A EPA
EPA/635/R-19/243
IRIS Assessment Protocol
www.epa.gov/iris
Systematic Review Protocol for the Methylmercury IRIS Assessment
(Preliminary Assessment Materials)
[CASRN 22967-92-6]
May 2020
Integrated Risk Information System
Center for Public Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
DISCLAIMER
This document is a public comment draft for review purposes only. This information is
distributed solely for the purpose of public comment It has not been formally disseminated by
EPA. It does not represent and should not be construed to represent any Agency determination or
policy. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use
This document is a draft for review purposes only and does not constitute Agency policy.
ii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
CONTENTS
CONTENTS iii
AUTHORS | CONTRIBUTORS | REVIEWERS vii
1. INTRODUCTION 1
2. SCOPING AND INITIAL PROBLEM FORMULATION SUMMARY 2
2.1. BACKGROUND 2
2.2.SCOPING SUMMARY 4
2.3. PROBLEM FORMULATION 5
2.4. ASSESSMENT APPROACH 7
2.5. KEY SCIENCE ISSUES 7
3. OVERALL OBJECTIVES, SPECIFIC AIMS, AND POPULATIONS, EXPOSURES, COMPARATORS,
AND OUTCOMES (PECO) CRITERIA 9
3.1. SPECIFIC AIMS 10
3.2. POPULATIONS, EXPOSURES, COMPARATORS, AND OUTCOMES (PECO) CRITERIA 11
4. LITERATURE SEARCH AND SCREENING STRATEGIES 13
4.1. LITERATURE SEARCH STRATEGIES 13
4.2. NON-PEER-REVIEWED DATA 14
4.3. LITERATURE SCREENING STRATEGY 15
4.3.1. Multiple Publications of the Same Cohort 16
4.3.2. Literature Flow Diagram 16
5. REFINED EVALUATION PLAN 19
6. STUDY EVALUATION (REPORTING, RISK OF BIAS, AND SENSITIVITY) STRATEGY 20
6.1.STUDY EVALUATION OVERVIEW FOR EPIDEMIOLOGY HEALTH EFFECT STUDIES 20
6.2. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL DESCRIPTIVE SUMMARY
AND EVALUATION 33
6.2.1. Pharmacokinetic/Physiologically Based Pharmacokinetic (PK/PBPK) Model
Descriptive Summary 35
6.2.2. Pharmacokinetic/Physiologically Based Pharmacokinetic (PK/PBPK) Model
Evaluation 37
7. DATA EXTRACTION OF STUDY METHODS AND RESULTS 39
8. DOSE-RESPONSE ASSESSMENT: STUDY SELECTION AND QUANTITATIVE ANALYSIS 40
This document is a draft for review purposes only and does not constitute Agency policy.
iii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
8.1. SELECTING STUDIES FOR DOSE-RESPONSE ASSESSMENT 40
8.2. CONDUCTING DOSE-RESPONSE ASSESSMENTS 42
8.2.1. Dose-response Analysis in the Range of Observation 42
8.2.2. Extrapolation: Slope Factors and Unit Risks 43
8.2.3. Extrapolation: Reference Values 43
9. PROTOCOL HISTORY 45
APPENDICES 46
APPENDIX A. ELECTRONIC DATABASE SEARCH STRATEGIES 46
APPENDIX B. DATA EXTRACTION FIELDS 49
APPENDIX C. CRITERIA FOR EVALUATION OF ANALYTICAL CHEMISTRY METHODS USED FOR
ANALYSIS OF MERCURY/METHYLMERCURY IN BLOOD AND HAIR 50
REFERENCES 68
This document is a draft for review purposes only and does not constitute Agency policy.
iv DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
TABLES
Table 1. EPA program and regional office interest in a methylmercury assessment 4
Table 2. PECO criteria for epidemiology data 11
Table 3. PECO criteria for PBPK studies 12
Table 4. Questions to guide the development of criteria for each domain in epidemiology
studies 25
Table 5. Information relevant to evaluation domains for epidemiology studies 33
Table 6. Classic one-compartment PK and PBPK models for MeHg in humans since 2001 35
Table 7. Criteria for evaluating PBPK models 37
Table 8. Attributes used to evaluate studies for derivation of toxicity values 41
Table A-l. Overall database search strategy 46
Table A-2. SWIFT Review search of titles and abstracts to identify epidemiology dose-response
articles 47
Table A-3. Database search strategies (PBPK studies) 48
Table B-l. Key data extraction elements to summarize study design, experimental model,
methodology, and results 49
Table C-l. Evaluation of analytical methods for blood total and methylmercury in epidemiology
studies 50
Table C-2. Evaluation of analytical methods for hair total and methylmercury in epidemiology
studies 56
FIGURES
Figure 1. Methylmercury IRIS systematic review problem formulation and method documents 1
Figure 2. Simplified conceptual model of the reassessment of DNT resulting from exposure to
MeHg 10
Figure 3. Literature search for MeHg DNT dose-response studies 17
Figure 4. Literature search for MeHg PBPK models 18
Figure 5. Overview of IRIS study evaluation process for epidemiology studies: (a) an overview of
the evaluation process; (b) the evaluation domains and definitions for ratings
(i.e., domain and overall judgments, performed on an outcome-specific basis) 21
This document is a draft for review purposes only and does not constitute Agency policy.
v DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
ABBREVIATIONS
CPAD Chemical and Pollutant Assessment Division
CPHEA Center for Public Health and Environmental Assessment
DNT developmental neurotoxicity
EPA U.S. Environmental Protection Agency
HERO Health and Environmental Research Online
IAP IRIS Assessment Plan
IRIS Integrated Risk Information System
NAS National Academy of Sciences
NRC National Research Council
OLEM Office of Land and Emergency Management
PBPK physiologically based pharmacokinetic
PECO populations, exposures, comparators, and outcomes
PK pharmacokinetic
POD point of departure
RfD reference dose
SWIFT Sciome Workbench for Interactive computer-Facilitated Text-mining
UF uncertainty factor
This document is a draft for review purposes only and does not constitute Agency policy.
vi DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
AUTHORS | CONTRIBUTORS | REVIEWERS
Assessment Team
Leonid Kopylev, Ph.D. (co-Assessment Manager) U.S. EPA/ORD/CPHEA/CPAD
Deborah Segal, M.H. S. (co-Assessment Manager)
Johanna Congleton, Ph.D.
Yu-Sheng Lin, Ph.D.
Rebecca Nachman, Ph.D.
Elizabeth Radke-Farabaugh, Ph.D.
Executive Direction
Wayne Cascio, M.D. (CPHEA Director)
Samantha Jones, Ph.D. (CPHEA Associate Director)
Kristina Thayer, Ph.D. (CPAD Director)
Emma Lavoie, Ph.D. (CPHEA Senior Science Advisor
for Assessments)
Andrew Kraft, Ph.D. (CPAD Senior Science Advisor)
Paul White, Ph.D. (CPAD Senior Science Advisor)
Belinda Hawkins, Ph.D. (CPAD Senior Science Advisor)
Ravi Subramaniam, Ph.D. (CPAD Branch Chief)
Production Team
Hillary Hollinger
Ryan Jones
Samuel Thacker
Erin Vining
Vicki Soto
Dahnish Shams
Maureen Johnson
HERO Librarian
HERO Director
HERO Technical Information Specialist
HERO Data Specialist/ORAU Student Contractor
Project Management Team
Project Management Team
NCEA Webmaster
This document is a draft for review purposes only and does not constitute Agency policy.
vii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Systematic Review Protocol for the Methylmercury IRIS Assessment
1. INTRODUCTION
A draft assessment plan for methylmercury (MeHg) was presented at a public science
meeting on May 15, 2019 fhttps://www.epa.goy/iris/iris-public-science-meeting-mav-20191 to
seek input on the problem formulation components of the assessment plan. The assessment plan
summarizes the Integrated Risk Information System ( IRIS) Program's scoping and problem
formulation conclusions, specifies the objectives and specific aims of the assessment, provides draft
PECO (populations, exposures, comparators, and outcomes) criteria, and identifies key areas of
scientific complexity.
This protocol document presents the methods for conducting the systematic review and
dose-response analysis, including any adjustments made to the specific aims and PECO criteria for
the assessment in response to public input on the assessment plan. While the IRIS Assessment Plan
(IAP) describes what the assessment will cover, chemical-specific protocols describe how the
assessment will be conducted (see Figure 1 for specific aims of the MeHg assessment). The IRIS
Program posts assessment protocols on its website and in repositories such as Zenodo
(https: //zenodo.org/). Public comments will be considered as part of developing the draft
assessment. Literature search results will also be posted in HERO (Health and Environmental
Research Online) when they are available.
Assessment
Initiated
IRIS Handbook Standard operating procedures and consider;
Identify Studies with ......
Systematic Sufficient Details of Study Select Studies for Use in Derive Toxicity
Scoping Review Protocol Quantitative Modeling Evaluation Deriving Toxicity Values Values
itions
\
^Assessment
Ceveloped
iducted
Initial Problen
Formulation
Assessment
Plans:
What the
assessment
will cover
4
Literature Refined Summarize Informative Extract data
Search Analysis Plan studies and Consider
Susceptibility
Protocols: How the assessment will be cor
Figure 1. Methylmercury IRIS systematic review problem formulation and
method documents.
This document is a draft for review purposes only and does not constitute Agency policy.
1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Systematic Review Protocol for the Methylmercury IRIS Assessment
2. SCOPING AND INITIAL PROBLEM
FORMULATION SUMMARY
2.1. BACKGROUND
Multiple health agencies fHealth Canada 7(107¦ I1NRP ?(!(!?¦II S FPA ?(W1lv ATSPR I WO
U.S. EPA. 19971 and the National Academy of Sciences' (NAS) National Research Council (NRC.
20001 have established that prenatal exposure to methylmercury in humans causes developmental
neurotoxicity (DNT). An existing IRIS reference dose (RfD) for methylmercury was published in
2001 (U.S. EPA. 2001b) and was based on an NAS assessment from 2000 fNRC. 20001. The
outcomes described by NAS included impaired cognitive function, motor function, visuospatial
performance, and abnormal (increased or decreased) muscle tone following in utero
methylmercury exposure fNRC. 20001. The RfD of 0.1 [ig/kg-day1 was derived from maternal daily
intakes of methylmercury of 0.86-1.47 [ig/kg-day, estimated to result in cord blood concentrations
of 46-79 |ig/L associated with multiple DNT measures (specifically, developmental
neuropsychological2 impairment) in a Faroe Island cohort described by Grandiean et al. (19971.
This epidemiology study found impaired cognitive function in 7-year-old children from the Faroe
Islands who were prenatally exposed to methylmercury (Budtz-l0rgensen etal.. 1999: Grandiean et
al.. 19971. IRIS's previous 1995 RfD for methylmercury was the same as the 2001 RfD and was also
based on DNT outcomes from in utero exposure using data from a 1971 Iraqi poisoning incident
[derivation described in U.S. EPA (1997)]. In both previous IRIS assessments, following
comprehensive literature searches and evaluations in each case, DNT outcomes were concluded to
be the most sensitive (other outcomes are discussed in Section 2.3).
Methylmercury is formed when inorganic mercury is methylated by biota in water and soil.
Gaseous elemental mercury is released into the atmosphere from natural (e.g., volcanoes) and
anthropogenic (e.g., fossil-fuel combustion) sources. Elemental mercury can be converted to
inorganic mercury, which then can be transported to land or water through wet or dry deposition
processes. Combustion processes can also release inorganic ionic mercury, which can adsorb to
particulate matter (Srivastava et al.. 2006). Inorganic divalent mercury adsorbed to particulates
can deposit after traveling relatively short distances, compared to elemental mercury vapor that
'Expressed as a concentration in whole maternal blood, the RfD is approximately 3.5 |ig/L fMahaffev et al..
20091.
2In the 2001 IRIS assessment of methylmercury, the term developmental neuropsychological impairment was
used to describe the adverse effects on the nervous system identified in humans following exposures to
methylmercury during developmental life stages. Developmental neuropsychological impairment is a type of
DNT, and is the term used in many epidemiological studies.
This document is a draft for review purposes only and does not constitute Agency policy.
2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Systematic Review Protocol for the Methylmercury IRIS Assessment
can travel long distances. Once deposited, microorganisms convert inorganic mercury to
methylmercury, which then bioaccumulates in fish tissue. Concentrations of methylmercury in fish
tissue, particularly predatory fish higher on the food chain (e.g., swordfish), can be much greater
than methylmercury concentrations found in ambient water (U.S. EPA. 20101.
Consumption of contaminated fish and other seafood is the major pathway for exposure to
methylmercury in humans (NRC. 20001: however, other foods, such as rice, can also expose humans
to methylmercury (Wells etal.. 2020: Cui etal.. 2017: Rothenberg etal.. 2017: Rothenberg etal..
20161. Between 2011 and 2014, average blood methylmercury levels in the U.S. population ranged
from 0.434 to 0.498 |ig/L fCDC. 20171. During this same period, average total blood mercury levels,
which often are used as a basis for determining methylmercury blood levels, ranged from 0.678 to
0.703 |ig/L between 2011 and 2016 (CDC. 20181. Males had slightly higher methylmercury blood
levels than females. For example, the average methylmercury blood level in 2013-2014 was
0.448 |ig/L and 0.422 |ig/L for males and females, respectively. Blood methylmercury levels were
also found to increase with age. In 2011 and 2012, the most recent years that methylmercury blood
levels were available for several age groups, the average for children 6 to 11 years of age was
0.209 |ig/L; for 12 to 19 year-olds, it was 0.276 |ig/L; and for adults over 19, it was 0.624 |ig/L
(CDC. 20181. The estimated mean daily intake of total mercury for women older than 20 years in
the United States is approximately 1 [ig/day3 (CDC. 2016a: Birch etal.. 20141.
Methylmercury readily crosses the placenta and concentrates in cord blood at
approximately 1.7 times the levels in maternal blood fStraka etal.. 2016: Stern and Smith. 2003:
Yang etal.. 19971. It is also transferred from mothers to children via breastmilk (CDC. 2009:
ATSDR. 19991. As noted earlier, the developing nervous system is particularly sensitive to
methylmercury, so these gestational, lactational, and other postnatal exposures are of great
concern. Methylmercury exposures to women of childbearing age who could become pregnant
might be harmful as well, as studies have reported an average half-life of methylmercury in the
body of 50 days, which might then result in fetal exposure early in pregnancy (CDC. 2016bl. A one-
compartment toxicokinetic model estimated a longer half-life for methylmercury, 80 days, on the
basis of blood samples from an adult population (To etal.. 20151. The half-life of methylmercury
varies among individuals, as some individuals have longer clearance times than others. For
example, EPA's 2001 assessment reported half-lives for methylmercury ranging from 32 to
189 days after evaluating data from 5 studies (Smith etal.. 1994: Sherlock etal.. 1984: Kershaw et
al.. 1980: Al-ShahristaniandShihab. 1974: Miettinenetal.. 19711.
Subsistence fishing communities and other populations with high dietary intakes of
predatory fish species could be exposed to higher-than-average levels of methylmercury.
Therefore, women of childbearing age and children in these communities could have high
3Based on the calculated average monthly mercury intake using 2009-2010 NHANES (National Health and
Nutrition Examination Survey] data reported by Birch et al. and CDC's anthropometric reference values for
2011-2014 (CDC. 2016a: Birch etal.. 20141.
This document is a draft for review purposes only and does not constitute Agency policy.
3 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Systematic Review Protocol for the Methylmercury IRIS Assessment
methylmercury exposures during susceptible life stages. People who consume fish from habitats
with high methylmercury concentrations due to large microbial populations that convert inorganic
mercury to methylmercury also might have particularly high exposures. This includes people
eating fish from certain types of wetlands, rivers with a high proportion of wetlands in their
watersheds, dilute and low-pH lakes in the Northeast and Northcentral United States, parts of the
Florida Everglades, newly flooded reservoirs, and coastal wetlands particularly along the Gulf of
Mexico, Atlantic Ocean, and San Francisco Bay (U.S. Department of the Interior. 20001. In some
regions of the world, consumption of fish from waters polluted by mercury from small-scale and
artisanal gold mining also might result in high methylmercury exposures. Contaminated rice and
rice-based food products, such as infant cereals, also can be a source of methylmercury exposure
(Cui etal.. 2017: Rothenberg etal.. 2017: Rothenbergetal.. 2016],
2.2. SCOPING SUMMARY
During the scoping process, the IRIS Program met with EPA program and regional offices
that had an interest in an IRIS reassessment of methylmercury to discuss specific needs. Table 1
provides a summary of input from this outreach.
Table 1. EPA program and regional office interest in a methylmercury assessment
EPA program
or regional
office
Oral
Inhalation
Statute/
Regulation
Anticipated uses/interest
Office of Land
and Emergency
Management
(OLEM)
EPA Regions 1-10
~
S a
Comprehensive
Environmental
Response,
Compensation
and Liability Act
(CERCLA)
Resource
Conservation and
Recovery Act
(RCRA)a
Clean Water Act
(CWA)
CERCLA authorizes EPA to conduct short- or long-term cleanups
at Superfund sites and later recover cleanup costs from
potentially responsible parties under section 107.
Methylmercury toxicological information may be used to make
risk determinations for such response actions (e.g., short-term
removals, long-term remedial response actions).
Mercury is listed under RCRA as a characteristic (40 CFR 261.24)
and hazardous waste (40 CFR 261.33). Methylmercury
toxicological information may be used to evaluate mercury
toxicity from releases of elemental mercury and mercury
compounds as environmental sources of methylmercury.
CWA requires EPA to develop water quality criteria for states and
tribes to use in developing water quality standards, requires
states and tribes to adopt water quality criteria that protect
designated uses such as fish consumption, and requires states
and authorized tribes to review water quality standards every
three years and modify them on the basis of updated health
effects studies derived by EPA.
aAlthough OLEM initially expressed the need for an inhalation reference concentration (RfC) for MeHg, this was de-prioritized
during subsequent scoping and problem formulation discussions on the basis of a lack of significant inhalation exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
4 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the Methylmercury IRIS Assessment
2.3. PROBLEM FORMULATION
Based on a preliminary survey of the methylmercury literature, including review of
assessments conducted by other agencies, potential health outcomes identified other than DNT
include the following:
Nervous system outcomes (non-developmental)
Developmental outcomes (other than nervous system effects)
Cardiovascular outcomes
Immune system outcomes
Reproductive outcomes
DNT resulting from oral exposure was selected as the focus of this first assessment module
because it is a well-established hazard and the two previous RfDs for methylmercury were derived
for oral exposure DNT outcomes. Some, but not all, recent epidemiology studies have reported DNT
adverse outcomes at exposure levels lower than exposure levels found in studies used to derive the
current reference dose. Many of these recent studies provide exposure-response information,
which justifies and enables reevaluation of the 2001 RfD fU.S. EPA. 2001b: NRC. 20001. Several
studies investigated cognitive function [e.g., Goldingetal. (2016): Tacobsonetal. (2015): Orenstein
etal. (2014): Sagivetal. (2012): Lederman etal. (2008): Oken etal. (2008): Oken etal. (2005)] and
motor function [e.g., Prpic etal. (2017): Goldingetal. (2016): Suzuki (2016): Ledermanetal.
(2008): Despres et al. (2005): Daniels etal. (2004)] at various ages following prenatal or postnatal
exposures to methylmercury. Other DNT outcomes (e.g., behavioral, structural, and
electrophysiological) following methylmercury exposures also have been evaluated [e.g., Tin etal.
f20161: Ngetal. f20151: Boucher etal. f20101]. This assessment will only reassess and update the
existing dose response for DNT outcomes. It will not reevaluate whether methylmercury causes
DNT outcomes because DNT is a well-established human hazard (as discussed in Section 2.1,
Background). Also, this assessment will not assess the potential for methylmercury exposure to
cause the other possible health outcomes of interest described above (see Section 2.4).
Once completed, the DNT dose-response assessment for oral exposure will undergo public
comment/peer review and finalization. After the assessment of DNT outcomes, EPA plans to assess
cardiovascular endpoints in a second assessment module and will also consider the need for
assessment of other endpoints, such as adult nervous system and reproductive effects.
Cardiovascular outcomes were identified as a specific priority during the public comment period
and science webinar on the methylmercury IAP (https: //www.regulations.gov/docket?D=EPA-HO-
QRD-2018-0655I The decision to evaluate other outcomes aside from DNT and cardiovascular
effects will be based on examining whether there is sufficient evidence to assess hazard, derive
This document is a draft for review purposes only and does not constitute Agency policy.
5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the Methylmercury IRIS Assessment
reference values, and consider whether the additional analyses are likely to impact EPA decision
making beyond what the DNT and cardiovascular assessments provide.
Because ingestion is the primary route of exposure for methylmercury (NRC. 20001.
inhalation and dermal routes of exposure are not addressed in this assessment. Although OLEM
initially expressed the need for an inhalation reference concentration (RfC) for MeHg, the need was
de-prioritized during subsequent scoping and problem formulation discussions on the basis of lack
of significant inhalation exposure.
The reassessment of DNT dose response will focus on human studies because the
availability of a large epidemiology database on methylmercury exposure and DNT outcomes
[e.g., review by Karagas etal. (20121] eliminates uncertainties associated with interspecies
extrapolation. During this reassessment, IRIS will evaluate epidemiology evidence for all types of
DNT outcomes resulting from exposure to the fetus, infants, children, or adolescents because during
development the brain is more vulnerable to the neurotoxicity of methylmercury. Targeted
literature searches might be conducted for animal toxicological or mechanistic studies to address
data gaps (e.g., susceptibility) or to replace default uncertainty factors (UFs) with data-derived
factors.
Public comments on the MeHg IAP also suggested the IRIS Program should conduct an
assessment to examine the adverse effects of developmental MeHg exposure and the beneficial
effects related to seafood consumption during pregnancy. Consideration of the health benefits of
fish consumption falls outside the traditional scope of the IRIS Program and was not identified as a
current EPA National Program priority. However, the IRIS Program understands the importance of
both types of effects, in particular, for providing fish consumption advice. In addition, as outlined in
the Specific Aims (Section 3.1), the assessment will seek to determine if the available data would
also support the derivation of dose-response relationships for DNT outcomes that would be useful
for analyses conducted by others to quantify the health impacts of actions to reduce exposures to
MeHg. The IRIS Program is also communicating with staff at the U.S. Department of Agriculture
supporting the 2020 Dietary Guidelines Advisory Committee (Dietary Fats and Seafood
Subcommittee) regarding their review of "What is the relationship between types of dietary fat
consumed and neurocognitive development fbirth to 18 vearsl or neurocognitive health ffor those
18 years and olderl?" fhttps://www.dietaryguidelines.gov/dietary-fats-and-neurocognitive-
healthl. Although stakeholders indicated a need for evaluation of other forms of mercury
(e.g., elemental and inorganic salts), these mercury forms would need to be evaluated through
separate assessments. Currently, IRIS is developing an assessment of inorganic mercury salts;
however, no other forms of mercury were identified as an Agency priority.
Other public comments on the methylmercury IAP focused on the following topics: an
agreement that DNT should be the focus of the first module, the need to consider confounding by
fish nutrients and the need to consider biomarker imprecision (see Key Science issues), and the
need to consider individual variation (see Specific Aims).
This document is a draft for review purposes only and does not constitute Agency policy.
6 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
1
2
3 evaluate DNT outcomes associated with oral exposure in the first assessment module. DNT was
4 selected for the first module because it is a well-established, sensitive hazard and the two previous
5 RfDs for methylmercury were derived for oral exposure DNT outcomes (see Section 2.1). In
6 addition to DNT, EPA plans to assess cardiovascular outcomes in a second assessment module and
7 will also consider the need for assessing other endpoints, such as the adult nervous system and
8 reproductive effects in additional module(s). Once completed, the draft assessment addressing the
9 DNT dose-response relationship for oral exposure will undergo public comment/peer-review and
10 finalization, rather than waiting for the cardiovascular module to be completed. The decision to
11 evaluate other outcomes aside from DNT and cardiovascular toxicity will be based on examining
12 whether evidence is sufficient to assess hazard, derive reference values, and consider whether the
13 additional analyses are likely to impact EPA decision making beyond what the DNT and
14 cardiovascular assessments provide.
15 While completing the DNT module, EPA will survey the available hazard information for
16 cardiovascular and other adverse health outcomes (see Section 2.3 for list), primarily by reviewing
17 methylmercury assessments by other agencies and organizations, and recent epidemiology studies.
18 For health effects for which hazard has not been established on the basis of epidemiology evidence,
19 animal and mechanistic studies will also be surveyed. The cardiovascular and any other
20 assessment modules will have their own IAPs that will be released separately. Because inhalation
21 exposure to MeHg is not a significant route of exposure, only oral exposure studies will be
22 evaluated.
24 scientific issues were identified and will be addressed in this assessment as indicated below. The
25 assessment will consider:
23
26
27
The accuracy and reliability of measures of the different types of biomarkers (e.g., hair,
maternal blood, cord blood) to quantify methylmercury exposure.
28
29
EPA plans to use analytical chemistry criteria to evaluate the accuracy and reliability of
measures of the different types of biomarkers (see Appendix C),
30
31
32
33
34
How to best use the different biomarkers measured in PECO-relevant epidemiology studies
to inform estimates of the relationship between methylmercury exposure and
neurodevelopmental effects. For example, some epidemiology studies measure
methylmercury directly in human blood, hair, or nails, while other studies rely on measures
of total mercury to estimate methylmercury exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Systematic Review Protocol for the Methylmercury IRIS Assessment
EPA plans to consider several approaches for utilizing studies that rely on total mercury
measures to estimate methylmercury exposure: Regression modeling of the methylmercury
based on total mercury and, possibly, covariates; accepting that total mercury is an adequate
proxy for methylmercury (e.g., for hair); or deriving RfDsfor both total mercury and
methylmercury,
How potential confounding [e.g., Budtz-lorgensen etal. f20071] in studies will be accounted
for in the analyses. For example, many fish species that contain methylmercury also have
beneficial nutrients, such as selenium and polyunsaturated fatty acids, which are important
to brain development In addition, fish could contain other contaminants that might be
harmful to brain development, such as polychlorinated biphenyls. Accounting for
confounders will be assessed during study evaluation.
EPA plans to assess whether confounders were appropriately accounted for as part of study
evaluation,
The differences in DNT evaluation methods and how their results could be used in this
assessment For example, developmental scores are consistently higher for both term and
preterm infants when using the Bayley III test versus the Bayley II test, and some suggest
using an adjustment factor to compare the two scores (Lowe etal.. 20121.
EPA plans to use criteria (currently in development in collaboration with NTP using
contractors who are experts in the field) for evaluating DNT tests and their appropriate use in
epidemiology studies evaluating DNT effects of methylmercury exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Systematic Review Protocol for the Methylmercury IRIS Assessment
3. OVERALL OBJECTIVES, SPECIFIC AIMS, AND
POPULATIONS, EXPOSURES, COMPARATORS,
AND OUTCOMES (PECO) CRITERIA
The overall objective of this assessment is to characterize the dose-response relationship
between methylmercury exposure and DNT outcomes and then use this information to update the
existing RfD. Because the current RfD for methylmercury was posted by IRIS in 2001 and was
based on an NAS (NRC. 20001 assessment, evaluation of studies since 1998 is expected to capture
literature that was not considered in the earlier assessments. Any new health risk assessments for
methylmercury identified in the search for newer literature will be reviewed as secondary
literature sources. The relevant dose-response analyses included in these previous assessments
will also be considered in this reassessment Studies that evaluated the relationships between
methylmercury exposures to women of childbearing age and the developing child and DNT
outcomes that become apparent at any life stage (infancy through the elderly) will be considered. A
conceptual model is presented below to illustrate the focus of the planned assessment (Figure 2). A
critical effect will be selected for derivation of a RfD that will be protective against all DNT effects
that occur at any age following prenatal to adolescent methylmercury exposure.
Systematic review methods will be used to evaluate the epidemiology literature on DNT
outcomes, and the analysis conducted will be consistent with all relevant EPA guidance.4 As part of
this systematic review, potentially susceptible populations, for example, populations with certain
genetic polymorphisms, and life stages will be considered. This Systematic Review Protocol reflects
scoping and problem formulation changes made to the specific aims and PECO criteria in response
to public input received on the MeHg IAP.
4EPA guidance documents: http://www.epa.gov/iris/basic-information-about-integrated-risk-information-
svstem#guidance/.
This document is a draft for review purposes only and does not constitute Agency policy.
9 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Systematic Review Protocol for the Methylmercury IRIS Assessment
Figure 2. Simplified conceptual model of the reassessment of DNT resulting
from exposure to MeHg.
3.1. SPECIFIC AIMS
Identify epidemiology literature examining effects of exposure to methylmercury as outlined
in the PECO criteria (see Section 3.2, Table 2). Develop and execute a literature search
strategy to broadly capture data from MeHg epidemiology studies published since 1998, and
screen results for relevance,
Use predefined criteria to identify epidemiology studies from the screened results that
provide exposure-response information for DNT outcomes.
Conduct study evaluations (risk of bias and sensitivity ) for identified epidemiology studies.
This will include an assessment of the proper consideration of confounders (e.g., fish
nutrients such as polyunsaturated fatty acids). Studies with critical deficiencies generally
will be considered uninformative and not considered further.
Summarize study methods and results from epidemiology studies on DNT outcomes,
including explicit identification and discussion of issues concerning susceptible populations
and life stages, including potentially important genetic polymorphisms.
Evaluate whether dose conversion [i.e., physiologically based pharmacokinetic (PBPK)
modeling] is needed. Depending on the biomarker (e.g., cord blood), conduct a search and
review of the relevant literature as needed to determine if calculations used in the previous
assessment (to convert from cord blood to oral exposure) need to be updated. If necessary,
This document is a draft for review purposes only and does not constitute Agency policy,
10 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
1 individual PBPK models will be evaluated using predefined criteria, and their strengths and
2 uncertainties will be summarized.
3 Characterize uncertainties, including individual variability in MeHg toxicokinetics where
4 data are available to do so, thereby reducing reliance on default UFs. Identify key data gaps
5 and research needs, such as limitations of the evidence base and the systematic review.
6 Derive a toxicity value (e.g., RfD) for DNT outcomes as supported by the available data.
7 Assemble the available data to support analyses conducted by others to quantify the health
8 impacts of actions to reduce exposures to MeHg.
3.2. POPULATIONS, EXPOSURES, COMPARATORS, AND OUTCOMES
(PECO) CRITERIA
9 The PECO criteria are used to identify the evidence that addresses the specific aims of the
10 assessment and to focus the search terms and inclusion/exclusion criteria in a systematic review.
11 The draft PECO criteria for this MeHg assessment (Tables 2 and 3) were based on (1) basis for the
12 chemical's prioritization for assessment, (2) discussions with scientists in EPA program and
13 regional offices to determine the scope of the assessment that will best meet Agency needs, and
14 (3) preliminary review of the DNT literature for MeHg (primarily reviews and authoritative health
15 assessment documents).
Table 2. PECO criteria for epidemiology data
PECO element
Evidence
Populations
Human populations exposed during life stages ranging from the fetus through adolescence.
Exposures
Any quantitative exposure to MeHg based on biomonitoring data (e.g., hair, nails, blood).
Measurements must be either direct MeHg measurements or measurements of total mercury
(not other forms of mercury, e.g., mercury salts).
Comparators
Referent populations exposed to lower (within the study) levels of MeHg will be used to
examine specific effects. The results of the comparisons must be presented with sufficient
detail of quantitative modeling (e.g., regression coefficients presented with statistical measure
of variation).
Outcomes
DNT outcomes measured at any age, includingbut not limited totests or measures of
cognition, motor function, behavior, vision, and hearing.
This document is a draft for review purposes only and does not constitute Agency policy.
11 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table 3. PECO criteria for PBPK studies
PECO element
Evidence
Populations
Human populations exposed during life stages ranging from the fetus through adolescence.
Exposures
Implemented for relevant information of exposure (defined by route, time of exposure,
intensity, and frequency) that informs toxicokinetic modeling to improve estimation
procedures for dietary intake of MeHg using biomonitoring data.
Comparators
Any comparison that helps improve the estimation of body burden of MeHg in humans
including absorption, distribution, metabolism, and elimination (ADME) processes.
Outcomes
Any examination of MeHg deposition (ADME) and dose metrics (e.g., peak concentration of
blood MeHg) that inform the evaluation of MeHg DNT outcomes.
This document is a draft for review purposes only and does not constitute Agency policy.
12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Systematic Review Protocol for the Methylmercury IRIS Assessment
4. LITERATURE SEARCH AND SCREENING
STRATEGIES
4.1. LITERATURE SEARCH STRATEGIES
Literature search strategies were developed for epidemiology studies published since 1998
using key terms and words related to the PECO criteria. Of note, the 2001 RfD for MeHg was
derived using the toxicokinetic model [one-compartment pharmacokinetic (PK) model]
recommended by the NRC (2000). As PK/PBPK studies published prior to 2001 had been
extensively evaluated in the 2001 EPA MeHg assessment (U.S. EPA. 2001bl. a literature search was
conducted for PK/PBPK studies published since 2001. The search strategy involved identifying
relevant search terms through the following approaches: (1) extracting key terminology from
relevant reviews and (2) consulting with the HERO librarian. Relevant subject headings and text-
words were crafted to maximize the sensitivity and specificity of the search results. No language
restrictions were applied. The four databases listed below were searched (PK/PBPK search did not
include Toxline). Because each database has its own search architecture, the resulting search
strategy was tailored to account for each database's unique search functionality (the detailed
search strategies are presented in Appendix A, Tables A-l and A-3).
PubMed (National Library of Medicine)
Web of Science (Thomson Reuters)
Toxline (National Library of Medicine)
Science Direct (Elsevier)
Literature searches were conducted using EPA's HERO database.5 Because PECO criteria
focused only on studies that report data amenable to dose-response modeling, the literature search
for epidemiology studies was organized as follows. First, all MeHg literature was searched in the
four databases listed above and duplicates were removed by HERO. After deduplication in HERO,
these studies were imported into SWIFT Review software fHoward etal.. 20161 to identify
epidemiology studies most likely to be suitable for dose-response analysis. In brief, SWIFT Review
has preset literature search strategies ("filters") developed by information specialists that can be
applied (or modified) by the user to identify PECO-relevant studies. The filters function like a
typical search strategy in which studies are tagged as belonging to a certain filter if the terms in the
5Health and Environmental Research Online: https: //hero.epa.gov/hero/.
This document is a draft for review purposes only and does not constitute Agency policy.
13 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the Methylmercury IRIS Assessment
filter literature search strategy appear in title, abstract, keyword, or medical subject headings
(MeSH) fields content. The SWIFT Review filter for human evidence was modified slightly to
identify epidemiology studies (see Appendix A, Table A-2). In addition, a new search filter
containing dose-response terms (e.g., regression, p-value) was developed and applied to identify
those epidemiology studies most likely to be relevant to the MeHg PECO. Studies that included one
or more of the search terms in the title, abstract, keyword, or MeSH fields for epidemiology and
dose-response search strings were exported as an RIS (Research Information Systems) file for
screening in DistillerSR. as described below. Application of the SWIFT Review filters reduced the
number of studies for title and abstract screening from 15,277 to 2,905. In addition, because some
articles do not contain abstracts and sometimes abstracts are not imported from the HERO
database to SWIFT Review, epidemiology articles without abstracts were identified from the SWIFT
Review and screened for relevance by one person.
Because the PBPK literature search resulted in relatively few papers (284 unique references
after deduplication), the screening was performed by one person using Endnote. The search
strategies developed by SWIFT Review for PBPK studies were used to guide the screening.
The literature search will be updated throughout draft development to identify literature
published during the course of review. The last full literature search update will be conducted less
than 1 year before the planned release of the draft document for public comment. The results
returned (i.e., the number of "hits" from each electronic database or other literature source),
including the results of any literature search updates, are documented in the literature flow
diagrams (see Section 4.3.2), which also reflect the literature screening decisions (see Section 4.4).
The IRIS Program takes extra steps to ensure identification of pertinent studies by
encouraging the scientific community and the public to identify additional studies and ongoing
research and by considering late-breaking studies that would affect the credibility of the
conclusions, even during the review process. Studies identified after peer review begins will only be
considered for inclusion if they meet the PECO criteria and are expected to fundamentally alter the
assessment's conclusions. Release of the PECO-screened literature in parallel with release of the
protocol for public comment provides an opportunity for stakeholders to identify any missing
studies, which if identified, will be screened as outlined above for adherence to the PECO criteria.
4.2. NON-PEER-REVIEWED DATA
IRIS assessments rely mainly on publicly accessible, peer-reviewed studies. However, it is
possible that unpublished data directly relevant to the PECO may be identified during assessment
development. Depending on the potential impact of the study on assessment conclusions, EPA
might obtain external peer review if the owners of the data are willing to have the study details and
results made publicly accessible fU.S. EPA. 20151. This independent, contractor-driven, peer
review would include an evaluation of the study similar to that for peer review of a journal
publication. The contractor would identify and select two or three scientists knowledgeable in
This document is a draft for review purposes only and does not constitute Agency policy.
14 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the Methylmercury IRIS Assessment
scientific disciplines relevant to the topic as potential peer reviewers. Persons invited to serve as
peer reviewers would be screened for conflict of interest. In most instances, the peer review would
be conducted by letter review. The study authors would be informed of the outcome of the peer
review and given an opportunity to clarify issues or provide missing details. The study and its
related information, if used in the IRIS assessment, would become publicly available. In the
assessment, EPA would acknowledge that the document underwent external peer review managed
by EPA, and the names of the peer reviewers would be identified. In certain cases, IRIS will conduct
an assessment for utility and data analysis based on having access to a description of study
methods and raw data that has undergone rigorous quality assurance/quality control review
(e.g., ToxCast/Tox21 data, results of National Toxicology Program studies) but that have not yet
undergone external peer review.
Unpublished data from personal author communication can supplement a peer-reviewed
study provided the information is made publicly available (typically through documentation in
HERO).
4.3. LITERATURE SCREENING STRATEGY
The PECO criteria were used to determine inclusion or exclusion of a reference as a primary
source of health effects data or a published PBPK model. Targeted literature searches might be
conducted for animal, mechanistic, or ADME studies to address data gaps (e.g., susceptibility) or to
replace default UFs with data-derived factors.
Title and abstract-level screening (epidemiology studies). Following a pilot phase to calibrate
screening guidance, two screeners independently conducted a title and abstract screen of the
search results to identify records that appear to meet the PECO criteria using a structured form in
DistillerSR (Evidence Partners; https://www.evidencepartners.com/products/distillersr-
svstematic-review-software/].
Screening conflicts were resolved by discussion among the primary screeners with
consultation by a third reviewer to resolve any remaining disagreements. Eligibility status of
non-English studies was assessed using the same approach, and online translation tools were used
to assess eligibility at the title and abstract levels.
Studies not meeting the PECO criteria but identified as "potentially relevant supplemental
material" were tagged during the title and abstract screening process (see Figures 3 and 4). Conflict
resolution was not required during the screening process to identify supplemental information
(i.e., tagging by a single screener is sufficient to identify the study as potentially relevant
supplemental material that might be considered during draft development).
Full-text level screening (epidemiology studies). Records not excluded on the basis of the title
and abstract were advanced to full-text review. Full-text copies of these potentially relevant
records were retrieved, exported from the HERO database to Distiller, and independently assessed
by two screeners to confirm eligibility according to the PECO criteria. Screening conflicts were
This document is a draft for review purposes only and does not constitute Agency policy.
15 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Systematic Review Protocol for the Methylmercury IRIS Assessment
resolved by discussion between the primary screeners with consultation by a third reviewer (as
needed to resolve any remaining disagreements). Studies that advanced to full-text review could
also be tagged as "potentially relevant supplemental material."
For the PBPK studies PECO, one person conducted title and abstract and full-text screening
using Endnote, because relatively few studies were identified.
The results of this screening process were posted on the project page for this assessment in
the HERO database https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2589. These
studies will be "tagged" with appropriate category descriptors (e.g., studies eligible for study
evaluation, potentially relevant supplemental material, excluded). Results are also annotated and
reported in a literature flow diagram (see Figures 3 and 4).
It is important to emphasize that being tagged as supplemental material does not mean the
study would necessarily be excluded from consideration in the assessment. The initial screening
level distinctions between a study that meets the PECO criteria and a supplemental study are
designed to ensure the supplemental studies are categorized for easy retrieval while conducting the
assessment The impact on the assessment conclusions of individual studies tagged as
supplemental material is often difficult to assess during the screening phase of the assessment.
These studies might be critical to the assessment, and if so, they will be summarized at the
individual study level. Alternatively, they could be cited because they provide context or they might
not be cited at all in the assessment (e.g., individual studies that contribute to a well-established
scientific conclusion). In addition, studies might be tagged as supplemental material during either
title and abstract or full-text screening.
Release of the PECO-screened literature in the protocol (or protocol update) for public
comment provides an opportunity for stakeholders to identify any missing studies. If identified,
those studies will be screened as outlined above for adherence to the PECO criteria.
4.3.1. Multiple Publications of the Same Cohort
When a cohort is the subject of multiple publications, all publications focused on the cohort
will be included. For each cohort, several primary publications could be selected.
4.3.2. Literature Flow Diagram
Flow diagrams for literature searches for epidemiology and PBPK studies are presented in
Figures 3 and 4.
This document is a draft for review purposes only and does not constitute Agency policy.
16 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
MeHg Literature Searches (1998 to present)
PubMed
(n = 5,846)
Web of Science
(n = 10,246)
ToxNet
(n = 6,801)
*
Following duplicate removal, SWIFT Review used to search 15,277 recordsfrom database searches
Identification of potentially relevant records based on application of SWIFT-Review evidence stream tags and
customized terms for dose-response terms, n = 2,905
Records identified from Other Sources
Pre-1998 studies
r
Studies from other
identified in NAS (2000)
sources
(n = 9)
CTi
IN
II
C
Title & Abstract Screening
(n = 2,943)
2,905 from databases + 38 from other
sources)
T
FULL TEXT SCREENING
Full-Text Screening
(n = 479)
Dose-Response Studies Considered
Further (n = 269)
~
Excluded (n= 1,698)
Tagged as Supplemental* (n= 768)
Sum of excluded or supplemental*(n = 2,466)
Excluded as not relevant to PECO (n = 64)
Tagged as Supplemental* (n= 116)
Sum of excluded or supplemental* (n = 180)
Figure 3. Literature search for MeHg DNT dose-response studies.
aBecause the literature search was first performed for dose-response epidemiology studies and only then
screened, the supplemental literature is not comprehensive and thus not categorized further.
This document is a draft for review purposes only and does not constitute Agency policy.
17 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Literature Searches for PK/PBPK Studies (Jan 2001- Oct 2019)
PubMed
(n =209)
Science Direct
(n =19)
Web of Science
(n =151)
Following duplicates removed, 284 records were identified potentially
relevant for PBPK or PK (human) studies based on application of SWIFT-
Review evidence stream tags
TITLE AND ABSTRACT
Title & Abstract Screening
(n = 284)
FULL TEXT SCREENING
Full-Text Screening
(n = 51)
i
Studies Considered Further (n = 26)
One-compartment model (n = 8)
PBPK (n =18)
Excluded (n = 164 )
Tagged as Supplemental (n = 69)
Sum of excluded or supplemental (n = 233)
~ |^^Exdudednot^reJevanttoJ>ECO^(n^=J5)^^
Sum of excluded or supplemental (n = 25)
Tagged as Supplemental (n = 86)
ADME/QSAR models (n =41), exposure and
risk characterization (n =28), Toxicity /
Adverse Health Effects (n =6),
toxicodynamics (n = 11)
Figure 4. Literature search for MeHg PBPK models.
This document is a draft for review purposes only and does not constitute Agency policy.
18 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
5. REFINED EVALUATION PLAN
1
2
3
4
6 Public comments can be found at https://www.regulations.gov/docket?D=EPA-HQ-QRD-2018-0655.
This document is a draft for review purposes only and does not constitute Agency policy.
19 DRAFT-DO NOT CITE OR QUOTE
Public comments6 on the assessment plan for the DNT module did not suggest a change was
warranted to the specific aims or PECO; thus, no refined evaluation plan was pursued (i.e., all DNT
outcomes in all the studies that met the PECO criteria will be evaluated in this assessment module)
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the Methylmercury IRIS Assessment
6. STUDY EVALUATION (REPORTING, RISK OF
BIAS, AND SENSITIVITY) STRATEGY
6.1. STUDY EVALUATION OVERVIEW FOR EPIDEMIOLOGY HEALTH
EFFECT STUDIES
Evaluation of epidemiology studies of health effects to assess risk of bias and study
sensitivity will be conducted for the following domains: exposure measurement, outcome
ascertainment, participant selection, potential confounding, analysis, study sensitivity, and selective
reporting. Bias can result in false positives and false negatives, while study sensitivity is typically
concerned with identifying the latter.
A key concern for the review of epidemiology studies is risk of bias, which is the assessment
of internal validity (factors that affect the magnitude or direction of an effect in either direction)
and insensitivity (factors that limit the ability of a study to detect a true effect; low sensitivity is a
bias toward the null when an effect exists). Reporting quality is evaluated to determine the extent
the available information allows for evaluating these concerns. The study evaluations are aimed at
discerning the expected magnitude of any identified limitations (focusing on limitations that could
substantively change a result), considering also the expected direction of the bias. The study
evaluation considerations described below can be refined to address a range of study designs,
health effects, and chemicals. The general approach for reaching an overall judgment for the study
(or a specific analysis in a study) regarding confidence in the reliability of the results is illustrated
in Figure 5.
At least two reviewers will independently evaluate the studies to identify characteristics
that bear on the informativeness (i.e., validity and sensitivity) of the results and provide additional
chemical- or outcome-specific knowledge or methodological concerns.
Considerations for evaluating studies will be developed in consultation with topic-specific
technical experts and existing guidance documents when available, including EPA guidance for
neurotoxicity, reproductive toxicity, and developmental toxicity fU.S. EPA. 1998.1996.19911. The
independent evaluations include a pilot phase to assess and refine the evaluation process. During
this phase, decisions will be compared and a consensus reached between reviewers, and when
necessary, differences will be resolved by discussion between the reviewers, the chemical
assessment team, or technical experts. As reviewers examine a group of studies, additional
chemical-specific knowledge or methodological concerns could emerge, and a second pass might
become necessary. Refinements to the study evaluation process made during the pilot phase and
subsequent implementation will be acknowledged as updates to the protocol.
This document is a draft for review purposes only and does not constitute Agency policy.
20 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Study evaluation process (b)
Refined evaluation plan
Individual evaluation
Domains
Criteria development
Pilot testing/refine criteria
Epidemiology
Participant selection
Confounding
Exposure misclassification
Outcome ascertainment
Analysis
Other sensitivity
0
Domain judgments
Evaluation by two
reviewers
$
Conflict resolution
Final domain judgments
and overall study rating
Judgment
Good
Adequate
Deficient
Critically
Deficient
Interpretation
Appropriate study conduct relating to the domain
and minor deficiencies not expected to influence
results.
A study that may have some limitations relating to
the domain, but they are not likely to be severe or to
have a notable impact on results.
Identified biases or deficiencies interpreted as likely
to have had a notable impact on the results or
prevent reliable interpretation of study findings.
A serious flaw identified that makes the observed
effect(s) uninterpretable. Studies with a critical
deficiency will almost always be considered
"uninformative" overall.
Overall study rating for an outcome
Rating
Interpretation
High
No notable deficiencies or concerns identified; potential
for bias unlikely or minimal; sensitive methodology.
Medium
Possible deficiencies or concerns noted, but resulting
bias or lack of sensitivity is unlikely to be of a notable
degree.
Low
Deficiencies or concerns were noted, and the potential
for substantive bias or inadequate sensitivity could have
a significant impact on the study results or their
interpretation.
Uninformative
Serious flaw(s) makes study results unusable for hazard
identification or dose response.
Figure 5. Overview of IRIS study evaluation process for epidemiology studies:
(a) an overview of the evaluation process; (b) the evaluation domains and
definitions for ratings (i.e., domain and overall judgments, performed on an
outcome-specific basis).
This document is a draft for review purposes only and does not constitute Agency policy.
21 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the Methylmercury IRIS Assessment
For studies that examine more than one outcome, the evaluation process will be performed
separately for each outcome because the utility of a study can vary for different outcomes. If a
study examines multiple endpoints for the same outcome,7 evaluations might be performed at a
more granular level if appropriate, but these measures could still be grouped for evidence
synthesis.
Authors might be queried to obtain missing critical information, particularly when
reporting quality information or data are missing (e.g., content that would be required to conduct a
meta-analysis or other quantitative integration) or to provide additional analyses that could
address potential limitations. The decision on whether to seek missing information includes
considering what additional information would be useful, specifically with respect to any
information that could result in a reevaluation of the overall study confidence. Outreach to study
authors should be documented and considered unsuccessful if researchers do not respond to an
email or phone request within 1 month of the attempt to contact
For each outcome in a study,8 reviewers will reach a consensus judgment of Good, Adequate,
Deficient, Not reported, or Critically deficient for each evaluation domain. If a consensus is not
reached, a third reviewer will perform conflict resolution. That these evaluations are performed in
the context of the study's utility for dose-response analysis is important to stress. These categories
are applied to each evaluation domain for each study as follows:
Good represents a judgment that the study was conducted appropriately in relation to the
evaluation domain, and any deficiencies, if present, are minor and would not be expected to
influence the study results.
Adequate indicates a judgment that there are methodological limitations relating to the
evaluation domain, but that those limitations are not likely to be severe or to have a notable
impact on the results.
Deficient denotes identified biases or deficiencies that are interpreted as likely to have had a
notable impact on the results or that may prevent reliable interpretation of the study
findings.
Not reported indicates that the information necessary to evaluate the domain in question was
not available in the study. Generally, this term carries the same functional interpretation as
Deficient for the purposes of the study confidence classification (described below).
Depending on the number and severity of other limitations identified in the study, it may or
may not be worth reaching out to the study authors to obtain this information (see
discussion above).
7"Outcome" will be used throughout these methods; the same methods also apply to an endpoint within a
larger outcome.
8"Study" is used instead of a more accurate term (e.g., "experiment") throughout these sections owing to an
established familiarity within the field for discussing a study's risk of bias or sensitivity, etc. However, all
evaluations discussed herein are explicitly conducted at the level of an individual outcome within an
(un)exposed group of animals or humans, or to a sample of the population within a study.
This document is a draft for review purposes only and does not constitute Agency policy.
22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Systematic Review Protocol for the Methylmercury IRIS Assessment
Critically deficient reflects a judgment that the study conduct introduced a serious flaw that
makes the study uninterpretable. Studies with a determination of critically deficient in an
evaluation domain will almost always be considered overall "uninformative". For example,
in assessing MeHg DNT studies, the studies using DNT tests that are considered deficient or
using inappropriate analytical chemistry methods will be considered critically deficient.
Once the evaluation domains have been rated, the identified strengths and limitations will
be considered to reach a study confidence classification of high, medium, or low confidence, or
uninformative for each specific health outcome. This classification is based on the reviewer
judgments across the evaluation domains and includes consideration of the likely impact the noted
deficiencies in bias and sensitivity or inadequate reporting have on the results. The classifications,
which reflect a consensus judgment between reviewers, are defined as follows:
High confidence: A well-conducted study with no notable deficiencies or concerns identified;
the potential for bias is unlikely or minimal, and the study used sensitive methodology. High
confidence studies generally reflect judgments of good across all or most evaluation
domains.
Medium confidence: A satisfactory (acceptable) study where deficiencies or concerns are
noted, but the limitations are unlikely to be of a notable degree. Generally, medium
confidence studies include adequate or good judgments across most domains, with the
impact of any identified limitation not being judged as severe.
Low confidence: A substandard study where deficiencies or concerns are noted, and the
potential for bias or inadequate sensitivity could have a significant impact on the study
results or their interpretation. Typically, /ow-confidence studies have a deficient evaluation
for one or more domains, although some medium-confidence studies may have a deficient
rating in domain(s) considered to have less influence on the magnitude or direction of effect
estimates. Generally, /ow-confidence results are given less weight compared to high- or
medium- confidence results during evidence synthesis and integration, and are generally not
used as the primary sources of information for derivation of toxicity values unless they are
the only studies available. Studies rated as low confidence only because of sensitivity
concerns about bias towards the null would require additional consideration during
evidence synthesis. Observing an effect in these studies may increase confidence, assuming
the study is otherwise well conducted (see Section 9).
Uninformative: An unacceptable study where serious flaw(s) make the study results
unusable for informing dose response. Studies with critically deficient judgments in any
evaluation domain are almost always classified as uninformative (see explanation above).
Studies with multiple deficient judgments across domains may also be considered
uninformative. Uninformative studies will not be considered further in the dose-response
analysis, but may be used to highlight possible research gaps.
Study evaluation determinations reached by each reviewer and the consensus judgment
between reviewers will be documented, and final study evaluations will be made available when
the draft is publicly released. The study confidence classifications and their rationales will be
This document is a draft for review purposes only and does not constitute Agency policy.
23 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Systematic Review Protocol for the Methylmercury IRIS Assessment
carried forward and considered as part of selecting studies for dose-response, to aid in the
interpretation of results across studies.
The principles and framework used for evaluating epidemiology studies are adapted from
the principles in the Cochrane Risk of Bias in Nonrandomized Studies of Interventions [ROBINS-I;
(Sterne etal.. 20161]. modified to address environmental and occupational exposures. The
underlying philosophy of ROBINS-I is to describe attributes of an "ideal" study with respect to each
of the evaluation domains (e.g., exposure measurement, outcome classification). Core and
prompting questions are used to collect information to guide evaluation of each domain.
Core and prompting questions, as well as additional considerations that apply to most
outcomes for each domain are presented in Table 4. Core questions represent key concepts, while
the prompting questions help the reviewer focus on relevant details under each key domain.
Exposure- and outcome-specific criteria to use during evaluation of studies will be developed using
the core and prompting questions and refined during a pilot phase with engagement from
topic-specific experts. The types of information that might be the focus of those criteria are listed in
Table 5.
This document is a draft for review purposes only and does not constitute Agency policy.
24 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table 4. Questions to guide the development of criteria for each domain in epidemiology studies
Domain and core
question
Prompting questions
Follow-up questions
Considerations that apply to most exposures and outcomes
Exposure
measurement
Does the exposure
measure reliably
distinguish
between levels of
exposure in a time
window
considered most
relevant for a
causal effect with
respect to the
development of
the outcome?
For all:
Does the exposure measure capture the
variability in exposure among the participants,
considering intensity, frequency, and duration of
exposure?
Does the exposure measure reflect a relevant
time window? If not, can the relationship
between measures in this time and the relevant
time window be estimated reliably?
Was the exposure measurement likely to be
affected by a knowledge of the outcome?
Was the exposure measurement likely to be
affected by the presence of the outcome
(i.e., reverse causality)?
For case-control studies of occupational exposures:
Is exposure based on a comprehensive job
history describing tasks, setting, time period, and
use of specific materials?
For biomarkers of exposure, general population:
Is a standard assay used? What are the intra-
and inter-assay coefficients of variation? Is the
assay likely to be affected by contamination?
Are values less than the limit of detection dealt
with adequately?
What exposure time period is reflected by the
biomarker? If the half-life is short, what is the
correlation between serial measurements of
exposure?
Is the degree of exposure
misclassification likely to
vary by exposure level?
If the correlation
between exposure
measurements is
moderate, is there an
adequate statistical
approach to ameliorate
variability in
measurements?
If there is a concern
about the potential for
bias, what is the
predicted direction or
distortion of the bias on
the effect estimate (if
there is enough
information)?
These considerations require customization to the exposure and outcome
(relevant timing of exposure)
Good
Valid exposure assessment methods used, which represent the etiologically
relevant time period of interest.
Exposure misclassification is expected to be minimal.
Adequate
Valid exposure assessment methods used, which represent the etiologically
relevant time period of interest.
Exposure misclassification may exist but is not expected to greatly change
the effect estimate.
Deficient
Valid exposure assessment methods used, which represent the etiologically
relevant time period of interest. Specific knowledge about the exposure
and outcome raise concerns about reverse causality, but there is
uncertainty whether it is influencing the effect estimate.
Exposed groups are expected to contain a notable proportion of unexposed
or minimally exposed individuals, the method did not capture important
temporal or spatial variation, or there is other evidence of exposure
misclassification that would be expected to notably change the effect
estimate.
Critically deficient
Exposure measurement does not characterize the etiologically relevant time
period of exposure or is not valid.
There is evidence that reverse causality is very likely to account for the
observed association.
Exposure measurement was not independent of outcome status.
This document is a draft for review purposes only and does not constitute Agency policy.
25 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Domain and core
question
Prompting questions
Follow-up questions
Considerations that apply to most exposures and outcomes
Outcome
ascertainment
Does the outcome
measure reliably
distinguish the
presence or
absence(or
degree of severity)
of the outcome?
For all:
Is outcome ascertainment likely to be affected by
knowledge of, or presence of, exposure
(e.g., consider access to health care, if based on
self-reported history of diagnosis)?
For case-control studies:
Is the comparison group without the outcome
(e.g., controls in a case-control study) based on
objective criteria with little or no likelihood of
inclusion of people with the disease?
For mortality measures:
How well does cause of death data reflect
occurrence of the disease in an individual? How
well do mortality data reflect incidence of the
disease?
For diagnosis of disease measures:
Is the diagnosis based on standard clinical
criteria? If it is based on self-report of the
diagnosis, what is the validity of this measure?
For laboratory-based measures (e.g., hormone levels):
Is a standard assay used? Does the assay have
an acceptable level of inter-assay variability? Is
the sensitivity of the assay appropriate for the
outcome measure in this study population?
Is there a concern that
any outcome
misclassification is
nondifferential,
differential, or both?
What is the predicted
direction or distortion of
the bias on the effect
estimate (if there is
enough information)?
These considerations require customization to the outcome
Good
High certainty in the outcome definition (i.e., specificity and sensitivity),
minimal concerns with respect to misclassification.
Assessment instrument was validated in a population comparable to the
one from which the study group was selected.
Adequate
Moderate confidence that outcome definition was specific and sensitive,
some uncertainty with respect to misclassification but not expected to
greatly change the effect estimate.
Assessment instrument was validated but not necessarily in a population
comparable to the study group.
Deficient
Outcome definition was not specific or sensitive.
Uncertainty regarding validity of assessment instrument.
Critically deficient
Invalid/insensitive marker of outcome.
Outcome ascertainment is very likely to be affected by knowledge of, or
presence of, exposure.
Note: Lack of blinding should not be automatically construed to be critically
deficient.
This document is a draft for review purposes only and does not constitute Agency policy.
26 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Participant
selection
Is there evidence
that selection into
or out of the study
(or analysis
sample) was
jointly related to
exposure and to
outcome?
For longitudinal cohort:
Did participants volunteer for the cohort based
on knowledge of exposure and/or preclinical
disease symptoms? Was entry into the cohort or
continuation in the cohort related to exposure
and outcome?
For occupational cohort:
Did entry into the cohort begin with the start of
the exposure?
Was follow-up or outcome assessment
incomplete, and if so, was follow-up related to
both exposure and outcome status?
Could exposure produce symptoms that would
result in a change in work assignment/work
status ("healthy worker survivor effect")?
For case-control study:
Were controls representative of population and
time periods from which cases were drawn?
Are hospital controls selected from a group
whose reason for admission is independent of
exposure?
Could recruitment strategies, eligibility criteria,
or participation rates result in differential
participation relating to both disease and
exposure?
For population based- survey:
Was recruitment based on advertisement to
people with knowledge of exposure, outcome,
and hypothesis?
Were differences in
participant enrollment
and follow-up evaluated
to assess bias?
If there is a concern
about the potential for
bias, what is the
predicted direction or
distortion of the bias on
the effect estimate (if
there is enough
information)?
Were appropriate
analyses performed to
address changing
exposures over time in
relation to symptoms?
Is there a comparison of
participants and
nonparticipants to
address whether
differential selection is
likely?
These considerations may require customization to the outcome. This could
include determining what study designs effectively allow analyses of associations
appropriate to the outcome measures (e.g., design to capture incident vs.
prevalent cases, design to capture early pregnancy loss).
Good
Minimal concern for selection bias based on description of recruitment
process (e.g., selection of comparison population, population-based random
sample selection, recruitment from sampling frame including current and
previous employees).
Exclusion and inclusion criteria specified and would not induce bias.
Participation rate is reported at all steps of study (e.g., initial enrollment,
follow-up, selection into analysis sample). If rate is not high, there is
appropriate rationale for why it is unlikely to be related to exposure
(e.g., comparison between participants and nonparticipants or other
available information indicates differential selection is not likely).
Adequate
Enough of a description of the recruitment process to be comfortable that
there is no serious risk of bias.
Inclusion and exclusion criteria specified and would not induce bias.
Participation rate is incompletely reported but available information
indicates participation is unlikely to be related to exposure.
Deficient
Little information on recruitment process, selection strategy, sampling
framework and/or participation OR aspects of these processes raises the
potential for bias (e.g., healthy worker effect, survivor bias).
Critically deficient
Aspects of the processes for recruitment, selection strategy, sampling
framework, or participation result in concern that selection bias resulted in
a large impact on effect estimates (e.g., convenience sample with no
information about recruitment and selection, cases and controls are
recruited from different sources with different likelihood of exposure,
recruitment materials stated outcome of interest and potential participants
are aware of or are concerned about specific exposures).
This document is a draft for review purposes only and does not constitute Agency policy.
27 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Confounding
Is confounding of
the effect of the
exposure likely?
Is confounding adequately addressed by
considerations in:
Participant selection (matching or restriction)?
Accurate information on potential confounders
and statistical adjustment procedures?
Lack of association between confounder and
outcome, or confounder and exposure in the
study?
Information from other sources?
Is the assessment of confounders based on a
thoughtful review of published literature, potential
relationships (e.g., as can be gained through directed
acyclic graphing), and minimizing potential
overcontrol (e.g., inclusion of a variable on the
pathway between exposure and outcome)?
If there is a concern
about the potential for
bias, what is the
predicted direction or
distortion of the bias on
the effect estimate (if
there is enough
information)?
These considerations require customization to the exposure and outcome, but
this may be limited to identifying key covariates.
Good
Conveys strategy for identifying key confounders. This may include: a priori
biological considerations, published literature, causal diagrams, or statistical
analyses; with recognition that not all "risk factors" are confounders.
Inclusion of potential confounders in statistical models not based solely on
statistical significance criteria (e.g., p < 0.05 from stepwise regression).
Does not include variables in the models that are likely to be influential
colliders or intermediates on the causal pathway.
Key confounders are evaluated appropriately and considered to be unlikely
sources of substantial confounding. This often will include:
o Presenting the distribution of potential confounders by levels of
the exposure of interest and/or the outcomes of interest (with
amount of missing data noted);
o Consideration that potential confounders were rare among the
study population, or were expected to be poorly correlated with
exposure of interest;
o Consideration of the most relevant functional forms of potential
confounders;
o Examination of the potential impact of measurement error or
missing data on confounder adjustment.
Adequate
Similar to Good but may not have included all key confounders, or less detail may
be available on the evaluation of confounders (e.g., sub-bullets in Good). It is
possible that residual confounding could explain part of the observed effect, but
concern is minimal.
Deficient
Does not include variables in the models that are likely to be influential
colliders or intermediates on the causal pathway.
And any of the following:
The potential for bias to explain some of the results is high based on an
inability to rule out residual confounding, such as a lack of demonstration
This document is a draft for review purposes only and does not constitute Agency policy.
28 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Domain and core
question
Prompting questions
Follow-up questions
Considerations that apply to most exposures and outcomes
that key confounders of the exposure-outcome relationships were
considered;
Descriptive information on key confounders (e.g., their relationship relative
to the outcomes and exposure levels) are not presented; or
Strategy of evaluating confounding is unclear or is not recommended
(e.g., only based on statistical significance criteria or stepwise regression
[forward or backward elimination]).
Critically deficient
Includes variables in the models that are colliders and/or intermediates in
the causal pathway, indicating that substantial bias is likely from this
adjustment; or
Confounding is likely present and not accounted for, indicating that all of
the results were most likely due to bias.
o Presenting a progression of model results with adjustments for
different potential confounders, if warranted.
This document is a draft for review purposes only and does not constitute Agency policy.
29 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Analysis
Does the analysis
strategy and
presentation
convey the
necessary
familiarity with the
data and
assumptions?
Are missing outcome, exposure, and covariate
data recognized, and if necessary, accounted for
in the analysis?
Does the analysis appropriately consider variable
distributions and modeling assumptions?
Does the analysis appropriately consider
subgroups of interest (e.g., based on variability in
exposure level or duration or susceptibility)?
Is an appropriate analysis used for the study
design?
Is effect modification considered, based on
considerations developed a priori?
Does the study include additional analyses
addressing potential biases or limitations
(i.e., sensitivity analyses)?
If there is a concern
about the potential for
bias, what is the
predicted direction or
distortion of the bias on
the effect estimate (if
there is enough
information)?
These considerations may require customization to the outcome. This could
include the optimal characterization of the outcome variable and ideal statistical
test (e.g., Cox regression).
Good
Use of an optimal characterization of the outcome variable.
Quantitative results presented (effect estimates and confidence limits or
variability in estimates) (i.e., not presented only as a p-value or
"significant"/"not significant").
Descriptive information about outcome and exposure provided (where
applicable).
Amount of missing data noted and addressed appropriately (discussion of
selection issuesmissing at random vs. differential).
Where applicable, for exposure, includes LOD (and percentage below the
LOD), and decision to use log transformation.
Includes analyses that address robustness of findings, e.g., examination of
exposure-response (explicit consideration of nonlinear possibilities,
quadratic, spline, or threshold/ceiling effects included, when feasible);
relevant sensitivity analyses; effect modification examined based only on
a priori rationale with sufficient numbers.
No deficiencies in analysis evident. Discussion of some details may be
absent (e.g., examination of outliers).
Adequate
Same as Good, except:
Descriptive information about exposure provided (where applicable), but
may be incomplete; might not have discussed missing data, cutpoints, or
shape of distribution.
Includes analyses that address robustness of findings (examples in Good),
but some important analyses are not performed.
Deficient
Does not conduct analysis using optimal characterization of the outcome
variable.
This document is a draft for review purposes only and does not constitute Agency policy.
30 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Domain and core
question
Prompting questions
Follow-up questions
Considerations that apply to most exposures and outcomes
Descriptive information about exposure levels not provided (where
applicable).
Effect estimate and p-value presented, without standard error or confidence
interval.
Results presented as statistically "significant"/"not significant."
Critically deficient
Results of analyses of effect modification examined without clear a priori
rationale and without providing main/principal effects (e.g., presentation
only of statistically significant interactions that were not hypothesis driven).
Analysis methods are not appropriate for design or data of the study.
Selective
reporting
Is there reason to
be concerned
about selective
reporting?
Were results provided for all the primary
analyses described in the methods section?
Is there appropriate justification for restricting
the amount and type of results that are shown?
Are only statistically significant results
presented?
If there is a concern
about the potential for
bias, what is the
predicted direction or
distortion of the bias on
the effect estimate (if
there is enough
information)?
These considerations generally do not require customization and may have fewer
than four levels.
Good
The results reported by study authors are consistent with the primary and
secondary analyses described in a registered protocol or methods paper.
Adequate
The authors described their primary (and secondary) analyses in the
methods section and results were reported for all primary analyses.
Deficient
Concerns were raised based on previous publications, a methods paper, or a
registered protocol indicating that analyses were planned or conducted that
were not reported, or that hypotheses originally considered to be secondary
were represented as primary in the reviewed paper.
Only subgroup analyses were reported suggesting that results for the entire
group were omitted.
Only statistically significant results were reported.
This document is a draft for review purposes only and does not constitute Agency policy.
31 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Domain and core
question
Prompting questions
Follow-up questions
Considerations that apply to most exposures and outcomes
Sensitivity
Is there a concern
that sensitivity of
the study is not
adequate to detect
an effect?
Is the exposure range adequate to detect
associations and exposure-response
relationships?
Was the appropriate population included?
Was the length of follow-up adequate? Is the
time/age of outcome ascertainment optimal
given the interval of exposure and the health
outcome?
Are there other aspects related to risk of bias or
otherwise that raise concerns about sensitivity?
These considerations may require customization to the exposure and outcome,
and may have fewer than four levels. Some study features that affect study
sensitivity may have already been included in the other evaluation domains.
Other features that have not been addressed should be included here. Some
examples include:
Adequate
The range of exposure levels provides adequate variability to evaluate the
relevant associations.
The population was exposed to levels expected to have an impact on
response.
The study population was sensitive to the development of the outcomes of
interest (e.g., ages, life stage, sex).
The timing of outcome ascertainment was appropriate given expected
latency for outcome development (i.e., adequate follow-up interval).
The study was adequately powered to observe an association based on
underlying population sensitivity and exposure contrasts.
No other concerns raised regarding study sensitivity.
Deficient
Concerns were raised about the issues described for adequate that are
expected to notably decrease the sensitivity of the study to detect
associations for the outcome.
This document is a draft for review purposes only and does not constitute Agency policy.
32 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table 5. Information relevant to evaluation domains for epidemiology studies
Domain
Types of information that might need to be collected or are important for evaluating
the domain
Exposure
measurement
Source(s) of exposure (e.g., consumer products, occupational, an industrial accident) and
source(s) of exposure data, blinding to outcome, level of detail for job history data, when
measurements were taken, type of biomarker(s), assay information, reliability data from repeat
measures studies, validation studies.
Outcome
ascertainment
Source of outcome (effect) measure, blinding to exposure status or level, how
measured/classified, incident vs. prevalent disease, evidence from validation studies, prevalence
(or distribution summary statistics for continuous measures).
Participant
selection
Study design, where and when was the study conducted, and who was included? Recruitment
process, exclusion and inclusion criteria, type of controls, total eligible, comparison between
participants and nonparticipants (or followed and not followed), and final analysis group. Does
the study include potential susceptible populations or life stages (see discussion in Section 9)?
Confounding
Background research on key confounders for specific populations or settings; participant
characteristic data, by group; strategy/approach for consideration of potential confounding;
strength of associations between exposure and potential confounders and between potential
confounders and outcome; and degree of exposure to the confounder in the population.
Analysis
Extent (and if applicable, treatment) of missing data for exposure, outcome, and confounders;
approach to modeling; classification of exposure and outcome variables (continuous vs.
categorical); testing of assumptions; sample size for specific analyses; and relevant sensitivity
analyses.
Sensitivity
What are the ages of participants (e.g., not too young in studies of pubertal development)?
What is the length of follow-up (for outcomes with long latency periods)? Choice of referent
group, the exposure range, and the level of exposure contrast between groups (i.e., the extent to
which the "unexposed group" is truly unexposed, and the prevalence of exposure in the group
designated as "exposed").
Selective
reporting
Are results presented with adequate detail for all endpoints and exposure measures reported in
the methods section, and are they relevant to the PECO? Are results presented for the full
sample and for specified subgroups? Were stratified analyses (effect modification) motivated by
a specific hypothesis?
1 Evaluation of MeHg epidemiology studies includes evaluation of the analytical chemistry
2 methods and the associated QA/QC procedures that were employed. For this purpose, criteria were
3 developed for assessing the analytical chemistry methods used for analysis of mercury/MeHg in
4 blood and in hair. These criteria are presented in Appendix C.
6.2. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL
DESCRIPTIVE SUMMARY AND EVALUATION
5 PBPK (and/or classic PK) models should be used in an assessment when an applicable one
6 exists and no equal or better alternative for dosimetric extrapolation is available. Any models used
7 should represent current scientific knowledge and accurately translate the science into
8 computational code in a reproducible, transparent manner. For a specific target organ/tissue, using
This document is a draft for review purposes only and does not constitute Agency policy.
33 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the Methylmercury IRIS Assessment
or adapting an existing PK/PBPK model or developing a new model or an alternative quantitative
approach might be possible following the EPA Quality Assurance Plan for PBPK models fU.S. EPA.
20181. Data for PK/PBPK models might come from studies across various species and could be
from in vitro or in vivo model systems.
Because the aim of this assessment is to update the existing RfD by reevaluating the DNT
effects associated with MeHg exposure as assessed by MeHg biomarkers (e.g., cord blood), applying
pharmacokinetic models in conjunction with biomarker data to estimate MeHg intake doses is
essential. In general, two major types of pharmacokinetic models are available for studying MeHg
toxicokinetics: a one-compartment (classic PK) model and a multicompartment PBPK model,
hereinafter referred to as the PBPK model (U.S. EPA. 2001a). The one-compartment model is the
simplest form of a pharmacokinetic model in that the entire body is assumed to act like a single,
uniform entity that uses only one volume term, the apparent volume of distribution. As shown in
controlled human studies, the absorption rate of MeHg generally is much faster than its elimination
rate. In general, the one-compartment model describes existing MeHg data reasonably well and
therefore has been the model most often used for estimating MeHg intake doses since 2001.
In 2001, EPA employed a one-compartment (classic PK) model, adapted from the NRC
(2000) model, to derive the existing RfD for MeHg by estimating ingestion doses (mg/kg-day) with
the use of measured hair mercury from a female population with a consistently high consumption
of fish and whale meat in the Faroe Islands (Grandiean et al.. 1997). For better characterization of
uncertainty and variability in estimating MeHg intake, the 2001 EPA assessment adopted a range of
46-79 ppb of total Hg in maternal blood instead of using a fixed cord blood level of 58 ng/L used by
the NRC f20001 model. The Agency for Toxic Substances and Disease Registry fATSDR. 19991. on
the basis of a study of the population with constant fish consumption in the Seychelles Islands,
likewise adopted a one-compartment model with an overall UF of 3 to estimate a chronic oral
Minimal Risk Level for MeHg. That MeHg levels could reach a steady-state after approximately five
elimination half-lives for those who have constant exposure to MeHg has been estimated, as is the
case for the populations in the Seychelles and Faroes Islands fNRC. 20001.
In comparison, the PBPK model typically includes several pharmacokinetic parameters
(e.g., tissue volumes, partition coefficients, rate constants for metabolism and elimination) that can
vary from one individual to another within the subpopulation of interest These compartments
represent organs and tissues that are interconnected in the body via blood flow. PBPK models are
conceptually more accurate in predicting body burden of MeHg (e.g., internal dose) as compared to
one-compartment models. PBPK modeling is typically considered to be more complex and data
intensive than a PK model as it requires more comprehensive ADME data for model development
and validation (e.g., for characterization of uncertainty and variability associated with the model
parameters and model outputs).
As the pharmacokinetic modeling needs to reflect a balance between the principles of model
parsimony and plausibility (i.e., reflective of physiological reality), both the one-compartment
This document is a draft for review purposes only and does not constitute Agency policy.
34 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
1 (classic PK) model and the PBPK model will be considered in this assessment. To ensure
2 consistency, both types of toxicokinetic models will be evaluated against measured biomarker data
3 (e.g., hair and maternal blood) following the EPA guidelines, as articulated in Section 6.3.2.
6.2.1. Pharmacokinetic/Physiologically Based Pharmacokinetic (PK/PBPK) Model
Descriptive Summary
4 Classic one-compartment PK and PBPK models for analyzing MeHg in humans that have
5 been published since 2001 are summarized in Table 6.
Table 6. Classic one-compartment PK and PBPK models for MeHg in humans
since 2001
Reference
Notes
Classic (one-compartment) PK model
fStern. 2005: Stern and Smith.
2003: Stern et al.. 20021
This series of papers extends the 2001 EPA MeHg assessment work to improve
the uncertainty and variability analysis. The refinements include the adoption
of the Bayesian approach and the use of a range of ratios for cord blood to
maternal blood for the maternal dose reconstruction.
fSirot etal.. 20081
This analysis estimates MeHg intake dose using an EPA one-compartment
model and compared it with the intake estimated by the food frequency
questionnaire in French frequent seafood consumers.
fAlbert et al.. 20101
This studv describes a modified one-comDartment model based on the WHO
fl9901 model to estimate integrated variabilitv in dietarv MeHe intake and
MeHg half-life in blood and to predict mercury level in hair among pregnant
women who consumed seafood.
("Yamnuma-Sakurai et al..
2012)
These authors used a one-compartment model described previously by NRC
f20001a and a two-wav analvsis of variance approach to estimate time-
dependent hair-to-blood ratio and half-life of blood mercury in a controlled
human study.
Ho et al.. 20151
This study estimates the between-person variability of the MeHg half-life in
Korean adults using the same model as described bv Swartoutand Rice
(2000).a
CLi etal.. 20151
This studv uses the same one-compartment model of Stern (2005) to evaluate
the relationship between MeHg intake and blood and hair MeHg levels in a
rice-consuming population in China.
PBPK model
fBvczkowski and Lipscomb.
2001)
This studv describes an extension and a refinement of Clewell et al. (1999)
model for gestational transfer, along with lactation transfer of MeHg from the
exposed mother to the fetus. The results from model simulation were
compared with experimental data obtained from rodents for which the model
parameters were scaled to humans using allometric procedures.
This document is a draft for review purposes only and does not constitute Agency policy.
35 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Reference
Notes
fCarrier etal.. 2001b: Carrier
et al.. 2001al
These authors first developed a biologically based dynamic multicompartment
model for predicting MeHg using animal data Carrier et al. (2001b) and then
reparametrized it for humans. Although the model prediction is generally
comparable to that of Clewell et al. (1999), this model is considered less
informative due to lack of sensitivity and uncertainty analysis.
fYoune et al.. 20011
This analysis describes the deposition of mercury (both MeHg and inorganic Hg)
in humans using empirical animal data (hamster, rat, guinea pig, cat, rabbit,
monkey, sheep, pig, goat, cow). An allometric approach was used to estimate
key kinetic parameters (e.g., metabolism rate constants) in the development of
the human model, followed by model validation using human autopsy data.
fLeggett et al.. 20011
This study is a literature review of biokinetic models for the deposition of
inhaled mercury vapor in the respiratory tract and different patterns of
absorption in blood in animals and humans. As the focus of the current
assessment is focused on oral exposure, this study is considered irrelevant.
fMcnallv and Loizou. 2015: Lu
etal.. 2012: Ruiz etal.. 2010:
Kirman et al.. 2 00 3:
Pierrehumbert et al.. 20021
These studies propose a generic PBPK model platform with no specific focus on
MeHg. They do not add value to the MeHg PBPK database.
fNoisel et al.. 2011: Gosselin
etal.. 20061
These authors used the model of Carrier et al. (2001b) and data sets on
measured total Hg in hair and blood for reconstruction of the likely monthly
MeHg intakes among fish consumption populations in Brazil and Canada. No
pregnancy and lactation compartments are included in the analysis.
fAllen et al.. 20071
This studv uses the same model structure as Clewell et al. (1999) but
incorporates a Bayesian approach, implemented using Markov Chain Monte
Carlo analysis to characterize interindividual variation in exposure to MeHg and
in the pharmacokinetics (distribution, clearance) of MeHg.
fBerthet et al.. 20101
This study presents a generic two-compartment model for 14 chemicals
(e.g., mercury, arsenic, cadmium) to characterize biological monitoring
variability.
fLee et al.. 20171
This studv uses the Clewell et al. f19991 model as the template but removes
the gestation-related compartments (e.g., uterus, fetus) from the model.
fAbass etal.. 20181
This work consists of several linear toxicokinetic equations for depicting
toxicokinetics for MeHg, inorganic Hg, and metallic Hg. Compared to the
estimated intake of Hg using a food frequency questionnaire, the predicted
intake using toxicokinetic modeling based on total mercury levels in the blood
tended to be higher.
fOuetal.. 20181
This study examines a model derived on the basis of a reparameterization of the
Clewell model (1999) and organized into three sub-models: (1) the pregnancy
model, (2) the lactation model (for lactating mothers), and (3) the infant model
(for suckling infants) with the data on repeated measurements of MeHg in
children's hair up to 1 year of age.
aBoth NRC (2000) and Swartout and Rice (2000) models were derived from the WHO (1990) model.
This document is a draft for review purposes only and does not constitute Agency policy.
36 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Systematic Review Protocol for the Methylmercury IRIS Assessment
6.2.2. Pharmacokinetic/Physiologically Based Pharmacokinetic (PK/PBPK) Model
Evaluation
Once available PBPK models are summarized, the assessment team will evaluate the models
in accordance with criteria outlined in U.S. EPA (20181. Judgments on the suitability of a model are
separated into two categories: scientific and technical (Table 7). The scientific criteria focus on
whether the biology, chemistry, and other information available for chemical mode(s) of action are
justified (i.e., preferably with citations to support use) and represented by the model structure and
equations. The scientific criteria are judged on the basis of information presented in the
publication or report that describes the model and do not require evaluation of the computer code.
Initial technical criteria include availability of the computer code and completeness of parameter
listing and documentation. Studies that meet the preliminary scientific and technical criteria are
then subjected to an in-depth technical evaluation, which includes a thorough review and testing of
the computational code. The in-depth technical and scientific analyses focus on the accurate
implementation of the conceptual model in the computational code, use of scientifically supported
and biologically consistent parameters in the model, and reproducibility of model results reported
in journal publications and other documents. This approach stresses (1) clarity in the
documentation of model purpose, structure, and biological characterization; (2) validation of
mathematical descriptions, parameter values, and computer implementation; and (3) evaluation of
each plausible dose metric. The in-depth analysis is used to evaluate the potential value and cost of
developing a new model or substantially revising an existing one. PBPK models EPA develops
during the course of the assessment will be peer reviewed, either as a component of the draft
assessment or by publication in a journal article.
Table 7. Criteria for evaluating PBPK models
Category
Specific criteria
Scientific
Biological basis for the model is accurate.
Consistent with mechanisms that significantly impact dosimetry.
Predicts dose metric(s) expected to be relevant.
Applicable for relevant route(s) of exposure.
Consideration of model fidelity to the biological system strengthens the scientific basis of the assessment relative
to standard exposure-based extrapolation (default) approaches.
Ability of model to describe critical behavior, such as nonlinear kinetics in a relevant dose range, better
than the default (i.e., BW3/4 scaling).
Model parameterization for critical life stages or windows of susceptibility. Evaluation of these criteria
should also consider the model's fidelity vs. default approaches and possible use of an intraspecies
uncertainty factor in conjunction with the model to account for variations in sensitivity between life
stages.
This document is a draft for review purposes only and does not constitute Agency policy.
37 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Category
Specific criteria
Predictive power of model-based dose metric vs. default approach, based on exposure
o Specifically, model-based metrics may correlate better than the applied doses with animal/human
dose-response data.
o The degree of certainty in model predictions vs. default is also a factor. For example, while target
tissue metrics are generally considered better than blood concentration metrics, lack of data to
validate tissue predictions when blood data are available may lead to choosing the latter.
Principle of Parsimony
Model complexity or biological scale, including number and parameterization of (sub)compartments (e.g.,
tissue or subcellular levels) should be commensurate with data available to identify parameters.
Model describes existing PK data reasonably well, both in "shape" (matches curvature, inflection points, peak
concentration time, etc.) and quantitatively (e.g., within factor of 2-3).
Model equations are consistent with biochemical understanding and biological plausibility.
Initial
Well-documented model code is readily available to EPA and public.
technical
Set of published parameters is clearly identified, including origin/derivation.
Parameters do not vary unpredictably with dose (e.g., any dose dependence in absorption constants is
predictable across the dose ranges relevant for animal and human modeling).
Sensitivity and uncertainty analysis have been conducted for relevant exposure levels (local sensitivity analysis is
sufficient, but global analysis provides more information).
If a sensitivity analysis was not conducted, EPA may decide to independently conduct this additional work
before using the model in the assessment.
A sound explanation should be provided when sensitivity of the dose metric to model parameters differs
from what is reasonably expected based on experience.
This document is a draft for review purposes only and does not constitute Agency policy.
38 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the Methylmercury IRIS Assessment
7. DATA EXTRACTION OF STUDY METHODS AND
RESULTS
Data extraction and content management will be carried out using HAWC (Health
Assessment Workspace Collaborative). Data extraction elements that may be collected from
epidemiology studies are listed in Appendix B. The content of the data extraction may be revised
following the identification of the studies included in the review as part of a pilot phase to assess
the data extraction workflow. Not all studies that meet the PECO criteria go through data
extraction. Studies evaluated as being uninformative are not considered further and would,
therefore, not undergo data extraction. The same may be true for /ow-confidence studies if
sufficient medium- and /jzg/j-confidence studies are available. All findings are considered for
extraction, regardless of statistical significance, although the level of extraction for specific
outcomes within a study may differ (i.e., ranging from a narrative to full extraction of dose-response
effect size information). Similarly, decisions about data extraction for low-confidence studies are
typically made during implementation of the protocol based on consideration of the quality and
extent of the available evidence. The version of the protocol released with the draft assessment will
outline how low-confidence studies were treated for extraction and evidence synthesis.
The data extraction results for included studies will be presented in the assessment and
made available for download from EPA HAWC in Excel format when the draft is publicly released.
Data extraction will be performed by one member of the evaluation team and checked by one or
two other members. Discrepancies in data extraction will be resolved by discussion or consultation
with a third member of the evaluation team if needed. Once the data have been verified, they will
be "locked" to prevent accidental changes. Digital rulers, such as WebPlotDigitizer
fhttps: //automeris.io/WebPlotDigitizer]. are used to extract numerical information from figures.
Use of digital rulers is documented during extraction.
As previously described, routine attempts will be made to obtain information missing from
human health effect studies, if it is considered influential during study evaluations (see Section 6)
or when it can provide information required to conduct a meta-analysis (e.g., missing group size or
variance descriptors such as standard deviation or confidence interval). Missing data from
individual mechanistic (e.g., in vitro) studies will generally not be sought Outreach to study
authors should be documented and considered unsuccessful if researchers do not respond to an
email or phone request within 1 month of the attempt to contact.
This document is a draft for review purposes only and does not constitute Agency policy.
39 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the Methylmercury IRIS Assessment
8. DOSE-RESPONSE ASSESSMENT: STUDY
SELECTION AND QUANTITATIVE ANALYSIS
This section of the protocol provides an overview of considerations for conducting the
dose-response assessment, particularly statistical considerations specific to dose-response analysis
that support quantitative risk assessment Importantly these considerations do not supersede
existing EPA guidance.
A MeHg oral RfD will be derived. An RfD is an estimate, with uncertainty spanning perhaps
an order of magnitude, of an exposure to the human population (including susceptible subgroups)
that is likely to be without an appreciable risk of deleterious health effects over a lifetime ("U.S. EPA.
2002, §4.2). Reference values are not predictive risk values; that is, they provide no information
about risks at higher or lower exposure levels.
8.1. SELECTING STUDIES FOR DOSE-RESPONSE ASSESSMENT
The dose-response assessment begins with a review of the DNT effects, particularly among
the studies of highest quality and that exemplify the study attributes summarized in Table 7. This
review also considers whether there are opportunities for quantitative evidence integration.
Examples of quantitative integration, from simplest to more complex, include (1) characterizing
overall toxicity, as in combining effects that comprise a syndrome, or occur on a continuum (e.g.,
precursors and eventual overt toxicity) and (2) conducting a meta-analysis or meta-regression of
all studies addressing a category of important health effects.
Studies of low sensitivity may be less useful if they fail to detect a true effect or yield points
of departure with wide confidence limits, but such studies would be considered for inclusion in a
meta-analysis.
Studies most useful for dose-response analysis generally have at least one exposure level in
the region of the dose-response curve near the benchmark response (the response level to be used
for deriving toxicity values), to minimize low-dose extrapolation, and more exposure levels and
larger sample sizes overall (U.S. EPA. 2012). These attributes support a more complete
characterization of the shape of the exposure-response curve and decrease the uncertainty in the
associated exposure-response metric (RfD) by reducing statistical uncertainty in the point of
departure and minimizing the need for low-dose extrapolation. In addition to these more general
considerations, specific issues that may impact the feasibility of dose-response modeling for
individual data sets are described in more detail in the Benchmark Dose Technical Guidance fU.S.
EPA. 20121.
This document is a draft for review purposes only and does not constitute Agency policy.
40 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table 8. Attributes used to evaluate studies for derivation of toxicity values
Study attributes
Considerations for human studies
Study confidence
High- or med/um-confidence studies are highly preferred over /ow-confidence studies. The available high and medium confidence studies are further
differentiated based on the study attributes below as well as a reconsideration of the specific limitations identified and their potential impact on
dose-response analyses.
Rationale for choice of species
Human data are preferred over animal data to eliminate interspecies extrapolation uncertainties (e.g., in toxicodynamics, relevance of specific health
outcomes to humans).
Relevance of
exposure
paradigm
Exposure
durations
Studies involving human environmental exposures (oral, inhalation).
Exposure levels
When developing a chronic toxicity value, chronic or subchronic studies are preferred over studies of acute exposure durations. Exceptions exist, such
as when a susceptible population or life stage is more sensitive in a particular time window (e.g., developmental exposure).
Exposure route
Exposures near the range of typical environmental human exposures are preferred. Studies with a broad exposure range and multiple exposure levels
are preferred to the extent that they can provide information about the shape of the exposure-response relationship [see the EPA Benchmark Dose
Technical Guidance (U.S. EPA. 2012, §2.1.1)1 and facilitate extrapolation to more relevant (generally lower) exposures.
Subject selection
Studies that provide risk estimates in the most susceptible groups are preferred.
Controls for possible
confounding3
Studies with a design or analysis (e.g., covariates or other procedures for statistical adjustment) that adequately address the relevant sources of
potential critical confounding for a given outcome are preferred.
Measurement of exposure
Studies that can reliably distinguish between levels of exposure in a time window considered most relevant for development of a causal effect are
preferred. Exposure assessment methods that provide measurements at the level of the individual and that reduce measurement error are preferred.
Measurements of exposure should not be influenced by knowledge of health outcome status.
Measurement of health
outcome(s)
Studies that can reliably distinguish the presence or absence (or degree of severity) of the outcome are preferred. Outcome ascertainment methods
using generally accepted or standardized approaches are preferred.
Studies with individual data are preferred in general. Examples include: to characterize experimental variability more realistically, to characterize
overall incidence of individuals affected by related outcomes (e.g., phthalate syndrome).
Among several relevant health outcomes, preference is generally given to those with greater biological significance.
Study size and design
Preference is given to studies using designs reasonably expected to have power to detect responses of suitable magnitude.15 This does not mean that
studies with substantial responses but low power would be ignored, but that they should be interpreted in light of a confidence interval or variance for
the response. Studies that address changes in the number at risk (through decreased survival, loss to follow-up) are preferred.
aAn exposure or other variable that is associated with both exposure and outcome, but is not an intermediary between the two.
bPower is an attribute of the design and population parameters, based on a concept of repeatedly sampling a population; it cannot be inferred post hoc using data from one
experiment (Hoenig and Heisev. 2001).
This document is a draft for review purposes only and does not constitute Agency policy.
41 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the Methylmercury IRIS Assessment
8.2. CONDUCTING DOSE-RESPONSE ASSESSMENTS
EPA uses a two-step approach for dose-response assessment that distinguishes analysis of
the dose-response data in the range of observation from any inferences about responses at lower
environmentally relevant exposure levels fU.S. EPA. 2012: 2005, §3):
1) Within the observed dose range, the preferred approach is to use dose-response modeling
to incorporate as much of the data set as possible into the analysis. This modeling yields a
POD, an exposure level ideally near the lower end of the range of observation, without
significant extrapolation to lower exposure levels. See Section 8.2.1 for more details.
2) Derivation of reference values nearly always involves extrapolation to exposures lower than
the POD and is described in more detail in Section 8.2.3.
For reference values, IRIS assessments typically derive a candidate value from each suitable
data set. Evaluating these candidate values grouped within a particular DNT domain yields a single
DNT value for each domain under consideration. Next, evaluation of these domain values results in
the selection of a single overall reference value to cover all DNT domains. While this overall
reference value is the focus of the assessment, the domain values can be useful for subsequent
cumulative risk assessments that consider the combined effect of multiple agents acting on a
common domain, regardless of the domain selected as the basis for an RfD.
8.2.1. Dose-response Analysis in the Range of Observation
For conducting a dose-response assessment, toxicodynamic ("biologically based") modeling
can be used when there are sufficient data to ascertain the mode of action and quantitatively
support model parameters that represent rates and other quantities associated with the key
precursor events of the mode of action. Toxicodynamic modeling is potentially the most
comprehensive way to account for the biological processes involved in a response. Such models
seek to reflect the sequence of key precursor events that lead to a response. Toxicodynamic models
can contribute to dose-response assessment by revealing and describing nonlinear relationships
between internal dose and response. Such models may provide a useful approach for analysis in
the range of observation, provided the purpose of the assessment justifies the effort involved.
When a toxicodynamic model is not available for dose-response assessment or when the
purpose of the assessment does not warrant developing such a model, empirical modeling should
be used to fit the data (on the apical outcome or a key precursor event) in the range of observation.
For this purpose, EPA has developed a standard set of models (http: //www.epa.gov/bmdsl that can
be applied to typical data sets, including those that are nonlinear. In situations where there are
alternative models with significant biological support, the decision maker can be informed by the
presentation of these alternatives along with the models' strengths and uncertainties. The EPA has
developed guidance on modeling dose-response data, assessing model fit, selecting suitable models,
This document is a draft for review purposes only and does not constitute Agency policy.
42 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the Methylmercury IRIS Assessment
and reporting modeling results [see the EPA Benchmark Dose Technical Guidance fU.S. EPA. 20121].
Additional judgment or alternative analyses are used if the procedure fails to yield reliable results,
for example, if the fit is poor, modeling may be restricted to the lower doses, especially if there is
competing toxicity at higher doses.
For each modeled response, a POD from the observed data should be estimated to mark the
beginning of extrapolation to lower doses. The POD is an estimated dose (expressed in
human-equivalent terms) near the lower end of the observed range without significant
extrapolation to lower doses. The POD is used as the starting point for subsequent extrapolations
and analyses.
The response level at which the POD is calculated is guided by the severity of the endpoint
For dichotomous data, a response level of 10% extra risk is generally used for minimally adverse
effects, 5% or lower for more severe effects. For continuous data, a response level is ideally based
on an established definition of biological significance. In the absence of such definition, one control
standard deviation from the control mean is often used for minimally adverse effects, one-half
standard deviation for more severe effects. The POD is the 95% lower bound on the dose
associated with the selected response level.
8.2.2. Extrapolation: Slope Factors and Unit Risks
A cancer assessment is not included in the scope of this assessment for MeHg. Accordingly,
this assessment will not derive an oral slope factor or inhalation unit risk.
8.2.3. Extrapolation: Reference Values
Reference value derivation is EPA's most frequently used type of nonlinear extrapolation
method and is most commonly used for noncancer effects.
For each data set selected for reference value derivation, reference values are estimated by
applying relevant adjustments to the PODs to account for the conditions of the reference value
definitionfor human variation, extrapolation from animals to humans (not necessary for this
assessment as only human studies are being evaluated), extrapolation to chronic exposure
duration, and extrapolation to a minimal level of risk (if not observed in the data set). Increasingly,
data-based adjustments fU.S. EPA. 20141 and Bayesian methods for characterizing population
variability (NRC. 20141 are feasible and can be distinguished from the UF considerations outlined
below. The assessment will discuss the scientific bases for estimating these data-based
adjustments and UFs:
Human variation: The assessment accounts for variation in susceptibility across the human
population and the possibility that the available data may not represent individuals who are
most susceptible to the effect, by using a data-based adjustment or UF or a combination of
the two. Where appropriate data or models for the effect or for characterizing the internal
dose are available, the potential for data-based adjustments for toxicodynamics or
This document is a draft for review purposes only and does not constitute Agency policy.
43 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Systematic Review Protocol for the Methylmercury IRIS Assessment
toxicokinetics is considered (U.S. EPA. 2014. 2002).9 10 When sufficient data are available, an
intraspecies UF either less than or greater than 10-fold may be justified (U.S. EPA. 20021.
This factor may be reduced if the POD is derived from or adjusted specifically for susceptible
individuals [not for a general population that includes both susceptible and nonsusceptible
individuals; fU.S. EPA. 2002. §4.4.5: 1998. §4.2: 1996. §4: 1994. §4.3.9.1: 1991. §3.41], When
the use of such data or modeling is not supported, a UF with a default value of 10 is
considered.
LOAEL to NOAEL: If a POD is based on a LOAEL (lowest-observed-adverse-effect level), the
assessment includes an adjustment to an exposure level where such effects are not expected.
This can be a matter of great uncertainty if no evidence is available at lower exposures. A
factor of 3 or 10 is generally applied to extrapolate to a lower exposure expected to be
without appreciable effects (NOAEL, or no-observed-adverse-effect level). A factor other
than 10 may be used, depending on the magnitude and nature of the response and the shape
of the dose-response curve (U.S. EPA. 2002.1998.1996.1994.1991).
Subchronic-to-chronic exposure: When using subchronic studies to make inferences about
chronic/lifetime exposure, the assessment considers whether lifetime exposure could have
effects at lower levels of exposure. A factor of up to 10 may be applied to the POD,
depending on the duration of the studies and the nature of the response fU.S. EPA. 2002.
1998.19941.
Database deficiencies: In addition to the adjustments above, if database deficiencies raise
concern that further studies might identify a more sensitive effect, organ system, or life
stage, the assessment may apply a database UF (U.S. EPA. 2002.1998.1996.1994.1991).
The size of the factor depends on the nature of the database deficiency. For example, the
EPA typically follows the recommendation that a factor of 10 be applied if both a prenatal
toxicity study and a two-generation reproduction study are missing and a factor of
10V2 (i.e., 3) if either one or the other is missing (U.S. EPA. 2002. §4.4.5).
The POD that is used for an RfD is divided by the product of these factors. U.S. EPA (2002.
§4.4.5) recommends that any composite factor that exceeds 3,000 represents excessive uncertainty,
and recommends against relying on the associated RfD.
The derivation of an RfD for DNT health effects for MeHg conducted as part of the current
module will be performed consistent with EPA guidance summarized above.
9Examples of adjusting the toxicokinetic portion of interhuman variability include the IRIS boron
assessment's use of nonchemical-specific kinetic data [e.g., glomerular filtration rate in pregnant humans as a
surrogate for boron clearance (U.S. EPA. 20041] and the IRIS trichloroethylene assessment's use of population
variability in trichloroethylene metabolism, via a PBPK model, to estimate the lower 1st percentile of the dose
metric distribution for each POD (U.S. EPA. 20111.
10Note that when a PBPK model is available for relating human internal dose to environmental exposure,
relevant portions of this UF may be more usefully applied prior to animal-to-human extrapolation, depending
on the correspondence of any nonlinearities (e.g., saturation levels) between species.
This document is a draft for review purposes only and does not constitute Agency policy.
44 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
9. PROTOCOL HISTORY
1 Release date:
2 Revisions history:
This document is a draft for review purposes only and does not constitute Agency policy.
45 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
APPENDICES
APPENDIX A. ELECTRONIC DATABASE SEARCH STRATEGIES
Table A-l. Overall database search strategy
Search
Search strategy
Date and results
PubMed
Chemical terms
(((((("methylmercury"[All Fields] OR "methyl mercury"[AII Fields]) OR
"methyl-mercury"[All Fields]) OR "MeHg"[AII Fields]) OR
"monomethylmercury"[AII Fields]) OR "22967-92-6"[EC/RN Number])
OR "Methylmercure"[AII Fields]) OR " Methylquecksilber"[AII Fields])
AND ("1998"[PDAT]: "3000"[PDAT])
1998-March 2017: 5,028
May 2019 update: 818
Web of Science
Chemical terms
(TS="methylmercury" OR TS="methyl mercury" OR TS="methyl-
mercury" OR TS="Methylmercury (MeHg)" OR
TS="monomethylmercury" OR TS="22967-92-6" OR
TS="Methylmercure" OR TS="Methylquecksilber" OR
TS="methylmercury ii" OR TS="MeHg") AND PY=(1998-2017)
1998-March 2017: 8,962
May 2019 update: 1,284
Toxline
Chemical terms
@SYNl+@AND+@OR+(methylmercury+"methyl+mercury"+monomethy
lmercury+Methylmercure+Methylquecksilber+MeHg+@TERM+@rn+229
67-92-6)+@RANGE+yr+1998+2017+@NOT+@org+"nih+reporter"
1998-March 2017: 5,714
May 2019 update: 1,087
Science Direct
Chemical terms
(methylmercury OR "methyl mercury" OR methyl-mercury OR
"Methylmercury (MeHg)" OR monomethylmercury OR "22967-92-6" OR
Methylmercure OR Methylquecksilber OR "methylmercury ii" OR MeHg)
1998-March 2017: 5,330
May 2019 update: 0
(HERO could not search
Science Direct)
Total
Total unique records from database searches with duplicates removed:
15,277
This document is a draft for review purposes only and does not constitute Agency policy.
46 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table A-2. SWIFT Review search of titles and abstracts to identify
epidemiology dose-response articles
Target
Search string
Epidemiology study3
(mesh_mh:( humans OR "human development") OR tiab: (human* OR person* OR people) OR
mesh_mh:( age groups) OR tiab: (pediatric* OR paediatric* OR baby OR babies OR toddler* OR
child* OR youth* OR youngster* OR tween* OR teen OR teens OR teenager*) OR (tiab:("in utero"
OR prenat* OR perinat* OR neonat* OR postnat*) AND NOT tiab: (mice OR mouse OR rat OR rats))
OR
tiab:(preschool* OR "pre-school*" OR kindergarten* OR schoolchild* OR student*) OR
tiab:("middle age*" OR elder* OR "senior citizen*" OR seniors OR retiree* OR septuagenarian* OR
octagenarian* OR sexagenarian* OR nonagenarian* OR centenarian*) OR
mesh_mh:("nuclear family") OR tiab:(famil* OR parent* OR father* OR mother* OR sibling* OR
brother* OR sister* OR twin OR twins OR "stepfather*" OR "step father*" OR "stepmother*" OR
"step mother*" OR "stepdaughter*" OR "step daughter*" OR "stepson*" OR "step son*" OR aunt*
OR uncle* OR niece* OR nephew* OR grandparent* OR grandfather* OR "grand father*" OR
grandmother* OR "grand mother*" OR
grandchild* OR granddaughter* OR grandson* OR spouse* OR partner* OR husband* OR wife OR
wives OR guardian* OR caregiver* OR "care giver*") OR
mesh_mh:(men OR women) OR tiab:(men OR man OR boy OR boys OR boyhood OR women OR
woman OR girl OR girls OR girlhood) OR
mesh_mh:("population groups" OR "vulnerable populations") OR tiab:("african american*" OR
"asian american*" OR hispanic* OR latina* OR latino* OR "mexican american*" OR underserved
OR disadvantaged) OR
mesh_mh:("epidemiologic studies" OR "double-blind method" OR "single-blind method") OR
mesh_sh:(epidemiology) OR tiab:("case control*" OR cohort OR "cross sectional" OR "follow-up
study" OR longitudinal OR prospective OR retrospective) OR
mesh_pubtype:("case reports" OR "clinical trial" OR "observational study" OR "randomized control
trial" OR "twin study") OR tiab:("clinical trial*" OR observational OR "randomized control trial*")
OR
mesh_mh:("research subjects" OR "human experimentation" OR patients OR "Patient
Participation") OR tiab:("human subject*" OR "research subject*" OR client* OR patient* OR
inpatient* OR outpatient* OR participant* OR volunteer*) OR
mesh_mh:("occupational groups" OR "occupational exposure") OR tiab:(occupation* OR workplace
OR "work place" OR "work-related" OR administrator* OR aides OR assistant* OR crew OR crews
OR employee* OR personnel OR professional OR staff OR technician* OR worker* OR educator*
OR instructor* OR teacher* OR clinician* OR doctor* OR physician* OR pharmacist* OR nurs* OR
residents OR veterinarian*))
Dose-response
("meta-analysis" OR "Systematic review" OR
tiab_punct:"P <" OR tiab_punct:"p <" OR tiab_punct:"P <=" OR tiab_punct:"p <=" OR
tiab_punct:"P >*" OR tiab_punct:"p >*"
OR tiab_punct:"p=*" OR tiab_punct:"P =" OR tiab_punct:"p =" OR tiab_punct:"p>*" OR
tiab_punct:"p<*"
OR tiab:"significan*" OR tiab:"nonsignificant" OR RR OR RRs OR SMR OR SMRs OR "rate ratio*"
OR "prevalence ratio*" OR "hazard ratio*" OR "odds ratio*" OR "risk ratio*" OR "relative risk*"
OR "prevalence ratio*"
OR tiab:"covariate*" OR tiab:"adjust*" OR tiab:"control* for" OR tiab:"associat*" OR
tiab:"confound*"
OR CI OR "confidence interval*" OR "credible interval" OR regression* OR "explanatory variable*"
OR tiab:"dose-response")
aThe search strategy from epidemiology studies is adapted from standard SWIFT Review search strategies for humans.
This document is a draft for review purposes only and does not constitute Agency policy.
47 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table A-3. Database search strategies (PBPK studies)
Search
Search string
Date and results
PubMed
(pbpk[tiab] OR "pb-pk"[tiab] OR pk[tiab] OR tk[tiab] OR pbtk[tiab] OR
"pb-tk"[tiab] OR httk[tiab] OR pk-model*[tiab] OR tk-model*[tiab] OR
(pharmacokinetic*[tiab] OR pharmacokinetics[mh:noexp] OR
pharmacokinetics[sh] OR toxicokinetic*[tiab] OR
toxicokinetics[mh:noexp] OR "physiologically based"[tiab] OR
"biologically based"[tiab])) AND (model*[tiab] OR models[tiab] OR
modeling[tiab]) AND (methylmercury[tiab] OR "methyl mercury"[tiab]
OR mercury[tiab])
Jan 2001-Nov 2019: 209
Science Direct
(pbpk OR pb-pk OR pk OR tk OR pbtk OR pb-tk OR httk) OR
(pharmacokinetic* OR toxicokinetic* OR physiologically OR
"biologically)) AND (model* OR models OR modeling) AND
(methylmercury OR "methyl mercury" OR mercury)
Jan 2001-Nov 2019:19
Web of Science
(pbpk OR "pb-pk" OR pk OR tk OR pbtk OR "pb-tk" OR httk OR pk-
model* OR tk-model* OR (pharmacokinetic* OR
pharmacokinetics[mh:noexp] OR pharmacokinetics[sh] OR
toxicokinetic* OR toxicokinetics[mh:noexp] OR "physiologically
based" OR "biologically based")) AND (model* OR models OR
modeling) AND (methylmercury OR "methyl mercury" OR mercury)
Jan 2001-Nov 2019:151
This document is a draft for review purposes only and does not constitute Agency policy.
48 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
APPENDIX B. DATA EXTRACTION FIELDS
Table B-l. Key data extraction elements to summarize study design,
experimental model, methodology, and results
Field label
Possible data extraction elements
Epidemiology studies
Funding
Funding source(s)
Reporting of conflict of interest by authors
Subjects
Study population name/description
Dates of study and sampling time frame
Geography (country, region, state, etc.)
Demographics (sex, race/ethnicity, age or life stage at exposure, and at outcome assessment)
Number of subjects (target, enrolled, n per group in analysis, and participation/follow-up rates)
Inclusion/exclusion criteria/recruitment strategy
Description of reference group
Methods
Study design (e.g., prospective or retrospective cohort, nested case-control study, cross-sectional,
population-based case-control study, intervention, case report)
Length of follow-up
Health outcome category (e.g., cardiovascular)
Health outcome (e.g., blood pressure)
Diagnostic or methods used to measure health outcome
Confounders or modifying factors and how considered in analysis (e.g., included in final model,
considered for inclusion but determined not needed)
Chemical name
Exposure assessment (e.g., blood, hair, placenta)
Methodological details for exposure assessment (e.g., analysis method, limit of detection)
Statistical methods
Results
Exposure levels (e.g., mean, median, measures of variance as presented in paper, such as standard
deviation, standard error of the mean, 75th/90th/95th percentile, minimum/maximum); range of
exposure levels, number of exposed cases
Statistical findings (e.g., adjusted P, standardized mean difference, adjusted odds ratio, standardized
mortality ratio, relative risk) or description of qualitative results. When possible, convert measures of
effect to a common metric with associated 95% confidence intervals. Most often, measures of effect for
continuous data are expressed as mean difference, standardized mean difference, and percentage
control response. Categorical data are typically expressed as odds ratio, relative risk (also called risk
ratio), or P values, depending on what metric is most commonly reported in the included studies and
ability to obtain information for effect conversions from the study or through author query.
Observations on dose-response (e.g., trend analysis, description of whether dose-response shape
appears to be monotonic, nonmonotonic)
Other
Documentation of author queries, use of digital rulers to estimate data values from figures, exposure
unit, and statistical result conversions, etc.
This document is a draft for review purposes only and does not constitute Agency policy.
49 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
APPENDIX C. CRITERIA FOR EVALUATION OF ANALYTICAL CHEMISTRY
METHODS USED FOR ANALYSIS OF MERCURY/METHYLMERCURY IN
BLOOD AND HAIR
Table C-l. Evaluation of analytical methods for blood total and
methylmercury in epidemiology studies
Level
Criteria
Good
Acceptable proportion of samples (>50%) above the limit of detection (LOD) OR LOD less
than median blood mercury concentration in the study population
Method
All papers must include a description of methodological factors:
All papers must indicate the type of sample analyzed; for Good quality,
should be liquid whole blood (collected in tubes with
ethylenediaminetetraacetic acid [EDTA] anticoagulant, stored in the
refrigerator within 1 year) or a freeze-dried (within 2 years) blood sample.
NOTE: If samples have been stored for longer than noted above, recovery of storage
stability spiked samples must be reported but results will not be evaluated.
Additional information regarding storage stability samples is provided in "Useful
Terms Defined."
For laboratories using a standard method: The paper will include the
method number and a citation for a standard method for measuring
mercury levels in blood (e.g., Centers for Disease Control and Prevention
method ITB003A or National Health and Nutrition Examination Survey
methods 3001.1 or 3016.8 for total mercury, method DLS-3020.5 for
methylmercury) OR a description of the method discussed as follows.
For laboratories using a nonstandard, validated method: The paper will
provide a citation for a peer-reviewed report of the validation in the main
body of the paper or in supplementary information (refer to "Useful Terms
Defined" for more information on validation) OR a description of the
method discussed, as follows.
For laboratories using nonstandard, nonvalidated methods: Nonstandard
methods may perform acceptably if sufficient evidence of data quality
(quality control [QC] results) is provided (as follows). For nonstandard
methods, the paper must provide a description of the method, including
the following:
o Analytical limits including LOD and/or the limit of quantitation
(LOQ) must be reported for nonstandard, nonvalidated methods
to obtain a Good rating.
o Combustion Atomic Absorption Spectroscopy (CAAS) refers to
direct mercury analysis and may use several instruments, such as
the Milestone Direct Mercury Analyzer (DMA) or the Nippon
Mercury Analyzer (MA).
¦ For studies using CAAS for total mercury analysis, critical
method variables that should be noted are time and
This document is a draft for review purposes only and does not constitute Agency policy.
50 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
temperature of sample drying step, charring/decomposition
step of analysis, and amalgamator flash time and temperature.
o Sample stabilization measures (e.g., addition of gold or thiols;
refer to "Useful Terms Defined" for additional details).
o Sample preparation method including, but not limited to, the
following:
¦ Derivatization steps for gas chromatography-inductively
coupled plasma-mass spectrometry (GC-ICP-MS) analysis or
another speciation technique for methylmercury.
¦ Oxidizing agents for cold vapor atomic fluorescence
spectroscopy (CVAFS) and cold vapor atomic absorption
spectroscopy (CVAAS) (e.g., potassium permanganate [KMn04],
nitric acid, or sulfuric acid for total mercury analysis).
¦ Reducing agent for cold vapor atomic fluorescence
spectroscopy (CVAFS) and cold vapor atomic absorption
spectroscopy (CVAAS) (e.g., tin chloride [SnCI2], sodium
borohydride [NaBH4], sodium hypochlorite [NaCIO]).
¦ Digestion reagents, temperatures (<65°C), and times.
¦ Extraction solvent composition.
For speciation analysis (i.e., methvlmercurv): Separation steps before
detection. Examples include the following:
o For liauid chromatography (LC; e.g., high-performance liauid
chromatography [HPLC1 or ultra-high-performance liauid
chromatography rUHPLCl): Column tvpe, eluent composition
(e.g., mercaptoethanol, EDTA, cysteine), eluent gradient (or
indicate whether an isocratic program [where the same eluent
content was used through the run] was used), and injection
volume.
o For gas chromatographv: Transfer line temperature, column
temperature program, and column identity.
o For distillation: Solvent and temperature.
Use of an appropriate internal standard (e.g., bismuth [Bi], praseodymium
[Pr], holmium [Ho]; ICP-MS only; other organomercury compound
[e.g., propylmercury, butylmercury; speciation by LC- or GC-ICP-MS only]):
o NOTE: Internal standard is not generally included in
CVAAS/CVAFS/CAAS methods.
NOTE: The items listed earlier should be included, but because of the great variability in
nonstandard methods, it is not necessary to evaluate the quality of all methodological
factors, only to ensure that they are included in the method description.
Qua litv Control
All papers (standard methods, validated methods, and nonstandard methods) must include a
description of QC procedures and results performed to verify method performance and data
quality for the study:
This document is a draft for review purposes only and does not constitute Agency policy.
51 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
At least three laboratory QC procedures and results in the methods
section, supplementary materials, cited papers, or standardized laboratory
protocol, including, but not limited to, the following:
o Blood-based standard reference materials (SRMs) (955C, levels 1
and 3 only for total mercury, level 3 for methylmercury only; 966
if the study was done before December 2015, level 2 for
methylmercury only): % recovery information, 90%-110%. If the
study was done before 2000 (the year SRM 966 was issued),
acceptable reference materials include "Control Blood for Metals"
by the Behring Institute, OSSD 20/21.
o Duplicate sample preparation and relative percent difference
(RPD) results, <15% RPD.
o Representative blank sample analyses or chromatograms to
demonstrate the absence of interferences.
o Recovery of spiked study samples, 90%-110%.
o Post-extraction spiked sample recovery or method blank spiked
(MBS) samples, 90%-110%.
o Replicate QC precision (also called uncertainty, repeatability, or
reproducibility; or percent relative standard deviation [%RSD] or
coefficient of variation [%CV], <15%, or correlation coefficient,
>0.90).
o Participation in an interlaboratory testing program with
documented results for mercury in blood samples and
"satisfactory" results.
NOTE: A wide range of interlaboratory testing programs is available for trace
metals, but only ones that monitor mercury in whole blood, serum, or plasma,
or that use standard methods, are relevant for assessing data quality.
o Incurred sample reanalysis (ISR) to demonstrate reproducibility
on different days, and RPD results, <15% RPD; refer to "Useful
Terms Defined" for more information.
o Control charts (e.g., Bland-Altman, Levey-Jennings, Harrell-Davis,
Shewhart) for QC samples of the same sample type, prepared by
the cited method, that show method performance over time
(large sample populations only).
NOTE: If any of the above method parameters or quality control measures
exhibit values that fall outside the range of value noted above, then the
entire study should be categorized at the lowest quality level of the
individual method parameter(s) or QC measure(s).
This document is a draft for review purposes only and does not constitute Agency policy.
52 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Adequate
Criteria
Method
Same as Good, with the following exceptions:
All papers must indicate the type of sample analyzed; for Adequate quality,
may be liquid whole blood (collected in tubes with
ethylenediaminetetraacetic acid [EDTA] anticoagulant, stored in the
refrigerator within 1 year) or a freeze-dried (within 2 years) blood sample.
NOTE: If samples have been stored for longer than noted above, recovery
of storage stability spiked samples must be reported but results will not be
evaluated.
For laboratories using standard methods: Samples analyzed by
Environmental Protection Agency (EPA) method 7473 (total mercury only).
For nonstandard, nonvalidated methods: The paper will comment on two
or three methodological details listed earlier such as extraction times,
cleaning times between sample analyses, instrumental technique, and so
on.
Analytical limits including LOD and/or the limit of quantitation (LOQ) must
be reported for nonstandard, nonvalidated methods to obtain an
Acceptable rating.
For measurements of methylmercury: Determined methylmercury by
subtracting inorganic mercury from total mercury.
For ICP-MS only: Use of less common or less appropriate internal
standards such as terbium (Tb), rhodium (Rh), gallium (Ga), thallium (Tl),
indium (In), yttrium (Y), or scandium (Sc).
Quality Control
Any Adequate analysis must include the following:
For standard methods and validated methods (with literature citation):
Discussion of one or two laboratory QC procedures and results in the
methods section, supplementary materials, cited papers, or standardized
laboratory protocol, including either of the following:
o Reference material percentage recovery information (may be the
National Institute of Standards and Technology [NIST] SRM
mentioned earlier or another such as NIST 3133 or 3177 for total
mercury only, or commercial blood SRM [e.g., ClinCheck or
Seronorm for total mercury]); recovery will fall in the range 85%-
90% or 110%-115%.
o Replicate sample preparation information (e.g., %CV or RPD, for
duplicate preparation, of 10%-15%).
For nonstandard, nonvalidated studies: Discussion of at least three
laboratory QC procedures and results in the methods section,
supplementary materials, cited papers, or standardized laboratory
protocol, including a combination of the procedures mentioned under
"Good" and "Adequate" levels.
This document is a draft for review purposes only and does not constitute Agency policy.
53 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Deficient
Criteria
Method
Same as Adequate, with the following exceptions:
All papers must indicate the type of sample analyzed; for Deficient quality,
may be liquid whole blood (collected in vacutainer tubes with
ethylenediaminetetraacetic acid [EDTA] anticoagulant, 1 year storage in
the refrigerator, or lithium heparin anticoagulant due to the lower stability
of heparin) or a freeze-dried (2 years) blood sample.
NOTE: If samples have been stored for longer than noted above, recovery
of storage stability spiked samples must be reported but results will not be
evaluated. Additional information regarding storage stability samples is
provided in "Useful Terms Defined".
Minimal methodological details provided in the paper, supplemental
information, or cited papers; only one or two of the items from the
aforementioned method detail list.
o For example, mentioning only the instrumental technique used
for analysis, omitting all collection and storage procedures,
analytical limits, or references for the sample preparation
method.
NOTE: The analytical limits for nonstandard, nonvalidated
methods must be provided for papers to be considered Good or
Adequate, as it is necessary to determine whether a method was
used with sensitivity levels appropriate to the matrix and that the
method has been appropriately optimized by the analytical
laboratory. If analytical limits for nonstandard, nonvalidated
methods are not provided, the maximum rating possible is
Deficient.
Quality Control
Studies will be evaluated as Deficient if they include the following:
Use of NIST standard reference materials or commercial certified reference
materials (CRMs) in non-blood-based matrices (e.g., NIST 2976, 3668,
1641e; several European Union Joint Research Centre [JRC] CRMs).
QC results fall outside of "Adequate" acceptance ranges but are not severe
enough to warrant exclusion of the study (e.g., recovery of QC samples in
the range 80%-85% or 115%-120%; variability of replicate samples 15%-
20% RPD).
OR
QC procedures and results not discussed in the paper or supplemental
information.
NOTE: Where QC procedures are not described, requesting additional information from the
corresponding author will be necessary because QC results are instrumental in gauging
reliability of reported sample data.
Low proportion of samples (<50%) above the LOD OR LOD greater than median blood
mercury concentration in the study population.
Critically
Deficient
This document is a draft for review purposes only and does not constitute Agency policy.
54 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
Method
If any of the following are true:
Analysis of dried blood spots (DBS).
NOTE: To date, there have been virtually no studies that have explored the
analysis of dried blood spot samples for mercury or methylmercury.
Additionally, the field of dried blood spot analysis of metals is still in
development, especially for quantitative applications.
Use of inappropriate standard method (e.g., methods EPA 245.2 or 7471B).
NOTE: The above methods describe the analysis of bulk volumes of water
for mercury content, so those and similar methods cannot be directly used
for analysis of mercury in blood. It may be possible to adapt such methods
for use with blood samples, but they should be treated as nonstandard,
nonvalidated methods and detailed descriptions of the preparation and
analysis methods must be provided.
No collection, preparation, or analysis method details described in the
paper, supplementary information, or cited reports.
NOTE: If no description of the preparation and analysis methods are
provided, it is not possible to confirm that appropriate measures were
taken to ensure the accuracy of results. Quality control results alone may
not account for all essential method parameters in study samples (e.g.,
collection and storage measures).
For ICP-MS or GC-ICP-MS studies: No internal standard reported.
NOTE: The use of internal standards is an essential aspect of ICP-MS
analysis to appropriately account for instrument drift and matrix impact on
analyte signals.
Qualitv Control
QC results reveal concerns about reliability of measurements
(e.g., recovery of QC samples outside 80%-120%, variability of replicate
samples >20% RPD).
If after inquiry, it is found that no QC samples were analyzed.
NOTE: Analysis of quality control samples is essential to demonstrate the
accuracy of chemical analyses and provide confidence in data quality and is
generally considered a standard practice in analytical laboratories. QC
samples can demonstrate that analytical accuracy is maintained even in
complex sample matrices, such as blood. If no quality control samples
were prepared by the same method and analyzed alongside study samples,
then it is not possible to have confidence in the quality of analytical data
generated in support of an epidemiology study.
This document is a draft for review purposes only and does not constitute Agency policy.
55 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Table C-2. Evaluation of analytical methods for hair total and methylmercury
in epidemiology studies
Level
Criteria
Good
Acceptable proportion of samples (>50%) above the limit of detection (LOD) OR LOD less
than median hair mercury concentration in study population
Method
All papers must include a description of methodological factors:
All papers must indicate the sample collection considerations including:
age and sex of study participants, hair segment characteristics (e.g., length
from scalp, length of hair analyzed). Studies should clearly indicate the
area of the body from which hair was collected (e.g., scalp, underarm,
pubis), which is important for interpreting results and comparing across
studies.
Sample storage and shipment procedures must be described including
sample containers/bags, tying procedures, etc.
Cleaning procedures for hair samples must be described. Acceptable
washing procedures include deionized water, ionic or nonionic detergent
solution, acetone, methanol, etc. Multiple washes or heat, or both, may
be used below 65°C. An example standard cleaning technique is presented
in IAEA Report 50 (IAEA/RL/50), "Activation analysis of hair as an indicator
of contamination of man by environmental trace element pollutants."3
For laboratories using a standard method: The paper will include the
method number and a citation for a standard method for measuring
mercury levels in hair (e.g., United States Department of Agriculture
(USDA) Food Safety Inspection Service [FSIS] method MER for total
mercury, Environmental Protection Agency (EPA) SW-846 Method 3200 for
extraction only, EPA SW-846 method 6800 for Hg speciation only,
European Union Consortium to Perform Human Biomonitoring on a
European Scale [COPHES] D3.6 "Mercury: Determination in Scalp Hair" for
total mercury only) OR a method description discussed as follows.
For laboratories using a nonstandard, validated method: The paper will
provide a citation for a peer-reviewed report of the validation in the main
body of the paper or in supplementary information (refer to "Useful Terms
Defined" for more information on validation) OR a description of the
method discussed, as follows.
For laboratories using nonstandard, nonvalidated methods: Nonstandard
methods may perform acceptably if sufficient evidence of data quality
(quality control [QC] results) is provided (as follows). For nonstandard
methods, the report must provide a description of the method including
the following:
o Analytical limits including LOD and/or the limit of quantitation
(LOQ) must be reported for nonstandard, nonvalidated methods
to obtain a Good rating.
o Combustion Atomic Absorption Spectroscopy (CAAS) refers to
direct mercury analysis and may use several instruments, such as
This document is a draft for review purposes only and does not constitute Agency policy.
56 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
the Milestone Direct Mercury Analyzer (DMA) or the Nippon
Mercury Analyzer (MA).
¦ For studies using CAAS for total mercury analysis, critical
method variables that should be noted are time and
temperature of sample drying step, charring/decomposition
step of analysis, and amalgamator flash time and temperature.
o Sample stabilization measures (e.g., addition of gold or thiols
during or after extraction; refer to "Useful Terms Defined" for
additional details).
o Sample preparation method including, but not limited to, the
following:
¦ Hair sample drying method and conditions (e.g., temperature,
time).
¦ Derivatization steps for gas chromatography-inductively
coupled plasma-mass spectrometry (GC-ICP-MS) analysis or
another speciation technique for MeHg.
¦ Oxidizing agents for cold vapor atomic fluorescence
spectroscopy (CVAFS) and cold vapor atomic absorption
spectroscopy (CVAAS) (e.g., potassium permanganate [KMnOj,
nitric acid, or sulfuric acid for total mercury analysis).
¦ Reducing agent for CVAFS and CVAAS (e.g., tin chloride [SnCI2],
sodium borohydride [NaBH4], sodium hypochlorite [NaCI02],
hydroxylamine hydrochloride).
¦ For total mercury analysis only: Digestion reagents,
temperatures (<65°C for open vessel digestion or digestion
without stabilizers mentioned previously), and times.
¦ Extraction solvent composition.
For speciation analysis (i.e., methvlmercurv): Separation steps before
detection. Examples include the following:
o For liauid chromatography (LC; e.g., high-performance liauid
chromatography [HPLC1 or ultra-high-performance liauid
chromatography rUHPLCl): Column tvpe, eluent composition
(e.g., mercaptoethanol, ethylenediaminetetraacetic acid [EDTA],
cysteine), eluent gradient (or indicate whether an isocratic
program [where the same eluent content was used through the
run] was used), and injection volume.
o For gas chromatography: Transfer line temperature, column
temperature program, and column identity.
o For distillation: Solvent and temperature.
Use of an appropriate internal standard (e.g., bismuth [Bi], praseodymium
[Pr], holmium [Ho]; ICP-MS only; other organomercury compound
[e.g., propylmercury, butylmercury; speciation by LC- or GC-ICP-MS only]):
o NOTE: Internal standard is not generally included in
CVAAS /CVAFS/ CAAS methods.
This document is a draft for review purposes only and does not constitute Agency policy.
57 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
NOTE: the items listed above should be included, but due to the great variability in
nonstandard methods, it is not necessary to evaluate the quality of all methodological
factors, only ensure that they are included in the method description.
Quality Control
All papers (standard methods, validated methods, and nonstandard methods) must include a
description of QC procedures and results performed to verify method performance and data
quality for their study:
At least three laboratory QC procedures and results in methods section,
supplementary materials, cited papers, or standardized laboratory
protocol including, but not limited, to the following:
o Hair-based certified reference material (CRM) (NIES CRM 13:
Human Hair, IAEA-085 or -086: Human Hair (Methyl Mercury),
Joint Research Centre CRM BCR-397: Trace elements in human
hair for total mercury, Chinese CRM GBW 07601): % recovery
information, 85-115%.
NOTE: As of this time, no NIST (National Institute of Standards and
Technology) Standard Reference Materials are available for hair
mercury.
o Duplicate sample preparation and relative percent difference
(RPD) results, <20% RPD.
o Representative blank sample analyses or chromatograms to
demonstrate the absence of interferences.
o Recovery of spiked study samples, 85%-115%.
o Post-extraction spiked sample recovery or method blank spiked
(MBS) samples, 85%-115%.
o Replicate QC precision (also called uncertainty, imprecision,
repeatability, or reproducibility; or percent relative standard
deviation [%RSD] or coefficient of variation [%CV], <20%, or
correlation coefficient >0.85).
o Participation in an interlaboratory testing program with
documented results for mercury in hair samples and "satisfactory"
results.
NOTE: A wide range of interlaboratory testing programs is available
for trace metals, but only ones that monitor mercury in hair, or that
use standard methods, are relevant for assessing data quality.
o Incurred sample reanalysis (ISR) to demonstrate reproducibility on
different days, and RPD results, <20% RPD; refer to "Useful Terms
Defined" for more information.
o Control charts (e.g., Bland-Altman, Levey-Jennings, Harrell-Davis,
Shewhart) for QC samples of the same sample type, prepared by
the cited method that show method performance over time (large
sample populations only).
This document is a draft for review purposes only and does not constitute Agency policy.
58 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
Note: If any of the above method parameters or quality control measures
exhibit values that fall outside the range of values noted above, then the
entire study should be categorized at the lowest quality level of the
individual method parameter(s) or QC measure(s).
Adequate
Method
Same as Good, with the following exceptions:
All papers may indicate the sample collection considerations including age
and sex of study participants, hair segment characteristics (e.g., length
from scalp, length of hair analyzed). Studies should clearly indicate the
area of the body from which hair was collected (e.g. scalp, underarm,
pubis), which is important for interpreting results and comparing across
studies.
Sample storage and shipment procedures must be described including
sample containers/bags, tying procedures, etc.
Cleaning procedures for hair samples must be described. Acceptable
washing procedures include deionized water, ionic or nonionic detergent
solution, acetone, methanol, etc. Multiple washes or heat, or both, may be
used below 65°C. An example standard cleaning technique is presented in
IAEA Report 50 (IAEA/RL/50), "Activation analysis of hair as an indicator of
contamination of man by environmental trace element pollutants."3
For laboratories using standard methods: Samples analyzed bv EPA SW-
846 method 6800 (Hg speciation only) or method 7473 (total mercury
only).
For nonstandard, nonvalidated methods: The report will comment on two
or three methodological details above such as extraction times, cleaning
times between sample analyses, instrumental technique, and so on.
Analytical limits including LOD and/or the limit of quantitation (LOQ) must
be reported for nonstandard, nonvalidated methods to obtain an
Acceptable rating.
For measurements of methvlmercurv: Determined methvlmercurv bv
subtracting inorganic mercury (iHg) from total mercury (THg).
For ICP-MS onlv: Use of less common or less appropriate internal standards
such as terbium (Tb), rhodium (Rh), gallium (Ga), thallium (Tl), indium (In),
yttrium (Y), or scandium (Sc).
Qualitv Control
Any Adequate analysis must include:
For standard methods and validated methods (with literature citation):
Discussion of one or two laboratory QC procedures and results in methods
section, supplementary materials, cited paper, or standardized laboratory
protocol including either:
o Reference material percentage recovery information (may be the
certified reference materials [CRMs] mentioned earlier or another
This document is a draft for review purposes only and does not constitute Agency policy.
59 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
such as NIST Standard Reference Material [SRM] 3133 or 3177 for
use in spiking total Hg only), recovery will fall in the range 80%-
85% or 115%-120%.
o Replicate sample preparation information (e.g., %CV or RPD) for
duplicate preparation of 20%-25%.
For nonstandard, nonvalidated studies: Discussion of at least three
laboratory QC procedures and results in the methods section,
supplementary materials, cited papers, or standardized laboratory
protocol, including a combination of the procedures mentioned under
"Good" and "Adequate" quality levels.
Deficient
Method
Same as Adequate except:
For Deficient quality: All papers may indicate 1-2 sample collection
considerations including age and sex of study participants, hair segment
characteristics (e.g., length from scalp, length of hair analyzed). Studies
should clearly indicate the area of the body from which hair was collected
(e.g. scalp, underarm, pubis), which is important for interpreting results
and comparing across studies.
Sample storage and shipment procedures must be described including
sample containers/bags, tying procedures, etc.
Cleaning procedures for hair samples must be described. Acceptable
washing procedures include deionized water, ionic or nonionic detergent
solution, acetone, methanol, etc. Multiple washes or heat, or both, may be
used below 65°C. An example standard cleaning technique is presented in
IAEA Report 50 (IAEA/RL/50), "Activation analysis of hair as an indicator of
contamination of man by environmental trace element pollutants."3
Minimal methodological details provided in the paper, supplemental
information, or cited papers; onlv one or two of the items from the
aforementioned method detail list.
o For example, only mentioning the instrumental technique used
for analysis, omitting all collection, storage, and cleaning
procedures, analytical limits or references for the sample
preparation method.
NOTE: The analytical limits for nonstandard, nonvalidated
methods must be provided for papers to be considered Good or
Adequate, as it is necessary to determine whether a method was
used with sensitivity levels appropriate to the matrix and that the
method has been appropriately optimized by the analytical
laboratory. If analytical limits for nonstandard, nonvalidated
methods are not provided, the maximum rating possible is
Deficient.
Qualitv Control
Studies will be evaluated as Deficient if they include the following:
This document is a draft for review purposes only and does not constitute Agency policy.
60 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
Use of NIST standard reference materials or commercial certified reference
materials in non-hair-based tissue matrices (e.g., NIST 1575a [total
mercury only], 1946,1947, 2976, several European Union Joint Research
Centre [JRC] CRMs).
QC results fall outside of "Adequate" acceptance ranges but are not severe
enough to warrant exclusion of the study (e.g., recovery of QC samples in
the range 75%-80% or 120%-125%; variability of replicate samples 25%-
50% RPD).
OR
QC Procedures and results not discussed in the paper or supplemental
information.
NOTE: Where QC procedures are not described, requesting additional information from the
corresponding author will be necessary because QC results are instrumental in gauging
reliability of reported sample data.
Critically
Deficient
Low proportion of samples (<50%) above the LOD OR LOD greater than median blood
mercury concentration in the study population.
Method
If any of the below are true:
Use of inappropriate standard method (e.g., methods EPA 245.1, EPA
245.2, EPA 245.7, EPA 7470A, EPA 1630, EPA 1631E).
NOTE: The above methods describe the analysis of bulk volumes of water
for mercury content, so those and similar methods cannot be directly used
for analysis of mercury in hair. It may be possible to adapt such methods
for use with hair samples, but they should be treated as nonstandard,
nonvalidated methods and detailed descriptions of the preparation and
analysis methods must be provided.
No collection, preparation, or analysis method details described in the
paper, supplementary information, or cited reports.
NOTE: If no description of the preparation and analysis methods are
provided, it is not possible to confirm that appropriate measures were
taken to ensure the accuracy of results. Quality control results alone may
not account for all essential method parameters in study samples (e.g.,
collection and storage measures).
Cleaning, drying, or open-vessel extraction of samples at any temperature
greater than 65°C except sample preparation for speciation analysis by
distillation and closed-vessel digestion of hair samples for total mercury
analysis in the presence of sulfur-containing stabilizing agents.
NOTE: Mercury volatilizes at temperatures above 65°C in the absence of a
stabilizer during sample preparation such as thiols or gold. Cleaning and
drying procedures are not generally performed in the presence of such
stabilizers and open vessel digestions will readily allow release of
volatilized mercury.
This document is a draft for review purposes only and does not constitute Agency policy.
61 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
Level
Criteria
For ICP-MS or GC-ICP-MS studies: No internal standard reported.
NOTE: The use of internal standards is an essential aspect of ICP-MS
analysis to appropriately account for instrument drift and matrix impact on
analyte signals.
Qualitv Control
QC results reveal concerns about reliability of measurements
(e.g., recovery of QC samples outside 75%-120% or variability of replicate
samples >20% RPD).
If after inquiry, it is found that no QC samples were analyzed.
NOTE: Analysis of quality control samples is essential to demonstrate the
accuracy of chemical analyses and provide confidence in data quality and is
generally considered a standard practice in analytical laboratories. QC
samples can demonstrate that analytical accuracy is maintained even in
complex sample matrices, such as blood. If no quality control samples were
prepared by the same method and analyzed alongside study samples, then
it is not possible to have confidence in the quality of analytical data
generated in support of an epidemiology study.
aRyabukhin, Y. S. (1976). Activation analysis of hair as an indicator of contamination of man by environmental trace element
pollutants (No. IAEA-RL-50). International Atomic Energy Agency.
1
2 In comparison of the quality control procedures between hair and blood, it should be noted
3 that the criteria for each rating category are different for the two matrices, specifically the
4 recovery/relative error (accuracy) of spiked QC samples and percent difference (precision) of
5 duplicate samples. The differences serve as an acknowledgment that analysis of hair is subject to
6 greater variability than analysis of blood as a result of several factors, primarily sample collection,
7 instrumental error when weighing samples, segmenting of hair samples, etc.
This document is a draft for review purposes only and does not constitute Agency policy.
62 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the Methylmercury IRIS Assessment
Useful Terms Defined
Acceptance criteria: Predetermined values for accuracy or precision that demonstrate the
generation of data of acceptable quality or accuracy. If a measurement falls outside predefined
acceptance criteria, then the reliability of data generated during analysis or in the same analytical
batch may be affected.
Coefficient of variation: A measure of variability of repeated measurement of a value, or
precision. Calculated as the ratio of standard deviation of a set of measurements to the average
value; often expressed as a percentage:
Standard deviation (SD,g)
Coefficient of variation (%CV) = x 100%
Mean concentration
Also called relative standard deviation, %RSD, and uncertainty.
Control chart: A graphical representation of quality control results over time, intended to
demonstrate reliable method performance for longer analysis projects, usually encompassing
anywhere from several days to several months. Many types of control charts can be designed to
show instrument or method performance measures. The most relevant charts for assessing
confidence in analytical data quality show method performance measures, such as quality control
sample recovery (accuracy) or precision over time. Specific examples of types of control charts
include Bland-Altman, Levey-Jennings, Harrell-Davis, Shewhart, and others.
Correlation coefficient: Statistical analysis showing agreement between replicate
measurements over a range of concentrations. Calculated by plotting the concentration of duplicate
samples over several concentrations, then plotting a line of best fit to the pairs of data. The
correlation coefficient of the line shows how closely the points fall to the ideal line with a slope of 1.
Derivatization: A sample preparation step involving chemical reaction of the target analyte
to enhance a particular chemical property for the purposes of analysis. Examples include reacting
mercury and methylmercury with sodium tetraethylborate (NaBEU). The resulting products are
more reactive and thus better suited to analysis, in this case, by gas chromatography. Products can
also still be separated and accurately analyzed.
Digestion: A sample preparation method involving degradation of the sample, with acid or
base, to break down the matrix and efficiently separate the analyte from the sample matrix.
Digestion methods are generally very harsh and are less able to preserve analyte speciation, often
making them preferable for total metal analysis (e.g., total mercury analysis).
Dried blood spots: A sample collection technique for analyzing blood samples collected for
screening purposes on a filter paper or similar material, often done in newborn screening but also
increasingly implemented in public health studies because of ease of collection and lower concerns
about sample stability or storage. Sometimes abbreviated as DBS. Also called dried matrix spots
(DMS).
This document is a draft for review purposes only and does not constitute Agency policy.
63 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the Methylmercury IRIS Assessment
Extraction: A sample preparation method involving the separation of a target analyte from a
sample by selective dissolution in a liquid solvent (organic or aqueous). Extraction methods are
often gentler than digestion methods and are often more suitable for analyzing speciation of
samples. Most common extraction techniques are distillation and alkaline extraction (using a basic
pH to gently break down tissues or solids).
Incurred sample reanalvsis: Reanalysis of select samples on a second analytical day to
demonstrate reproducibility of results over time. This quality control measure differs from
duplicate analysis, which is generally performed on the same day.
Interference: Any substance besides the target analyte that results in inaccurate
measurements of a chemical signal during analysis. There are several types of interferences,
including chemical interferences, spectral interferences, and so on. Interferences may result in
higher or lower measured concentrations of a chemical depending on the nature of the interference.
Instrumental factors are generally not included in the definition of interferences.
Internal standard: Chemical substance added to standards and samples in ICP-MS analysis
to allow monitoring of instrument drift or effects on analyte signal due to sample composition. For
mercury analysis, usually relevant only to ICP-MS methods.
Laboratory accreditation: A certification for laboratories to demonstrate the application of
quality systems and laboratory practices that support the generation of high-quality data.
Programs are organized by a range of commercial, government, or academic entities and are often
administered annually.
Matrix: A specific type of sample, such as blood, hair, water, or biological tissue.
Matrix blank: A type of quality control sample prepared containing only extraction reagents
or solvents and control biological matrix that are carried through the sample preparation process;
analyzed to monitor endogenous concentration of analyte in unexposed samples. In toxicology
studies, these are expected to be low unless the target analyte is a pervasive environmental
contaminant
Memory effect: A chemical phenomenon specific to mercury and a few other elements
where the metal interacts strongly with the surface of plastic materials of an instrument and adsorb
to the surface, only to be released slowly over time. This phenomenon results in a gradual decrease
of mercury signal to zero between samples and requires special considerations in the design of an
experiment including special wash solutions, longer cleaning times, or substitution of alternate
materials to prevent the effect from biasing measured results.
Method blank: A type of quality control sample prepared containing only extraction
reagents or solvents that are carried through the sample preparation process; analyzed to monitor
contribution to a background analyte signal arising from the sample preparation or reagents. Also
called reagent blank.
Recovery: A measure of method accuracy; comparison of the measured concentration of an
analyte against the expected or "known" concentration of the analyte. Usually calculated as follows:
This document is a draft for review purposes only and does not constitute Agency policy.
64 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the Methylmercury IRIS Assessment
Measured concentration
Recovery = x 100%
Expected concentration
Reference material: A well-characterized standard sample containing known concentrations
of one or more analytes, intended to be processed and analyzed alongside study samples to verify
accurate measurement of an analyte in samples when using an analytical method.
Certified reference material (CRM): A reference material that has been repeatedly
characterized for use as a quality control sample. Usually produced in smaller batches and
sold by commercial entities, with a less rigorous characterization than standard reference
materials.
Standard reference material fSRMl: A reference material that has been produced in large
single batches and characterized to a very high degree, usually by an official organization
such as the National Institute of Standards and Technology, for use as a quality control
sample. Considered the "gold standard" of reference materials for use as a quality control
sample.
Optimal reference material selection will include the exact combination of analyte and
matrix being analyzed, but if such a combination is not available, then similar matrices containing
the target analyte may be used.
Relative error: An alternate measure of method accuracy, demonstrating the variance of a
measured concentration of an analyte from the expected or "known" concentration of the analyte.
Usually calculated as follows:
("Measured concentration - Expected concentration]
Relative error (%RE) = - x 100%
Expected concentration
Replicate: Preparation of a quality control or study sample two or more times for repeated
determination of the concentration of an analyte, intended to demonstrate precision of analytical
measurements, usually performed on the same day to demonstrate reproducibility in an analytical
batch.
Speciation: Study of the different chemical forms of an analyte; in the case of mercury,
referring to the analysis of the concentration and percentage of inorganic mercury and organic
mercury, such as methylmercury, dimethylmercury, ethylmercury, and so on, present in a sample.
Important considerations in speciation methods are the sample preparation methods, chemical
composition of eluents, temperatures, and column types, all of which control the separation of
different forms and the ability to maintain the forms of mercury throughout a speciation analysis.
Spike: Addition of the analyte to a method blank, matrix blank, or study sample, intended to
demonstrate accurate measurement of a known concentration of the analyte after processing
alongside study samples.
This document is a draft for review purposes only and does not constitute Agency policy.
65 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the Methylmercury IRIS Assessment
Stabilization: Ensuring that the concentration and species of analytes remain constant over
time. This process can apply to sample storage before analysis or sample preparation, to ensure
that no mercury is lost or converted before analysis. For analysis of mercury, this is most often
performed through addition of gold to interact with mercury, addition of hydrochloric acid (specific
to total mercury analysis after digestion), addition of a metal binding agent like EDTA, or addition
of sulfur-containing chemicals (thiols). This is usually only necessary for long-term storage of
blood samples or during extraction and digestion processes to prevent volatilization. One part of a
method validation may be to characterize chemical stability in the sample tissue or in digests or
extracts.
Storage Stability Samples: A specific type of quality control sample that is prepared and
analyzed to demonstrate the impact of long storage periods on recovery of mercury in samples.
Storage stability samples are a set of control samples of the same type as study samples (blood for
the purposes of this document) spiked with the analyte of interest (mercury in this case) and stored
alongside study samples from time of receipt until time of analysis. The recovery of mercury in
storage stability samples provides a measure of potential loss of mercury as a result of long storage
times.
Uncertainty: A measure of precision in analytical measurements, encompassing several
types of calculations including relative standard deviation, coefficient of variation, reproducibility,
and repeatability.
Validation: Systematic performance of the preparation and analysis method, to demonstrate
consistent accurate performance of a method under a range of conditions. Includes repeated
analysis of replicates of standard samples or spiked matrix samples, often over several days or with
multiple analysts. Characterizes linear range, accuracy, precision, analytical limits (LOD and LOQ),
specificity (the absence of analyte signal due to interferences), stability of the analyte in samples or
extracts. One approach to validation can be found in the FDA publication Bioanalytical Method
Validation: Guidance for Industry (FDA. 2018). which discusses parameters to be validated.
However, this publication is specifically aimed at the design of studies regulated by Good
Manufacturing Practices and Good Laboratory Practices; alternative approaches can be used for
research purposes that address the critical parameters discussed earlier.
This document is a draft for review purposes only and does not constitute Agency policy.
66 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the Methylmercury IRIS Assessment
Useful Abbreviations Defined
CAAS combustion atomic absorption spectroscopy
CRM certified reference material
CV coefficient of variation OR cold vapor (depending on context)
CVAAS cold vapor atomic absorption spectroscopy
CVAFS cold vapor atomic fluorescence spectroscopy
DBS dried blood spot
DMS dried matrix spot
EDTA ethylenediaminetetraacetic acid
EPA Environmental Protection Agency
FDA Food and Drug Administration
GC gas chromatography
HPLC high-performance liquid chromatography
ICP-MS inductively coupled plasma-mass spectrometry
ISR incurred sample reanalysis
JRC European Commission Joint Research Centre
LC liquid chromatography
LOD limit of detection
LOQ limit of quantitation
MBS method blank spiked
NIST National Institute of Standards and Technology
QC quality control
RE relative error
RPD relative percent difference
RSD relative standard deviation
SRM standard reference material
U H PLC ultra-high-performance liquid chromatography
This document is a draft for review purposes only and does not constitute Agency policy.
67 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the Methylmercury IRIS Assessment
REFERENCES
Abass. K: Huusko. A: Knutsen. HK: Nieminen. P: Myllvnen. P: Meltzer. HM: Vahakangas. K: Rautio. A.
(2018). Quantitative estimation of mercury intake by toxicokinetic modelling based on total
mercury levels in humans. Environ Int 114: 1-11.
http://dx.doi.Org/10.1016/i.envint.2018.02.028
Al-Shahristani. H: Shihab. KM. (1974). Variation of biological half-life of methylmercury in man.
Arch Environ Occup Health 28: 342-344.
http://dx.doi.Org/10.1080/00039896.1974.10666505
Albert. I: Villeret. G: Paris. A: Verger. P. (2010). Integrating variability in half-lives and dietary
intakes to predict mercury concentration in hair. Regul Toxicol Pharmacol 58: 482-489.
http: / /dx. do i. or g /10.1016/i .vrtph. 2010.08.020
Allen. BC: Hack. CE: Clewell. HI. (2007). Use of Markov Chain Monte Carlo analysis with a
physiologically-based pharmacokinetic model of methylmercury to estimate exposures in
US women of childbearing age. Risk Anal 27: 947-959. http: //dx.doi.org/10.1111 /i. 1539-
6924.2007.00934.x
ATSDR (Agency for Toxic Substances and Disease Registry). (1999). Toxicological profile for
mercury [ATSDR Tox Profile], Atlanta, GA: U.S. Department of Health and Human Services,
Public Health Service. http: //www.atsdr.cdc.gov/toxprofiles/tp.asp?id=115&tid=24
Berthet. A: de Batz. A: Tardif. R: Charest-Tardif. G: Truchon. G: Vernez. D: Droz. PP. (2010). Impact
of biological and environmental variabilities on biological monitoring-an approach using
toxicokinetic models. J Occup Environ Hyg 7: 177-184.
http://dx.doi.Org/10.1080/15459620903530052
Birch. RT: Bigler. I: Rogers. TW: Zhuang. Y: Clickner. RP. (2014). Trends in blood mercury
concentrations and fish consumption among U.S. women of reproductive age, NHANES,
1999-2010. Environ Res 133: 431-438. http://dx.doi.org/10.1016/i.envres.2014.02.001
Boucher. 0: Bastien. CH: Saint-Amour. D: Dewaillv. E: Avotte. P: Tacobson. TL: Tacobson. SW: Muckle.
G. (2010). Prenatal exposure to methylmercury and PCBs affects distinct stages of
information processing: An event-related potential study with Inuit children.
Neurotoxicology 31: 373-384. http://dx.doi.Org/10.1016/i.neuro.2010.04.005
Budtz-l0rgensen. E: Keiding. N: Grandiean. P. (1999). Benchmark modeling of the Faroese
methylmercury data: Final report to U.S. EPA. (Research Report 99/5). Copenhagen,
Denmark: University of Copenhagen.
Budtz-Torgensen. E: Keiding. N: Grandiean. P: Weihe. P. (2007). Confounder selection in
environmental epidemiology: Assessment of health effects of prenatal mercury exposure.
Ann Epidemiol 17: 27-35. http://dx.doi.Org/10.1016/i.annepidem.2006.05.007
This document is a draft for review purposes only and does not constitute Agency policy.
68 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Systematic Review Protocol for the Methylmercury IRIS Assessment
Bvczkowski. TZ: Lipscomb. TC. (2001). Physiologically based pharmacokinetic modeling of the
lactational transfer of methylmercury. Risk Anal 21: 869-882.
http://dx.doi.org/10.llll/0272-4332.215158
Carrier. G: Bouchard. M: Brunet. RC: Caza. M. (2001a). A toxicokinetic model for predicting the
tissue distribution and elimination of organic and inorganic mercury following exposure to
methyl mercury in animals and humans. II. Application and validation of the model in
humans. Toxicol Appl Pharmacol 171: 50-60. http: / /dx.doi.org/10.1006/taap.2000.9113
Carrier. G: Brunet. RC: Caza. M: Bouchard. M. (2001b). A toxicokinetic model for predicting the
tissue distribution and elimination of organic and inorganic mercury following exposure to
methyl mercury in animals and humans. I. Development and validation of the model using
experimental data in rats. Toxicol Appl Pharmacol 171: 38-49.
http://dx.doi.Org/10.1006/taap.2000.9112
CDC (Centers for Disease Control and Prevention). (2009). Mercury fact sheet [Fact Sheet], Atlanta,
GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic
Substances and Disease Registry.
CDC (Centers for Disease Control and Prevention). (2016a). Anthropometric reference data for
children and adults: United States, 2011-2014. Hyattsville, MD: U.S. Department of Health
and Human Services, Centers for Disease Control and Prevention, National Center for Health
Statistics.
CDC (Centers for Disease Control and Prevention). (2016b). Mercury biomonitoring summary.
Available online at
https://www.cdc.gov/biomonitoring/Mercury BiomonitoringSummary.html
CDC (Centers for Disease Control and Prevention). (2017). Table blood methyl mercury (2011-
2014), fourth national report on human exposure to environmental chemicals, updated
tables, January 2017, volume one (pp. 278). Atlanta, GA: U.S. Department of Health and
Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry.
https://www.cdc.gov/exposurereport/pdf/FourthReport UpdatedTables Volumel Tan201
7.pdf
CDC (Centers for Disease Control and Prevention). (2018). Fourth national report on human
exposure to environmental chemicals, updated tables, March 2018, volume one. Atlanta, GA:
U.S. Department of Health and Human Services.
https://www.cdc.gov/exposurereport/pdf/FourthReport UpdatedTables Volumel Mar20
18.pdf
Clewell. HI: Gearhart. TM: Gentry. PR: Covington. TR: Van Landingham. CB: Crump. KS: Shipp. AM.
(1999). Evaluation of the uncertainty in an oral reference dose for methylmercury due to
interindividual variability in pharmacokinetics. Risk Anal 19: 547-558.
http://dx.doi.Org/10.1023/a:1007017116171
Cui. W: Liu. G: Bezerra. M: Lagos. DA: Li. Y: Cai. Y. (2017). Occurrence of methylmercury in rice-
based infant cereals and estimation of daily dietary intake of methylmercury for infants. J
Agric Food Chem 65: 9569-9578. http://dx.doi.org/10.1021/acs.iafc.7b03236
This document is a draft for review purposes only and does not constitute Agency policy.
69 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Systematic Review Protocol for the Methylmercury IRIS Assessment
Daniels. TL: Longnecker. MP: Rowland. AS: Golding. T: Team. AS. (2004). Fish intake during
pregnancy and early cognitive development of offspring. Epidemiology 15: 394-402.
http://dx.doi.Org/10.1097/01.ede.0000129514.46451.ce
Despres. C: Beuter. A: Richer. F: Poitras. K: Veilleux. A: Avotte. P: Dewaillv. E: Saint-Amour. D:
Muckle. G. (2005). Neuromotor functions in Inuit preschool children exposed to Pb, PCBs,
and Hg. Neurotoxicol Teratol 27: 245-257. http://dx.doi.Org/10.1016/i.ntt.2004.12.001
FDA (U.S. Food and Drug Administration). (2018). Bioanalytical method validation guidance for
industry. Rockville, MD: Food and Drug Administration, https: //www.fda.gov/regulatorv-
information/search-fda-guidance-documents/bioanalytical-method-validation-guidance-
industrv
Golding. 1: Gregory. S: Iles-Caven. Y: Hibbeln. 1: Emond. A: Taylor. CM. (2016). Associations between
prenatal mercury exposure and early child development in the ALSPAC study.
Neurotoxicology 53: 215-222. http://dx.doi.Org/10.1016/i.neuro.2016.02.006
Gosselin. NH: Brunet. RC: Carrier. G: Bouchard. M: Feelev. M. (2006). Reconstruction of
methylmercury intakes in indigenous populations from biomarker data. J Expo Sci Environ
Epidemiol 16: 19-29. http://dx.doi.org/10.1038/si.iea.7500433
Grandiean. P: Weihe. P: White. RF: Debes. F: Araki. S: Yokovama. K: Murata. K: S0rensen. N: Dahl. R:
l0rgensen. PI. (1997). Cognitive deficit in 7-year-old children with prenatal exposure to
methylmercury. Neurotoxicol Teratol 19: 417-428. http://dx.doi.org/10.1016/SQ892-
0362C97100097-4
Health Canada. (2007). Mercury and human health. Ottawa, Ontario, Canada.
https://www.canada.ca/en/health-canada/services/healthy-living/your-
health/environment/mercury-human-health.html
Hoenig. TM: Heisev. DM. (2001). The abuse of power: The pervasive fallacy of power calculations for
data analysis. Am Stat 55: 19-24.
Howard. BE: Phillips. 1: Miller. K: Tandon. A: Mav. D: Shah. MR: Holmgren. S: Pelch. KE: Walker. V:
Roonev. AA: Macleod. M: Shah. RR: Thayer. K. (2016). SWIFT-Review: a text-mining
workbench for systematic review. Syst Rev 5: 87. http: //dx.doi.org/10.1186/sl3643-016-
0263-z
Tacobson. TL: Muckle. G: Avotte. P: Dewaillv. E: Tacobson. SW. f2015I Relation of prenatal
methylmercury exposure from environmental sources to childhood IQ. Environ Health
Perspect 123: 827-833. http://dx.doi.org/10.1289/ehp.14Q8554
Tin. L: Liu. M: Zhang. L: Li. Z: Yu. 1: Liu. 1: Ye. R: Chen. L: Ren. A. (2016). Exposure of methyl mercury
in utero and the risk of neural tube defects in a Chinese population. Reprod Toxicol 61: 131-
135. http://dx.doi.Org/10.1016/i.reprotox.2016.03.040
To. S: Woo. HP: Kwon. HI: Oh. SY: Park. ID: Hong. YS: Pvo. H: Park. KS: Ha. M: Kim. H: Sohn. ST: Kim.
YM: Lim. TA: Lee. SA: Eom. SY: Kim. BG: Lee. KM: Lee. TH: Hwang. MS: Kim. T. (2015).
Estimation of the biological half-life of methylmercury using a population toxicokinetic
model. Int J Environ Res Public Health 12: 9054-9067.
http://dx.doi.Org/10.3390/iierphl20809054
This document is a draft for review purposes only and does not constitute Agency policy.
70 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the Methylmercury IRIS Assessment
Karagas. MR: Choi. AL: Oken. E: Horvat. M: Schoenv. R: Kamai. E: Cowell. W: Grandjean. P: Korrick. S.
(2012). Evidence on the human health effects of low-level methylmercury exposure
[Review], Environ Health Perspect 120: 799-806. http://dx.doi.org/10.1289/ehp.1104494
Kershaw. TG: Clarkson. TW: Dhahir. PH. (1980). The relationship between blood levels and dose of
methylmercury in man. Arch Environ Occup Health 35: 28-36.
Kirman. CR: Sweeney. LM: Meek. ME: Gargas. ML. (2003). Assessing the dose-dependency of
allometric scaling performance using physiologically based pharmacokinetic modeling.
Regul Toxicol Pharmacol 38: 345-367. http://dx.doi.Org/10.1016/i.yrtph.2003.07.004
Lederman. SA: Tones. RL: Caldwell. KL: Rauh. V: Sheets. SE: Tang. D: Viswanathan. S: Becker. M:
Stein. TL: Wang. RY: Perera. FP. (2008). Relation between cord blood mercury levels and
early child development in a World Trade Center cohort Environ Health Perspect 116:
1085-1091. http://dx.doi.org/10.1289/ehp.10831
Lee. S: Tan. YM: Phillips. MB: Sobus. TR: Kim. S. (2017). Estimating methylmercury intake for the
general population of South Korea using physiologically based pharmacokinetic modeling.
Toxicol Sci 159: 6-15. http://dx.d0i.0rg/l0.109 3/toxsci/kfxl 11
Leggett. RW: Munro. NB: Eckerman. KF. (2001). Proposed revision of the ICRP model for inhaled
mercury vapor. Health Phys 81: 450-455. http://dx.doi.org/10.1097/00004Q32-
200110000-00010
Li. P: Feng. X: Chan. H: Zhang. X: Du. B. (2015). Human body burden and dietary methylmercury
intake: The relationship in a rice-consuming population. Environ Sci Technol 49: 9682-
9689. http://dx.doi.org/10.1021/acs.est.5b00195
Lowe. TR: Erickson. ST: Schrader. R: Duncan. AF. (2012). Comparison of the Bayley II Mental
Developmental Index and the Bayley III Cognitive Scale: Are we measuring the same thing?
Acta Paediatr 101: e55-e58. http://dx.doi.Org/10.llll/i.1651-2227.2011.02517.x
Lu. G: Abdulialil. K: Tamei. M: Tohnson. TN: Soltani. H: Rostami-Hodiegan. A. (2012). Physiologically-
based pharmacokinetic (PBPK) models for assessing the kinetics of xenobiotics during
pregnancy: Achievements and shortcomings [Review], Curr Drug Metab 13: 695-720.
http://dx.doi.Org/10.2174/138920012800840374
Mahaffev. KR: Clickner. RP: Teffries. RA. (2009). Adult women's blood mercury concentrations vary
regionally in the United States: Association with patterns of fish consumption (NHANES
1999-2004). Environ Health Perspect 117: 47-53. http: //dx.doi.org/10.1289/ehp.11674
Mcnallv. K: Loizou. GD. (2015). A probabilistic model of human variability in physiology for future
application to dose reconstruction and QIVIVE. Front Pharmacol 6: 213.
http://dx.doi.org/10.3389/fphar.2015.0Q213
Miettinen. IK: Rahola. T: Hattula. T: Rissanen. K: Tillander. M. (1971). Elimination of 203-Hg-
methylmercury in man. Ann Clin Res 3: 116-122.
Ng. S: Lin. CC: Teng. SF: Hwang. YH: Hsieh. WS: Chen. PC. (2015). Mercury, APOE, and child behavior.
Chemosphere 120: 123-130. http://dx.doi.Org/10.1016/i.chemosphere.2014.06.003
This document is a draft for review purposes only and does not constitute Agency policy.
71 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Systematic Review Protocol for the Methylmercury IRIS Assessment
Noisel. N: Bouchard. M: Carrier. G: Plante. M. (2011). Comparison of a toxicokinetic and a
questionnaire-based approach to assess methylmercury intake in exposed individuals. J
Expo Sci Environ Epidemiol 21: 328-335. http://dx.doi.org/10.1038/ies.2010.33
NRC (National Research Council). (2000). Toxicological effects of methylmercury. Washington, DC:
National Academy Press, http: //dx.doi.org/10.17226/9899
NRC (National Research Council). (2014). Review of EPA's Integrated Risk Information System
(IRIS) process. Washington, DC: The National Academies Press.
http://www.nap.edu/catalog.php7record id=18764
Oken. E: Osterdal. ML: Gillman. MW: Knudsen. VK: Halldorsson. TI: Strom. M: Bellinger. DC:
Hadders-Algra. M: Michaelsen. KF: Olsen. SF. (2008). Associations of maternal fish intake
during pregnancy and breastfeeding duration with attainment of developmental milestones
in early childhood: A study from the Danish National Birth Cohort Am J Clin Nutr 88: 789-
796.
Oken. E: Wright. R0: Kleinman. KP: Bellinger. D: Amarasiriwardena. CI: Hu. H: Rich-Edwards. TW:
Gillman. MW. (2005). Maternal fish consumption, hair mercury, and infant cognition in a
U.S. Cohort. Environ Health Perspect 113: 1376-1380. http://dx.doi.org/10.1289/ehp.8041
Orenstein. ST: Thurston. SW: Bellinger. DC: Schwartz. ID: Amarasiriwardena. CI: Altshul. LM:
Korrick. SA. (2014). Prenatal organochlorine and methylmercury exposure and memory
and learning in school-age children in communities near the New Bedford Harbor
Superfund site, Massachusetts. Environ Health Perspect 122: 1253-1259.
http://dx.doi.org/10.1289/ehp.1307804
Ou. L: Wang. H: Chen. C: Chen. L: Zhang. W: Wang. X. (2018). Physiologically based pharmacokinetic
(PBPK) modeling of human lactational transfer of methylmercury in China. Environ Int 115:
180-187. http://dx.doi.Org/10.1016/i.envint.2018.03.018
Pierrehumbert. G: Droz. PO: Tardif. R: Charest-Tardif. G: Truchon. G. f2002I Impact of human
variability on the biological monitoring of exposure to toluene, phenol, lead, and mercury: II.
Compartmental based toxicokinetic modelling. Toxicol Lett 134: 165-175.
http://dx.doi.org/10.1016/S0378-4274r02100186-8
Prpic. I: Milardovic. A: Vlasic-Cicvaric. I: Spiric. Z: Radic Nisevic. I: Vukelic. P: Snoi Tratnik. I: Mazei.
D: Horvat. M. (2017). Prenatal exposure to low-level methylmercury alters the child's fine
motor skills at the age of 18 months. Environ Res 152: 369-374.
http://dx.doi.0rg/lO.lOl6/i.envres.2Ol6.lO.Oll
Rothenberg. SE: Tackson. BP: Carlv McCalla. G: Donohue. A: Emmons. AM. (2017). Co-exposure to
methylmercury and inorganic arsenic in baby rice cereals and rice-containing teething
biscuits. Environ Res 159: 639-647. http://dx.doi.Org/10.1016/i.envres.2017.08.046
Rothenberg. SE: Yu. X: Liu. 1: Biasini. FT: Hong. C: Tiang. X: Nong. Y: Cheng. Y: Korrick. SA. (2016).
Maternal methylmercury exposure through rice ingestion and offspring neurodevelopment:
A prospective cohort study. Int J Hyg Environ Health 219: 832-842.
http://dx.doi.Org/10.1016/i.iiheh.2016.07.014
This document is a draft for review purposes only and does not constitute Agency policy.
72 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the Methylmercury IRIS Assessment
Ruiz. P: Fowler. BA: Osterloh. TP: Fisher. 1: Mumtaz. M. (2010). Physiologically based
pharmacokinetic (PBPK) tool kit for environmental pollutants-Metals. SAR QSAR Environ
Res 21: 603-618. http://dx.doi.org/10.1080/1062936X.2010.528942
Sagiv. SK: Thurston. SW: Bellinger. DC: Amarasiriwardena. C: Korrick. SA. (2012). Prenatal exposure
to mercury and fish consumption during pregnancy and attention-deficit/hyperactivity
disorder-related behavior in children. Arch Pediatr Adolesc Med 166: 1123-1131.
http://dx.doi.org/10.1001/archpediatrics.2012.1286
Sherlock. 1: Hislop. 1: Newton. D: Topping. G: Whittle. K. (1984). Elevation of mercury in human-
blood from controlled chronic ingestion of methylmercury in fish. Hum Toxicol 3: 117-131.
Sirot. V: Guerin. T: Mauras. Y: Garraud. H: Volatier. 1: Leblanc. TC. (2008). Methylmercury exposure
assessment using dietary and biomarker data among frequent seafood consumers in France
- CALIPSO study. Environ Res 107: 30-38. http://dx.doi.Org/10.1016/i.envres.2007.12.005
Smith. TC: Allen. PV: Turner. MP: Most. B: Fisher. HL: Hall. LL. (1994). The kinetics of intravenously
administered methylmercury in man. Toxicol Appl Pharmacol 128:251-256.
http://dx.doi.org/10.1006/taap.1994.1204
Srivastava. RK: Hutson. N: Martin. B: Princiotta. F: Staudt. 1. (2006). Control of mercury emissions
from coal-fired electric utility boilers. Environ Sci Technol 40: 1385-1393.
Stern. AH. (2005). A revised probabilistic estimate of the maternal methyl mercury intake dose
corresponding to a measured cord blood mercury concentration. Environ Health Perspect
113: 155-163. http://dx.doi.org/10.1289/ehp.7417
Stern. AH: Clewell. HI: Swartout. 1. (2002). An objective uncertainty factor adjustment for
methylmercury pharmacokinetic variability. Hum Ecol Risk Assess 8: 885-894.
Stern. AH: Smith. AE. (2003). An assessment of the cord blood:maternal blood methylmercury ratio:
implications for risk assessment. Environ Health Perspect 111: 1465-1470.
http://dx.doi.org/10.1289/ehp.6187
Sterne. TAC: Hernan. MA: Reeves. BC: Savovic. 1: Berkman. ND: Viswanathan. M: Henry. D: Altman.
DG: Ansari. MT: Boutron. I: Carpenter. TR: Chan. AW: Churchill. R: Peeks. 11: Hrobiartsson. A:
Kirkham. 1: liini. P: Loke. YK: Pigott. TP: Ramsay. CR: Regidor. D: Rothstein. HR: Sandhu. L:
Santaguida. PL: Schiinemann. HI: Shea. B: Shrier. I: Tugwell. P: Turner. L: Valentine. TC:
Waddington. H: Waters. E: Wells. GA: Whiting. PF: Higgins. TPT. (2016). ROBINS-I: A tool for
assessing risk of bias in non-randomised studies of interventions. Br Med J 355: i4919.
Straka. E: Ellinger. I: Balthasar. C: Scheinast. M: Schatz. 1: Szattler. T: Bleichert. S: Saleh. L: Knofler.
M: Zeisler. H: Hengstschlager. M: Rosner. M: Salzer. H: Gundacker. C. (2016). Mercury
toxicokinetics of the healthy human term placenta involve amino acid transporters and ABC
transporters. Toxicology 340: 34-42. http://dx.doi.Org/10.1016/i.tox.2015.12.005
Suzuki. C. (2016). Assessing change of environmental dynamics by legislation in Japan, using red
tide occurrence in Ise Bay as an indicator. Mar Pollut Bull 102: 283-288.
http://dx.doi.Org/10.1016/i.marpolbul.2015.08.010
This document is a draft for review purposes only and does not constitute Agency policy.
73 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
Systematic Review Protocol for the Methylmercury IRIS Assessment
Swartout. I: Rice. G. (2000). Uncertainty analysis of the estimated ingestion rates used to derive the
methylmercury reference dose. Drug Chem Toxicol 23: 293-306.
http://dx.doi.Org/10.1081/DCT-100100116
U.S. Department of the Interior. (2000). Mercury in the environment fact sheet 146-00 (October)
[Fact Sheet], Washington, D.C. https://www2.usgs.gov/themes/factsheet/146-00/
U.S. EPA (U.S. Environmental Protection Agency). (1991). Guidelines for developmental toxicity risk
assessment (pp. 1-71). (EPA/600/FR-91/001). Washington, DC: U.S. Environmental
Protection Agency, Risk Assessment Forum.
http: / /cfpub.epa.gov/ncea/cfm /recordisplay.cfm?deid=2 3162
U.S. EPA (U.S. Environmental Protection Agency). (1994). Methods for derivation of inhalation
reference concentrations and application of inhalation dosimetry [EPA Report],
(EPA/600/8-90/066F). Research Triangle Park, NC: U.S. Environmental Protection Agency,
Office of Research and Development, Office of Health and Environmental Assessment,
Environmental Criteria and Assessment Office.
https://cfpub.epa.gov/ncea/risk/recordisplay.cfm?deid=71993&CFID=51174829&CFTOKE
N=25006317
U.S. EPA (U.S. Environmental Protection Agency). (1996). Guidelines for reproductive toxicity risk
assessment (pp. 1-143). (EPA/630/R-96/009). Washington, DC: U.S. Environmental
Protection Agency, Risk Assessment Forum.
https://www.epa.gov/sites/production/files/2014-
11/documents/guidelines repro toxicity.pdf
U.S. EPA (U.S. Environmental Protection Agency). (1997). Mercury study report to congress. Volume
4. An assessment of exposure to mercury in the United States. (EPA/452/R-97-006). U.S.
EPA, Office of Air Quality Planning and Standards and Office of Research and Development.
U.S. EPA (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicity risk
assessment [EPA Report] (pp. 1-89). (EPA/630/R-95/001F). Washington, DC: U.S.
Environmental Protection Agency, Risk Assessment Forum.
http://www.epa.gov/risk/guidelines-neurotoxicitv-risk-assessment
U.S. EPA (U.S. Environmental Protection Agency). (2001a). IRIS summary for methylmercury
(MeHg) (CASRN 22967-92-6). Available online at
http://www.epa.gov/ncea/iris/subst/0Q73.htm (accessed September 13, 2010).
U.S. EPA (U.S. Environmental Protection Agency). (2001b). Methylmercury chemical assessment
summary. Washington, D.C.: Integrated Risk Information System, National Center for
Environmental Assessment
U.S. EPA (U.S. Environmental Protection Agency). (2002). A review of the reference dose and
reference concentration processes. (EPA/630/P-02/002F). Washington, DC: U.S.
Environmental Protection Agency, Risk Assessment Forum.
https://www.epa.gov/sites/production/files/2014-12/documents/rfd-final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2004). Toxicological review of boron and
compounds. In support of summary information on the Integrated Risk Information System
(IRIS) [EPA Report], (EPA/635/04/052). Washington, DC: U.S. Environmental Protection
Agency, IRIS. http://nepis.epa.gov/exe/ZyPURL.cgi?Dockey=P1006CK9.txt
This document is a draft for review purposes only and does not constitute Agency policy.
74 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Systematic Review Protocol for the Methylmercury IRIS Assessment
U.S. EPA (U.S. Environmental Protection Agency). (2005). Guidelines for carcinogen risk assessment
[EPA Report], (EPA/630/P-03/001F). Washington, DC: U.S. Environmental Protection
Agency, Risk Assessment Forum, https://www.epa.gov/sites/production/files/2013-
09/documents/cancer guidelines final 3-25-05.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2010). Guidance for implementing the January
2001 methylmercury water quality criterion, April 2010. (NTIS/11880158).
U.S. EPA (U.S. Environmental Protection Agency). (2011). Toxicological review of trichloroethylene
(CASRN 79-01-6) in support of summary information on the Integrated Risk Information
System (IRIS) [EPA Report], (EPA/635/R-09/011F). Washington, DC.
https: //cfpub.epa.gov/ncea/iris /iris documents/documents/toxreviews /0199tr /0199tr.p
df
U.S. EPA (U.S. Environmental Protection Agency). (2012). Benchmark dose technical guidance.
(EPA/100/R-12/001). Washington, DC: U.S. Environmental Protection Agency, Risk
Assessment Forum, https: //www.epa.gov/risk/benchmark-dose-technical-guidance
U.S. EPA (U.S. Environmental Protection Agency). (2014). Guidance for applying quantitative data to
develop data-derived extrapolation factors for interspecies and intraspecies extrapolation
[EPA Report], (EPA/100/R-14/002F). Washington, DC: Risk Assessment Forum, Office of
the Science Advisor, https://www.epa.gov/sites/production/files/2015-
01/documents/ddef-final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2015). Science policy council peer review
handbook [EPA Report] (4th ed.). (EPA/100/B-15/001). Washington, DC: U.S.
Environmental Protection Agency, Science Policy Council, https: //www.epa.gov/osa/peer-
revie w-handbook- 4th-edition- 2015
U.S. EPA (U.S. Environmental Protection Agency). (2018). An umbrella Quality Assurance Project
Plan (QAPP) for PBPK models [EPA Report], (0RD QAPP ID No: B-0030740-QP-1-1).
Research Triangle Park, NC.
UNEP (United Nations Environment Programme). (2002). Global mercury assessment. Geneva,
Switzerland: UNEP Chemicals.
https://www.unenvironment.org/explore-topics/chemicals-waste/what-we-do/
mercury/global-mercury-assessment
Wells. EM: Kopvlev. L: Nachman. R: Radke. EG: Segal. D. (2020). Seafood, wine, rice, vegetables, and
other food items associated with mercury biomarkers among seafood and non-seafood
consumers: NHANES 2011-2012. J Expo Sci Environ Epidemiol.
http://dx.doi.Org/10.1038/s41370-020-0206-6
WHO (World Health Organization). (1990). Methylmercury.
Yaginuma-Sakurai. K: Murata. K: Iwai-Shimada. M: Nakai. K: Kurokawa. N: Tatsuta. N: Satoh. H.
(2012). Hair-to-blood ratio and biological half-life of mercury: experimental study of
methylmercury exposure through fish consumption in humans. J Toxicol Sci 37: 123-130.
http://dx.doi.org/10.2131/its.37.123
Yang. TM: Tiang. ZZ: Wang. YL: Oureshi. IA: Wu. XD. (1997). Maternal-fetal transfer of metallic
mercury via the placenta and milk. Ann Clin Lab Sci 27: 135-141.
This document is a draft for review purposes only and does not constitute Agency policy.
75 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the Methylmercury IRIS Assessment
1 Young. IF: Wosilait. WD: Luecke. RH. (2001). Analysis of methylmercury disposition in humans
2 utilizing a PBPK model and animal pharmacokinetic data. J Toxicol Environ Health A 63: 19-
3 52. http://dx.doi.org/10.1080/152873901750128344
This document is a draft for review purposes only and does not constitute Agency policy.
76 DRAFT-DO NOT CITE OR QUOTE
------- |