Title
Application and validation of approaches for the predictive hazard
assessment of realistic pesticides mixtures (Junghans et al., 2006)
Journal
Aquatic Toxicology, 76, 93-100
Authors
Marion Junghans, Thomas Backhaus, Michael Faust, Martin Scholze, L.H
Grimme
Year
2006
Background
In freshwater systems located in agricultural areas, organisms are
&
exposed to a multitude of toxicologically and structurally different
Objectives
pesticides. For regulatory purposes it is of major importance whether the
combined hazard of these substances can be predictively assessed from
the single substance toxicity. This study aimed to analyse whether both
concepts of CA or IA may be used to predict the toxicity of
environmentally realistic mixtures, including a mixture of chemicals
acting by similar and dissimilar MOA. In order to do so, the mixture was
studied for its effect on the reproduction of the freshwater algae
Scedenesmus vacuolatus. The predictability of CA (Σ TU) and IA was then
assessed, by comparing the predicted results to the actual measured
toxicity.
Substances
A defined mixture of pesticides (25 single substance) reflecting a realistic
exposure scenario
Exposure
Field run-off water leading to exposure of aquatic organisms in edge-of-
Scenario
field surface waters
Problem
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Formulation
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
(according
MIXTURE ITSELF?
to
Exposure of algae to field run-off water. Key components are known
WHO/IPCS
because artificial mixture data available on the hazard of the mixture
mixture
itself to be compared with theoretical calculated toxicity.
assessment 2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
framework)
Yes (for algae)
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT TIME
FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
35
No assessment groups.
Information Exposure: modelled; the physico-chemical characteristics for all active
sources
ingredients were collected from registration dossier
Toxicity:
Single
substance
and
mixture
concentration-response
relationships were determined experimentally
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
The exposure scenario is based on emission patterns from growing three
Assessment major crops (cereals, maize, sugar beet), and modelled according to the
standard FOCUS scenario "R1", in order to reflect the median load of
pesticides in field run-off in central European agricultural areas after pre-
emergence treatment in spring. A reasonable worst case (WC)
application scenario based on common crop protection strategies has
been assumed.
Hazard
To allow for a comparison of observed mixture toxicity with the prediction
Assessment according to CA and IA, single substance concentration-response (CR)
relationships were determined for all mixture components in a bio-test
(24h of exposure). Those data were used to calculate the mixture toxicity
according to the CA and IA models.
However for 3 out of 25 substances this prerequisite was not or only
partially fulfilled: for isoxaflutole algal toxicity, the maximum effect
observed at the limit of solubility did not exceed 45%. Therefore, WC
estimates of higher effect concentrations were extrapolated from the CR
function.
For carbosulfan and clopyralid, no CR relationship could be determined
within the limits of solubility and within concentration ranges causing no
strong acidification of the algal growth medium, respectively. Therefore,
they were left out and total mixture toxicity predictions are based on 23
(out of 25) substances only.
RA for
When all components are present at their PEC in run-off water, the
Algae
growth of the algal population was inhibited by 46%, which is
significantly higher than 17% caused alone by the PEC of the most active
component atrazine. When comparing this measured mixture toxicity
with the predictions, it is slightly lower than predicted according to CA
(49%), but higher than predicted according to IA (39%).
Overall
-The resulting mixture proved to be clearly more toxic than any individual
summary of
component
outcome
-CA shows a good predictive quality over the complete range of effect.
This is consistent with the finding that the toxicity was dominated by a
group of similarly acting photosystem II inhibitors (they contribute 0.80
TUs to the total sum of 0.98 TUs), although the mixture included
substances with diverse and partly unknown mechanisms of action.
-IA underestimates the mixture toxicity slightly; however this
underestimation is significant only with increasing effect levels. At the
50% effect level the confidence interval of the EC50 predicted according
to IA still overlap with the confidence interval of the EC50 derived from
the measured concentration-response data. Moreover, the EC50 values
that can be derived from each prediction only differed by a factor of 1.3.
36
Those results suggest that CA provides a precautious but not
overprotective approach to the predictive hazard assessment of pesticide
mixtures under realistic exposure scenarios, irrespective of the similarity
or dissimilarity of their mechanisms of action.
• Problems: Substance specific degradation and sorption processes are
Future
not taken into account. Therefore, a conclusive assessment of the
perspec-
expectable mixture toxicity in receiving water might require a second
tives /
step of fate and effect modelling.
Outlook
• The major limitation for such modelling exposure is the restricted
availability of reliable information on pesticide use.
• Two circumstance can challenge the precautionary character of the
approach:
-
If due to rather flat concentration-response curves IA predict a
higher toxicity than CA and if the mixture is dominated by
dissimilarly acting components, the mixture toxicity can be
expected to comply better with IA than with CA.
-
In case of interaction of the mixtures components, which leads to
a mixture toxicity that is higher than predicted by CA. In this
case, a more detailed hazard assessment has to be performed.
37
ID
2
Title
Pesticide Toxicity Index—A tool for assessing potential toxicity of
pesticide mixtures to freshwater aquatic organisms (Nowell et al., 2014)
Journal
Science of the Total Environment
Authors
Lisa H. Nowell, Julia E. Norman, Patrick W. Moran, Jeffrey D. Martin,
Wesley W. Stone.
Year
2014
Background
The Pesticide Toxicity Index (PTI) is a screening tool to assess potential
&
aquatic toxicity of complex pesticide mixtures by combining measures of
Objectives
pesticide exposure and acute toxicity in an additive toxic-unit model. This
paper addresses exposure to pesticide mixtures and presents the
Pesticide Toxicity Index (PTI) as a robust and readily applicable screening
tool for interpreting the biological significance of concentration data for
pesticide mixtures in hydrologic systems and expands the number of
pesticides and degradates included in previous editions of the PTI from
124 to 492 pesticides and degradates, and includes two types of PTI for
use in different applications, depending on study objectives.
Substances
Pesticides mixtures (active ingredients and degradates)
Exposure
Exposure of aquatics organisms
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Exposure of freshwater organisms; key components known;
to
information on the hazards of the sample itself known from literature.
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
Yes, mixture of pesticides are frequently present in freshwater system
assessment
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
framework)
TIME FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No assessment groups
Information Toxicity data: A master list of 484 pesticides were compiled from
sources
agricultural pesticide use lists for 1992 to 2011, for which toxicity data
has been searched in the USEPA ECOTOX DB, the USEPA registration and
RA documents cited in support of the OPP aquatic-life benchmarks, and
the Pesticides Properties Database.
Exposure data: Published data on concentrations of pesticides in ambient
stream water
Published data on organism survival in toxicity test conducted in the
laboratory with undiluted ambient water were also used to compare with
the calculated PTI.
Exposure
No exposure data. Literature data (toxicity of environmental sample to
38
Assessment
C.dubia) were aggregated and used to test how well the PTI approach
predicted the incidence of observed toxicity.
Hazard
The following data were used in order of priority:
Assessment
• Standardized toxicity test data from the ECOTOX database
• Toxicity test data from core or supplemental studies underlying OPP
aquatic-life benchmarks or summarized in registration documents
• Non-standard toxicity test data from the ECOTOX database
• Toxicity values compiled from the PPDB
Two approaches were used: used of the MTC (median toxicity
concentration, calculated as the median of the toxicity value for each
compounds toward the taxonomic group) or the STC (sensitive toxicity
concentration, either the 5th percentiles if more than 13 data were
available, or the minimum toxicity value for each compounds toward the
taxonomic group )
• The PTI approach is used
RA for
aquatic
=
organisms
,
With t : the taxonomic group, Ei the concentration of the pesticide i, n the
number of detected pesticides, TCit the toxicity concentration for the
pesticide i for the taxonomic group t.
The Median-PTI and the Sensitive-PTI were calculated for each sample.
• MTC and STC values are significantly correlated with one another
Overall
within a taxonomic group.
summary of
• When MTCs (medians) are used, the purpose of the index is to
outcome
represent the relative toxicity of sites, samples, or individual
pesticides - MTC values are relatively robust to outliers. When the STC
values are used, the index is better suited for use as a screening
level, because it is a more conservative (protective) indicator of the
potential for toxicity.
•
C. dubia survival was reduced to ≤50% of controls in 44% of samples
with Median-PTI values in the range of 0.1 to 1, and to 0% in 96% of
samples with Median-PTI values 1. For the Sensitive-PTI,
C. dubia survival was reduced to ≤50% of controls in 81% of samples in the
range of 0.1 to 1, and in 89% of samples with Sensitive-PTI values
>1.
Future
Limitation:
perspec-
• The PTI is a relative ranking system that indicates that one sample is
tives /
likely to be more or less toxic than another sample, but does not indicate
Outlook
that toxicity will necessarily occur.
• Toxicity values are based on short-term laboratory data with EC50 or
LC50 endpoints and do not reflect long-term/chronic exposure or
incorporate sublethal endpoints.
• The PTI does not account for environmental factors (dissolved organic
carbon, particulates, pH, T°…) which can affect the toxicity and
bioavailability of pesticides.
• The PTI assumes that pesticide toxicity is additive and there is no
chemical interaction which may not be the case for complex mixtures of
pesticides from different chemical classes and with different MOAs across
39
all taxonomic groups and life stages.
• The PTI does not take into account the dose–response curves of either
single-chemical or mixtures exposures.
• The PTI is limited to pesticides measured in the water column;
hydrophobic pesticides may be underrepresented in terms of potential
toxicity, especially to benthic organisms.
• Uncertainty in the relative toxicity of compounds is high for pesticides
with relatively few bioassays available. The 10,837 bioassays in this data
set are divided among 440 pesticides and 52 degradates, 559 different
species, and three taxonomic groups, making the number in each group
relatively small. Although this does not preclude the use of the data as
the best available, it demonstrates the sparseness of available data on
the toxicity of many currently used pesticides.
• The PTI is a relative, but quantitative, indicator of potential toxicity that
can be used in study design or to interpret water quality data, relate
pesticide exposure to biological condition, and prioritize future
assessments
IDEAS FOR IMPROVEMENT OF METHODOLOGY:
A more rigorous test of the PTI model is needed, but this will require the
availability of data for pesticides from multiple classes and MOA,
concurrent with data on aquatic toxicity and(or) ecological condition.
40
ID
3
Title
A European model and case studies for aggregate exposure assessment
of pesticides (Kennedy et al., 2015)
Journal
Food and Chemical Toxicology
Authors
Marc C. Kennedy, C. Richard Glass, Bas Bokkers, Andy D.M. Hart, Paul Y.
Hamey,
Johannes W. Kruisselbrink, Waldo J. de Boer, Hilko van der Voet, David
G. Garthwaite, Jacob D. van Klaveren
Year
2015
Background
To assess aggregated exposure and risk to pesticides a new aggregate
&
model/general framework is described, which allows individual users to
Objectives
include as much, or as little, information as is available or relevant for
their particular scenario. Depending on the inputs provided, the outputs
can range from simple deterministic values through to probabilistic
analyses including characterizations of variability and uncertainty.
Exposures can be calculated for multiple compounds, routes and sources
of exposure, and this aggregate model links to the cumulative dietary
exposure model developed in parallel. It is implemented in the web-based
software tool MCRA. This work presents case studies to illustrate the
potential of this model, with inputs drawn from existing European data
sources and models.
Substances
Pesticides mixtures (active ingredients and degradates) from the conazole
group
Exposure
Human exposure:
Scenario
exposure to UK arable spray operators, Italian vineyard spray operators,
Netherlands users of a consumer spray and UK bystanders/residents, and
finally a hypothetical population performing a combination of these
activities.
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Human exposure to pesticides within different scenarios: aggregate
to
exposure combines dietary and non-dietary sources and example of
WHO/IPCS
exposure are occupational farming activities, use of amateur or
mixture
consumer products, or incidental exposures experienced by residents
assessment
or bystanders; key components known; no data available on the
framework)
hazard of the mixture itself.
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
Yes, humans are exposed to pesticides via dietary and non-dietary
routes, being consumer, operator, worker or bystander.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Compounds are grouped into a cumulative assessment group (CAGs)
41
if they have a similar toxicological effect.
Information Toxicity data: from the literature or pesticides registration data
sources
Exposure: Already existing models, databases and Software have been
used (Operator activities: EUROPOEM Databases; Worker activities: BEAT
Models and ART; exposure from non-professional uses: ConsExpo
Software; Bystander and resident activities: BREAM models).
MIXTURE ASSESSMENT/METHODOLOGY
HRA
Exposure assessment:
The model implies several step:
• To define an exposure question (selection of an appropriate
population, health effects and relevant compounds)
• The estimation of non-dietary exposure from one or more activities
• Matching non-dietary exposures with dietary exposures at the
individual level
• Aggregation of those exposures to an appropriate common unit. Each
compound is aggregated separately, before a suitable weighted sum is
derived to give a total exposure. The weights are derived based on
relative potency factors (RPFs, toxic potencies expressed relative to a
selected index compound)
• If a chronic assessment is required, average daily exposure is
calculated per individual. In the acute case, exposure values per
individual day should be calculated.
Simple absorption factors are used rather than more detailed
dosimetry/toxicokinetic modelling, as they are more compatible with the
available data in EU.
• The outputs available from the aggregate model provide estimates of
Overall
the relative exposure from various sources, which may be more
summary of
effective for communication. A comparison of risks is easier to process
outcome
than an individual exposure or probability value in isolation
• Those case studies demonstrate how the relative contributions to
exposure can be shown to differ between particular scenarios and
populations. For example, based on those hypothetical scenarios the
main routes of exposure are seen to be inhalation for the spray user,
and dermal for the UK operator; for child bystanders, exposure
through non-dietary dermal exposure is estimated to be small
compared with dietary exposure
• Alterative model specifications can greatly influence the results
• When interpreting the results, care must be taken to recognize
possible differences in the degree of conservatism between dietary
and non-dietary exposure models
• Data gap: A useful extension of this case study would be to obtain
Future
information about realistic frequency of use of this type of products
perspec-
by amateur, and this would be essential for chronic assessments
tives /
• To reliably assess exposure related to some activities, survey of non-
Outlook
dietary activities would be required
• The model could also be adapted to handle non-PPP compounds, if
they can be weighted relative to some reference compounds
• A significant challenge in this area is the communication of risks and
probabilities
• In future assessment, the selected scenario parameters and
distributions would require specific and detailed justification,
42
regarding their impact on the results obtained.
• Further refinements will be made based on feedback from stakeholder
groups testing and using the model in practice. Particular computation
issues may arise as larger CAGs become available and are included
43
ID
4
Title
Cumulative dietary exposure to a selected group of pesticides of the
triazole group in different European countries according to the EFSA
guidance on probabilistic modelling (Boon et al., 2015).
Journal
Food and Chemical Toxicology
Authors
Polly E. Boon, Gerda van Donkersgoed, Despo Christodoulou, Amélie
Crépet, Laura D’Addezio, Virginie Desvignes, Bengt-Göran Ericsson,
Francesco Galimberti, Eleni Ioannou-Kakouri, Bodil Hamborg Jensen,
Irena Rehurkova, Josselin Rety, Jiri Ruprich, Salomon Sand, Claire
Stephenson, Anita Strömberg, Aida Turrini, Hilko van der Voet, Popi
Ziegler, Paul Hamey, Jacob D. van Klaveren
Year
2014
Background A cumulative dietary exposure assessment according to the
&
requirements of the EFSA guidance (EFSA Panel on Plant Protection
Objectives
Products and their residues (PPR), 2012) on probabilistic modelling was
performed in order to assess the practicality of the guidance.
Substances
Pesticides residues mixture from the triazole group
Exposure
Human exposure
via food consumption
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Human exposure to pesticides via food consumption; key
to
components known; no hazard data available on the mixture itself.
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
Yes, human are frequently exposed to mixture of pesticides via food
assessment
consumption
framework) 3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Compounds are grouped into a cumulative assessment group (CAGs)
if they have a similar toxicological effect.
Information Exposure: The acute and chronic cumulative exposure to triazole
sources
pesticides was estimated using national food consumption data part of
the Comprehensive database of EFSA, and monitoring data on pesticide
residue of eight European countries in the period 2007-2010 (acute
exposure) and 2 countries for chronic exposure.
MIXTURE ASSESSMENT/METHODOLOGY
HRA
Exposure assessment:
• Both the acute and chronic cumulative dietary exposures were
44
calculated according to two model runs (optimistic and pessimistic)
as recommended in the EFSA guidance: national food consumption
data were combined with national monitoring data, including
available information on the effect of processing on pesticide
residues if appropriate (those data were coming from the German
Database developed by the federal Institute for Risk Assessment).
• Information on unit variability was included in the pessimistic model
run. Additionally, for two countries the chronic cumulative exposure
was calculated for the group of triazole pesticides of the chronic CAG
(chronic effect: hepatoxicity) according to both model runs.
• Calculations were performed with the Monte Carlo Risk Assessment
(MCRA) software, developed to facilitate cumulative exposure
assessments. Those calculations followed the EFSA guidance
procedure for performing an acute or chronic cumulative
assessment, and consist in the conversion of single compound
concentration databases to cumulative concentration database
containing cumulative residues levels per sample, by using Relative
Potency Factors.
• Acute exposure: In the optimistic model run, none of the simulated
Overall
exposures per country and age class exceeded the ARfD (acute
summary of
reference dose), whereas in the pessimistic model run person-days
outcome
with simulated exposures exceeding the ARfD were observed for IT:
10 person-days per million in adults and 20 in adolescents
• Chronic exposure: In the optimistic model run no exposures were
simulated exceeding the ADI. In the pessimistic model run, 6,% and
4,3% of the population had a simulated chronic exposure that
exceeded the ADI in Denmark and Italy respectively, with a 97.5%
upper confidence limit of 71,900/48,900. The P99.9 of chronic
exposure exceeded in both countries the ADI in the pessimistic
model run.
• The exposures obtained with these model runs differed substantially
for all countries, with the highest exposures obtained with the
pessimistic model run. In this model, animal commodities including
cattle milk and different meat types (entered in the exposure
calculations at the level of the maximum residue limit, MRL),
contributed most to the exposure.
• In this study the uncertainties due to sampling uncertainty of the
food consumption and residue concentration data were quantified in
both models run via central 95% confidence intervals around the
number of person-days or persons exceeding a toxicological
reference value and around the P99.9 of exposure. However,
exposure assessments are affected by many other uncertainties
(e.g. food conversion factors, monitoring data…) which should also
be evaluated. This evaluation, based on the experience of the
authors, was therefore subjective.
• The authors conclude that application of the optimistic model run on
a routine basis for cumulative assessments is feasible; however, the
resulting exposure estimates are very likely underestimates of the
real exposure.
• The pessimistic model run is laborious and the exposure results
could be too far from reality.
Future
•
Differences in exposures results between models and countries are
very likely due to the dissimilarities in the approaches/models taken
perspec-
tives /
•
The link with processing information should be improved to further
optimize the application of the optimistic model run.
Outlook
• More experience with the pessimistic model run is needed to
45
stimulate the discussion of the feasibility of all the requirements,
especially the inclusion of MRLs of animal commodities which seem
to result in unrealistic conclusions regarding their contribution to the
dietary exposure. Furthermore, tools are needed to standardised
pessimistic residue database.
• A database with authorised uses of pesticides worldwide that will be
updated and maintained over the years would be needed to make it
feasible to perform CA according to the pessimistic model on a
routine basis.
• The use of common effects in CRA with much higher reference
values than the most sensitive effect of the index compound may
result in conclusions that are contrary to past conclusions based on
single compound assessments. Risk assessors and managers should
keep this in mind when evaluating the outcomes of cumulative
exposure assessments.
• More experience is needed with some kind of intermediate 'realistic'
scenario combining the optimistic and pessimistic model run in such
a way that it results in more realistic acute and chronic exposures
which would be conservative enough (precautionary principle) but
not over-conservative such as the pessimistic model run.
46
ID
5
Title
Examining the feasibility of mixture risk assessment: A case study using
a tiered approach with data of 67 pesticides from the joint FAO/WHO
meeting on pesticide residues (JMPR)
Journal
Food and Chemical Toxicology
Authors
Evans RM, Scholze M, Kortenkamp A.
Year
2015
• Case study to illustrate the application of the WHO/ICPS
Background
framework for MRA
& Objectives
• Applied to a mixture of 67 pesticides, going through the tiered
approach
• Illustrate data needs and gaps for refinements at the different
tiers
Substances
67 pesticides
Exposure
Exposure to pesticides via food residues
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Dietary exposure to pesticides, based on likely exposure to individual
WHO/IPCS
pesticides.
mixture
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
assessment
International estimated daily intakes were used for the individual
framework)
pesticides,
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes, co-exposure via multiple food residues is possible, even if only
theoretically assumed in this case.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No classical grouping based on common effects/MoA was applied, but
a surrogate based on PPDB health issue categories was performed.
• JMPR reports reporting Acceptable Daily intakes (ADIs) and
Information
International estimated daily intakes (IEDIs)
sources
• Analysis also done for 13 WHO food cluster diet regions
MIXTURE ASSESSMENT/METHODOLOGY
• HI approach according to WHO/ICPS tiered scheme
HRA
• For Tier 0, all pesticides were classified as Cramer Class III, with a
TTC of 90 µg/person per day.
• For Tier 1, HI calculated using ADI values
• Tier 2 calculation based on specific endpoints not feasible due to
limitation in relevant data availability. However, a Tier 2 like
refinement with a surrogate data set was performed.
• A low-tier assessment identified a potential risk. For the 67
Overall
pesticides HI>1 was calculated for all 13 food cluster diet regions
summary of
and exceeded 10 in one region (range for all 13 regions was 2.8-11).
47
outcome
The HI was never driven by just 1 chemical. 80% of the HI are
contributed by each 9-18 chemicals in the mixture.
• A tier 0 assessment was performed even if not needed due to the
availability of ADI values, to investigate the differences in the
resulting HI. HI values based completely on TTC ranged 37.5-146
and were up to 16 times greater than ADI-based HI calculations.
• Tier 2 refinement was not possible due to a lack of relevant input
data for the refinement, however, a surrogate refinement based on
human health issues categories of the Pesticide Property Database
(PPDB) was performed.
• In lower tiers, investigating further the individual HQs allows to
Future
identify the drivers of the mixture risk (chemicals contributing most
perspec-
to combined effect)
tives /
• Data requirements in higher tiers are high and relevant input often
Outlook
unavailable, which represents a major obstacle in MRA
• In this case study, an HI>1 would be reached if depending on the
food cluster diet region co-exposure would occur to a minimum of 6-
24 compounds assuming for each the average HQ individually.
• In such a mixture, not all chemicals will have a common effect and
contribute to a combined effect; it is however not implausible that 6-
7 compounds in a mixture of 67 compounds might have a common
effect.
48
ID
6
Title
Pesticides in the Ebro River basin: Occurrence and risk
assessment
Journal
Environmental Pollution 211:414-424
Authors
Alexander Ccanccapa, Ana Masiá, Alícia Navarro-Ortega, Yolanda Picó,
Damià Barceló
Year
2016
Background
Previous studies performed in the Ebro River linking occurrence of
& Objectives pollutants, concentrations and toxicity, but most of them have focused
on a single chemical family or select one environmental matrix (water,
soils, sediments or biota). The objective of this study was to establish
pesticide's occurrence, spatial distribution and transport and to evaluate
the ecotoxicological risk in three trophic levels (Algae, daphnia and
fish), using RQs for each pesticide and sumTUs for each sampling site.
Substances
Pesticides: 42 and some of their degradation products
Azol
(Imazalil,
Prochloraz),
Benzimidazole
(Carbendazim,
Thiabendazole),
Carbamates
(3-hydroxycarbofuran,
Methiocarb),
Chloroacetanilide (Metoalachlor),
Juvenile Hormone Mimics (
Pyriproxyphen),
Neonicotinoid (Imidacloprid
), Organophosphorus (Azinphos
Methyl,
Chlorfenvinphos,
Chlorpyrifos,
Diazinon,
Diclofenthion, Dimethoate, Fenitrothion, Fenoxon, Fenoxon Sulfone,
Fenoxon Sulfoxide, Fenthion, Fenthion Sulfone, Fenthion Sulfoxide,
Malathion, Omethoate, Parathion-Ethyl, Parathion-Methyl, Tolclophos-
Methyl),
Other Pesticides (Buprofezin, Hexythiazox),
Triazines (Atrazine, Deisopropylatrazine, Deethylatrazine, Propazine, Simazine,
Terbumeton, Terbumeton-Deethyl, Terbuthylazine
Terbuthylazine Deethyl, Terbuthylazine-2 Hydroxy, Terbutryn),
Triazole (Tebuconazole),
Ureas (Diuron, Isoproturon)
Exposure
Exposure (to biota: fish, algae and Daphnia) via water. Sediment
Scenario
concentrations are used to predict pore water concentrations.
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Chemical exposure is estimated based on chemical analysis of the
WHO/IPCS
matrices water and sediment. Other components might be present.
mixture
Based on the compounds analysed, the sumTU was calculated,
assessment
based on the acute toxicity values. If possible, also the Risk
framework)
Quotients was calculated
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
Yes, because the analysis is referring to water, and pore water in the
sediment. Exposure is very likely for fish, algae and Daphnids.
Chemicals analysis in fish also show the relevance of exposure.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes. All compounds are analysed in the environment, and exposure
49
at the same time is very likely.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
All compounds were taken together, based on acute toxicity.
Although in the presentation of the compounds they were grouped
according to family or mode of action, this is not taking into account
in the summation of the effect.
Information
• Concentrations in water and sediment were measured.
sources
• EC50 values collected from the PPDB database
http://sitem.herts.ac.uk/aeru/ppdb/en/atoz.htm
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Exposure is assumed to be via water. Water concentrations are
Assessment
measured directly, while concentrations in the sediment are used to
calculate the pore water concentration based on the partitioning
coefficient Kd (Cpw = Cs/Kd) where Kd = Koc x foc
Hazard
Acute 48 h EC50 for D. magna, 72 h EC50 for algae and 96 h LC50 for
Assessment
fish, as well as Chronic 96 h NOEC data for algae and 21 days NOEC for
fish and D. magna of each chemical was collected from Hazard is based
on acute toxicity values.
ERA
The calculated sumTU is the sum of all the individual TUs which are
calculated by TU = ci/EC50i.
To evaluate the impact of the pesticides on the Ebro River basin
ecosystems, the risk quotient (RQ) method was used employing,
whenever possible, the NOEC values obtained from chronic toxicity tests
for producing the corresponding PNECs. (RQ = EC/PNEC).
ERA was performed for fish, algae and Daphnia.
Overall
The obtained Sum TUs for water and sediment were <1 in all sites,
summary of
evidencing that there is no acute risk associated with pollution either in
outcome
water or sediments. The Toxic Unit for water and sediments showed the
daphnia was the most sensitive (in 2010).
Several pesticides showed a RQ > 1 indicating that pesticide risk to the
aquatic communities needs further study.
Future
A long-term chronic study on assessment of these mixtures is required.
perspec-
Not all chronic effects could be calculated due to missing information
tives /
(NOECs)
Outlook
50
A.2 Phthalates
ID
7
Title
Estimated daily intake and cumulative RA of phthalate diesters in a
Belgian general population (Dewalque et al., 2014).
Journal
Toxicology Letters
Authors
Lucas Dewalque, Corinne Charlier, Catherine Pirard
Year
2014
Background The 5 phthalate diesters taken into consideration in this work are known
&
to exhibit ED properties, especially anti-androgenic effects. The aims of
Objectives
this study were (1) to estimate, in a Belgian general population, the daily
intake (DI) of those phthalates based on their urinary measurement, (2)
to investigate the diet contribution to the total exposure, (3) to assess
the risk of exposure to phthalates by comparing their intake to well-
recognized reference values, (4) to assess the risk of cumulative
exposure based on anti-androgenic endpoints to several phthalate
compounds and (5) finally to compare the risk assessment results in
adults and children.
Substances
Phthalates diesters: diethyl phthalate (DEP), di-n-butyl phthalate (DnBP),
di-iso-butyl phthalate (DiBP), butylbenzyl phthalate (BBzP) and di-2-
ethylhexyl phthalate (DEHP)
Exposure
Human exposure to phthalates from food consumption and other sources
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Exposure of human to phthalate via food consumption and other
to
sources. The key components are known. No data on the hazard of
WHO/IPCS
the mixture itself.
mixture
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
assessment
Yes, exposure data comes from biomonitoring data.
framework) 3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Phthalates are a structurally similar group of chemicals that have
been shown to exhibit similar toxicological action, thus additive effect
should be expected when considering this assessment group.
Information Exposure data: measurement from biomonitoring study
sources
Toxicity value: EFSA TDI and RVs from literature
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Daily intake (DI) was based on the urinary measurements of the
51
Assessment corresponding metabolites, and estimated using the volumetric model
developed by Knoch et al (2003)
Hazard
Reference value chosen were tolerable daily intakes (TDI) for phthalates
Assessment established by EFSA or a reference dose for anti-androgenicity (RfD AA)
recently developed
HRA
The HI approaches was used:
-HQ was calculated (HQ=DI/TDI)
-HI=ΣHQ
• Although very few participants exceeded the threshold of 1
Overall
considered as safe for individual HQ, 6.2% of the adults and 25% of
summary of
the children showed a HI
outcome
TDI higher than 1. These high HI values
warranted further investigations since several studies suggested that
anti-androgenic effects of phthalate exposure on reproductive health
could occur at all life stages and because phthalates are not the only
anti-androgenic chemicals to which humans are exposed.
• The HITDI was 3-4 fold higher than HI RfDAA showing that CEA results
are very dependent of the reference value taken into account
• This biomonitoring approach has relevant advantages, such as
Future
integrating all routes and sources of exposure, and avoiding the
perspec-
external contamination due to the widespread presence of the
tives /
phthalate diesters in the lab environment. However, it does not
Outlook
provide detailed information concerning exposure pathways.
• DEHP would be the only phthalate congener studied for which the
main contributor to the daily exposure would be the ingestion of food.
For all other congeners, dietary intake seemed to be a minor pathway
of exposure, suggesting that other routes should occur.
• The TDI and RfD AA determination were based on animals exposed
by gavage and therefore did not take into account other route of
exposure. The estimation of DI was based on urinary levels measured
in spot urine samples and extrapolated to a daily excretion with an
estimated urine volume excreted daily. This also implies that these
spot samples were considered as representative in terms of daily
phosphate levels excreted while more and more studies highlighted
the within-person variability of the urinary levels for these
compounds.
• Larger biomonitoring studies including pertinent biomarkers of
exposure of other anti-androgenic compounds should be performed.
52
ID
8
Title
Human biomonitoring of phthalate exposure in Austrian children
and adults and cumulative risk assessment
Journal
International Journal of Hygiene and Environmental Health
Authors
Christina Hartmann, Maria Uhl, Stefan Weiss, Holger M. Koch, Sigrid
Scharf, Jürgen König
Year
2015
Background Assessment of population exposure to phthalates used in consumer
&
products through a biomonitoring campaign, estimation of daily intake,
Objectives
estimation of cumulative risk assessment.
Substances
14 metabolites of 11 parent phthalate compounds
Exposure
Exposure through consumer products (and home environment) is
Scenario
assumed. Daily intakes are calculated from measured metabolites
concentration in urine.
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Exposure occurs mainly though consumer products and house dust.
WHO/IPCS
Differences in urine levels show that environmental exposure matters
mixture
as well as a difference in phthalates metabolites concentrations is
assessment
observed between samples collected from (sub)urban and rural
framework)
areas.
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
Yes. It occurs through different consumer products (e.g. toys, school
supplies, plastic gloves, or paints, as well as food and cosmetic
products)
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes. Metabolites of different parent compounds were detected in the
same population group, showing that consumer products imply
exposure to a mixture of phthalates.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Antiandrogenic activity
• Individual phthalates daily intakes are estimated from metabolites
Case
detected in urine samples and compared with acceptable exposure
Specific
levels
Information
• TDI and reference dose for anti-androgenic activity are used for te
sources
calculation of the HI for each population class (adults, children,
elderly)
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Total daily intake is calculated depending on metabolites concentrations
Assessment detected in urine
53
Hazard
Dose-addition by using the hazard index for the anti-androgenic
Assessment phthalates related to the Reference Dose for Anti-Androgenicity or
related to the Tolerable Daily Intake.
HRA
Cumulative risk assessment is calculated through the HI for anti-
androgenic phthalates.
Overall
Median HIs based on all acceptable levels of exposure used are far below
summary of the value of 1. The highest values were identified among children,
outcome
Exceedances of the HI of 1 existed among all age groups for tolerable
daily intake based values, whereas no exceedance was identified for the
reference dose for anti-androgenicity (reference doses for anti-
androgenicity are higher than tolerable daily intake). Authors report that
assuming other exposure to androgenic chemicals (e.g. pesticides
residues and cosmetic products) there is potential indication of cause of
concern.
Future
Inclusion of a larger set of phthalates secondary metabolites.
perspec-
tives /
Outlook
54
A.3 PBDEs
ID
9
Title
Example Case study A: PBDEs - Annex A (Meek et al., 2011)
Journal
Regulatory Toxicology and Pharmacology 60 S1-S14
Authors
Bette Meek
Year
2011
Background
A screening level RA of PBDEs was conducted under the Canadian
& Objectives Environmental Protection Act and slightly modified to illustrate the
WHO/IPCS framework for combined exposure to multiple chemicals (Tier
0 and Tier 1).
Substances
Polybrominated diphenyl ethers (PBDEs) used as flame retardants in a
wide variety of consumer products; three main commercial mixtures
containing
seven
isomers
were
subject
of
assessment:
pentabromodiphenyl ether (PeBDE), or ComPeBDE (usually containing a
mixture of PBDEs with 4–6 bromines); commercial octabromodiphenyl
ether, (OcBDE), or ComOcBDE (usually containing a mixture of PBDEs
with 6–9 bromines); and commercial decabromodiphenyl ether
(DeBDE), or ComDeBDE (usually containing PBDEs with 9–10 bromines)
Exposure
Exposure of general population through consumer products and via the
Scenario
environment
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Focus of the case study on exposure of the population in the general
WHO/IPCS
environment including through consumer products. The majority of
mixture
data relevant of human health risk relate to commercial mixtures
assessment
with much less information on individual congeners.
framework)
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
Yes. Direct contact via consumer products containing PBDEs is
possible, also via the environment through the use and disposal of
PBDEs.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes. There is overlap in congeners within the commercial mixtures
and reason to believe that their kinetics will be similar based on
similarity in physicochemical properties.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
The assessment group contains seven isomers with identical base
structure, overlap in congeners within the commercial mixtures,
similarities in uses and common target organs. Trends in physic-
chemical properties and toxicity vary consistently with increasing
55
degree of bromination.
• Exposure data available from the assessment under the Canadian
Information
Environmental Protection Act (Tier 0); for Tier 1 estimated from
sources
available data
• Hazard data no tolerable intakes or concentrations were available
(Tier 0); for Tier 1 from literature.
MIXTURE ASSESSMENT/METHODOLOGY
Tier 0
Tier 1
• Limited data were available, therefore
Exposure
Semiquantitative measure
a conservative estimate was based on
Assessment
available from Canadian
maximum levels in air, water, dust,
Assessment;
determined
food, human breast milk
based
on
volume
of • Standard intake values for six age
production, numbers of
groups in the Canadian population.
producing
and
using • Thus upper-bound estimates of daily-
companies, and the sum of
intake of total PBDEs were estimated.
"expert
ranked
uses"
(based on potential to Estimates considered conservative since
create exposure for each they were based on summed estimates of
use)
all congeners for which data were
available
and
highest
measured
concentrations for many media.
• Most toxicity data found were for
Hazard
No
reference
tolerable
commercial mixtures, less for the
Assessment
intakes or concentrations
individual congeners;
for
relevant
congeners • From all data the critical effect level
were available, thus a
was selected at 0.8,g/kg body weight
hazard index could not be
(PeBDE) based on neurobehavioural
developed;
as
a
effects
conservative measure, the
LOEL for the most toxic
congener was considered.
• Comparison of critical effect level with
HRA
The
sum
of
upper-bound estimate of exposure to
semiquantitative estimates
total PBDEs for the potentially most
of exposure exceeded the
exposed subgroup.
LOEL of the most toxic • Resulted in Margin of Exposure of
congener, so additional
approximately 300.
assessment
was • Margins based on available
considered
biomonitoring data were approx. 10-
necessary Tier1
fold less, but less confident due to
uncertainty in back-calculation of
exposure from biomonitoring data
• Food represented the principal source of exposure for the majority
Overall
of age groups, highest for breastfed infants with breast milk
summary of
accounting for 92% of the exposure. Estimates of intake from
outcome
dermal contact with dust or oral contact with household products
were negligible in comparison to uptake via food.
• Uncertainties: Degree of conservatism in the derived margin is
relevant to its interpretation. One critical aspect is the large
interindividual variability in levels of PBDEs in breast milk (mean
and median levels observed in the general population were 400 and
200 fold less, respectively, than the maximum levels on which the
exposure estimate was based. The hazard was based on the most
56
sensitive effect for the most toxic congener. In other studies the
effect levels were 100 times higher than the one used in this
assessment. However, continuing increase in body burden was not
considered due to limited information availability.
• In view of the smaller margin between the most conservative
Future
estimated critical values for exposure and effects on the
perspec-
environment in comparison with that for human health and resulting
tives /
recommended action to protect the environment, in-depth
Outlook
evaluation of PBDEs from a human health perspective was
considered a low priority at this time.
57
A.4 Parabens
ID
10
Title
Aggregate exposure approaches for parabens in personal care products:
a case assessment for children between 0 and 3 years old (Gosens et
al., 2013).
Journal
Journal of Exposure Science and Environmental Epidemiology
Authors
Ilse Gosens, Christiaan J.E. Delmaar, Wouter ter Burg, Cees de Heer and
A. Gerlienke Schuur
Year
2013
Background
Many chemical substances in consumer products are used in multiple
& Objectives product categories, leading to multiple source of exposure, but in risk
assessment, aggregation of exposures from different sources is not
common practice, especially when these sources are regulated under
different legal frameworks.
Objective is to assess aggregate exposure (exposure to a substance
from different sources via different pathways) to the four most common
parabens in personal care products for children between 0 and 3 years
old. A deterministic approach with conservative assumptions (tiers 1)
followed by a person-oriented probabilistic (tier 2) approach for
exposure assessment was applied, to gain more insight into the
feasibility and necessity of refining an aggregate exposure approach.
Parabens are used in a wide variety of products: personal care products
for adults and children, in consumer products such as dog shampoo, in
pharmaceutical products such as antibiotics and they are used as food
additives.
Given the estrogenic effects of parabens and the potential severity of
the effects during early human child development, the aggregate
exposure for children between 0 and 3 years of age was assessed.
Substances
Methyl-, ethyl-, propyl- and butylparaben.
Exposure
Human exposure to personal care product
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Human exposure, oral and dermal absorption.
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
assessment
Yes. The parabens considered are the 4 most frequent paraben in
framework)
personal care product from children from 0 to 3 years old.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
This study considers aggregate exposure (several sources) of one
compound and therefore does not include co-exposure.
58
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No assessment group
• Exposure, Tier 1: Product composition and ConsExpo default value
Information
• Exposure, Tier 2: more detailed information on product use has been
sources
obtained from a small survey on product use of consumers.
• Toxicity: NOAEL
MIXTURE ASSESSMENT/METHODOLOGY
• Tier 1: worst case, deterministic approach
Exposure
Assessment
Parameters used for the exposure calculations: 1) maximum amount of
paraben found in a product, 2) default use amounts of PCP as
reasonable worst case estimates from ConsExpo 3) ConsExpo defaults
of frequency of use as reasonable worst-case estimate. When
application on body surface area was involved, the default value was
extrapolated to children using a correction factor that account for the
smaller total body surface area of children.
• Tier 2: Person-oriented probabilistic approach is performed to
estimate the variability and uncertainty of exposure in a population.
Raw data on weight fraction measurements in 12 product types by the
Dutch Food and Product Safety Authority and information from a pilot
survey have been used to estimate exposure. The aggregate exposure
per day is determined by adding all exposure on the same day for one
person and subsequently averaging the daily aggregate exposure for
each individual. The result is a distribution of the daily average
aggregate exposure for all persons in the population.
Hazard
NOAEL
Assessment
HRA
Percentiles of the distribution of exposure can be compared against the
NOAEL. It gives an indication on the fraction of the population with
average exposures above a certain Margin of Exposure (MoE).
• The internal exposure for each paraben calculated in Tiers 2 is below
Overall
the level determined in Tier 1. However, for propyl- and
summary of
butylparaben, the percentile of the population with an exposure
outcome
probability above the assumed ‘‘safe’’ MoE of 100, is 13% and 7%,
respectively (MoE of 8 and 10 respectively) indicating that further
evaluation of the exposure calculations is necessary.
• In conclusion, a Tier 1 approach can be performed using simple
equations and default point estimates, and serves as a starting point
for exposure and risk assessment. If refinement is required, the
more data demanding person-oriented probabilistic approach should
be used. This probabilistic approach results in a more realistic
exposure
estimate,
including
the
uncertainty,
and
allows
determining the main drivers of exposure. Furthermore, it allows to
estimate the percentage of the population for which exposure is
likely to be above a specific value.
• Refinement is difficult as detailed data on the use of PCP is scarce,
Future
and it is unknown whether extrapolation from adult use by scaling
perspec-
the amount of product used to body surface is appropriate.
tives /
• Steps need to be taken before aggregate exposure can be assessed
Outlook
routinely: it would be useful to perform an extended personal care
59
product use survey for children
• Uncertainty in the exposure assessment for propyl- and
butylparaben could be reduced by collecting more suitable data.
• Pharmaceutical products contributed as the second largest product
group toward the total paraben exposure. More exposure data via
these products would be needed to obtain an even more accurate
aggregate estimate
60
A.5 Pharmaceuticals
ID
11
Title
Screening level mixture RA of pharmaceuticals in STP effluents
(Backhaus & Karlsson, 2014a).
Journal
Water Research
Authors
Thomas Backhaus, Maja Karlsson
Year
2014
Background Pharmaceuticals do not occur as isolated, pure substances in an
&
environmental compartment. A broad range of different substances is
Objectives
used simultaneously in human and veterinary medicine, hence
pharmaceuticals often occur in the environment as multi-component
mixtures. The joint ecotoxicity of such chemical cocktails is typically
higher than the toxicity of each individual compound, and even if the
compounds of a mixture are present only below their respective toxicity
threshold, a joint toxic effect cannot be ruled out
a priori. Both
approaches of the mixture quotient and of the STUs were used for
providing a screening level assessment of the environmental risks of
pharmaceutical mixtures previously determined in European sewage
treatment plant effluents.
The aim was to determine whether the detected pharmaceutical
cocktails might pose a risk to aquatic organisms, how this relates to the
toxicities of the individual pharmaceuticals, which group of organisms
(trophic levels) is most sensitive and which are the ecotoxicologically
most important compounds. Standard regulatory environmental risk
assessment approaches for individual pharmaceuticals were followed as
closely as possible.
Substances
Pharmaceuticals
Exposure
Exposure of aquatics organisms from sewage treatment plant (STP)
Scenario
effluents
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Exposure of freshwater organisms; key components known from
to
previous published data; no information on the hazards of the
WHO/IPCS
mixture itself.
mixture
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
assessment
Yes, mixture of pharmaceuticals are frequent in freshwater system
framework) 3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No assessment groups, mixture contains compounds with similar
and dissimilar mode of action
61
Information Exposure data are based on a comparative exposure assessment of a
sources
range of pharmaceuticals previously published in the literature.
Hazard data were compiled in the published literature and/or database.
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Exposure is based on previously published data on pharmaceuticals
Assessment mixtures: data analysis of 26 pharmaceuticals present in 7 STP
effluents was used as a basis.
Hazard
Toxicity data for chemicals were compiled from reviews, electronic
Assessment databases, and if needed, primary literature.
The European Medicines Agency guideline for the ERA of human
pharmaceuticals (EMA, 2006) stated that environmental hazard
assessments of pharmaceuticals should be based on chronic data, using
an AF of 100 or lower. However, such chronic data are only available
for a minority of the pharmaceuticals used in this work; the assessment
was then based on acute data for algae, daphnids and fish, following
the approach outlined in the REACH guidance document to estimate a
PNEC on the basis of acute data, using an AF of 1000 (ECHA, 2008).
If more than one EC50 was available for a given compound, the lowest
value found for that particular species group was used.
If no experimental toxicity data were found for a given trophic level,
QSAR estimates were used for the EC50 values, assuming a common
MOA of compounds from a similar chemical class. Differences in toxicity
between members of a chemical class are then assumed to be caused
by differences in lipophilicity-driven uptake rates.
RA for
The concept of CA has been used
via two approaches:
aquatic
organisms
1) Estimation of the risk quotient of a mixture:
RQ=Σ (MECs/PNEC)i
MEC: measured environmental concentration
2) Calculation of the sum of toxic units (STU, with a toxic unit being
TU = MEC/EC50) in a first step for each of the main trophic levels
(usually algae, invertebrates, fish).
The final risk quotient (RQSTU) for the mixture equals the sum of toxic
units of the most sensitive trophic level multiplied with the
corresponding AF, which is set to 1000 if data represent EC50 values
from short-term toxicity studies with algae, invertebrates and fish
(ECHA, 2008).
• The risk quotient of a single, randomly selected pharmaceutical is
Overall
often more than a factor of 1000 lower than the mixture risk, clearly
summary of
indicating that a mixture risk assessment is indispensable for an
outcome
environmentally realistic risk assessment when it comes to
pharmaceuticals. The MCR varies between 1.2 and 4.2, depending on
the actual scenario and species group under consideration.
• The mixture risk quotients, based on acute data and an assessment
factor of 1000, regularly exceed 1, indicating a potential risk for the
environment, depending on the dilution in the recipient stream.
• The top 10 mixture components explain more than 95% of the
62
mixture risk in all cases. However, the ranking profile strongly
depends on the considered group of organisms.
• Regarding the relative sensitivity of the three trophic level, algae are
the most sensitive group, followed by invertebrate, fish being always
least sensitive.
• The ratio between the RQMEC/PNEC and RQSTU never exceeds 1.3 for the
7 effluents, if identical assessment factors are used.
Future
• Ignoring Independent Action or using the sum of individual risk
perspec-
quotients as a rough approximation of Concentration Addition does not
tives /
have a major impact on the final risk estimate
Outlook
• The lack of data on the chronic toxicity of most pharmaceuticals as
well as the very few data available for
in vivo fish toxicity has to be
regarded as a major knowledge gap in this context
63
ID
12
Title
Environmental risk assessment of antibiotics including synergistic and
antagonistic combination effects
Journal
Science of the Total Environment
Authors
Conrad Marx, Viktoria Mühlbauer, Peter Krebs, Volker Kuehn
Year
2015
• Aim of this study is to make a solid estimate on the possible
Background
synergistic potential of combined antibiotics
& Objectives • To quantify the subsequent effect for the case of the receiving river
Elbe, Germany.
Substances
Antibiotics
Exposure
Exposure of aquatic organisms in receiving waters of waste water
Scenario
treatment plants (WWTP). Exposures calculated based on 15 most
prescribed antibiotics in the investigated catchment area.
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Exposure of aquatic organisms via receiving waters of WWTPs.
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
Yes, exposure estimates based on antibiotic prescription information
assessment
and known rates of degradation in WWTP.
framework)
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes, co-occurrence in Elbe river.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Share of different categories of antibiotics in the overall HI was
assessed.
• Exposure based on available information on antibiotic prescriptions
Information
and literature data.
sources
• Predicted no effect concentrations (PNEC) from own studies and
literature. Information on binary interactions from literature.
MIXTURE ASSESSMENT/METHODOLOGY
• Exposure was predicted based on ambulant and hospital prescription
Exposure
data in the study area, excretion ratio, WWTP outflow, daily flow of
Assessment
receiving river Elbe, elimination rates in WWTP. Veterinary uses of
antibiotics play minor role in the study area and were therefore
neglected.
• Predicted no effect concentrations (PNEC) from own studies and
Hazard
literature. Information on binary interactions from literature for
Assessment
bacteria, algae and daphnia, aggregated for the different antibiotic
classes.
• HI
RA for
add (based on concentration addition only) and HIint (including
interactions) were calculated.
aquatic
organisms
64
• HI
Overall
add was calculated over a 7 years period with a mean value of
0.37. 20% of all weeks exceeded HI
summary of
add of 0.5, HIadd>1 was calculated
only for 1 week in the 7 years period.
outcome
• The hazard share of different classes of antibiotics changed over
time (for some doubled), while the overall HIadd did not noticeably
change over time.
• Considering HIint showing in a worst-case scenario a 50% risk
increase, the threshold of HI>1 would is exceeded in 25 weeks over
7 years.
• Most underlying data on binary interactions were gained using much
Future
higher than environmentally relevant concentrations and using a
perspec-
variety of organisms (algae, daphnia, bacteria).
tives /
• Some underlying data show that the probability for synergistic
Outlook
interactions increases at lower antibiotic concentration in contrast to
many other studies. In summary, since the concentration influence
on synergisms depends on the target organism and the combination
of substances, no general statement on concentration dependency
can be made for antibiotic mixtures.
• Different scenarios applied in the HIint lead to the conclusion that
antibiotic mixtures tend to exhibit an overall synergistic effect. The
resulting increase was between 20-50%.
65
A.6 Food Contact Materials
ID
13
Title
Assessing the safety of co-exposure to food packaging migrants in food
and water using the maximum cumulative ratio and an established
decision tree (Paul Price et al., 2014)
Journal
Food additives and contaminants: Part A
Authors
Paul Price, Rosemary Zaleski, Hans Hollnagel, Hans Ketelslegers &
Xianglu Han
Year
2014
Background Food contact materials (FCM) can release low levels of multiple
&
chemicals (migrants) into foods and beverages, to which individuals can
Objectives
be exposed through food consumption. This paper investigates the
potential for non-carcinogenic effects from exposure to multiple
migrants using the Cefic Mixtures Ad hoc Team (MIAT) decision tree.
This assessment aims to demonstrate how the decision tree can be
applied to concurrent exposures to multiple migrants using either
hazard or structural data on the specific components, i.e. based on the
acceptable daily intake (ADI) or the threshold of toxicological concern
(TTC).
Substances
Food packaging migrants
Exposure
Human exposure via food and water consumption
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Human exposure via food and human consumption; key
to
components known from previous published data.
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
Yes
assessment
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
framework)
TIME FRAME?
Yes
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No assessment groups
Information Exposure: Data on co-exposure to migrants were previously reported in
sources
a study on non-intentionally added substances (NIAS) eluting from food
contact-grade plastic and two studies of water bottles (one on organic
compounds and the other on ionic forms of various elements).
Toxicity data: Existing ADI value or TTC approach.
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Exposure is based on 3 examples previously published of NIAS eluting
66
Assessment from food contact material (food and water bottle)
Hazard
Reference values were based either on existing ADI or on a TTC
Assessment approach, according to the Cramer class of the chemical, for organics
chemicals. Inorganics without RVs were not considered in the analysis.
HRA
Determination of the HQ of each compounds, of the HI and MCR of the
mixture
• The two first examples were assigned to the group II (low concern)
Overall
by the decision tree.
summary of
• The cumulative olefins and saturated hydrocarbons for the NIAS
outcome
study and ethyl-4-ethoxybenzoate for the water bottles study
provided the largest toxicity of any of the migrants.
• The MCR value in example 1 is greater than 2, but this is impacted
by the fact that the driving components are not single compounds
but each represent a group of compounds falling into a structure-
based chemical class. HIs are not affected by this grouping since
the same RV is applied to all compounds.
• In example 2, a single compound dominates the toxicity of the
mixture (MCR<2).
• The co-exposure assessment indicated that while multiple substance
were extracted from FCM samples, the risk of adverse effects to
individuals was very low (HI<1).
Future
Outcomes :
perspec-
• Those three example show that sufficient data are available for the
tives /
safety evaluation of many co-exposure to migrants that occurs in food
Outlook
and water, except for the inorganics for which 30 RVs were missing on
the 57 chemicals.
• When RVs are not available for organic compounds, the Cramer class
predictions might be used, since the levels of exposure of migrants are
low and often fall below the conservative estimates of RVs produced by
the Cramer class approach.
• The decision tree demonstrated that given the available screening
toxicity data, exposures to the reported migrants both separately and in
combination were unlikely to cause adverse health effects.
• Future work on combined exposure to ionic forms would benefit from
additional toxicity information.
67
A.7 Dioxin-like compounds (DLCs) including dioxins, furans
and PCBs
ID
14
Title
Applying the maximum cumulative ratio methodology to biomonitoring
data on dioxin-like compounds in the general public and two
occupationally exposed populations
Journal
Journal of Exposure Science and Environmental Epidemiology
Authors
Xianglu Han and Paul S. Price
Year
2013
MCR values were calculated for three groups of individuals based on
Background
monitoring data and the WHO toxic equivalency factors (TEFs) for dioxin
& Objectives like compounds (DLCs)
Substances
Dioxin-like compounds (DLCs) including polyhalogenated dioxins,
furans, and polychlorinated biphenyls (PCBs)
Exposure
2 occupationally exposed groups and one group of general population
Scenario
based on human biomonitoring data
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
3 biomonitoring study groups, 2 with relevant occupational
to
exposure, 1 with general exposure
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
assessment
Yes, as this is based on biomonitoring data.
framework)
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes, as this is based on biomonitoring data.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Selection of included chemicals based on common dioxin-like
characteristics
• NHANES biomonitoring data plus 2 biomonitoring studies on
Information
occupational worker exposure
sources
• WHO TEF values
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Lipid-adjusted concentrations of DLCs in serum were used from human
Assessment
biomonitoring studies. NHANES was used for one group of individuals
reflecting current and historical DLC exposure in the general population.
Further 2 groups of workers occupationally exposed to dioxins were
included (trichlorophenol workers in Michigan (MI dataset) and New
Zealand (NZ dataset)).
Hazard
35 DLCs were analysed in the 2 worker groups, but not for all of them
68
Assessment
TEFs are available, thus only 26 DLCs were used in the analysis and
calculation of overall TEQs. This was done by multiplying the serum
levels with the respective TEFs for each individual person.
HRA
Since the investigated mixture components share the same MoA, the
toxic equivalency (TEQ) approach is preferred over the HI approach.
TEFs are used to convert doses of each component into an equivalent
dose of the index chemical 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD).
These equivalent doses are then summed up resulting in a
toxicologically equivalent exposure to TCDD.
The MCR is then calculated as the ratio of the person’s cumulative
toxicologically equivalent exposures for the mixture divided by the
person’s maximum chemical-specific equivalent.
Total TEQ values were calculated for each individual in the studies.
Within each of the 3 study groups, the mean TEQ was calculated for
each DLC and ranked from high to low. Subjects with one or more
missing values for the top 5 chemicals were excluded from the dataset.
Non-detects (NDs) were assumed to be present at the limit of
detection/20.5. Persons where NDs would have contributed >50% to the
MCR or where the primary chemical was a non-detect, were excluded
from the dataset.
• The top five major contributors to total TEQs in the NHANES dataset
Overall
were 12378-PeCDD, 123678-HxCDD, PCB 126, TCDD, and 23478-
summary of
PeCDF. On average they accounted for 76% of the total TEQ.
outcome
• Total TEQs were higher in the MI and NZ datasets than in the
NHANES dataset (58,96 fg/g for MI, 25.5 fg/g for NZ, 19,72 fg7/g for
NHANES). Part of the difference is however also explained by the
different age distributions, i.e. for persons >45 of age, NHANES total
TEQs were lower than in the MI dataset but higher than for the NZ
dataset.
• Average MCR values (including 2.5th percentile and 97.5th percentile)
were: for NHANES 3.5 (2.2/5.7), for MI 3.6 (1.6/5.1), and for NZ 3.2
(1.4/4.6). This indicates that for all 3 groups a small number of DLCs
drives the total TEQ.
• MCR showed decreasing trend with increasing total TEQ values.
Overall more highly exposed people tend to have lower MCR values
for the MI and NZ dataset, but not for NHANES.
• Age and total TEQ are positively correlated. In the NHANES dataset
two groups of age > or < 45 years can be distinguished with persons
< 45 years showing generally lower DCL levels and higher MCR
values.
• For all three groups, the MCR values were larger than in previous
Future
investigations of MCR of different mixtures, indicating a greater need
perspec-
for CRA for DLCs. A single substance RA based on the largest
tives /
contributor only would underestimate the total TEQ by a factor of 2-
Outlook
6.
• In the case of occupational or local sources of exposure, the impact
of performing a CRA compared to single substance RA decreases.
• Only 2-5 of the DLCs make significant contribution to the total TEQ.
It might thus be sufficient to focus the CRA on the 5 highest ranking
DLCs.
69
A.8 Cross-sectorial mixtures from consumer product and
environmental exposure
ID
15
Title
Combined exposures to anti-androgenic chemicals: steps towards
cumulative risk assessment (Kortenkamp & Faust, 2010)
Journal
International Journal of Andrology
Authors
Kortenkamp, A., Faust, M.
Year
2010
Background There is widespread exposure to anti-androgens. Substances of concern
&
include certain phthalates, pesticides and chemicals used in cosmetics
Objectives
and personal care products. However, chemicals risk assessment
normally does not take account of the effects of combined exposure
although
a
disregard
for
combination
effects
may
lead
to
underestimations of risks. For this reason, this work aims at assessing
the feasibility of conducting cumulative risk assessment, where the focus
is on considering the effects of exposure to multiple chemicals, via
multiple routes and pathways.
Substances
Anti-androgenic chemicals: a total of 15 substances including phthalates
and other chemicals
Exposure
Human exposure from all known sources and by all known routes and
Scenario
pathways
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Exposure of human to anti-androgenic chemicals via all kind of
to
exposure. The key components are known. No data on the hazard of
WHO/IPCS
the mixture itself.
mixture
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
assessment
Yes.
framework) 3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
No information
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Phthalates and agents capable of inducing the androgen insufficiency
syndrome were grouped together.
Information •
Human exposure estimates from literature and publicly available
sources
assessment reports.
• Reference doses for anti-androgenicity from literature
MIXTURE ASSESSMENT/METHODOLOGY
• Data were taken from peer-reviewed literature, or from publicly
Exposure
available reports of the European Scientific Committees and
Assessment
international regulatory bodies.
70
• A distinction was made between median human intake value and
intake value for highly exposed population groups.
• Only toxicological endpoint with relevance to anti-androgenicity were
Hazard
considered
Assessment • Dose describing "point of departure" normally used for RA (NOAELs,
Benchmark doses) were taken from the peer reviewed literature and
combined with uncertainty factors to derive acceptable level (AL);
ADI were used when existing and derived from toxicological endpoint
with relevance to anti-androgenicity
HRA
The HI approaches was used:
-HQ was calculated for each chemical i (HQi=ELi/ALi)
-HI=ΣHQ
EL: exposure level
AL: acceptable level (e.g ADI)
• The cumulative risks from anti-androgen exposures exceed
Overall
acceptable levels for people on the upper end of exposure levels. The
summary of
value obtained for median exposures to the 15 substances can be
outcome
judged tolerable (HI=0.38), whereas the value obtained for highly
exposed population reaches 2.01. In this case, butyl paraben alone
made up 50% of the HI.
• Those results suggest that combined exposures to anti-androgens
have reached levels of concern, especially among highly exposed
groups of the population.
• The authors suggest that risk reductions can be achieved by limiting
exposures to the plasticizer diethyl hexyl phthalate, the cosmetic
ingredients butyl- and propyl paraben, the pesticides vinclozolin,
prochloraz and procymidone and bisphenol A.
• One assumption underlying the use of the HI approach is that the
Future
joint action of anti-androgens can be approximated by dose-addition;
perspec-
however synergism was observed with a mixture of androgens with
tives /
diverse mode-of-action for particular endpoints; further work should
Outlook
be done to know whether this is a phenomenon of concern that
should be taken into account in RA.
• The summing of HQs implies that human population is exposed to
each of the listed chemicals at the same time, which might not be
very likely, especially in the high-intake scenario. Information about
the co-occurrence of several chemicals in one and the same
individual would be needed.
• Significant knowledge gaps exist that prevent from arriving at
definitive conclusions, i.e the absence of appropriate
in vivo toxicity
data about large numbers of
in vitro androgen receptor antagonists.
This assessment was restricted to chemicals where information about
in vivo anti-androgenic effects was available, however many more
substances with known human exposure are likely to contribute to
cumulative anti-androgenic risks.
• At this stage, too little is known about correlations between
in vitro
AR anti-agonists and their ability to induce disruption of male sexual
differentiation
in vivo to make meaningful extrapolations
• p,p’-DDE and BDE 99 are highly lipophilic and build up in human
tissues. By using intake values to calculate HQs these accumulating
effects are not taken into consideration, thus the effective internal
dose of these substances may be higher than suggested and the
resulting risks may have been underestimated. To deal with this
71
issue, it would be necessary to employ a different dose metric, and
to relate intake values for all chemicals to their corresponding tissue
concentrations. The data necessary for such calculations are
currently not available.
72
ID
16
Title
An application of a decision tree for assessing effects from exposures to
multiple substances to the assessment of human and ecological effects
from combined exposures to chemicals observed in surface waters and
wastes water effluents
Journal
Environmental Sciences Europe
Authors
Paul Price, Xianglu Han, Marion Junghans, Petra Kunz, Chris Watts, Dean
Leverett
Year
2012
Background In 2010, Cefic has published a decision tree for the RA of chemical
&
mixture, based on concepts taken from a number of published
Objectives
approaches including those developed by the joint group of three non-
food Scientific Committees to the European Commission (SCs), the World
Health
Organisation/International
program
on
Chemical
Safety
(WHO/IPCS), and recent publication on new quantitative tools (Maximum
Cumulative Ratio, MCR), use of the Threshold of Toxicological Concern
(TTC) in the assessment of risk from combined exposure.
This paper applies the CEFIC decision tree to real world examples of
exposures to multiple chemicals, for both human health and
environmental risk assessment.
Substances
559 mixtures analysed for up to 222 substances measured in surface
water samples (362) and effluent samples (197). The samples contained
detectable levels of 2 to 49 substances, reported from water monitoring
programmes in Europe, and include a wide range of inorganics, and polar
and non-polar organic chemicals.
Exposure
Exposure via surface water or effluent from wastewater treatment plants
Scenario
(WWTPs)
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Exposure from surface water or water effluent. Key component
WHO/IPCS
known, according to monitoring data. No data available on the
mixture
hazard of the mixture itself.
assessment
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
framework)
Yes, data are coming from monitoring data
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes, data are coming from monitoring data
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No assessment groups
73
Information •
Exposure: From monitoring data programme in Europe. Include
sources
seven data sets, differing in the number of compounds analysed in
each samples and the water surveyed.
• Reference Values: literature and internet based search.
MIXTURE ASSESSMENT/METHODOLOGY
• Similar approach for HRA and ERA
Exposure
• Conservative assumption:
Assessment
- Sampled surface water is assumed to be used directly as a water
supply; individuals would be exposed from the consumption of drinking
water.
- A 10-fold dilution of the effluent has been assumed before the water
would be used as drinking water.
- A consumption of 2l per day for a 60 kg adult has been assumed.
• It has been assumed that none of the components have non-additive
Hazard
interaction
Assessment
• MOA have not been researched therefore an additive models was
used as the default assumption
HH: If the RVs were not available, the Cramer classes provided an
alternative source of conservative estimate of oral toxicity in order to
determine the HI (WHO Tier 0).
ERA: RVs available to determine HI (WHO Tier 1).
HRA
The HQ/HI approach is used: HQ=Dose/RV; HI=ΣHQ
• If HI>1 the MCR is determined
• For non-detected chemicals (NDs), which might be present at level
<LOD, two assumptions has been made:
-NDs=0
-NDs=LOD/20.5
ERA
Similar approach as for Human RA, with HQ=Concentration/RV
• For HH effects, 2% of the mixtures were of concern, 98% had a
Overall
HI<1. For ERA, 68% of the mixture were of concern with one or
summary of
more substance that had an individual HQ>1, 19% of the mixture
outcome
had a HI<1, and about 12% were predicted to have toxicity of
concern that would not have been identified unless a combined
assessment has been performed (HI>1 but HQ<1). This means that
the HH effects of the combined measured substances would have
been sufficiently addressed by chemical-by-chemical approaches and
had little need for a separate assessment of the combined exposure,
which is not the case for ERA.
• The majority of the toxicity came from one chemical in 44% of the
case (HH) and 60% of the exposure (ERA).
• The tree identified chemicals where data on the MOA would be most
useful in refining an assessment.
• Chemicals with exposure levels exceeding their RVs, which would be
Future
subject to a refined chemical-specific RA, were not considered in this
perspec-
case-study.
tives
/ • The assumption of a 10-fold dilution of the effluents can be wrong
Outlook
for small rivers under low-flow condition; in addition, for rivers
74
receiving multiple discharges the receiving water might already
contain one or more of the compounds from discharges that occur
upstream
75
ID
17
Title
Determining the maximum cumulative ratios for mixtures observed in
ground water wells used as drinking water supplies in the United States
Journal
Environmental Research and Public Health
Authors
Xianglu Han and Paul Price
Year
2011
Background
Data from water samples taken from groundwater wells from the public
& Objectives
water system across the USA 1993-2007 were used. These samples
have been analysed for a wide variety of chemicals including PPPs,
VOCs, metals and other inorganics.
The aim of this study was to further explore the usefulness of the MCR
(Maximum Cumulative Ratio) and to investigate in detail (1) the pattern
of the MCR and its ranges when applied to different types of
samples/mixtures, (2) to explore the relationship between the MCR,
number of substances in a mixture (n) and HI, and (3) to detect the
impact of non-detects on the MCR values.
Substances
Dataset for 932 samples of ground water with measured compounds
(number in brackets) being major ions (11), trace elements (23), PPPs
and their metabolites/degradates (83), and VOCs (85). Not all 200
substances were analysed in all samples. 58 of the 200 substances
were never detected and therefore the cases study focused on the
remaining 142 compounds.
Samples were excluded from further assessment if any of the 3 highest
ranking chemicals in mean HQ was not measured (option 1) or if any of
the first 6 highest ranking chemicals in mean HQ was not measured
(option 2). Furthermore only mixtures including at least 5 compounds
measured were included in the assessment. 2 options to deal with
compounds below the LOD were compared.
Exposure
Exposure via ground water used for human consumption as drinking
Scenario
water without prior treatment as worst case assumption.
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Exposure from groundwater used as drinking water. Key component
WHO/IPCS
known, according to chemical analysis data. No data available on
mixture
the hazard of the mixture itself.
assessment
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
framework)
Yes.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
76
No assessment groups
• Hazard information (permitted doses PD) were taken from US
Information
EPA, ATSDR databases and other sources.
sources
• Exposure data from USGS groundwater monitoring data set
MIXTURE ASSESSMENT/METHODOLOGY
• Assumption:
Exposure
Assessment
- groundwater directly consumed as drinking water (i.e. without prior
treatment)
- drinking water consumption rate 2L/day, 100% oral absorption, body
weight 60 kg
• Chronic RfD (for non-PPPs). For PPPs chronic Population Adjusted
Hazard
Doses (PADs) were used, and acute PADs if no chronic PAD was
Assessment
available.
• No MoA and grouping considered
Concentration addition assumed for all components using HI
• The HQ/HI approach is used : HQ=Dose/PD (PD=permitted dose)
HRA
HI=ΣHQ
• If HI>1 the MCR is determined
• For non-detected chemicals (NDs), which might be present at level
<LOD, two assumptions have been made:
-NDs=0
-NDs=LOD/20.5
• MCR has a negative correlation to HI (i.e. for mixtures with high HI
Overall
the effect is driven by fewer compounds).
summary of
• The effect of in- or excluding non-detects has a large influence on
outcome
MCR for mixtures with small HI, but little impact on MCR for
mixtures with HI>1.
• A positive correlation of MCR with the number of analytes n was
shown for both cases considering and not considering non-detects.
E.g. in samples with 5-10 detects the MCR ranged from 1.0-2.0,
while in samples with 15-25 detects the MCR range was 1.0-5.0.
• The average MCR in all samples was 2.2-3.1, indicating that HI of
most mixtures are dominated by just a few chemicals.
• MCR values decreased with increase in toxicity (fewer compounds
driving the risk in more toxic mixtures).
• The authors state that the toxicity of environmental mixtures is
Future
usually dominated by a relatively small number of components,
perspec-
• The MCR is a useful tool for screening and ranking on where mixture
tives /
effects need to considered and where a single substance RA might
Outlook
be sufficient.
77
ID
18
Title
Example Case study B: Tier 0 – Substances potentially detectable in
surface water - Annex B (Meek et al., 2011)
Journal
Regulatory Toxicology and Pharmacology 60 S1-S14
Authors
Boobis, Budinsky, Crofton, Emry, Felter, Mihlan , Mumtaz, Price,
Solomon, Zaleski
Year
2011
Background
Surface water represents a real-world example of a complex mixture.
& Objectives Many of the substances present do not have established chronic health
standards or health-based guidance values, indeed, for some of the
components there might be little or no information on their toxicity.
Investigation of these mixtures using higher-tier assessments would
require considerable resources and a significant number of data. The
intent of this case study is to illustrate the potential utility of applying
the threshold of toxicological concern (TTC) approach in a Tier 0
assessment to prioritize the need for further evaluation of a chemical
mixture.
Substances
Data are based on surface water monitoring data, but to create an
example a similar hypothetical mixture of 10 compounds was created.
The 10 chemicals are form different classes (fragrances, pesticides,
surfactants, personal care products, solvents, petrochemicals)
Exposure
Human exposure via the consumption of water is the considered
Scenario
exposure pathway.
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Data are available from surface water monitoring but no data on
WHO/IPCS
the hazard of the mixture itself are available. Human exposure via
mixture
the consumption of water is the considered exposure pathway.
assessment
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
framework)
Yes. For the purpose of the case study it is assumed to be possible
via the consumption of surface water as drinking water.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes. The 10 substances used for the case study were detected in
the same survey. They are therefore assumed to occur
simultaneously and continuously.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
\
• Exposure data available from monitoring of surface water.
Information
• Use of TTC
78
sources
MIXTURE ASSESSMENT/METHODOLOGY
• Assumed that surface water is directly consumed without treatment
Exposure
• Worst case by choosing exposure of children and lifetime chronic
Assessment
exposure using maximum detected levels.
Hazard
It was assumed for the case example that no data would be available
Assessment
and the TTC was applied using ToxTree.
• Concentration Addition was assumed using the HI (HI=sum of HQ)
HRA
• Resulting HI was 0.2
(Tier 0)
• Given the conservative choices made to address the uncertainties,
Overall
a HI<1 is considered to trigger no need for higher tier analysis.
summary of
outcome
• This hypothetical case study demonstrated the utility of using the
Future
TTC approach as a Tier 0 assessment tool for chemical co-
perspec-
exposures.
tives /
Outlook
79
ID
19
Title
Organic chemicals jeopardize the health of freshwater ecosystems on
the continental scale (Malaj et al., 2014)
Journal
Proceedings of the National Academy of Sciences of the United States
of America
Authors
Egina Malaj, Peter C. von der Ohe, Matthias Grote, Ralph Kühne, Cédric
P. Mondy, Philippe Usseglio-Polatera, Werner Brack, Ralf B. Schäfer
Year
2014
Background
To investigate new spatial scales in chemical RA and to achieve a RA of
& Objectives
organic chemical on the continental scale, including 4000 European
monitoring sites.
To compare the chemical risk with the ecological status of the site,
when possible
Substances
Organic chemicals. Data are based on surface water monitoring data
Exposure
Exposure of aquatics organisms (fish, invertebrates, and algae,
Scenario
represented
by
Pimephales
promelas,
Daphnia
magna,
and
Pseudokirchneriella subcapitata, respectively).
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according to
WHO/IPCS
Data are available from surface water monitoring but no data on
mixture
the hazard of the mixture itself are available.
assessment
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
framework)
Yes, as this is based on monitoring data.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes, as this is based on monitoring data.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No assessment groups
• Exposure data available from monitoring of surface water
Information
(Waterbase dataset of the European Environmental Agency).
sources
• Hazard data collected from database
MIXTURE ASSESSMENT/METHODOLOGY
• Measured concentration of 223 chemicals for 4001 sites distributed
Exposure
over 91 European river
Assessment
• The chemical concentrations (µg/l) for each monitoring site were
reported as mean (Cmean), and maximum (Cmax) annual values,
typically used to characterize chronic and acute exposure,
respectively.
• Short term toxicity values were collected for each chemical and
Hazard
80
Assessment
each of the three species: (
P. promelas (96 h);
D. magna (48 h);
and
P. subcapitata (48–96 h). In a sequential order, LC
50 values
were compiled by using experimental, predicted, or baseline (from
the octanol–water partitioning coefficient) toxicity data.
• Those toxicity data allowed the calculation of risk threshold for each
organisms group, defined as:
1) Acute risk threshold (ART):
1/10 of the LC50 values for each of the
three standard test organisms
2) Chronic risk threshold (CRT):
1/1,000, 1/100,
and 1/50 of the LC50 values for invertebrates, fish, and algae,
respectively.
• Chemical risk (CR): the CR index for each organism group per river
RA for
basin was calculated:
aquatic
organisms
CRj,o,b = Nj,o,b/Ntotal,b,
N : number of sites for which one of the chemical concentrations
exceeded the risk threshold j (ART or CRT) for each organism group o
within a river basin b,
Ntotal: total number of sites within that river basin.
• Maps of distribution of the chemical risk (divided into 5 classes from
low to high CR) in Europe were created.
• To compare with the ecological status of the sites another approach
was used:
For each site within a rivers basin for which an ecological status was
available:
1) Cmax was compared to the ART;
2) Cmean was compared to the CRT.
As concentrations exceeding these thresholds may cause acute and
chronic ecological effects, respectively. Those sites were divided into
three classes:
(i) Chemical concentration > ART,
sites acutely affected by chemicals;
(ii) Chemical concentration > CRT, but <ART,
sites chronically affected
by chemicals;
(iii) Chemical concentrations < CRT,
sites with no or negligible risk from
chemicals.
The frequency of sites with high or good ecological status was
calculated per class.
• Organic chemicals were likely to exert acute lethal and chronic long-
Overall
term effects on sensitive fish, invertebrate, or algae species in 14%
summary of
and 42% of the sites, respectively.
outcome
• Of the 223 chemicals monitored, pesticides, tributyltin, polycyclic
aromatic hydrocarbons, and brominated flame retardants were the
major contributors to the chemical risk (pesticides were responsible
for 81%, 87% and 96% of the observed exceedances of the ART for
fish, invertebrates and algae respectively)
• The risk of potential acute and chronic long-term effects increased
with the number of ecotoxicologically relevant chemicals (ARCs)
analysed at each site.
• As most monitoring programs considered in this study only included
81
a subset of these chemicals, this assessment likely underestimates
the actual risk. Moreover, the results also depend on monitoring
practice: a dense monitoring network and the inclusion of most
ecotoxicologically relevant chemicals trigger a higher risk.
• Chemical risk strongly depended on the land use in the upstream
catchments of the monitoring sites.
• Increasing chemical risk was associated with deterioration in the
quality status of fish and invertebrate communities. Those results
clearly
indicate
that
chemical
pollution
is
a
large-scale
environmental problem and requires far-reaching, holistic mitigation
measures to preserve and restore ecosystem health
• There is a theoretical risk predicted based on the exposure
Future
concentrations monitored. This risk is increasing with the number of
perspec-
chemicals, as CA is the model used. However, in this study no
tives /
mixture testing has been done and therefore it is not possible to
Outlook
compare the real toxicity with the predicted toxicity.
• Those results are probably underestimating the risk for the following
reason:
1) The significantly increasing trend of the CR with the number of ARCs
that were analysed suggested that the acute and chronic risks would be
higher if more ARCs were analysed. River basins with more than 15
ARCs analysed exhibited generally higher chemical risks.
2) For 18% of the analysed chemicals, in the majority of cases
(>50%), the reported LOQ (smallest concentrations that can be reliably
quantified) values were above the CRT. Thus, analytical measurements
with higher sensitivity are required.
3) Whereas pesticides are designed to acutely affect invertebrates and
algae, fish typically suffer from compounds affecting development,
fitness, or reproduction (e.g., by endocrine disruptors), which are not
covered here, but might increase the risk to fish communities
4) Other considerations could increase the chemical risk:
(i) chemicals usually occur in mixtures, which might exhibit stronger
combined adverse effects
(ii) transformation products may be more ecotoxicologically potent than
their parent compounds
(iii) current monitoring relies on point grab water samples at monthly
or quarterly intervals, which are very likely to underestimate the real
maximum concentrations
For a more realistic prospective risk assessment, monitoring programs
should be designed to measure at least all ARCs, unless there is strong
evidence that a specific ARC is ecotoxicologically irrelevant in a basin.
However, emerging chemicals other than those frequently monitored
are likely to be present in ecotoxicologically relevant concentrations in
water samples and should be progressively identified and included in
monitoring programs.
82
ID
20
Title
Should the scope of human mixture risk assessment span legislative /
regulatory silos?
Journal
Science of the Total Environment
Authors
Evans RM, Martin OV, Faust M, Kortenkamp A.
Year
2016
Background
Based on the fact that most of current chemical legislation addresses
& Objectives potential risks based on single substance assessments, it was
investigated whether there is a concern that this approach is not
sufficiently protective. The need for a mixture risk assessment (MRA)
spanning different regulatory sectors is discussed based on two aspects:
(1) evidence that combined effects have been shown for chemical
mixtures containing substances regulated under different legislation and
(2) evidence for human co-exposure to chemicals regulated under
different legislation.
One case study example is included to illustrate the potential risk, based
on data published by Schlumpf et al 2010.
Substances
UV filters, fragrances, parabens, phthalates, organochlorine pesticides,
PDBEs, and PCBs
-
Exposure of breast-fed children through human milk
Exposure
Scenario
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
to
Exposure data from human biomonitoring of breast milk. Range of
WHO/IPCS
POPs and cosmetic product ingredients measured. No measured data
mixture
on the hazard of the mixture itself available.
assessment
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
framework)
Yes, data from human biomonitoring in human milk.
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
TIME FRAME?
Yes, co-exposure of breast-fed children to chemicals detected in
human milk.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
Risk is calculated for the whole range of compounds as one group,
but also for individual subgroups based on chemicals classes.
• Exposure data from Schlumpf et al 2010 from mother/child cohorts
Information
where for the first time a large number of POPs and cosmetic product
sources
ingredients were measured
• Hazard information: reference doses collected from authorities and
literature as in Schlumpf et al 2010
MIXTURE ASSESSMENT/METHODOLOGY
83
• A HI approach was used. Individual substance RQ were calculated,
HRA
HI for different chemical groups and the HI for the whole mixture.
• HI>1 was identified for several chemical classes (i.e. organochlor
Overall
pesticides and PCBs)
summary of
• The overall HI for the whole mixture was 66, indicating a potential
outcome
risk.
• The different chemical components were mapped on different
regulations and it is shown that some of them are covered under
several pieces of legislation and the overall mixtures span a wide
range of relevant regulatory silos.
• There is evidence underlining the co-exposure of humans to
Future
substances regulated under different "regulatory silos" and evidence
perspec-
of combined effects.
tives /
• Several examples are shown of chemicals regulated under different
Outlook
legislation that elicit common effects (e.g. (developmental)
neurotoxicants, substances potentially harmful to the developing
brain).
• Options to address a MRA across regulatory silos are discussed, e.g.
extending the EFSA pesticide residue cumulative assessment group
approach to other regulatory sectors.
84
ID
21
Title
Application of the maximum cumulative ratio (MCR) as a screening tool
for the evaluation of mixtures in residential indoor air
Journal
Science of the Total Environment
Authors
Katleen De Brouwere, Christa Cornelis, Athanasios Arvanitis, Terry
Brown, Derrick Crump, Paul Harrison, Matti Jantunen, Paul Price, Rudi
Torfs
Year
2014
Four datasets of residential indoor air exposure were used to calculate
Background
HI and MCR based on chronic inhalation toxicity values.
& Objectives
Substances
Volatile Organic Carbons (VOCs) and NO2 / residential indoor air
Exposure
Exposure to mixtures via residential indoor air (volatile organic carbons
Scenario
VOCs, and NO2)
1. WHAT IS THE NATURE OF EXPOSURE? ARE THE KEY COMPONENTS
Problem
KNOWN? ARE THERE DATA AVAILABLE ON THE HAZARD OF THE
Formulation
MIXTURE ITSELF?
(according
Exposure via indoor air; composition not fully know, only for
to
monitored compounds, not data available on whole mixture
WHO/IPCS
2. IS EXPOSURE LIKELY, TAKING INTO ACCOUNT THE CONTEXT?
mixture
Yes, considering the time humans spent indoors.
assessment
3. IS THERE A LIKELIHOOD OF CO-EXPOSURE WITHIN A RELEVANT
framework)
TIME FRAME?
Yes.
4. WHAT IS THE RATIONALE FOR CONSIDERING COMPOUNDS IN AN
ASSESSMENT GROUP?
No grouping
• Reference values retrieved by a structured review
Information
• Indoor air monitoring data from 5 European datasets including 1800
sources
records
MIXTURE ASSESSMENT/METHODOLOGY
Exposure
Exposure from monitoring data measuring VOCs and NO2. Flemish
Assessment
school and home survey, OQAI French home indoor air study, EXPOLIS
personal sampling and indoor residential air across European cities.
Hazard
Chronic inhalation RVs for non-cancer endpoints were collected from an
Assessment
array of sources (starting from authorities documents). For some data-
poor substances, they were derived from occupational exposure limits.
Chronic inhalation RVs could be identified for 44 substances. Large
variations were found for RVs from different agencies ranging up to
factor 300.
HRA
Calculating the HI using air concentrations/inhalation RVs
Calculating the MCRs
• Average MCR was 1.8, with a range from 1 to 5.8. MCR was found to
Overall
be small compared to the number of chemicals in the mixtures,
85
summary of
indicating that generally the overall effect was driven by only a few
outcome
chemicals.
• MCR is significantly declining with increasing HI.
• Large majority from Flemish school survey are categorised in the low
concern group II, while Flemish homes to the concern for combined
effects group III, and to the single substance concern group I. Most
of the OQAI data are assigned to single substance concern group I.
• Substances identified as biggest contributors were NO2,
trichloroethylene, acrolein, xylenes. These were however, not
consistently measured in all the studies, so comparison of datasets
and overall drivers is difficult.
• Study shows that there are a significant number of cases where
combined effects should be considered further and a chemical-by-
chemical approach would be insufficient. However, the mixtures
showing concern for combined effects were not those with the
highest HIs. Highest HI values were observed for samples where
single substances were dominating the overall risk.
• Personal measurements had generally a higher HI than indoor air
Future
measurements. Average ratio for HI was 1.5 (range 0.15-19). The
perspec-
use of indoor air versus personal monitoring could lead to some
tives /
underestimation.
Outlook
• The choice of the RV had a large impact on the overall results. Using
minimum RVs instead of the basic RVs moved most samples n to the
group of single substance of concern I.
86
Europe Direct is a service to help you find answers to your questions about the European Union
Free phone number (*): 00 800 6 7 8 9 10 11
(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.
A great deal of additional information on the European Union is available on the Internet.
It can be accessed through the Europa server http://europa.eu
How to obtain EU publications
Our publications are available from EU Bookshop (http://bookshop.europa.eu),
where you can place an order with the sales agent of your choice.
The Publications Office has a worldwide network of sales agents.
You can obtain their contact details by sending a fax to (352) 29 29-42758.
87
L
B
-NA-27968-EN-N
doi:10.2788/272583
ISBN 978-92-79-59146-4
88