A EPA
EPA/635/R-18/155
IRIS Assessment Protocol
www.epa.gov/iris
Systematic Review Protocol for the Hexavalent Chromium IRIS
Assessment
(Preliminary Assessment Materials)
[CASRN 18540-29-9]
March 2019
Integrated Risk Information System
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC

-------
Systematic Review Protocol for the Hexavalent Chromium 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ii	DRAFT-DO NOT CITE OR QUOTE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 Hexavalent Chromium IRIS Assessment
CONTENTS
AUTHORS | CONTRIBUTORS	ix
1.	INTRODUCTION	1
2.	SCOPING AND INITIAL PROBLEM FORMULATION SUMMARY	3
2.1. BACKGROUND	3
2.1.1. Previous IRIS Assessment	4
2.2.SCOPING SUMMARY	4
2.3. PROBLEM FORMULATION	5
3.	ASSESSMENT APPROACH, SPECIFIC AIMS, AND DRAFT POPULATIONS, EXPOSURES,
COMPARATORS, AND OUTCOMES (PECO) CRITERIA	11
3.1.	ASSESSMENT APPROACH	11
3.1.1.	Evaluation of the Potential Carcinogenicity of Inhaled Cr(VI)	11
3.1.2.	Evaluation of the Effects of Inhaled Cr(VI) on the Nasal Cavity	12
3.1.3.	Toxicokinetics of Cr(VI)	12
3.2.	SPECIFIC AIMS	13
3.3.	PECO CRITERIA	14
4.	LITERATURE SEARCH AND SCREENING STRATEGIES	16
4.1.	LITERATURE SEARCH STRATEGIES	16
4.2.	NON-PEER-REVIEWED DATA	17
4.3.SCREENING PROCESS	18
4.3.1.	Title- and Abstract-Level Screening	19
4.3.2.	Full-Text-Level Screening	21
4.3.3.	Multiple Publications of the Same Data	24
4.4.	SUMMARY-LEVEL LITERATURE INVENTORIES	24
4.4.1.	Studies Meeting PECO Criteria	24
4.4.2.	Potentially Relevant Supplemental Material	25
5.	REFINED EVALUATION PLAN	26
5.1.	AIRBORNE CHARACTERIZATION AND CHEMICAL PROPERTIES	26
5.2.TOXICOKINETIC	S	27
5.3.TOXICOGENOMIC	S	27
5.4.OUTCOMES CONSIDERED IN THE Cr(VI) ASSESSMENT	28
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iii	DRAFT-DO NOT CITE OR QUOTE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 Hexavalent Chromium IRIS Assessment
6.	STUDY EVALUATION (REPORTING, RISK OF BIAS, AND SENSITIVITY) STRATEGY	34
6.1.	STUDY EVALUATION OVERVIEW FOR HEALTH EFFECT STUDIES	34
6.2.	EPIDEMIOLOGY STUDY EVALUATION	38
6.3.	EXPERIMENTAL ANIMAL STUDY EVALUATION	48
6.4.	PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL DESCRIPTIVE SUMMARY
AND EVALUATION	60
6.4.1.	Pharmacokinetic (PK)/Physiologically Based Pharmacokinetic (PBPK) Model
Descriptive Summary	61
6.4.2.	Pharmacokinetic (PK)/Physiologically Based Pharmacokinetic (PBPK) Model
Evaluation	62
6.5.	MECHANISTIC STUDY EVALUATION	63
7.	ORGANIZING THE HAZARD REVIEW	66
8.	DATA EXTRACTION OF STUDY METHODS AND RESULTS	69
8.1.STANDARDIZING REPORTING OF EFFECT SIZES	70
8.2.STANDARDIZING ADMINISTERED DOSE LEVELS/CONCENTRATIONS	72
9.	SYNTHESIS WITHIN LINES OF EVIDENCE	73
9.1.SYNTHESES OF HUMAN AND ANIMAL HEALTH EFFECTS EVIDENCE	76
9.2. MECHANISTIC INFORMATION	77
10.	INTEGRATION ACROSS LINES OF EVIDENCE	81
10.1.	INTEGRATION WITHIN THE HUMAN AND ANIMAL EVIDENCE	82
10.2.	OVERALL EVIDENCE INTEGRATION CONCLUSIONS	92
10.3.	HAZARD CONSIDERATIONS FOR DOSE-RESPONSE	96
11.	DOSE-RESPONSE ASSESSMENT: SELECTING STUDIES AND QUANTITATIVE ANALYSIS	99
11.1.	SELECTING STUDIES FOR DOSE-RESPONSE ASSESSMENT	100
11.2.	CONDUCTING DOSE-RESPONSE ASSESSMENTS	103
11.2.1.	Dose-Response Analysis in the Range of Observation	104
11.2.2.	Extrapolation: Slope Factors and Unit Risks	105
11.2.3.	Extrapolation: Reference Values	106
12.	PROTOCOL HISTORY	108
REFERENCES	109
APPENDICES	121
APPENDIX A. ELECTRONIC DATABASE SEARCH STRATEGIES	121
APPENDIX B. TYPICAL DATA EXTRACTION FIELDS	128
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 Hexavalent Chromium IRIS Assessment
TABLES
Table 1.	EPA program and regional office interest in a reassessment of Cr(VI)	5
Table 2.	Cr(VI) values for inhalation exposure (ng/m3) from U.S. federal and state
agencies and international bodies (in reverse chronological order)	6
Table 3.	Cr(VI) cancer risk values for inhalation exposure from U.S. federal and state
agencies and international bodies (in reverse chronological order)	8
Table 4.	Cr(VI) values for oral exposure from U.S. federal and state agencies and
international bodies (in reverse chronological order)	9
Table 5.	Populations, exposures, comparators, and outcomes (PECO) criteria	15
Table 6.	Outcomes and associated endpoints to be considered for animal study
evaluation	29
Table 7.	Outcomes and associated endpoints to be considered for human study
evaluation	31
Table 8.	Inventory of selected reference topics screened as "potentially relevant
supplemental material" to be considered in the assessment	32
Table 9.	Questions to guide the development of criteria for each domain in epidemiology
studies	40
Table 10.	Information relevant to evaluation domains for epidemiology studies	48
Table 11.	Questions to guide the development of criteria for each domain in experimental
animal toxicology studies	50
Table 12.	Physiologically based pharmacokinetic models for Cr(VI)	62
Table 13.	Criteria for evaluating physiologically based pharmacokinetic (PBPK) models	64
Table 14.	Querying the evidence to organize syntheses for human and animal evidence	67
Table 15.	Information most relevant to describing primary considerations informing
causality during evidence syntheses	74
Table 16.	Individual and social factors that may increase susceptibility to exposure-related
health effects	76
Table 17.	Evidence profile table template	83
Table 18.	Considerations that inform judgments regarding the strength of the human and
animal evidence	85
Table 19.	Framework for evidence judgments from studies in humans	89
Table 20.	Framework for evidence judgments from studies in animals	91
Table 21.	Conclusions for the evidence integration narrative	94
Table 22. Attributes used to evaluate studies for derivation of toxicity values	101
Table A-l. Literature search query strings for computerized databases	121
Table A-2. Processes used to augment the search of core computerized databases for
Cr(VI)	124
Table B-l. Key data extraction elements to summarize study design, experimental model,
methodology, and results	128
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 Hexavalent Chromium IRIS Assessment
FIGURES
Figure 1.	Literature search flow diagram for Cr(VI)	23
Figure 2.	Overview of Integrated Risk Information System (IRIS) study evaluation process	35
Figure 3.	Relationship between ex vivo reduction models, in vivo gastric models, and
whole-body physiologically based pharmacokinetic (PBPK) models	60
Figure 4.	Process for evidence integration	81
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 Hexavalent Chromium IRIS Assessment
ABBREVIATIONS
ACGIH American Conference of Governmental	GRADE
Industrial Hygienists
ACToR Aggregated Computational Toxicology
Resource	HAP
AD ME absorption, distribution, metabolism,	HAWC
and excretion
AIHA American Industrial Hygiene	HERO
Association
ATSDR Agency for Toxic Substances and	HPV
Disease Registry	HPVIS
BMD benchmark dose
BMR benchmark response	HSDB
CAA Clean Air Act	HSNO
CalEPA California Environmental Protection
Agency	IAP
CASRN Chemical Abstracts Service registry	IARC
number
CCA chromated copper arsenate	IRIS
CCID Chemical Classification Information	IUCLID
Database
CCR Canadian Categorization Results	IUR
CCRMP Coordinated Chemicals Risk	J-CHECK
Management Programme Publications
CDAT Chemical Data Access Tool	JECDB
CEPA Canadian Environmental Protection Act	LOAEL
CESAR Canada's Existing Substances	LOEL
Assessment Repository	MOA
CHRIP Chemical Risk Information Platform	NAP
CPSC Consumer Product Safety Commission	NATA
Cr(III) trivalent chromium	NCEA
Cr(VI] hexavalent chromium
Cr042- chromate	NCI
Cr20 72" dichromate	NICNAS
DoCTER Document Classification and Topic
Extraction Resource	NIEHS
ECETOC European Centre for Ecotoxicology and
Toxicology of Chemicals	NIOSH
ECHA European Chemicals Agency
EnviChem Data Bank of Environmental Properties	NIOSHTIC
of Chemicals
EO	Executive Order
EPA Environmental Protection Agency	NMD
ERPG Emergency Response Planning	NOEL
Guidelines	NSCEP
ESIS European Chemical Substances
Information System	NTP
ESR Existing Substances Regulation	OECD
FDA Food and Drug Administration
GHS-J Globally Harmonized System-Japan	OEHHA
GI	gastrointestinal
OPP
Grading of Recommendations
Assessment, Development and
Evaluation
hazardous air pollutant
Health Assessment Workplace
Collaborative
Health and Environmental Research
Online
high production volume
High Production Volume Information
System
Hazardous Substances Data Bank
Hazardous Substances and New
Organisms
IRIS Assessment Plan
International Agency for Research on
Cancer
Integrated Risk Information System
International Uniform Chemical
Information Database
inhalation unit risk
Japan CHEmicals Collaborative
Knowledge
Japan Existing Chemical Data Base
lowest-observed-adverse-effect level
lowest-observed-effect level
mode of action
National Academies Press
National-Scale Air Toxics Assessment
National Center for Environmental
Assessment
National Cancer Institute
National Industrial Chemicals
Notification and Assessment Scheme
National Institute for Environmental
Health Sciences
National Institute for Occupational
Safety and Health
National Institute for Occupational
Safety and Health Technical
Information Center
normalized mean difference
no-observed-effect level
National Service Center for
Environmental Publications
National Toxicology Program
Organisation for Economic Cooperation
and Development
Office of Environmental Health Hazard
Assessment
Office of Pesticide Programs
This document is a draft for review purposes only and does not constitute Agency policy.
vii	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
ORD
Office of Research and Development
RoC
Report on Carcinogens
OSF
oral slope factor
RTECS
Registry of Toxic Effects of Chemical
OSHA
Occupational Safety and Health

Substances

Administration
SIDS
Screening Information Data Set
PBPK
physiologically based pharmacokinetic
SRS
Substance Registry Services
PEC
priority existing chemical
TCEQ
Texas Commission on Environmental
PECO
populations, exposures, comparators,

Quality

and outcomes
TSCA
Toxic Substances Control Act
PK
pharmacokinetic
TSCATS
Toxic Substances Control Act Test
POD
point of departure

Submissions
RED
registration eligibility decision
UK
United Kingdom
REL
reference exposure level
UNEP
United Nations Environment
RfC
reference concentration

Programme
RfD
reference dose
WEEL
Workplace Environmental Exposure
ROBINS-I
Risk of Bias in Nonrandomized Studies

Level

of Interventions
WOS
Web of Science
This document is a draft for review purposes only and does not constitute Agency policy.
viii	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
AUTHORS | CONTRIBUTORS
Assessment Team
Catherine Gibbons (co-Assessment Manager) U.S. EPA/ORD/NCEA
Alan Sasso (co-Assessment Manager)
Xabier Arzuaga
Johanna Congleton
Barbara Glenn
Rebecca Nachman
Larissa Pardo (formerly with EPA)
Elizabeth Radke
Paul Reinhart
Amina Wilkins
Erin Yost
David Lehmann	U.S. EPA/ORD/NHEERL
Contributors
Thomas Bateson	U.S. EPA/ORD/NCEA
Vincent Cogliano
Laura Dishaw
Leonid Kopylev
Roman Mezencev
Paul Schlosser
Executive Direction
Tina Bahadori
Mary Ross
Emma Lavoie
Samantha Jones
Kristina Thayer
James Avery
Andrew Kraft
Susan Rieth
NCEA Center Director
NCEA Deputy Center Director
NCEA Assistant Center Director for Scientific Support
NCEA Associate Director for Health (acting)
NCEA/IRIS Division Director
NCEA/IRIS Deputy Division Director (acting)
NCEA/IRIS Associate Director for Science (acting)
NCEA/IRIS Quantitative Modeling Branch Chief
This document is a draft for review purposes only and does not constitute Agency policy.
ix	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Production Team
Vicki Soto	Project Management Team
Dahnish Shams	Project Management Team
Maureen Johnson	NCEA Webmaster
Contractor Support
Kim Osborn	ICF International
Robyn Blain
Michelle Cawley
Alessandria Schumacher
This document is a draft for review purposes only and does not constitute Agency policy.
x	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 Hexavalent Chromium IRIS Assessment
1. INTRODUCTION
The Integrated Risk Information System (IRIS) Program is undertaking a reassessment of
the health effects of hexavalent chromium (Cr[VI]). Significant new epidemiologic and
experimental animal toxicity information for Cr(VI) has become available since EPA's IRIS
assessment for Cr(VI) was posted in 1998, including updates of occupational cohort studies
(Proctor etal.. 2016: Gibb etal.. 20151 and a National Toxicology Program (NTP) bioassay that
reported increased incidences of tumors in rats and mice exposed to Cr(VI) in drinking water (NTP.
20081. The dose-response information from epidemiologic and experimental animal studies
published since 1998 could result in changes to current toxicity values. Cr(VI) was included on the
December 2015 IRIS Program multiyear agenda fhttps://www.epa.gov/iris/iris-agendal as a
chemical having high priority for assessment development It was also included in the December
2018 IRIS Program Outlook that provides an updated outlook of IRIS program activities
(https://www.epa.gov/sites/production/files/2018-
12/documents/iris program outlook december 2018.pdf). Given the widespread exposure to
Cr(VI) and the availability of studies that provide significant new health effects information, the
IRIS Program is developing an updated assessment of Cr(VI).
Preliminary materials for the Cr(VI) reassessment were released to the public in April and
August 2014, and public meetings were held in June and October 2014 to seek input regarding the
Cr(VI) assessment from the scientific community and interested parties fU.S. EPA. 2014b. c). The
preliminary materials included a summary of the IRIS Program's scoping and problem formulation
conclusions, information on the approaches used to identify pertinent literature, results of the
literature search, approaches for selection of studies for hazard identification, and presentation of
studies eligible for study evaluation in evidence tables and exposure-response arrays. A
preliminary summary of toxicokinetic and mechanistic studies pertinent to the assessment was also
presented.
The protocol is a new document adopted by the IRIS Program as part of its full
implementation of systematic review [see presentation materials for the NAS Workshop "Review of
Advances Made to the IRIS Process" (Bahadori and Thayer. 20181 and the NASEM (20181 report
Progress Toward Transforming the Integrated Risk Information System]. The chemical-specific
protocol describes how the assessment will be conducted, while the IRIS Assessment Plan (IAP),
typically released early in the assessment process, describes what the assessment plans to cover
based on scoping and problem formulation. As noted above, scoping and problem formulation
documents were previously released for public comment; this protocol summarizes and updates
those earlier materials (e.g., see Sections 1-4). Because development of the chromium assessment
began before the introduction of these early stage documents to the IRIS process, EPA is
This document is a draft for review purposes only and does not constitute Agency policy.
1	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
1	retroactively releasing the protocol, which presents the methods for conducting the systematic
2	review and dose-response analysis, to provide similar public engagement steps and documentation
3	as other assessments that started more recently. This protocol also includes specific aims and
4	populations, exposures, comparators, and outcomes (PECO) criteria that were not a part of the
5	2014 preliminary materials but are now apart of IRIS Systematic Review materials. The IRIS
6	Program posts assessment protocols on its website and in the Zenodo repository
7	fhttps://zenodo.org/l. Public comments will be considered as part of developing the draft
8	assessment. This protocol documents the studies identified during the initial literature searches
9	fU.S. EPA. 2014b. c) and updates to those literature searches. Additional literature search updates
10	will be posted to the IRIS website when they are available.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
2.SCOPING AND INITIAL PROBLEM FORMULATION
SUMMARY
2.1. BACKGROUND
Elemental chromium is a Group 6 transition metal (atomic number 24 and atomic weight
52) on the periodic table, existing in nature in the form of various oxide minerals (Anger etal..
20051. It is present in the Earth's crust and has oxidation states ranging from -2 to +6, with the +3
(trivalent) and +6 (hexavalent) states being the most common (Losi etal.. 1994). Chromium in the
environment can originate from both natural and anthropogenic sources. Atmospheric releases of
chromium from natural and anthropogenic sources are comparable in magnitude, while soil
releases are mostly anthropogenic, and all water releases are anthropogenic fUSGS. 19951.
Conversion of Cr(VI) to Cr(III) may occur in the environment under reducing conditions (by ferrous
iron, sulfides, and organic matter), while conversion of Cr(III) to Cr(VI) may occur under oxidizing
conditions [by manganese oxide minerals; (Hausladen andFendorf. 2017: McClain etal.. 2017:
Tardine etal.. 2011: Cummings etal.. 2007: Oze etal.. 2007: Oze etal.. 2004: Kim and Dixon. 2002:
Fendorf etal.. 2000: Fendorf. 19951], Most Cr(III) compounds are insoluble in water and immobile
in soils (which helps inhibit oxidation), while Cr(VI) compounds are readily soluble in water and
highly mobile and bioavailable f Fendorf etal.. 2000: Fendorf. 19951. In addition to being stabilized
by low solubility and mobility, Cr(III) compounds are more thermodynamically stable than Cr(VI)
compounds under mostpH values encountered in the environment (Fendorf. 1995).
Cr(VI) compounds are used for corrosion inhibition (including within water-cooling
systems), pigment manufacturing (including textile dyeing and printing inks), metal finishing
(chrome plating/electroplating), stainless steel production, leather manufacturing (leather
tanning), refractories (linings for high-temperature industrial furnaces), drilling muds,
pyrotechnics, chemical synthesis, and plastics fNIOSH. 2013b: NTP. 20111. Chromium compounds
have been used in wood preservatives [as chromated copper arsenate (CCA) in pressure-treated
wood; (ATSDR. 2012: Barnhart. 1997)]: however, this use began to decline in 2003 due to a
voluntary phaseout of all residential uses of CCA pressure-treated wood (Bedinger. 2015: NTP.
20111.
Occupational exposures to Cr(VI) occur primarily from inhalation or dermal contact
(NIOSH. 2013b). while general population exposures occur by inhalation of ambient air and
ingestion of food and drinking water fNTP. 20111. Dermal exposure may also occur from using
consumer products that contain chromium, such as some metals and wood or leather treated with
chromium-containing compounds (ATSDR. 2012: NTP. 2011). According to data collected between
2013 and 2015 under EPA's Third Unregulated Contaminant Monitoring Rule (UCMR3), Cr(VI) has
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
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
been reported above the minimum reporting limit (0.03 ng/L) by approximately 90% of public
water systems in the United States fU.S. EPA. 2014dl. Ambient air concentrations of Cr(VI) in the
United States typically range from 0.01 to 0.05 ng/m3 (U.S. EPA. 20161 but have been measured at
values above 1 ng/m3 in urban and industrial areas (Oregon DEO. 2016: Huang etal.. 2014: CalEPA.
2004. 20031. Cr(VI) concentrations measured in air downwind of industrial facilities emitting
Cr(VI) (such as chrome platers) have been found to be highly correlated with concentrations
measured at the facilities (OAOPS. 2012: CalEPA. 2004. 20031.
2.1.1. Previous IRIS Assessment
EPA's 1998 IRIS assessment classified Cr(VI) as "Group A—known human carcinogen by the
inhalation route of exposure" based on evidence of a causal relationship between inhalation of
Cr(VI) and increased incidence of lung cancer in humans. An inhalation unit risk (IUR) for Cr(VI) of
1.2 x 10"2 per |J.g/m3 was calculated based on increased incidence of lung cancer in chromate
workers fMancuso. 1997.19751. The 1998 assessment concluded that the carcinogenicity of Cr(VI)
"by the oral route of exposure cannot be determined and is classified as Group D." Accordingly, a
cancer slope factor for ingested Cr(VI) was not derived.
EPA's 1998 IRIS assessment derived two inhalation reference concentrations (RfCs) for
noncancer effects. An RfC of 8 x 10~3 |J.g/m3 was derived based on nasal effects observed in an
epidemiologic study of workers in chrome plating plants (Lindberg and Hedenstierna. 19831. and
was specific to chromic acid mists and dissolved Cr(VI) aerosols. An additional RfC of 0.1 |J.g/m3
was derived based on respiratory tract effects observed in subchronic duration rat studies fMalsch
etal.. 1994: Glaser etal.. 19901. and was specific to Cr(VI) particulates. EPA's 1998 IRIS assessment
also derived an oral reference dose (RfD) of 3 x 10~3 mg/kg-day for noncancer effects based on a
no-observed-adverse-effect level (NOAEL) reported in a 1-year drinking water study in rats
(MacKenzie etal.. 19581.
2.2. SCOPING SUMMARY
During scoping, the IRIS Program met with EPA program and regional offices that had
interest in an IRIS assessment for Cr(VI) to discuss specific assessment needs. As discussed in the
April 2014 preliminary materials document (U.S. EPA. 2014b). the scope of the IRIS assessment will
be limited to potential health effects by the inhalation and oral routes of exposure. EPA's Office of
Pesticide Programs (OPP) previously evaluated the dermal exposure pathway in its reregistration
eligibility decision (RED) for CCA pesticides (U.S. EPA. 2008c). and no priority needs related to
dermal exposure were identified by other EPA program and regional offices. Table 1 provides a
summary of EPA offices, programs, and regions that have interest in the assessment and what their
specific needs are.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 1. EPA program and regional office interest in a reassessment of Cr(VI)
EPA
program or
regional
office
Oral
Inh.
Statutes/regulations and anticipated uses/interest
OLEM
V
V
CERCLA and RCRA
Cr(VI) has been identified as a contaminant of concern at numerous contaminated
waste sites, including more than 100 NPL sites. 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. Cr(VI) toxicological information
may be used to make risk determinations for response actions (e.g., short-term
removals, long-term remedial response actions, RCRA Corrective Action).
EPA Regions
1-10
OW
V

SDWA
Currently, the EPA drinking water standard of 0.1 mg/L is for total chromium (Federal
Register, 2010). The SDWA requires EPA to periodically review the NPDWR for each
contaminant and revise the regulation, if appropriate. Cr(VI) toxicological information
may be used to inform risk determinations associated with revisiting the NPDWR.
Chromium is listed under the NPDWR.
CERCLA = Comprehensive Environmental Response, Compensation, and Liability Act; RCRA = Resource
Conservation and Recovery Act; Inh. = inhalation; NPDWR= National Primary Drinking Water Regulation;
NPL = National Priority List; OLEM = Office of Land and Emergency Management; OW = Office of Water;
SDWA = Safe Drinking Water Act.
2.3. PROBLEM FORMULATION
Problem formulation information pertaining to the reassessment of Cr(VI) was included in
the preliminary materials documents released to the public in April and August 2014 fU.S. EPA.
2014b. c); two public meetings were held in June and October 2014 to obtain public input on these
materials.
As discussed in the April 2014 preliminary materials document fU.S. EPA. 2014b). EPA
consulted federal, state, and international agency health assessments published since the U.S. EPA
f!998b) IRIS Toxicological Review of Hexavalent Chromium to identify studies and scientific issues
that may impact the reassessment of Cr(VI). EPA has continued to consult other agency health
assessments following the 2014 public meetings. These health agencies, and information regarding
the basis of any protective exposure values or health determinations, are presented in Tables 2 to 4.
Based on prior health agency assessments of Cr(VI) described in Tables 2 and 4, the health effects
of primary interest for evaluation in the current IRIS assessment are respiratory and
gastrointestinal (GI) effects. These health agencies also identify other potential target systems of
possible interest to the current IRIS assessment; these are discussed in Section 3.1 fU.S. EPA.
2014b).
This document is a draft for review purposes only and does not constitute Agency policy.
5	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 2. Cr(VI) values for inhalation exposure (|ig/m3) from U.S. federal and
state agencies3 and international bodies (in reverse chronological order)
Reference
Value
(pg/m3)
Time
adjustment
Chemical
note
Endpoints/basis
TCEQ (2014)
0.0043
Lifetime/chronic
Particulate
compounds
Excess lung cancer mortality risk of 1 x 10~5,
using risk value derived from Gibb et al.
(2000a) and Crump et al. (2003).

0.066
Lifetime/chronic
Particulate
compounds
Respiratory effect (increased relative lung
weight after 90 days of exposure) in rats
(Glaser et al., 1985).

0.39
Acute
Particulate
compounds
Respiratory effect (increased relative lung
weight after 30 days of exposure) in rats
(Glaser et al., 1990).
IPCS (2013)
0.03
Lifetime/chronic
Cr(VI) salts
Respiratory effects in rats (Glaser et al.,
1990).

0.005
Lifetime/chronic
Chromium
trioxide,
chromic acid
Upper respiratory effects in humans
(Lindberg and Hedenstierna, 1983).
NIOSH (2013a)
0.2
8-hour TWA,
40-hour
workweek
All Cr(VI)
compounds
Lung cancer and nonmalignant respiratory
effects. Based on analysis of Baltimore
cohort data bv Park et al. (2004).
ATSDR (2012)
0.005
Chronic
Dissolved
aerosols and
mists
Upper respiratory effects (nasal
irritation/ulceration, mucosal atrophy, and
decreases in spirometric parameters), based
on Lindberg and Hedenstierna (1983).

N/A
Chronic
Particulates
Insufficient data

0.005
Intermediate
Dissolved
aerosols and
mists
Upper respiratory effects (nasal
irritation/ulceration, mucosal atrophy, and
decreases in spirometric parameters), based
on Lindberg and Hedenstierna (1983).

0.3
Intermediate
Particulates
Respiratory tract (lung) and other effects.
Based on quantitative analysis of rat studies
(Glaser et al., 1990; Glaser et al., 1985)
performed bv Malsch et al. (1994).
OEHHA (2008)
0.2
Chronic
Soluble
compounds
Respiratory effect (bronchoalveolar
hyperplasia) in rats (Glaser et al., 1990).

0.002
Chronic
Chromic
trioxide (as
chromic acid
mist)
Respiratory effects in humans (Lindberg and
Hedenstierna, 1983).
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 Hexavalent Chromium IRIS Assessment
Reference
Value
(Hg/m3)
Time
adjustment
Chemical
note
Endpoints/basis
OSHA (2006)
5
8-hour TWA
All Cr(VI)
compounds
Lung cancer and nasal tissue damage. Based
on quantitative analysis of Baltimore cohort
data bv Gibb et al. (2000b) and Gibb et al.
(2000a).
RIVM (2001)
0.0025
Chronic
Inhalable dust
Excess lifetime lung cancer risk of 1 x 10"4,
based on analysis of human occupational
studies by the 1987 and 1994 World Health
Organization air quality guidelines.15
U.S. EPA (1998b)
0.008
Lifetime/chronic
Chromic acid
mists/dissolved
chromium
aerosols
Effects in the nasal cavity. Based on
Lindberg and Hedenstierna (1983).
0.1
Lifetime/chronic
Cr(VI)
particulates
Respiratory effects. Based on quantitative
analysis of rat studies (Glaser et al., 1990;
Glaser et al., 1985) performed bv Malsch et
al. (1994).
N/A = not applicable; TWA = time-weighted average.
aSelected values from states known by U.S. EPA to have derived independent values; most states typically adopt
values from U.S. EPA.
bRisk value rationale and studies unchanged in WHO (2000).
This document is a draft for review purposes only and does not constitute Agency policy.
7	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 3. Cr(VI) cancer risk values for inhalation exposure from U.S. federal
and state agencies3 and international bodies (in reverse chronological order)
Reference
Risk factor (ng/m3) 1
Rationale
TCEQ (2014)
Unit risk factor: 2.28 x 10~3
(particulate compounds)
Linearly extrapolated lung cancer risk based on a
weighted average of Gibb et al. (2000a) (Baltimore
cohort) and Crump et al. (2003) (Painesville
cohort).
IPCS (2013)
Occupational exposure risk:
6 x 10"3
Linearly extrapolated lung cancer risk based on
Gibb et al. (2000a).
Environmental exposure risk:
4 x 10"2
IARC (2012).
Carcinogenic to humans
(Group l)b
Lung cancer, based on multiple evidence streams.
Positive associations between Cr(VI) exposure and
cancer of the nose and nasal sinuses also cited.
NTP (2011)
Known to be human
carcinogen15
Cancers of the lung and sinonasal cavity, based on
studies in humans.
CalEPA (2011)
0.16 (95% upper confidence:
0.35)
Linearly extrapolated lung cancer risk based on
Gibb et al. (2000a).
1 x 10"2 (lower bound)
Linearly extrapolated lung cancer risk based on
Luippold et al. (2003).
WHO (2000)
4 x 10"2
Linearly extrapolated lung cancer risk based on
multiple human occupational studies.
U.S. EPA (1998b)
Inhalation unit risk: 1.2 x 10"2
Linearly extrapolated lung cancer risk based on
Mancuso (1997,1975).
aSelected values from states known by U.S. EPA to have derived independent values; most states typically adopt
values from U.S. EPA.
bAgency does not derive a quantitative risk factor.
This document is a draft for review purposes only and does not constitute Agency policy.
8	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 4. Cr(VI) values for oral exposure from U.S. federal and state agencies3
and international bodies (in reverse chronological order)
Reference
Risk value or limit
Rationale13
Health Canada (2016)
Maximum acceptable concentration:
50 ng/L
Cancer precursor, mouse small
intestine hyperplasia
TCEQ (2016)
RfD: 3.1 x 10"3 mg/kg-day
Cancer precursor, mouse small
intestine hyperplasia
IPCS (2013)
Tolerable daily intake: 9 x 10"4 mg/kg-day
Mouse small intestine noncancer
effects
ATSDR (2012)
Chronic MRL: 9 x 10"4 mg/kg-day
Mouse small intestine noncancer
effects
Intermediate MRL: 5 x 10"3 mg/kg-day
Hematological effects (rat data at
22 days)
CalEPA (2011)
Cancer PHG: 0.02 ng/L
1 x 10"6 cancer risk using OSF of
0.5 (mg/kg-day)"1 (mouse small
intestine tumors)
Noncancer PHG: 2 ng/L
Liver noncancer effects (rats)
SWRCB (2014); CDPH (2013)
Proposed MCL: 10 ng/L
Note: invalidated fsee CA State Water
Board (2017) fact sheetl
Cancer risk fsee CalEPA (2011)1
NJ DEP (2009)
Soil remediation criterion: 1 ppm soil
concentration
1 x 10"6 cancer risk using OSF of 0.5
(mg/kg-day)"1 (mouse small intestine
tumors)
U.S. EPA (2008a, 2008b)
OSF: 0.791 (mg/kg-day)"1
Upper-bound cancer risk estimate
(mouse small intestine tumors)
Values based on science or rules published prior to 2008 National Toxicology Program study
FDA (2013)
Allowable level in bottled water: 0.1 mg/L
(or 100 ppb) total chromium
Not specified
U.S. EPA [Federal Register
(2010)1
MCL: 100 ppb (total chromium)
Allergic dermatitis0
WHO (2003)
50 ng/L
Provisional value (nonspecific)
RIVM (2001)
5 x 10"3 mg/kg-day
Provisional noncancer effects, based
on no-effect level frats; (MacKenzie et
al„ 1958)1
This document is a draft for review purposes only and does not constitute Agency policy.
9	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Reference
Risk value or limit
Rationale13
U.S. EPA (1998b)
RfD: 3 x 10"3 mg/kg-day
No effect level for noncancer effects
[rats; (MacKenzie et al., 1958)1
MCL = maximum contaminant level; MRL = minimal risk level; OSF = oral slope factor; PHG = public health goal.
aSelected values from states known by U.S. EPA to have derived independent values; most states typically adopt
values from U.S. EPA (based on un-speciated total chromium).
bAII values based on mouse data from NTP (2008), unless otherwise noted.
cBased on rule promulgated in 1991 (National Primary and Secondary Drinking Water Regulations, 56 FR 3526,
1-30-91 and 54 FR 22062, 5-22-89).
This document is a draft for review purposes only and does not constitute Agency policy.
10	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
3.ASSESSMENT APPROACH, SPECIFIC AIMS, AND
DRAFT POPULATIONS, EXPOSURES,
COMPARATORS, AND OUTCOMES (PECO)
CRITERIA
3.1. ASSESSMENT APPROACH
The overall objective of this assessment is to identify adverse health effects and
characterize exposure-response relationships for the effects of Cr(VI) to support the development
of toxicity values. This assessment uses systematic review methods to evaluate the epidemiological
and toxicological literature for Cr(VI); relevant mechanistic evidence is also considered. The
evaluations conducted in this assessment are consistent with relevant EPA guidance.1
The specific approach taken to the reassessment of the health effects of Cr(VI) was based on
input received during scoping a survey of the health effects of Cr(VI) previously identified by
government health agencies (including EPA) and international health organizations, as well as
consideration of the physicochemical properties of Cr(VI). As discussed in the preliminary
materials released in 2014 (U.S. EPA. 2014b. c), the IRIS assessment will include evaluations of the
evidence relevant to all cancer outcomes, and will evaluate noncancer effects for the following
potential target systems: respiratory, gastrointestinal, hepatic, hematological, immunological,
reproductive, and developmental. As discussed further below, for cancer and nasal irritation via
the inhalation route, the systematic review will focus on data that may improve the quantitative
dose-response analysis, conducted in EPA's 1998 IRIS assessment, for these outcomes.
3.1.1. Evaluation of the Potential Carcinogenicity of Inhaled Cr(VI)
EPA's 1998 IRIS assessment classified Cr(VI) as "Group A—known human carcinogen by the
inhalation route of exposure" based on evidence of a causal relationship between inhalation of
Cr(VI) and increased incidence of lung cancer in humans. The same conclusion has since been
reached by other federal and state health agencies and international organizations fTCEO. 2014:
IPCS. 2013: NIOSH. 2013b: IARC. 2012: CalEPA. 2011: NTP. 2011: OSHA. 20061. Therefore, as
discussed in the preliminary materials released in 2014 (U.S. EPA. 2014b. c), this assessment will
focus on the review of the evidence for lung cancer to identify studies not included in the 1998
assessment that might improve the quantitative dose-response analysis for human lung cancer.
'EPA 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.
11	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 Hexavalent Chromium IRIS Assessment
3.1.2.	Evaluation of the Effects of Inhaled Cr(VI) on the Nasal Cavity
In the 1998 assessment (U.S. EPA. 1998b). EPA concluded that a number of occupational
epidemiological studies demonstrated an association between inhalation of Cr(VI) and upper
respiratory irritation and atrophy. Based on EPA's 1998 evaluation of the literature and the
determination that the effects of Cr(VI) on the nasal cavity have been well established [e.g., OSHA
(20061. U.S. EPA f2014cl], EPA will not reevaluate the qualitative evidence for an association
between Cr(VI) exposure and nasal irritation/atrophy. Rather, the review of the evidence for nasal
effects will focus on identifying studies that might improve quantitative dose-response analysis for
this outcome. This decision to focus the systematic review on studies useful for an improved
dose-response analysis is an update from the preliminary materials released in 2014 (U.S. EPA.
2014b. cl.
For noncancer effects occurring in the respiratory tract beyond the nasal cavity
(bronchopulmonary), and for systemic effects, both hazard identification and dose-response will be
evaluated.
3.1.3.	Toxicokinetics of Cr(VI)
The absorption and metabolism of Cr(VI) are topics that have been thoroughly reviewed in
previous health agency documents (IPCS. 2013: NIOSH. 2013a: ATSDR. 2012: CalEPA. 2011: OSHA.
20061. Briefly, chromium exists in multiple oxidation states, but the hexavalent and trivalent states
are most prevalent biologically. Following oral or inhalation exposure (and prior to systemic
absorption), Cr(VI) can be reduced to Cr(III) within the GI tract or the respiratory tract,
respectively. If reduced to the trivalent state prior to uptake, chromium is poorly absorbed by cells
and is not toxic. However, chromium in the hexavalent state can be readily absorbed by cells lining
the GI or respiratory tract After systemic absorption, Cr(VI) will continue to reduce to Cr(III)
within cells and tissues in the body. Only total chromium (Cr[VI] + Cr[III]) can be accurately
measured in biological tissues and excreta. This has implications for how human epidemiological
studies are evaluated for exposure, and how absorption, distribution, metabolism, or excretion
(ADME) studies are screened and inventoried.
The route of exposure affects the local and systemic distribution of chromium because
Cr(VI) will pass through different fluids and tissues of varying reduction capacity depending on the
site of absorption. Orally ingested Cr(VI) is likely to be absorbed in the GI tract and liver (both of
which will reduce Cr[VI] to Cr[III]). Due to the first-pass effect, less Cr(VI) may be available for
absorption to systemic circulation and other tissues following oral ingestion. Inhaled Cr(VI) is
likely to be absorbed in the respiratory tract and distributed to systemic circulation as Cr(VI)
because less extracellular reduction may occur. Cr(VI) administered by injection (intravenous or
intraperitoneal) or intratracheal instillation bypasses mechanisms that reduce and dampen
systemic Cr(VI) absorption and distribution. As a result, the toxicological effects induced by Cr(VI)
at both portal-of-entry and systemic tissues differ by exposure route. Exposures to Cr(VI) via oral
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
27
28
29
30
31
32
33
34
35
36
37
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
and inhalation routes will be considered more toxicologically relevant than other routes of
exposure (e.g., dermal, injection, or intratracheal). Criteria for the screening of studies that include
consideration of route of exposure are described in Section 3.3.
Extrapolating Cr(VI) dose-response data from animals to humans is complex in light of
these toxicokinetic properties (IPCS. 2013: ATSDR. 20121. The reassessment will consider the
available Cr(VI) toxicokinetic models for the quantitative analysis of toxicity data. As a result,
physiologically based pharmacokinetic (PBPK) models will undergo study evaluation.
3.2. SPECIFIC AIMS
The aims of the assessment are to:
•	Identify epidemiological (i.e., human) and toxicological (i.e., experimental animal) literature
reporting effects of exposure to Cr(VI) as outlined in the PECO. The assessment will include
evaluations of the evidence relevant to all cancer outcomes and will evaluate noncancer
effects for the following potential target systems: respiratory, GI, hepatic, hematological,
immunological, reproductive, and developmental. The systematic review will focus on
identifying data from inhalation exposures that are useful for deriving quantitative
estimates for lung cancer and nasal effects rather than revisiting the qualitative
identification of hazard for these outcomes.
•	Evaluate mechanistic events associated with exposure to Cr(VI) that inform the
development or progression of the health effects identified in humans and animals. The
scope of these analyses will be determined by the complexity and confidence in the
evidence in humans and animals, likelihood to impact evidence synthesis conclusions for
human health, and the directness or relevance of the model systems for understanding
potential human health hazards. The primary focus will be on the analysis of mechanistic
evidence for cancer and noncancer effects of the GI tract following oral exposures to Cr(VI).
Because the hazard identification of lung cancer and nasal effects will not be revisited, the
mechanistic analyses for these health effects will focus on evidence that may affect the
dose-response assessment.
•	Conduct study evaluations (risk of bias and sensitivity) for individual epidemiological and
toxicological studies and PBPK models as defined by the scoping decisions described in
Section 3.1.
•	Extract data on relevant health outcomes from selected epidemiological and toxicological
studies based on the study evaluations.
•	Synthesize the evidence across studies, assessing similar health outcomes using a narrative
approach.
•	For each health outcome, express strength of evidence conclusions from across studies (or
subsets of studies) separately for studies in humans and animals. If studies informing
mechanisms were synthesized, then mechanistic evidence from either human or animal
studies will be integrated with the health effects evidence.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
•	For each health outcome, integrate strength of evidence conclusions across evidence
streams (human and animal) to conclude whether a substance is hazardous to humans.
Identify and discuss issues concerning potentially susceptible populations and life stages.
Biological support from mechanistic studies and nonmammalian model systems will be
considered based on the iterative prioritization approach outlined in the PECO.
•	Derive toxicity values (e.g., RfDs, RfCs, cancer risk estimates) as supported by the available
data. Apply toxicokinetic and dosimetry modeling to account for interspecies differences.
•	Characterize uncertainties and identify key data gaps and research needs such as
limitations of the evidence base, limitations of the systematic review, and consideration of
dose relevance and pharmacokinetic differences when extrapolating findings from higher
dose animal studies to lower levels of human exposure.
3.3. PECO CRITERIA
The PECO is used to identify the evidence that addresses the specific aims of the assessment
and to focus the literature screening, including the inclusion/exclusion criteria, in a systematic
review. The PECO criteria for Cr(VI) (see Table 5) are based on (1) nomination of the chemical for
assessment, (2) discussions with scientists in EPA program and regional offices to determine the
scope of the assessment that will best meet Agency needs, (3) preliminary review of the health
effects literature for Cr(VI) (primarily reviews and authoritative health assessment documents) to
identify the major health hazards associated with exposure to Cr(VI) and key areas of scientific
complexity, and (4) input received during public discussion of preliminary materials released to the
public in 2014.
In addition to the PECO criteria, studies containing supplemental material that are
potentially relevant to the specific aims of the assessment were tracked during the literature
screening process. Although these studies did not meet PECO criteria, they were not excluded from
further consideration. The categories used to track studies as "potentially relevant supplemental
material" during screening and to prioritize these studies for consideration in the assessment based
on likelihood to impact evidence synthesis conclusions for human health are described in
Section 4.3.
This document is a draft for review purposes only and does not constitute Agency policy.
14	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 5. Populations, exposures, comparators, and outcomes (PECO) criteria
PECO element
Evidence
Populations
Human: Anv population and life stage (occupational or general population, including children and
other potentially sensitive populations).
Animal: Nonhuman mammalian animal species (whole organism) of anv life stage (including
preconception, in utero, lactation, peripubertal, and adult stages).
Exposures
Human: Anv exposure to Cr(VI), including occupational exposures, via oral or inhalation routes.
Exposures by the inhalation and oral routes may be assessed based on administered dose or
concentration, biomonitoring data (e.g., urine, blood, or other specimens), environmental or
occupational-setting measures (e.g., air, water, dust levels), or job title or residence. Some
relevant forms of compounds containing Cr(VI) (18540-29-9) are listed below:
•	Chromic acid (H2Cr04 [7738-94-5] and H2Cr207 [13530-68-2])
•	Salts of the chromate (CrC>42") and dichromate (Cr2C>72") anions: Sodium chromate
(7775-11-3), sodium dichromate (10588-01-9), sodium dichromate dihydrate (7789-12-0),
potassium chromate (7789-00-6), potassium dichromate (7778-50-9)
•	Chromium(VI) trioxide (commonly referred to as chromium oxide [1333-82-0])
•	Calcium chromate (13765-19-0)
Animal: Anv exposure to Cr(VI) via oral or inhalation routes based on administered dose or
concentration. Cr(VI) may be administered orally via gavage or ad libitum in diet or drinking water.
Cr(VI) may be administered by inhalation via whole-body or nose-only systems.
Relevant forms of Cr(VI) are listed above. Animal studies involving exposures to mixtures will be
included only if they include exposure to Cr(VI) alone.
Comparators
Human: A comparison or referent population exposed to lower levels (or no exposure/exposure
below detection limits) of Cr(VI), or exposure to Cr(VI) for shorter periods of time.
Animal: A concurrent control group exposed to vehicle-onlv treatment or an untreated control.
Outcomes
All cancer outcomes are considered; noncancer health outcomes are considered for the following
potential target systems: respiratory, Gl, hepatic, hematological, immunological, reproductive, or
developmental effects. As discussed above, EPA anticipates that a systematic review for other
health effect categories (e.g., nephrotoxicity, neurotoxicity) will not be undertaken unless a
significant amount of new evidence is identified.
PBPK models
Studies describing PBPK models for Cr(VI) will be included.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
4.LITERATURE SEARCH AND SCREENING
STRATEGIES
4.1. LITERATURE SEARCH STRATEGIES
Literature search strategies were developed using key terms and words related to the PECO
criteria. Relevant subject headings and text-words were crafted into a search strategy that was
designed to maximize the sensitivity and specificity of the search results. The search strategy was
run, and the results were assessed to ensure that all previously identified relevant primary studies
were retrieved in the search. Because each database has its own search architecture, the resulting
search strategy was tailored to account for the unique search functionality of each database.
The following databases were searched:
•	PubMed (National Library of Medicine)
•	Web of Science (Thomson Reuters)
•	Toxline (National Library of Medicine)
Searches were not restricted by publication date, and no language restrictions were applied.
Web of Science results were limited using the research areas filter. All Web of Science research
areas identified in the search results were prioritized by a technical advisor as high priority (e.g.,
toxicology), low priority (e.g., chemistry), and not relevant (e.g., forestry). Literature searches were
conducted in bibliographic databases as described in Appendix A and uploaded to EPA's Health and
Environmental Research Online (HERO) database.2
Additional relevant literature not found through database searching was sought by:
•	Manually searching citations from review articles and studies considered to meet PECO
criteria after screening ("included" studies).
•	Searches of gray literature, including primary studies that are not indexed in databases of
peer-reviewed literature (e.g., technical reports from government agencies or scientific
research groups; unpublished laboratory studies conducted by industry; working papers
from research groups or committees; and white papers), or other nontypical searches. Gray
literature is typically identified by searching the EPA Chemistry Dashboard
2Health and Environmental Research Online: https: //hero.epa.gov/hero/.
This document is a draft for review purposes only and does not constitute Agency policy.
16	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 Hexavalent Chromium IRIS Assessment
(https://comptox.epa.gov/dashboard) during problem formulation, by engaging with
technical experts, and during solicitation of Agency, interagency, and public comment at
multiple steps in the IRIS process.
• "Backward" searches (to identify articles cited by included studies, reviews, or prior
assessments by other agencies).
The initial search was performed in January 2013, and literature search updates were
conducted in July 2013, February 2014, April 2015, April 2016, May 2017, December 2017, and is
current through May 2018. The literature search will be updated throughout draft development to
identify literature published during the course of the review. The last full literature search update
will be conducted less than one 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, which also reflect the literature screening decisions (see Section 4.3).
The IRIS Program takes extra steps to ensure identification of pertinent studies by
(1) encouraging the scientific community and the public to identify additional studies and ongoing
research; (2) searching for publicly available data submitted under the Toxic Substances Control
Act and the Federal Insecticide, Fungicide, and Rodenticide Act; and (3) considering late-breaking
studies that would impact the credibility of the conclusions, even during the review process.3
Studies identified after peer review begins will only be considered for inclusion if they meet the
PECO criteria and may fundamentally alter the assessment's conclusions.
4.2. NON-PEER-REVIEWED DATA
IRIS assessments rely mainly on publicly accessible, peer-reviewed studies. However, it is
possible that gray literature (i.e., studies that are not reported in the peer-reviewed literature)
directly relevant to the PECO may be identified during assessment development (e.g., good
laboratory practice [GLP] studies submitted to EPA, dissertations, etc.). In this case, if the data
substantially affect assessment decisions or conclusions (i.e., potential to impact the PECO
statement, hazard conclusions, or dose-response analysis), EPA can obtain external peer review if
the owners of the data are willing to have the study details and results made publicly accessible.
This independent, contractor-driven peer review would include an evaluation of the study, similar
to a peer review of a journal publication. The contractor would identify and select two to three
scientists knowledgeable in scientific disciplines relevant to the topic as potential peer reviewers.
Persons invited to serve as peer reviewers would be screened for conflict of interest prior to
confirming their service. 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
3IRIS "stopping rules": https://www.epa.gov/sites/production/files/2014-
06/documents/iris stoppingrules.pdf.
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 Hexavalent Chromium IRIS Assessment
1	to clarify issues or provide missing details. The study and its related information, if used in the IRIS
2	assessment, would become publicly available. In the assessment, EPA would acknowledge that the
3	document underwent external peer review managed by the EPA, and the names of the peer
4	reviewers would be identified. In certain cases, IRIS will conduct an assessment for utility and data
5	analysis based on having access to a description of study methods and raw data that have
6	undergone rigorous quality assurance/quality control review (e.g., ToxCast/Tox21 data, results of
7	NTP studies) but that have not yet undergone external peer review.
8	Unpublished (e.g., raw) data from personal author communication can supplement a
9	peer-reviewed study if the information is made publicly available (typically through documentation
10	in HERO).
4.3. SCREENING PROCESS
11	The PECO criteria were used to determine inclusion or exclusion of a reference as a primary
12	source of health effects data or a published PBPK model. In addition to the PECO criteria, the
13	exclusion criteria noted below were applied, while also tagging studies as appropriate to allow for
14	later retrieval, dependent on assessment needs:
15	• Studies that were previously determined not to be pertinent, as described in the 2014
16	Supplemental Materials (U.S. EPA. 2014b. c);
17	• Study materials that have not been peer reviewed, unless they are expected to have a
18	substantial impact on the assessment (as described in Section 4.2);
19	• Records that do not contain original data, such as other agency assessments, informative
20	scientific literature reviews, grant submissions (from the National Institutes of Health [NIH]
21	reporter database), editorials, or commentaries;
22	• Chromium compounds that did not meet PECO criteria (e.g., metal chromates; animal
23	studies of exposures to mixtures containing Cr[VI]);
24	• Ecology studies;
25	• Studies appearing as abstracts only (e.g., conference abstracts); and
26	• Non-English studies in which the titles and abstracts (when available) did not suggest direct
27	relevance to the PECO or specific aims.
28	In addition to the inclusion of studies that meet PECO criteria, studies containing
29	supplemental material that is potentially relevant to the specific aims were tracked during the
30	screening process (see Section 4.4.2). Although not considered to directly meet PECO criteria, these
31	studies were not strictly excluded unless otherwise specified. Unlike studies that meet PECO
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 Hexavalent Chromium IRIS Assessment
1	criteria, supplemental studies may not be subject to systematic review unless predefined questions
2	are identified that focus the mechanistic (or other) analysis are added to the specific aims and PECO
3	criteria. Studies that were categorized as "potentially relevant supplemental material" include the
4	following:
5	• Mechanistic studies: Studies reporting measurements related to a health outcome that
6	informs the biological or chemical events associated with phenotypic effects, in both
7	mammalian and nonmammalian model systems, including in vitro, in vivo (by various
8	routes of exposure), ex vivo, and in silico studies.
9	• ADME studies: Studies designed to capture information regarding absorption, distribution,
10	metabolism, and excretion, including toxicokinetic studies (e.g., studies describing
11	quantitative models or data for Cr[VI] reduction kinetics in biological media [e.g., gastric
12	juice, red blood cells, lung, and GI tract epithelial cells]). Such information may be helpful in
13	updating or revising the parameters used in existing PBPK models.
14	• Exposure characteristics: Exposure studies that include data unrelated to toxicological
15	endpoints, but which provide information on exposure sources or measurement properties
16	of the environmental agent (e.g., demonstrating a biomarker of exposure).
17	• Susceptible populations: Studies that identify potentially susceptible subgroups, such as
18	studies that focus on a specific demographic, life stage, or genotype. (These are categorized
19	under "Mechanistic studies.").
20	• Related to included studies: Versions of other studies (e.g., updated cohort analyses) that
21	meetPECOcriteria.
22	• Human case reports or case series: In most cases, case reports and case series will be
23	tracked as potentially relevant supplemental information.
24	• Routes of exposure not pertinent to PECO: Studies using dermal, injection, or intratracheal
25	administration.
26	• Acute duration exposures: Animal studies of acute or short-term (less than 28 days)
27	exposure duration.
4.3.1. Title- and Abstract-Level Screening
28	Following a pilot phase to calibrate screening guidance, two screeners independently
29	conducted a title and abstract screen of the search results to identify records that appeared to meet
30	the PECO criteria using a structured form in DRAGON fICF Consulting. 20181. For non-English
31	studies, if the title and abstract were written in English, the eligibility status of these studies was
32	assessed using the same approach. For citations with no abstract, articles were screened based on
33	title relevance and page numbers (articles two pages in length or less may be assumed to be
This document is a draft for review purposes only and does not constitute Agency policy.
19	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
conference reports, editorials, or letters). All screening conflicts were resolved by a technical
advisor.
Studies not meeting PECO criteria but identified as "potentially relevant supplemental
material" were categorized (i.e., tagged) during the title and abstract screening process (further
described in Section 4.4). Conflict resolution is 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 may be considered during draft
development).
To ensure all relevant references were identified in the initial screening, the excluded
materials were reviewed to identify misclassified studies meeting PECO criteria or potentially
relevant supplemental material that may have been missed during the database searches. A subset
of excluded studies was prioritized for a second round of screening using text analytics. Supervised
clustering and machine learning using ICF's Document Classification and Topic Extraction Resource
(DoCTER) was conducted to ensure that all mechanistic studies were identified. Supervised
clustering is a form of semi-supervised machine learning that uses seeds or known-to-be-relevant
studies. DoCTER includes multiple text analytic algorithms (K-means and non-negative matrix
factorization) that can be used to find studies with titles and abstracts that are similar to "seed
studies" previously identified as relevant (Varghese etal.. 20171. These algorithms create a
user-defined number of clusters based on keyword similarities in the title and abstract, and each
algorithm is broadly accepted in the text analytics scientific field. Machine learning uses similar
algorithms, but requires a robust training set to predict the likelihood that a given unclassified
study is relevant. For this effort, both supervised clustering and machine learning were used to
prioritize a set of studies to rescreen. Training data and seeds were derived from the 806 studies
classified as mechanistic in the first round of screening. Results were rescreened for relevance to
mechanistic endpoints. In addition to tagging studies as mechanistic, screeners were also directed
to tag any additional supporting studies or health effect studies that were identified using the text
analytics prioritization methods described here.
Following the efforts to identify misclassified mechanistic studies and the literature search
updates described above, ICF identified 1,288 on-topic mechanistic references for screening. These
references were further screened using title and abstract information by two independent EPA staff
members, followed by conflict resolution if screening results were different. Due to the large
number of studies, it was necessary to develop deprioritization criteria to begin to set aside studies
that are potentially less impactful to the assessment of mechanistic events. These studies were
tagged so that they may be accessed later in the mechanistic analysis if needed.
The following types of mechanistic studies were deprioritized for further screening:
• Studies that were misidentified as on-topic during the first round of screening (e.g., studies
that did not include Cr[VI] or other oxidation states of chromium)
This document is a draft for review purposes only and does not constitute Agency policy.
20	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 Hexavalent Chromium IRIS Assessment
•	References only containing an abstract (i.e., conference abstracts)
•	Book chapters and reviews
•	Untranslated foreign language articles
•	Studies that only report chromium detection methods
•	Studies in less common model systems (e.g., plants, marine mammals)
•	Studies that are only relevant to a health effect not being evaluated (e.g., nephrotoxicity)
In addition, many studies were identified that used Cr(VI) as a positive control for new
assay validation or that were co-exposures (e.g., to investigate the antioxidant properties of a new
compound). Most of these studies did not contain information useful for the mechanistic analysis of
Cr(VI) and were deprioritized. However, studies were retained for full-text review if there was any
indication that they might be useful for mechanistic understanding or might report mechanistically
relevant information regarding a health effect not reported in human or animal studies (e.g.,
neurotoxicity). Studies were categorized and tagged based on the above criteria using DistillerSR to
record why each was deprioritized. This allows the assessors to revisit certain study categories if
deemed important later in the assessment process.
The mechanistic references that were prioritized for further consideration were categorized
by endpointtype using DistillerSR. Prioritized endpoints included studies relevant to cancer or
effects on the GI, respiratory, reproductive, developmental, hepatic, immune, or hematological
systems. Mechanistic references were also categorized if relevant to one or more of the 10 key
characteristics of carcinogens (Smith etal.. 2016). (intracellular) ADME, and/or contained
pathology findings. References were also tagged with the following: study type (in vivo, ex vivo, in
vitro), presence of "omics" data, relevance to a certain species based on whole organism or cell
type, and reported data using an acellular system. These tags allowed further prioritization and
organization for the next phase of screening.
Mechanistic references may be processed through an additional round of title and
abstract-based categorization to further assist with prioritization (for example, in vivo studies may
be categorized by route of exposure). This will allow additional narrowing of the mechanistic
studies of highest interest before the full text review and quality evaluation steps.
4.3.2. Full-Text-Level Screening
Records that were not excluded based on the title and abstract advanced to full-text review.
Full-text copies of these potentially relevant records were retrieved, stored in the HERO database,
and independently assessed by two screeners to confirm eligibility according to the PECO criteria.
Screening conflicts were resolved by discussion between the primary screeners with consultation
by a third reviewer or technical advisor (as needed to resolve any remaining disagreements).
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Studies that advanced to full-text review were also tagged as "potentially relevant supplemental
material" as appropriate.
The results of this screening process have been posted on the project page for this
assessment in the HERO database
(https://hero.epa.gov/hero/index.cfm/project/page/project id/22331, and studies have been
"tagged" with appropriate category descriptors (e.g., included, "potentially relevant supplemental
material," excluded). Results have also been annotated and reported in a literature flow diagram
(see Figure 1).
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, which, if
identified, will be screened as outlined above for adherence to the PECO criteria.
This document is a draft for review purposes only and does not constitute Agency policy.
22	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment


Database Searches,
n = 19,051



PubMed

WOS

Toxline

TSCATS


n = 7,396
V

n - 6,258
--

rt - 5,342

n = 55


(see Appendix A, TablesA-1 and A-2 forsearch strategies)

Additional Search Strategies
(n = 108)
(See Section 4.3.1 for methods)
Combined Dataset after electronic duplicate removal
(n = 12,728)

Title/Abstract/Full Text Screen
(n = 12,728)
Excluded (n= 10,328)
Did not meet PECO criteria:
•	Reviews (n = 292)
•	Other agency assessments (n = 33)
•	Other Cr compounds not meeting PECO
(e.g., metal chromates) (n = 172)
•	Not peer reviewed (n = 324)
•	Ecology studies (rt =415)
•	Other not pertinent (n = 8,841)
•	2014 Preliminary Materials—not pertinent
(" = 41)
•	Grant submissions (rt = 147)
•	Unable to determine (non-English) (rt = 98)
Potentially relevant supplemental
material {n = 2,587)
Did not meet PECO criteria, but tagged as
supplemental studies:
•	Dermal studies (n = 687)
•	Acute/short-term studies (rt = 54)
•	Injection/intratracheal studies (rt = 103)
•	Human case reports (rt = 113)
•	Mechanistic studies (n = 1291)
•	Exposure studies (n = 185)
•	ADME studies (n = 149)
•	Related to included study (n = 23)
Studies Eligible for Study Evaluation (n = 137)
•	Human health effects studies {n = 58)
•	Animal health effect studies (n = 69)
•	PBPK models (n = 11)
Figure 1. Literature search flow diagram for Cr(VI).
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
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
4.3.3. Multiple Publications of the Same Data
When there are multiple publications using the same or overlapping data, all publications
on the research will be included, with one selected for use as the primary study; the others will be
considered as secondary publications with annotation to indicate their relationship to the primary
record during data extraction. For epidemiology studies, the primary publication will generally be
the one with the longest follow-up, the largest number of cases, or the most recent publication date.
For animal studies, the primary publication will typically be the one with the longest duration of
exposure, or that assessed the outcome(s) most informative to the PECO. For both epidemiology
and animal studies, EPA will include relevant data from all publications of the study, although if the
same outcome is reported in more than one report, the data will only be extracted once.
4.4. SUMMARY-LEVEL LITERATURE INVENTORIES
During title/abstract or full-text level screening, studies tagged based on PECO eligibility
were further categorized based on features such as evidence type (human, animal, mechanistic,
PBPK, etc.), health outcome(s), and/or endpoint measure(s) included in the study. Literature
inventories for studies meeting PECO criteria were created to develop summary-level, sortable lists
that include some basic study design information (e.g., study population, exposure information such
as doses administered or biomarkers analyzed, age/life stage4 of exposure, endpoints examined,
etc.). These literature inventories facilitate subsequent review of individual studies or sets of
studies by topic-specific experts.
4.4.1. Studies Meeting PECO Criteria
The preliminary materials released in 2014 (U.S. EPA. 2014b. c) presented evidence tables
for the human and animal studies determined to be eligible for study evaluation. Following the
2014 public meetings, these data tables were maintained in Microsoft Word format and were
revised to correct errors identified by public commenters, EPA staff, and contractors. During this
revision process, additional data were added to the tables (both from studies already contained in
the tables and studies found in subsequent literature searches or public submissions). The
summary-level information in these tables was used as an inventory to prioritize data migration to
the Health Assessment Workplace Collaborative (HAWC; see Section 8), initiate HAWC study
entries, and identify subject matter experts for performing study evaluations. Depending on study
confidence (see Section 6) and data type, data from the inventories were migrated to HAWC. Any
studies identified as meeting the PECO criteria since the start of HAWC migration will be entered
directly into HAWC (and will not be added to the Microsoft Word inventory tables).
4Age/life stage of chemical exposure will be considered according to EPA's Guidance on Selecting Age Groups
for Monitoring and Assessing Childhood Exposures to Environmental Contaminants and EPA's A Framework for
Assessing Health Risk of Environmental Exposures to Children.
This document is a draft for review purposes only and does not constitute Agency policy.
24	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
4.4.2. Potentially Relevant Supplemental Material
Inventories were also created for studies that were tagged as "potentially relevant
supplemental material" during screening, including mechanistic studies (e.g., in vitro or in silico
models), ADME studies, and studies on endpoints or routes of exposure that do not meet the
specific PECO criteria but that may still be relevant to the research question(s). Here, the objective
is to create an inventory of studies that can be tracked and further summarized as needed—for
example, by model system, key characteristic [e.g., of carcinogens (Smith etal.. 20161]. mechanistic
endpoint, or key event—to support analyses of critical mechanistic questions that arise at various
stages of the systematic review (see Section 9.2 for a description of the process for determining the
specific questions and pertinent mechanistic studies to be analyzed).
ADME and mechanistic data (and related information) can be critical to the next steps of
prioritizing or evaluating individual PECO-specific studies, and thus these studies were reviewed by
subject matter experts early in the assessment process. ADME and mechanistic inventories
released in 2014 (U.S. EPA. 2014c) were revised to correct errors identified by public commenters
and will continue to be updated with new studies during assessment development (Note: PBPK
models are typically considered to meet PECO criteria, while ADME and toxicokinetic-related
studies are most commonly tracked as potentially relevant supplemental material). Cr(VI) ADME
studies will continue to be sorted into the following categories: (1) animal and human in vivo (oral,
inhalation, intratracheal, intravenous, intraperitoneal, subcutaneous, and multi-route),
(2) quantitative in vitro/ex vivo (gastric and red blood cell), (3) mechanistic distribution/reduction
(multiple system types), and (4) human biomonitoring. Summary information, such as species,
tissues examined, level of time-course sampling, and Cr(VI) reducing capacities, will continue to be
extracted from these studies. Mechanistic studies have been sorted according to the screening
criteria outlined in Section 4.3 to facilitate the analysis of mechanistic events.
This document is a draft for review purposes only and does not constitute Agency policy.
25	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
5.REFINED EVALUATION PLAN
The refined evaluation plan describes refinements made to the set of studies that met PECO
criteria and are to be carried forward to study evaluation. The process also helps determine which
studies tagged as "potentially relevant supplemental material" may need to be considered in the
assessment Refinements were based on (1) input from public comments on the preliminary
materials released in 2014 (U.S. EPA. 2014b. c), (2) literature screening and creation of the
inventories of studies meeting PECO criteria and potentially relevant supplemental material by EPA
staff and contractors, and (3) review of the inventories by subject matter experts. The refined
evaluation plan also identifies the endpoints, grouped by outcomes, that will be the primary focus
of the outcome-specific evaluations. These specifications will aid in implementing the
endpoint-specific study evaluation criteria (see Section 6).
5.1. AIRBORNE CHARACTERIZATION AND CHEMICAL PROPERTIES
Studies that met PECO criteria include those that provide data on inhaled Cr(VI) in a variety
of physical and chemical forms. Airborne Cr(VI) can exist in different sizes and forms (e.g.,
particulates, dusts, aerosols, fumes, or mists) that affect respiratory tract deposition. Furthermore,
the studies that met PECO criteria include compounds containing Cr(VI) that have different
chemical properties. All forms of Cr(VI) meeting PECO criteria will be evaluated for hazard
identification, regardless of chemical properties or airborne characteristics. However, the evidence
synthesis will consider the possibility that some forms or mixtures (such as Cr[VI] in extremely
acidic or alkaline solutions) may have properties that alter the toxicity or introduce uncertainties.
In addition, the physical and chemical properties of airborne Cr(VI) will be taken into consideration
when evaluating the suitability of studies for dose-response analysis.
Nine studies involving occupational exposure to welding fume were identified in the set of
studies meeting PECO criteria. Cr(VI) exposure via welding fume may occur if chromium is a
component of the base materials being joined (e.g., stainless steel), is present as a surface coating,
or is a component of materials consumed during the welding process, such as metal filler rod.
Occupational exposures to Cr(VI) in welding fume are variable due to differences in welding types,
practices, and duration of welding tasks (Shaw Environmental. 20061. Further, welding fume
components vary by the type of welding and base materials (Shaw Environmental. 2006). Because
variability in occupational exposure makes exposure to Cr(VI) difficult to quantify, toxicity data for
welding fume will not be considered for dose-response analysis. However, exposures to Cr(VI) are
high among stainless steel welders relative to workers performing other types of welding due to the
high chromium content of stainless steel compared with other base metals or alloys [e.g., mild steel;
This document is a draft for review purposes only and does not constitute Agency policy.
26	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 Hexavalent Chromium IRIS Assessment
fNIOSH. 2013a: Shaw Environmental. 20061]. Therefore, studies comparing stainless steel welders
to a less exposed reference group may be evaluated for noncancer hazard identification.
5.2.	TOXICOKINETICS
Information on the toxicokinetics of Cr(VI) is provided elsewhere in this document (see
Sections 3.1 and 6.4). Of the PBPK models available that met PECO criteria, evaluations will be
limited to those accounting for Cr(VI) reduction in the stomach compartment and interspecies
differences in gastric pH and physiology. Models must also include parameterization for mice, rats,
and humans. This narrows the evaluation to models that may be suitable for the dose-response
assessment Furthermore, based on the issues related to toxicokinetics outlined in Sections 3.1 and
6.4 and discussions and comments from public meetings (U.S. EPA. 2014c. 20131. route-to-route
extrapolations will not be considered.
5.3.	TOXICOGENOMICS
Eighteen studies reporting gene expression data following Cr(VI) exposures were identified
during screening as "potentially relevant supplemental material." Nine of these studies were
conducted in animals and will be subject to study evaluation using the criteria described in
Section 6.3. In addition, for both in vitro and in vivo toxicogenomic studies, the conduct of the
expression data generation and reporting will be evaluated using publicly available criteria based
on standard practices in the field fBourdon-Lacombe etal.. 20151: specifically, the Minimum
Information About a Microarray Experiment (MIAME) (Brazma etal.. 20011 and the Systematic
Omics Analysis Review (SOAR) tool fMcConnell etal.. 20141.
The applicability of the available microarray data to making toxicological inferences will be
assessed indirectly based on (1) comparison between the dose-response relationships derived from
transcriptomics data and apical outcomes and (2) evaluation of biological plausibility, as well as
external and internal consistency of the results of the gene expression analysis. Where appropriate,
tools such as BMDExpress 2.20.0148 beta fSciome. 20181 will be used to examine dose-response
relationships for gene expression and to identify pathways enriched with genes that demonstrate
significant dose-response trends and to determine the points of departure.
To use toxicogenomic data to inform biological processes associated with the exposure to
Cr(VI), the expression data will be analyzed using several complementary approaches. Pathways
and upstream regulators relevant to the genes identified as differentially expressed between
Cr(VI)-exposed versus unexposed controls will be explored using Ingenuity Pathway Analysis
fOiagen. 20181. Gene sets enriched in Cr(VI)-exposed versus unexposed control animals will be
determined by Gene Set Enrichment Analysis [Broad Institute; (Subramanian etal.. 20051],
Similarity of gene expression changes induced by Cr(VI) to public expression data corresponding to
various human and animal diseases and exposures to xenobiotics will be examined. This will be
This document is a draft for review purposes only and does not constitute Agency policy.
27	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 Hexavalent Chromium IRIS Assessment
done to identify conditions associated with gene expression profiles like those resulting from
animal exposure to Cr(VI).
Similarities between gene expression profiles will be examined using Basespace Correlation
Engine (Illumina. 2018) and Signature Search Tool [Genevestigator; f Kupershmidt et al.. 2010: Hruz
etal.. 20081]. Available genomic biomarkers will also be used to detect specific events. For
example, the TGx-DDI biomarker for DNA damage classification flackson etal.. 20171 will be used
as an auxiliary tool to detect the presence of DNA damage expression signatures in the analyzed
expression data set using the NTP web service (note that limitations due to differences between
actual and recommended specimen type/exposure time/species will be considered).
5.4. OUTCOMES CONSIDERED IN THE CR(VI) ASSESSMENT
As previously stated in Section 3.2, the assessment will evaluate evidence for all cancer
outcomes, and will evaluate noncancer effects for the following potential target systems:
respiratory, GI, hepatic, hematological, immunological, reproductive, and developmental. The
systematic review will focus on identifying data from inhalation exposures that are useful for
deriving quantitative estimates for lung cancer and nasal effects rather than revisiting the
qualitative identification of hazard for these outcomes. Additional details on how studies were
screened and sorted are contained in Sections 4.3 and 4.4.
The endpoints that will be the primary focus of the outcome-specific evaluations—grouped
by health outcome—are identified in Tables 6 and 7, along with the number of studies that
examined these endpoints. Identification of these endpoints will guide the development of
endpoint-specific study evaluation criteria (discussed further in Section 6). Table 8 provides an
inventory of a selection of categories used when screening studies identified as "potentially
relevant supplemental materials." This table is not comprehensive but provides a high-level
indication of the relative density of publications in these reference topic areas. A graphical
representation of the information in Table 8 for mechanistic studies identified from the "potentially
relevant supplemental materials" is provided in EPA's version of Health Assessment Workspace
Collaborative (HAWC), a free and open source web-based software application
f https://hawcprd.epa.gOv/lit/assessment/100500006/references/visualization/l.5'
5HAWC: A Modular Web-Based Interface to Facilitate Development of Human Health Assessments of
Chemicals. littps: //liawcproiect.org/portal /.
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 Hexavalent Chromium IRIS Assessment
Table 6. Outcomes and associated endpoints to be considered for animal
study evaluation
Health outcome and endpoints
Number of references
Gastrointestinal tract (oral)
4
Epithelial effects of small intestine
4
Stomach ulcer
2
Tumors of the Gl tract
2
Respiratory tract (inhalation)
7
Nasal
2
General respiratory and pulmonary
5
Tumors of the lung
2
Hepatic (oral)
13
Clinical chemistry changes
10
Histopathological changes
11
Organ-weight changes
7
Hepatic (inhalation)
4
Clinical chemistry changes
3
Histopathological changes
3
Organ-weight changes
3
General (including gross changes, liver disease mortality)
1
Hematological (oral)
9
Clinical chemistry changes
9
Hematological (inhalation)
4
Clinical chemistry changes
4
Immune (oral)
5
Clinical chemistry and functional assays
3
Histopathological changes
2
Organ-weight changes
2
Immune (inhalation)
3
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 Hexavalent Chromium IRIS Assessment
Health outcome and endpoints
Number of references
Clinical chemistry and functional assays
2
Organ-weight changes
3
Reproductive/developmental (oral)
40
Male reproductive
14
Female reproductive
10
Developmental (in utero and postnatal)
20
Reproductive/developmental (inhalation)
3
Male reproductive
3
PBPK modeling (see Section 6.4)
8
Note: Number of references indicates studies examining the outcome and associated endpoints, not the number of
observed effects. Some studies are counted in multiple categories.
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 Hexavalent Chromium IRIS Assessment
Table 7. Outcomes and associated endpoints to be considered for human
study evaluation
Health outcome and endpoints
Number of references
Lung cancer (inhalation)
10
Other cancer (inhalation)
l
Cancer (oral route of exposure)
7
Cancer in offspring (inhalation)
2
Respiratory noncancer, lung
6
Respiratory noncancer, nasal
11
Asthma
5
Hepatic
8
Hematological
5
Immunological
8
Reproductive and developmental
13
PBPK modeling (see Section 6.4)
7
Note: Number of references indicates studies examining the outcome and associated endpoints, not the number
of observed effects. Some studies are counted in multiple categories.
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 Hexavalent Chromium IRIS Assessment
Table 8. Inventory of selected reference topics screened as "potentially
relevant supplemental material" to be considered in the assessment
Reference topic
Number of references
Animal3
Human
In vivo toxicokinetics
48
6
Oral
8
6
Inhalation
3
0
Other
38
0
In vitro/ex vivo toxicokinetics
8
16
Gastric systems
4
6
Red blood cells
4
10
Mechanistic ADME
30
13
Liver
15
3
Gastrointestinal
2
0
Lung
4
6
Red blood cells
1
4
Other
10
0
Biomonitoring and biomarkersb
N/A
18
Blood/plasma/red blood cells
N/A
9
Urine
N/A
13
Other
N/A
6
Epidemiology studies related to included studies
N/A
18
Mechanistic studies (total number of studies)


Cancer(843)
358
334
Electrophilicity (144)
88
42
Genotoxicity (413)
183
172
Altered DNA repair (78)
27
50
Epigenetic alterations (24)
3
14
Oxidative stress (255)
100
105
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 Hexavalent Chromium IRIS Assessment
Reference topic
Number of references
Animal3
Human
Chronic inflammation (24)
9
9
Immunosuppression (3)
4
2
Receptor-mediated effects (111)
66
104
Immortalization/transformation (38)
16
26
Altered cell proliferation, death, or nutrient supply (224)
58
99
Gastrointestinal (31)
21
15
Respiratory (112)
53
128
Hepatic (59)
85
19
Hematological (11)
8
13
Immune (24)
27
27
Reproductive or developmental (38)
33
4
N/A = not applicable.
aCount does not include nonmammalian animal models or acellular systems.
bCount does not include epidemiology studies reporting human biomarker data.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
6.STUDY EVALUATION (REPORTING, RISK OF BIAS,
AND SENSITIVITY) STRATEGY
The general approach for evaluating primary health effect studies meeting PECO criteria for
all study types is described in Section 6.1; the specifics of applying the approach for evaluating
epidemiology and animal toxicology studies are described separately in Sections 6.2 and 6.3,
respectively. Different approaches are used for evaluating PBPK models (see Section 6.4) and
mechanistic studies (see Sections 6.5 and 9.2).
6.1. STUDY EVALUATION OVERVIEW FOR HEALTH EFFECT STUDIES
Key concerns for the review of epidemiology and animal toxicology studies are potential
bias (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 towards the
null when an effect exists). 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 2.
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 are developed in consultation with topic-specific
technical experts, and existing guidance documents will be used when available, including EPA
guidance for carcinogenicity, neurotoxicity, reproductive toxicity, and developmental toxicity fU.S.
EPA. 2005a. 2002.1998a. 1996.1991). These 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 methodologic
concerns may emerge, and a second pass may 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.
34	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
(a) Study evaluation process (b)
Individual evaluation domains
Refined evaluation plan

Criteria development
Pilot testing/refine criteria
Evaluation by 2 reviewers
Conflict resolution
£
Final domain judgments
and overall study rating
Animal
Epidemiology
Selection and performance
¦	Allocation
¦	Observational bias/blinding
Confounding/variable control
Participant selection
Confounding
Selective reporting and attrition
Exposure methods sensitivity
•	Chemical administration and
characterization
¦	Exposure timing, frequency, and duration
Outcome measures and results display
•	Endpoint sensitivity and specificity
¦	Results presentation
Reporting quality
Selective reporting
Exposure measurement
Outcome ascertainment
Analysis
Other Sensitivity
Domain judgments
Judgment
Interpretation
0 Good
Appropriate study conduct relating to the domain and
minor deficiencies not expected to influence results.
Adequate
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.
Deficient
^ Critically
™ Deficient
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
Low
Uninformative
Possible deficiencies or concerns noted, but resulting
bias or lack of sensitivity is unlikely to be of a notable
degree.
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
Serious flaw(s) makes study results unusable for hazard
identification or dose response.
Figure 2. Overview of Integrated Risk Information System (IRIS) study
evaluation process.
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
This document is a draft for review purposes only and does not constitute Agency policy,
35	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
study examines multiple endpoints for the same outcome,6 evaluations may be performed at a more
granular level if appropriate, but these measures may still be grouped for evidence synthesis.
Authors may be queried to obtain missing critical information, particularly when there is
missing reporting quality information or data (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 to seek missing information is largely based on the
likelihood that such information would affect the overall confidence of the study. Outreach to study
authors will be documented and considered unsuccessful if researchers do not respond to an email
or phone request within one month of the attempt to contact.
For each outcome in a study,7 reviewers will reach a consensus judgment of good, adequate,
deficient, not reported, or critically deficient for each evaluation domain (see Sections 6.2 and 6.3 for
a description of evaluation domains for epidemiology and experimental animal studies). If a
consensus is not reached, a third reviewer will perform conflict resolution. It is important to stress
that these evaluations are performed in the context of the study's use for identifying individual
hazards. Study limitations specific to the usability of the study for dose-response analysis may be
important for later decisions but do not contribute to the study confidence classifications. 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 minor deficiencies that are noted would not be expected to
influence the study results.
•	Adequate indicates a judgment that there may be 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 prevent interpretation of the study findings.
•	Not reported indicates that the information necessary to evaluate the domain 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 of unreported items and severity of other limitations identified in
the study, it may or may not be worth reaching out to the study authors for this information
(see discussion below).
6"Outcome" will be used throughout these methods; the same methods also apply to an endpoint within a
larger outcome.
7"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 study population within a study.
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
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
•	Critically deficient reflects a judgment that the study conduct relating to the evaluation
domain question introduced a serious flaw that is interpreted to be the primary driver of
any observed effect(s) or makes the study uninterpretable. Studies with a determination of
critically deficient in an evaluation domain will not be used for hazard identification or
dose-response but may be used to highlight possible research gaps. Examples include:
° An inhalation study of Cr(VI) in which the only control group is intentionally or
unintentionally infected with a respiratory virus (confounding/variable control).
° An oral ingestion study of Cr(VI) in which the chemical compound is not stated, drinking
water or gavage administration is not specified, control group exposure and husbandry
not specified, and the oral doses are not provided or cannot be verified due to missing
information (exposure methods sensitivity, reporting quality).
° A reproductive study of Cr(VI) in which rodents were administered high doses (known
to induce severe toxicity and death), and the numbers of dams in the results are less
than the sample sizes stated in the methods, with no documentation of animal deaths
(reporting or attrition)
Once the evaluation domains have been rated, the identified strengths and limitations will
be considered as a whole to reach a study confidence classification of high, medium, or low
confidence, or uninformative for a 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,
med/um-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 med/um-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
med/um-confidence results during evidence synthesis and integration (see Section 10.1,
Tables 19 and 20), and are generally notused as the primary sources of information for
hazard identification or to derive toxicity values unless they are the only studies available.
Studies rated as low confidence only because of sensitivity concerns about bias towards the
This document is a draft for review purposes only and does not constitute Agency policy.
37	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 Hexavalent Chromium IRIS Assessment
null will be asterisked or otherwise noted because these studies may 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 hazard identification. 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 synthesis and
integration of evidence for hazard identification or dose-response but may be used to
highlight possible research gaps.
Study evaluation determinations reached by each reviewer and the consensus judgment
between reviewers will be recorded in the EPA's version of HAWC. Final study evaluations housed
in HAWC will be made available when the draft is publicly released. The study confidence
classifications and their rationales will be carried forward and considered as part of evidence
synthesis (see Section 9) to aid in the interpretation of results across studies.
6.2. EPIDEMIOLOGY STUDY EVALUATION
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 negatives, while study sensitivity is typically
concerned with identifying the latter.
The principles and framework used for evaluating epidemiology studies are based on the
Cochrane Risk of Bias in Nonrandomized Studies of Interventions [ROBINS-I; fSterne etal.. 20161]
but 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, etc.). Core and prompting questions are used
to collect information to guide the evaluation of each domain.
Core and prompting questions for each domain, as well as additional considerations that
apply to most outcomes, are presented in Table 9. 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 the 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 may be the focus of those criteria are listed in
Table 10.
Exposures to Cr(VI) by the inhalation and oral routes may be assessed based on
administered dose or concentration, biomonitoring data (e.g., urine, blood, or other specimens),
environmental or occupational-setting measures (e.g., air, water, dust levels), or job title or
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
residence. Air concentration measurements are preferred to biomarker measurements for the
assessment of human exposure by inhalation in epidemiology studies. Studies in which human
exposure is quantified by measurements of total chromium in urine, blood, plasma, or erythrocytes
will be considered for determination of hazard if conducted in workers with known occupational
exposure to Cr(VI). Air concentrations of Cr(VI) are correlated with measurements of total
chromium in these biological matrices (Kuo etal.. 1997: Miksche and Lewalter. 19971. However,
uncertainty in biomarker measurements arises from reduction of Cr(VI) to Cr(III) throughout the
body fNIOSH. 2013al. The rate at which Cr(VI) is reduced to Cr(III) following exposure varies by
individual, further contributing to uncertainty in biomarker measurements.
When available, existing outcome-specific standard protocols for research studies will be
consulted in developing outcome-specific criteria for evaluating epidemiology studies. For
example, guidelines published by the American Thoracic Society for collecting spirometry
measurements will inform evaluations of epidemiology studies of pulmonary function f Culver etal..
2017: Miller etal.. 2005: ATS. 1995.19871. Likewise, EPA will refer to World Health Organization
(WHO) protocols when evaluating epidemiologic studies of semen parameters to assess toxicity to
the male reproductive system fWHO. 2010.19991.
This document is a draft for review purposes only and does not constitute Agency policy.
39	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 9. 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
interassay 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.
40	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium 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 interassay 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.
41	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Domain and
core
question
Prompting questions
Follow-up
questions
Considerations that apply to most exposures and outcomes
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
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 raise 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
This document is a draft for review purposes only and does not constitute Agency policy.
42	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Domain and
core
question
Prompting questions
Follow-up
questions
Considerations that apply to most exposures and outcomes


differential
selection is likely?
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).
Confounding
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 Considering that potential confounders were rare among the study
population or were expected to be poorly correlated with exposure
of interest;
o Considering the most relevant functional forms of potential
confounders; and
o Examining 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
Is confounding
of the effect of
the exposure
likely?
This document is a draft for review purposes only and does not constitute Agency policy.
43	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Domain and
core
question
Prompting questions
Follow-up
questions
Considerations that apply to most exposures and outcomes



(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 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
results were most likely due to bias.
o Presenting a progression of model results with adjustments for
different potential confounders, if warranted.
Analysis
Does the
analysis
strategy and
presentation
convey the
necessary
•	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?
If there is a
concern about the
potential for bias,
what is the
predicted
direction or
distortion of the
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.
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 Hexavalent Chromium IRIS Assessment
Domain and
core
question
Prompting questions
Follow-up
questions
Considerations that apply to most exposures and outcomes
familiarity
with the data
and
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)?
bias on the effect
estimate (if there
is enough
information)?
•	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 issues—missing at random vs. differential).
•	Where applicable, for exposure, includes limit of detection (and
percentage below the limit of detection), 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.
•	Descriptive information about exposure levels not provided (where
applicable).
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 Hexavalent Chromium IRIS Assessment
Domain and
core
question
Prompting questions
Follow-up
questions
Considerations that apply to most exposures and outcomes



•	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
•	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.
Is there reason
to be
concerned
about
selective
reporting?
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 Hexavalent Chromium IRIS Assessment
Domain and
core
question
Prompting questions
Follow-up
questions
Considerations that apply to most exposures and outcomes
Sensitivitv
Is there a
concern that
sensitivity of
the study is
not adequate
to detect an
effect?
•	Is the exposure range adequate?
•	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
primary hypotheses in study.
•	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 effect.
•	No other concerns raised regarding study sensitivity.
Deficient
•	Concerns were raised about the issues described for good 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.
47	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 10. Information relevant to evaluation domains for epidemiology
studies
Domain
Types of information that may 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; 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 the endpoints and exposure measures
reported in the methods section, and are they relevant to the PECO? Are results presented for
the full sample as well as for specified subgroups? Were stratified analyses (effect modification)
motivated by a specific hypothesis?
6.3. EXPERIMENTAL ANIMAL STUDY EVALUATION
1	The evaluation of experimental animal studies applies principles similar to those described
2	above for evaluating epidemiology studies. The evaluation process focuses on assessing aspects of
3	the study design and conduct through three broad types of evaluations: reporting quality, risk of
4	bias, and study sensitivity. A set of domains with accompanying core questions falls under each
5	evaluation type and directs individual reviewers to evaluate specific study characteristics. For each
This document is a draft for review purposes only and does not constitute Agency policy.
48	DRAFT-DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
domain and core question pairing, basic considerations provide additional guidance on how a
reviewer might evaluate and judge a study for that domain.
Table 11 provides the standard domains and core questions along with some basic
considerations for guiding the evaluation. Some domain considerations will need to be tailored to
the chemical and endpoint/outcome, while others are generalizable across assessments
(e.g., considerations for reporting quality). Assessment teams work with subject matter experts to
develop the assessment-specific considerations. These specific considerations are determined
prior to performing study evaluation, although they may be refined as the study evaluation
proceeds (e.g., during pilot testing). Assessment-specific considerations are documented and made
publicly available with the assessment
Each domain receives a consensus judgment of good, adequate, deficient, not reported, or
critically deficient (as described in Section 6.1) accompanied by a rationale for the judgment. Once
all domains are rated, an overall confidence classification of high, medium, or low confidence or
uninformative is assigned (as described in Section 6.1). The rationale for the classification,
including a brief description of any identified strengths and/or limitations from the domains and
their potential impact on the overall confidence determination, should be documented clearly and
consistently. This rationale should, to the extent possible, reflect an interpretation of the potential
influence on the results (including the direction and/or magnitude of influence).
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 Hexavalent Chromium IRIS Assessment
Table 11. Questions to guide the development of criteria for each domain in experimental animal toxicology
studies
Evaluation
concern
Domain—core question
Prompting questions
General considerations
cr
bjo
c
t
o
Q.
ai
cc
Reporting quality
Does the study report
information for evaluating
the design and conduct of
the study for the
endpoint(s)/outcome(s) of
interest?
Notes: Reviewers should
reach out to authors to
obtain missing information
when studies are considered
key for hazard evaluation
and/or dose-response.
This domain is limited to
reporting. Other aspects of
the exposure methods,
experimental design, and
endpoint evaluation
methods are evaluated
using the domains related
to risk of bias and study
sensitivity.
Does the study report the following?
•	Critical information necessary to
perform study evaluation:
o Species, test article name, levels
and duration of exposure, route
(e.g., oral; inhalation), qualitative
or quantitative results for at least
one endpoint of interest
•	Important information for
evaluating the study methods:
o Test animal: strain, sex, source,
and general husbandry
procedures
o Exposure methods: source,
purity, method of administration
o Experimental design: frequency
of exposure, animal age and life
stage during exposure and at
endpoint/outcome evaluation
o Endpoint evaluation methods:
assays or procedures used to
measure the
endpoints/outcomes of interest
These considerations typically do not need to be refined by
assessment teams, although in some instances the important
information may be refined depending on the endpoints/outcomes of
interest or the chemical under investigation.
A judgment and rationale for this domain should be given for the
study. Typically, these will not change regardless of the
endpoints/outcomes investigated by the study. In the rationale,
reviewers should indicate whether the study adhered to GLP, OECD,
or other testing guidelines.
•	Good: All critical and important information is reported or
inferable for the endpoints/outcomes of interest.
•	Adequate: All critical information is reported but some
important information is missing. However, the missing
information is not expected to significantly impact the study
evaluation.
•	Deficient: All critical information is reported but important
information is missing that is expected to significantly reduce
the ability to evaluate the study.
•	Critically deficient: Study report is missing any pieces of critical
information. Studies that are critically deficient for reporting are
uninformative for the overall rating and not considered further
for evidence synthesis and integration.
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 Hexavalent Chromium IRIS Assessment
Evaluation



concern
Domain—core question
Prompting questions
General considerations


Allocation
For each study:
These considerations typically do not need to be refined by


Were animals assigned to
• Did each animal or litter have an
assessment teams.


experimental groups using a
equal chance of being assigned to
A judgment and rationale for this domain should be given for each


method that minimizes
any experimental group (i.e., random
cohort or experiment in the study.

U)
.2
selection bias?
allocation3)?
• Good: Experimental groups were randomized and any specific


• Is the allocation method described?
randomization procedure was described or inferable

ai

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Evaluation
concern
Domain—core question
Prompting questions
General considerations
¦a
ai
o
u
.5
-a
ai
.5
ai
u
c
ro
E
a>
a.
¦a
c
ro
ai
1/1
Observational bias/blinding
Did the study implement
measures to reduce
observational bias?
For each endpoint/outcome or grouping
of endpoints/outcomes in a study:
•	Does the study report blinding or
other methods/procedures for
reducing observational bias?
•	If not, did the study use a design or
approach for which such procedures
can be inferred?
•	What is the expected impact of
failure to implement (or report
implementation) of these
methods/procedures on results?
These considerations typically do not need to be refined by the
assessment teams.
Note that it can be useful for teams to identify highly subjective
measures of endpoints/outcomes where observational bias may
strongly influence results prior to performing evaluations.
A judgment and rationale for this domain should be given for each
endpoint/outcome or group of endpoints/outcomes investigated in
the study.
•	Good: Measures to reduce observational bias were described
(e.g., blinding to conceal treatment groups during endpoint
evaluation; consensus-based evaluations of histopathology
lesionsb).
•	Adequate: Methods for reducing observational bias
(e.g., blinding) can be inferred or were reported but described
incompletely.
•	Not reported: Measures to reduce observational bias were not
described.
o (Interpreted as adequate) The potential concern for bias was
mitigated based on use of automated/computer driven
systems, standard laboratory kits, relatively simple, objective
measures (e.g., body or tissue weight), or screening-level
evaluations of histopathology.
o (Interpreted as deficient) The potential impact on the results
is major (e.g., outcome measures are highly subjective).
•	Critically deficient: Strong evidence for observational bias that
impacted the results.
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 Hexavalent Chromium IRIS Assessment
Evaluation
concern
Domain—core question
Prompting questions
General considerations
¦a
a>
o
u
.5
o
u
_a;
.q
.5
ro
>
o
u
Confounding
Are variables with the
potential to confound or
modify results controlled
for and consistent across all
experimental groups?
For each study:
•	Are there differences across the
treatment groups (e.g., co-
exposures, vehicle, diet, palatability,
husbandry, health status, etc.) that
could bias the results?
•	If differences are identified, to what
extent are they expected to impact
the results?
These considerations may need to be refined by assessment teams, as
the specific variables of concern can vary by experiment or chemical.
A judgment and rationale for this domain should be given for each
cohort or experiment in the study, noting when the potential for
confounding is restricted to specific endpoints/outcomes.
•	Good: Outside of the exposure of interest, variables that are
likely to confound or modify results appear to be controlled for
and consistent across experimental groups.
•	Adequate: Some concern that variables that were likely to
confound or modify results were uncontrolled or inconsistent
across groups but are expected to have a minimal impact on the
results.
•	Deficient: Notable concern that potentially confounding
variables were uncontrolled or inconsistent across groups and
are expected to substantially impact the results.
•	Critically deficient: Confounding variables were presumed to be
uncontrolled or inconsistent across groups and are expected to
be a primary driver of the results.
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 Hexavalent Chromium IRIS Assessment
Evaluation
concern
Domain—core question
Prompting questions
General considerations
-a
ai
.5
.5
¦a
c
ro
no
c
t
o
a.
ai
ai
'+¦»

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Evaluation
concern
Domain—core question
Prompting questions
General considerations
>
'>
a>
to
>
">
¦a
o
.c
a!
E
ai
o
a.
x
Chemical administration
and characterization
Did the study adequately
characterize exposure to
the chemical of interest and
the exposure administration
methods?
Note: Consideration of the
appropriateness of the
route of exposure is not
evaluated at the individual
study level. Relevance and
utility of the routes of
exposure are considered in
the PECO criteria for study
inclusion and during
evidence synthesis.
For each study:
•	Does the study report the source and
purity and/or composition
(e.g., identity and percent
distribution of different isomers) of
the chemical? If not, can the purity
and/or composition be obtained
from the supplier (e.g., as reported
on the website)?
•	Was independent analytical
verification of the test article purity
and composition performed?
•	Did the authors take steps to ensure
the reported exposure levels were
accurate?
•	Are there concerns about the
methods used to administer the
chemical (e.g., inhalation chamber
type, gavage volume, etc.)?
For inhalation studies:
•	Were target concentrations
confirmed using reliable analytical
measurements in chamber air?
It is essential that these considerations are considered, and
potentially refined, by assessment teams because the specific
variables of concern can vary by chemical (e.g., stability may be an
issue for one chemical but not another).
A judgment and rationale for this domain should be given for each
cohort or experiment in the study.
•	Good: Chemical administration and characterization is complete
(i.e., source, purity, and analytical verification of the test article
are provided). There are no concerns about the composition,
stability, or purity of the administered chemical, or the specific
methods of administration. For inhalation studies, chemical
concentrations in the exposure chambers are verified using
reliable analytical methods.
•	Adequate: Some uncertainties in the chemical administration
and characterization are identified but these are expected to
have minimal impact on interpretation of the results
(e.g., source and vendor-reported purity are presented, but not
independently verified; purity of the test article is suboptimal
but not concerning; for inhalation studies, actual exposure
concentrations are missing or verified with less reliable
methods).
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 Hexavalent Chromium IRIS Assessment
Evaluation



concern
Domain—core question
Prompting questions
General considerations


Chemical administration
For oral studies:
• Deficient: Uncertainties in the exposure characterization are


and characterization
• If necessary based on consideration
identified and expected to substantially impact the results


(continued)
of chemical specific-knowledge
(e.g., instability in solution; volatility)
and/or exposure design (e.g., the
frequency and duration of exposure),
were chemical concentrations in the
dosing solutions or diet analytically
confirmed?
(e.g., source of the test article is not reported; levels of
impurities are substantial or concerning; deficient administration
methods, such as use of static inhalation chambers or a gavage
volume considered too large for the species and/or life stage at
exposure).
• Critically deficient: Uncertainties in the exposure
characterization are identified, and there is reasonable certainty

¦a


that the results are largely attributable to factors other than

ai
3


exposure to the chemical of interest (e.g., identified impurities

£
'+¦»
c


are expected to be a primary driver of the results).
T3
a>
U

Exposure timing,
For each endpoint/outcome or grouping
Considerations for this domain are highly variable depending on the
3
C
+¦»
frequency, and duration
of endpoints/outcomes in a study:
endpoint(s)/outcome(s) of interest and must be refined by
£
o
l/l
r—
Was the timing, frequency,
• Does the exposure period include the
assessment teams.
u
>
ai
i/i
and duration of exposure
critical window of sensitivity?
A judgment and rationale for this domain should be given for each
,+¦»
1/1
c
i/i
T3
O
.c
ai
sensitive for the
endpoint(s)/outcome(s) of
interest?
• Was the duration and frequency of
exposure sensitive for detecting the
endpoint/outcome or group of endpoints/outcomes investigated in
the study.
ai
l/l
E
ai
*_
3
l/l
o
a.
X
LU
endpoint of interest?
•	Good: The duration and frequency of the exposure was sensitive,
and the exposure included the critical window of sensitivity (if
known).
•	Adequate: The duration and frequency of the exposure was
sensitive, and the exposure covered most of the critical window
of sensitivity (if known).
•	Deficient: The duration and/or frequency of the exposure is not
sensitive and did not include the majority of the critical window
of sensitivity (if known). These limitations are expected to bias
the results towards the null.
•	Critically deficient: The exposure design was not sensitive and is
expected to strongly bias the results towards the null. The
rationale should indicate the specific concern(s).
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 Hexavalent Chromium IRIS Assessment
Evaluation
concern
Domain—core question
Prompting questions
General considerations
-a
a>
>
,+j
>
ai
to
>
a.
ai
*_
T3
Endpoint sensitivity and
specificity
Are the procedures
sensitive and specific for
evaluating the
endpoint(s)/outcome(s) of
interest?
Note: Sample size alone is
not a reason to conclude an
individual study is critically
deficient.
Considerations related to
adjustments/ corrections to
endpoint measurements
(e.g., organ weight
corrected for body weight)
are addressed under results
presentation.
For each endpoint/outcome or grouping
of endpoints/outcomes in a study:
•	Are there concerns regarding the
sensitivity, specificity, and/or validity
of the protocols?
•	Are there serious concerns regarding
the sample size?
•	Are there concerns regarding the
timing of the endpoint assessment?
Considerations for this domain are highly variable depending on the
endpoint(s)/outcome(s) of interest and must be refined by
assessment teams.
A judgment and rationale for this domain should be given for each
endpoint/outcome or group of endpoints/outcomes investigated in
the study.
Examples of potential concerns include:
•	Selection of protocols that are insensitive or nonspecific for the
endpoint of interest.
•	Evaluations did not include all treatment groups (e.g., only
control and high dose).
•	Use of unreliable methods to assess the outcome.
•	Assessment of endpoints at inappropriate or insensitive ages, or
without addressing known endpoint variation (e.g., due to
circadian rhythms, estrous cyclicity, etc.).
•	Decreased specificity or sensitivity of the response due to the
timing of endpoint evaluation, as compared to exposure
(e.g., short-acting depressant or irritant effects of chemicals;
insensitivity due to prolonged period of nonexposure prior to
testing).
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 Hexavalent Chromium IRIS Assessment
Evaluation



concern
Domain—core question
Prompting questions
General considerations


Results presentation
For each endpoint/outcome or grouping
Considerations for this domain are highly variable depending on the

"D
Are the results presented in
of endpoints/outcomes in a study:
outcomes of interest and must be refined by assessment teams.

0)
3
a way that makes the data
• Does the level of detail allow for an
A judgment and rationale for this domain should be given for each

£
'+-»
£
usable and transparent?
informed interpretation of the
endpoint/outcome or group of endpoints/outcomes investigated in

o
u

results?
the study.

>
re

• Are the data analyzed, compared, or
Examples of potential concerns include:
0)
3
Q.
W

presented in a way that is
• Nonpreferred presentation (e.g., developmental toxicity data
C
'+¦»
U)
4-»

inappropriate or misleading?
averaged across pups in a treatment group, when litter
£
o
3
V)


responses are more appropriate; presentation of absolute
>
0)
"C


organ-weight data when relative weights are more appropriate).

c
(0


• Failing to present quantitative results either in tables or figures.
*l/>
c

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Evaluation
concern
Domain—core question
Prompting questions
General considerations
a>
¦c
a>
>
O
Overall confidence
Considering the identified
strengths and limitations,
what is the overall
confidence rating for the
endpoint(s)/outcome(s) of
interest?
Note: Reviewers should
mark studies that are rated
lower than high confidence
only due to low sensitivity
(i.e., bias towards the null)
for additional consideration
during evidence synthesis.
If the study is otherwise well
conducted and an effect is
observed, the confidence
may be increased.
For each endpoint/outcome or grouping
of endpoints/outcomes in a study:
•	Were concerns (i.e., limitations or
uncertainties) related to the
reporting quality, risk of bias, or
sensitivity identified?
•	If yes, what is their expected impact
on the overall interpretation of the
reliability and validity of the study
results, including (when possible)
interpretations of impacts on the
magnitude or direction of the
reported effects?
The overall confidence rating considers the likely impact of the noted
concerns (i.e., limitations or uncertainties) in reporting, bias, and
sensitivity on the results.
A confidence rating and rationale should be given for each
endpoint/outcome or group of endpoints/outcomes investigated in
the study. Confidence ratings are described above (see Section 6.1).
OECD = Organisation for Economic Co-operation and Development.
aSeveral studies have characterized the relevance of randomization, allocation concealment, and blind outcome assessment in experimental studies (Hirst et al.,
2014; Krauth etal.. 2013; Macleod. 2013; Higgins and Green. 2011; U.S. EPA. 2002).
bFor nontargeted or screening-level histopathology outcomes often used in guideline studies, blinding during the initial evaluation of tissues is generally not
recommended as masked evaluation can make "the task of separating treatment-related changes from normal variation more difficult" and "there is concern
that masked review during the initial evaluation may result in missing subtle lesions." Generally, blinded evaluations are recommended for targeted secondary
review of specific tissues or in instances when a predefined set of outcomes is known or predicted to occur (Crissman et al., 2004).
This document is a draft for review purposes only and does not constitute Agency policy.
59	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 Hexavalent Chromium IRIS Assessment
6.4. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL
DESCRIPTIVE SUMMARY AND EVALUATION
PBPK (or classical pharmacokinetic [PK]) models should be used in an assessment when an
applicable one exists and no equal or better alternative for dosimetric extrapolation is available.
Any models used should represent current scientific knowledge and accurately translate the
science into computational code in a reproducible, transparent manner. For a specific target
organ/tissue, it may be possible to employ or adapt an existing PBPK model or develop a new PBPK
model or an alternate quantitative approach. Data for PBPK models may come from studies across
various species and may be in vitro or in vivo in design. Because Cr(VI) can be reduced to Cr(III)
extracellularly by biological fluids (e.g., gastric juices) of humans and rodents fDe Flora et al.. 19971.
ex vivo studies and models are also available. The relationship between ex vivo and whole-body
toxicokinetic models of Cr(VI) for the oral route of exposure is presented below in Figure 3.
Whole-body model
Ex vivo
reduction
model
<-
Gl:
A
<-
z\

Stomach Duodenum Jejunum Ileum
Cr(VI)



->






->
Gastrointestinal tract model
Figure 3. Relationship between ex vivo reduction models, in vivo gastric
models, and whole-body physiologically based pharmacokinetic (PBPK)
models.
In the trivalent state, chromium is poorly absorbed by cells and has not been shown to
induce the same effects as Cr(VI) fCollins etal.. 20101. Thus, extracellular reduction is a pathway
for detoxification because it decreases the systemic uptake and distribution of Cr(VI) and the
exposure of epithelial cells to Cr(VI). Following oral ingestion, most extracellular reduction and
This document is a draft for review purposes only and does not constitute Agency policy.
60	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 Hexavalent Chromium IRIS Assessment
detoxification will occur in the stomach prior to systemic absorption due to the acidity of gastric
juice, and the length of time ingested water and food are stored in the stomach. However, this
mechanism is less important following inhalation exposure, because the thin layer of respiratory
tract lining fluid is less acidic and less effective at reducing Cr(VI) (Krawic etal.. 2017: Ngetal..
20041. Deposition in the lung is not uniform, and particulates may locally accumulate at high
quantities in susceptible areas such as airway bifurcation sites (Balashazv etal.. 20031. This is
supported by studies showing high chromium deposition at these sites in the lungs of chromate
workers, and a correlation between lung chromium burden and lung cancer fKondo etal.. 2003:
Ishikawa etal.. 1994a. b).
Because extracellular gastric reduction kinetics are expected to significantly impact
dosimetry, the scope of the PBPK model evaluations for this assessment will be limited to models
accounting for Cr(VI) reduction in the stomach compartment and interspecies differences in gastric
pH and physiology (mice, rats, and humans). For the inhalation route of exposure, the regional
deposited dose ratio (RDDR) for the respiratory tract region of interest, estimated by airway
particle deposition modeling, will be used to account for species differences fU.S. EPA. 19941.
Route-to-route extrapolation will not be considered.
6.4.1. Pharmacokinetic (PK)/Physiologically Based Pharmacokinetic (PBPK) Model
Descriptive Summary
PBPK models were identified from the literature search, screening, and survey processes
(see Table 12). The two models listed in the bottom two rows of Table 12 [referenced by Schlosser
and Sasso f20141. Sasso and Schlosser f20151. Kirmanetal. f20171. and Kirman et al. f20161] will
be evaluated for this assessment because they are the only models incorporating the effects of
gastric pH and physiology on Cr(VI) toxicokinetics of mice, rats, and humans.
Parameters and codes from the earlier models may still undergo limited evaluations due to
the shared lineage in derivation. Scientific or technical errors in earlier models may propagate to
the later versions. For example, Kirman etal. f20171 and Kirman et al. f 20161 supersedes Kirman
etal. (20121 and Kirman etal. (20131. and both sets of models use many of the same data sets,
codes, and parameters as the O'Flaherty models. The Sasso and Schlosser f20151 model uses codes
and parameters from the Kirmanetal. (20121 and Kirmanetal. (20131 models. PBPK parameters
that originated from the O'Flaherty models may need to be evaluated if they are used in the later
Kirman etal. (20171. Kirman etal. (20161. and Sasso and Schlosser (20151 models.
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 Hexavalent Chromium IRIS Assessment
Table 12. Physiologically based pharmacokinetic models for Cr(VI)
Reference
Species
Notes
O'Flahertv (1996)
O'Flahertv (1993)
O'Flahertv et al. (2001)
O'Flahertv and Radike
(1991)
Rat
Compartments include kidney, liver, bone, Gl tract, two lung pools (for
inhalation only), plasma, red blood cells, and lumped compartments for
remaining tissues (rapidly and slowly perfused). A single lumped
compartment represents the Gl tract, and reduction kinetics do not include
pH-reduction relationships. This model is not readily extendable to the
mouse.
O'Flahertv et al. (2001)
Human
Calibrated to data from exposure via intravenous injection, gavage,
inhalation (intratracheal), and drinking water (all data are from studies
dated 1985 and earlier). Background Cr(lll) exposure is simulated in the
model and contributes to predicted total chromium concentrations.
Kirman et al. (2012)
Rat,
mouse
Compartments include kidney, liver, bone, Gl tract, plasma, red blood cells,
and a lumped compartment for remaining tissues. A multicompartment
model represents the Gl tract (oral cavity, stomach, duodenum, jejunum,
ileum, large intestine), with reduction kinetics based on the model by
Proctor et al. (2012).
Kirman et al. (2013)
Human
Incorporates toxicokinetic data from experiments designed by the study
authors and data from other studies. Only data for drinking water and
dietary routes of exposure are incorporated. Total concentrations in
control groups are subtracted from exposure groups to account for
background Cr(lll) exposure.
Schlosser and Sasso
(2014); Sasso and
Schlosser (2015)
Rat,
mouse,
human
Simulates Cr(VI) reduction kinetics and transit in the stomach.
Incorporates toxicokinetic model of the stomach lumen bv Kirman et al.
(2012) and Kirman et al. (2013), but with a revised model for Cr(VI)
reduction based on reanalysis of ex vivo data to improve model/data fit.
Kirman et al. (2017) and
Kirman et al. (2016)
Rat,
mouse,
human
Same structure as Kirman et al. (2012) and Kirman et al. (2013), but
incorporates a revised model for Cr(VI) reduction based on additional
human gastric juice data. This model supersedes earlier models by the
same investigators.
6.4.2. Pharmacokinetic (PK)/Physiologically Based Pharmacokinetic (PBPK) Model
Evaluation
1	EPA will undertake model evaluation in accordance with criteria outlined by U.S. EPA
2	(2018b). Judgments on the suitability of a model are separated into two categories: scientific and
3	technical (see Table 13). The scientific criteria focus on whether the biology, chemistry, and other
4	information available for chemical modes of action (MOAs) are justified (i.e., preferably with
5	citations to support use) and represented by the model structure and equations. The scientific
6	criteria are judged based on information presented in the publication or report that describes the
7	model and does not require evaluation of the computer code. Preliminary technical criteria include
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
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. Further details of the initial and in-depth evaluation criteria
can be found in the Umbrella Quality Assurance Project Plan (QAPP) for PBPKModels (U.S. EPA.
2018b).
6.5. MECHANISTIC STUDY EVALUATION
Sections 9 and 10 outline an approach for considering information from mechanistic studies
(including in vitro, in vivo, ex vivo, and in silico studies) where the specific analytical approach is
targeted to the assessment needs depending on the extent and nature of the human and animal
evidence. In this way, the mechanistic synthesis might range from a high-level summary of
potential mechanisms of action to specific, focused questions needed to fill data gaps identified
from the human and animal syntheses and integration (e.g., shape of the dose-response curve,
applicability of the animal evidence to humans, identifying susceptible populations). Individual
study-level evaluation of mechanistic endpoints will typically be pursued only when the
interpretation of studies is likely to significantly affect hazard conclusions or assumptions about
dose-response analysis, and the issues that need resolution have not been sufficiently addressed in
previous assessments or reviews published in peer-reviewed journals. Toxicogenomic studies will
be evaluated using the criteria identified in the refined evaluation plan (see Section 5). If other
mechanistic endpoints require study-level evaluation using endpoint-specific criteria that have not
been predefined, these criteria will be described in the updated protocol released with the draft
assessment.
This document is a draft for review purposes only and does not constitute Agency policy.
63	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 13. Criteria for evaluating physiologically based pharmacokinetic
(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.
•	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.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Category
Specific criteria
Initial
technical
Well-documented model code is readily available to EPA and the public.
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 analyses 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.
BW3/4 = body weight raised to the % power.
Assessing potential bias in in vitro studies is an active area of method development in the
field of systematic review. Historically, most tools used to assess these studies have focused on
reporting quality; tools to assess risk of bias (internal validity) of mechanistic evidence are not well
established (NASEM. 2018: NTP. 20151. although current trends are to expand the assessment to
include methodological quality with consideration of potential bias (IRIS. 20151. One of the more
recently developed tools that has undergone user testing and refinement is the Science in Risk
Assessment and Policy fSciRAPl approach for the evaluation of reliability for in vitro studies
fBeronius et al.. 2018: Molander etal.. 2015: Beronius etal.. 2014: Agerstrand et al.. 2011: U.S. EPA.
20021. The IRIS Program is in the pilot phase of testing approaches for arriving at study level
judgments for in vitro studies. Currently, two methods for evaluating in vitro mechanistic studies
are being considered for use in IRIS assessments: (1) a tool used for conducting assessments under
the Toxic Substances Control Act (TSCA), which uses a numerical scoring approach to rate studies
fU.S. EPA. 2018al and (2) the SciRAP tool fBeronius et al.. 20181. which separately presents domain
judgments for reporting quality, methodological quality, and relevance. A comparison of study level
judgments based on use of both tools should assist in refining an approach for routine use in IRIS
assessments. The IRIS Program is aware of other tools being developed (NASEM. 20181 and will
monitor developments through its engagements with the systematic review community. The
tool(s) and/or criteria used for testing specific questions that arise during the evaluation of
mechanistic events in the chromium assessment will be described in the updated protocol released
with the draft assessment
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
7.0RGANIZING THE HAZARD REVIEW
The organization and scope of the hazard evaluation is determined by the available
evidence for the chemical regarding routes of exposure, metabolism and distribution, outcomes
evaluated, and number of studies pertaining to each outcome, as well as the results of the
evaluation of sources of bias and sensitivity. The hazard evaluations will be organized around
organ systems (e.g., respiratory, hepatic system) informed by one or multiple related outcomes, and
a decision will be made as to what level (e.g., organ system or subsets of outcomes within an organ
system) to organize the synthesis.
Table 14 lists some questions that may be asked of the evidence to assist with this decision.
These questions extend from considerations and decisions made during development of the refined
evaluation plan to include review of the concerns raised during individual study evaluations, as well
as the direction and magnitude of the study-specific results. Resolution of these questions will then
inform critical decisions about the organization of the hazard evaluation and what studies may be
useful in dose-response analyses.
This document is a draft for review purposes only and does not constitute Agency policy.
66	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 14. Querying the evidence to organize syntheses for human and animal
evidence
Evidence
Questions
Follow-up questions
ADME
Are absorption, distribution, metabolism, or excretion
different by route of exposures studied, life stage when
exposure occurred, or dosing regimens used?
Will separate analyses be needed by route
of exposure, or by methods of dosing
within a route of exposure (e.g., are large
differences expected between gavage and
dietary exposures)?
Which life stages and what dosing
regimens are more relevant to human
exposure scenarios?
Is there toxicity information for metabolites that also
should be evaluated for hazard?
What exposures will be included in the
evaluation?
Is the parent chemical or metabolite also produced
endogenously?
Outcomes
What outcomes are reported in studies? Are the data
reported in a comparable manner across studies (similar
output metrics at similar levels of specificity, such as
adenomas and carcinomas quantified separately)?
At what level (hazard, grouped outcomes,
or individual outcomes) will the synthesis
be conducted?
What commonalities will the outcomes be
grouped by:
•	health effect,
•	exposure levels,
•	functional or population-level
consequences (e.g., endpoints all
ultimately leading to decreased
fertility or impaired cognitive
function), or
•	involvement of related biological
pathways?
How well do the assessed human and
animal outcomes relate within a level of
grouping?
Are there inter-related outcomes? If so, consider
whether some outcomes are more useful and/or of
greater concern than others.
Does the evidence indicate greater sensitivity to effects
(at lower exposure levels or severity) in certain
subgroups (by age, sex, ethnicity, life stage)? Should the
hazard evaluation include a subgroup analysis?
Does incidence or severity of an outcome increase with
duration of exposure or a particular window of
exposure? What exposure time frames are relevant to
development or progression of the outcome?
Is there mechanistic evidence that informs any of the
outcomes and how might they be grouped together?
How robust is the evidence for specific outcomes?
•	What outcomes are reported by both human and
animal studies and by one or the other? Were
different animal species and sexes (or other
important population-level differences) tested?
•	In general, what are the study confidence
conclusions of the studies (high, medium, low,
uninformative) for the different outcomes? Is there
enough evidence from high- and
medium-confidence studies for particular outcomes
to draw conclusions about causality?
What outcomes should be highlighted?
Should the others be synthesized at all?
Would comparisons by specific limitations
be informative?
This document is a draft for review purposes only and does not constitute Agency policy.
67	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Evidence
Questions
Follow-up questions
Dose-
response
Did some outcomes include better coverage of exposure
ranges that may be most relevant to human exposure
than others?
What outcomes and study characteristics
are informative for development of
toxicity values?

Does the study have multiple dose levels for which you
can evaluate dose-response gradient? Are there
outcomes with study results of sufficient similarity
(e.g., an established linkage in a biological pathway) to
allow examination or calculation of common measures
of effect across studies? Do the mechanistic data
identify surrogate or precursor outcomes that are
sufficient for dose-response analysis?


Are there subgroups that exhibit responses at lower
exposure levels than others?


Are there findings from ADME studies that could inform
data-derived extrapolation factors, or link toxicity
observed via different routes of exposure, or link effects
between humans and experimental animals?
Is there a common internal dose metric
that can be used to compare species or
routes of exposure?
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
8.DATA EXTRACTION OF STUDY METHODS AND
RESULTS
Data extraction and content management will be carried out using HAWC. Data extraction
elements that may be collected from epidemiological and animal toxicological 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. In
addition, outcomes that are determined to be less relevant during PECO refinement may not go
through data extraction or may have only minimal 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
/ow-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 /ow-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.8
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. 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 and animal 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
8The following browsers are fully supported for accessing HAWC: Google Chrome (preferred), Mozilla Firefox,
and Apple Safari. There are errors in functionality when viewed with Internet Explorer.
This document is a draft for review purposes only and does not constitute Agency policy.
69	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
1	authors will be documented and considered unsuccessful if researchers do not respond to email or
2	phone requests after one or two attempts to contact
3	For animal data already extracted to evidence tables released in 2014 (U.S. EPA. 2014b 1 and
4	contained in Microsoft Word, data extraction procedures were followed according to data type.
5	• Dichotomous data: Following the 2014 public meetings, these data were revised to correct
6	errors identified by public commenters, EPA staff, and contractors. Additional dichotomous
7	data sets were extracted during this revision process. The revised dichotomous data tables
8	will be imported into HAWC from Microsoft Word.
9	• Continuous data: Because the evidence tables released in 2014 expressed continuous data
10	only as a percent control response, the values in those tables do not contain enough
11	information for quality revisions or HAWC importation. As a result, the raw data (means
12	and standard deviations or standard errors) will be re-extracted from the publications and
13	entered into HAWC. For chronic studies that collected data at multiple sampling times (e.g.,
14	4 days, 22 days, 3 months, 6 months, or 12 months), data extraction will be performed for
15	the chronic sampling time only. On a case-by-case basis, data extraction at earlier sampling
16	times will be performed (e.g., to illustrate the dynamic behavior of some endpoints).
17	• Qualitative results: Results that were only presented qualitatively by the study authors
18	and extracted for the evidence tables released in 2014 were imported into appropriate
19	HAWC text fields.
20	• Uninformative and /ow-confidence studies: Data and results from studies determined to
21	be uninformative or low confidence by study evaluation (further described in Section 6) will
22	generally not be imported into HAWC. HAWC entries for these studies may be limited to
23	basic information about the references. Additional study information or data may be
24	available in HAWC for these studies on a case-by-case basis (e.g., if HAWC data extraction
25	occurred before final study evaluation).
26	For human data already extracted to evidence tables released in 2014 (U.S. EPA. 2014c) and
27	contained in Microsoft Word, data extraction procedures will depend on the quality of the study
28	and the study design. In general, study summary information will be imported into appropriate
29	HAWC text fields for all studies that are evaluated in HAWC. If sufficient medium- and
30	/jzg/j-confidence studies are not available following study evaluation, data from /ow-confidence
31	studies will be extracted. Studies will undergo a more thorough data extraction than was
32	performed in 2014 (see Appendix B).
8.1. STANDARDIZING REPORTING OF EFFECT SIZES
33	In addition to providing quantitative outcomes in their original units for all study groups,
34	results from outcome measures will be transformed, when possible, to a common metric to help
35	compare distinct but related outcomes that are measured with different scales. These standardized
36	effect size estimates facilitate systematic evaluation and evidence integration for hazard
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
identification, whether or not meta-analysis is feasible for an assessment (see Section 9.1). Many
such data transformations can be performed automatically in HAWC. The following summary of
effect size metrics by data type outlines issues in selecting the most appropriate common metric for
a collection of related endpoints (Vesterinen et al.. 20141.
Common metrics for continuous outcomes include:
•	Absolute difference in means. This metric is the difference between the means in the control
and treatment groups, expressed in the units in which the outcome is measured. When the
outcome measure and its scale are the same across all studies, this approach is the simplest
to implement.
•	Percent control response (or normalized mean difference [NMD]). Percent control group
calculations are based on means. Standard deviation (or standard error) values presented
in the studies for these normalized effect sizes can also be estimated if sufficient
information has been provided. Note that some outcomes reported as percentages, such as
mean percentage of affected offspring per litter, can lead to distorted effect sizes when
further characterized as a percentage change from control. Such measures are better
expressed as absolute difference in means or are preferably transformed to incidences
using approaches for event or incidence data (see below).
•	Standardized mean difference. The NMD approach above is relevant to ratio scales, but
sometimes it is not possible to infer what a "normal" animal would score, such as when data
for animals without lesions are not available. In these circumstances, standardized mean
differences can be used. The difference in group means is divided by a measure of the
pooled variance to convert all outcome measures to a standardized scale with units of
standard deviations. This approach can also be applied to data for which different
measurement scales are reported for the same outcome measure (e.g., different measures of
lesion size such as infarct volume and infarct area).
Common metrics for event or incidence data include:
•	Percent change from control. This metric is analogous to the NMD approach described for
continuous data above.
•	For binary outcomes such as the number of individuals that developed a disease or died, and
with only one treatment evaluated, data can be represented in a 2 x 2 table. Note that when
the value in any cell is zero, 0.5 is added to each cell to avoid problems with the
computation of the standard error. For each comparison, the odds ratio (OR) and its
standard error can be calculated. ORs are normally combined on a logarithmic scale.
An additional approach for epidemiology studies is to extract adjusted statistical estimates
when possible rather than unadjusted or raw estimates.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
It is important to consider the variability associated with effect size estimates, with stronger
studies generally showing more precise estimates. Effect size estimation can be affected, however,
by such factors as variances that differ substantially across treatment groups, or by lack of
information to characterize variance, especially for animal studies in biomedical research
(Vesterinen etal.. 20141.
8.2. STANDARDIZING ADMINISTERED DOSE LEVELS/CONCENTRATIONS
Exposures will be standardized to common units. Exposure levels in oral studies will be
expressed in units of mg Cr(VI)/kg-day. When study authors provide exposure levels in
concentrations in the diet or drinking water, dose conversions will be made using study-specific
food or water consumption rates and body weights if available. When possible, time-weighted
average daily doses will be calculated from the start of the bioassay through the time of data
collection. Otherwise, EPA defaults will be used (U.S. EPA. 19881. addressing age and study
duration as relevant for the species/strain and sex of the animal of interest Exposure levels in
inhalation studies will be expressed in units of mg/m3. Assumptions used in performing dose
conversions will be documented.
Exposure levels will be converted to Cr(VI) equivalents depending on the chemical
compound. For example, doses of test material administered as sodium dichromate dihydrate
(Cr2H4Na20g) were expressed as Cr(VI) using a molecular weight conversion of approximately
0.3490 g Cr(VI) per g CrzH^azOg.
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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
9.SYNTHESIS WITHIN LINES OF EVIDENCE
For the purposes of this assessment, evidence synthesis and integration are considered
distinct, but related processes. The syntheses of separate lines of evidence (i.e., human, animal, and
mechanistic evidence) described in this section will directly inform the integration across the lines
of evidence to draw overall conclusions for each of the assessed human health effects (described in
Section 10). The phrase "evidence integration" used here is analogous to the phrase "weight of
evidence" used in some other assessment processes (EFSA. 2017: U.S. EPA. 2017: NRC. 2014: U.S.
EPA. 2005al.
For each potential health hazard or smaller subset of related outcomes, EPA will separately
synthesize the available human and animal health effect evidence. Mechanistic evidence is also
considered, although the specific analytical approach is targeted to the assessment needs
depending on the extent and nature of the human and animal evidence (see Sections 9.2 and 10).
Each synthesis will be written to provide a summaiy discussion of the available evidence that may
suggest causation adapted from considerations for causality introduced by Austin Bradford Hill
(Hill. 1965): consistency, exposure-response relationship, strength of the association, temporal
relationship, biological plausibility, coherence, and "natural experiments" in humans [fU.S. EPA.
2005a. 19941: see Table 15], Importantly, the evidence synthesis process explicitly considers and
incorporates the conclusions from the individual study evaluations (see Section 6).
Data permitting the syntheses will also discuss analyses relating to potential susceptible
populations9. These analyses will be based on knowledge about the health outcome or organ
system affected, demographics, genetic variability, life stage, health status, behaviors or practices,
social determinants, and exposure to other pollutants (see Table 16). This information will be used
to describe potential susceptibility among specific populations or subgroups in a separate section
(see Section 10.3) summarizing across lines of evidence and hazards to inform hazard identification
and dose-response analyses.
9Various terms have been used to characterize populations that may be at increased risk of developing health
effects from exposure to environmental chemicals, including "susceptible," "vulnerable," and "sensitive."
Further, these terms have been inconsistently defined across the scientific literature. The term susceptibility
is used in this protocol to describe populations at increased risk, focusing on biological (intrinsic) factors, as
well as social and behavioral determinants that can modify the effect of a specific exposure. However, certain
factors resulting in higher exposures to specific groups (e.g., proximity, occupation, housing) may not be
analyzed to describe potential susceptibility among specific populations or subgroups.
This document is a draft for review purposes only and does not constitute Agency policy.
73	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 15. Information most relevant to describing primary considerations
informing causality during evidence syntheses
Consideration
Description and synthesis methods
Study confidence
Description: Incorporates decisions about studv confidence within each of the
considerations.
Application: In evaluating the evidence for each of the causalitv considerations
described in the following rows, syntheses will consider study confidence decisions.
H/g/?-confidence studies carry the most weight. Syntheses will consider specific
limitations and strengths of studies and how they inform each consideration.
Consistency
Description: Examines the similarity of results (e.g., direction; magnitude) across
studies.
Application: Syntheses will evaluate the homogeneity of findings on a given outcome
or endpoint across studies. When inconsistencies exist, the syntheses consider
whether results were "conflicting" (i.e., unexplained positive and negative results in
similarly exposed human populations or in similar animal models) or "differing"
[i.e., mixed results explained by differences between human populations, animal
models, exposure conditions, or studv methods; (U.S. EPA, 2005a)l based on analyses
of potentially important explanatory factors such as:
•	Confidence in studies' results, including study sensitivity (e.g., some study results
that appear to be inconsistent may be explained by potential biases or other
attributes that affect sensitivity).
•	Exposure, including route (if applicable) and administration methods, levels,
duration, timing with respect to outcome development, and exposure
assessment methods (i.e., in epidemiology studies).
•	Specificity and sensitivity of the endpoint for evaluating the health effect in
question (e.g., functional measures can be more sensitive than organ weights).
•	Populations or species, including consideration of potential susceptible groups or
differences across life stage at exposure or endpoint assessment.
•	Toxicokinetic information explaining observed differences in responses across
route of exposure, other aspects of exposure, species, or life stages.
The interpretation of consistency will emphasize biological significance, to the extent
that it is understood, over statistical significance. Statistical significance from suitably
applied tests (this may involve consultation with an EPA statistician) adds weight when
biological significance is not well understood. Consistency in the direction of results
increases confidence in that association even in the absence of statistical significance.
Strength (effect
magnitude) and
precision
Description: Examines the effect magnitude or relative risk, based on what is known
about the assessed endpoint(s), and considers the precision of the reported results
based on analyses of variability (e.g., confidence intervals; standard error). This may
include consideration of the rarity or severity of the outcomes.
Application: Syntheses will analvze results both within and across studies and mav
consider the utility of combined analyses (e.g., meta-analysis). While larger effect
magnitudes and precision (e.g., p < 0.05) help reduce concerns about chance, bias, or
other factors as explanatory, syntheses should also consider the biological or
population-level significance of small effect sizes.
This document is a draft for review purposes only and does not constitute Agency policy.
74	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Consideration
Description and synthesis methods
Biological gradient/
dose-response
Description: Examines whether the results (e.g., response magnitude; incidence;
severity) change in a manner consistent with changes in exposure (e.g., level;
duration), including consideration of changes in response after cessation of exposure.
Application: Syntheses will consider relationships both within and across studies,
acknowledging that the dose-response (e.g., shape) can vary depending on other
aspects of the experiment, including the biology underlying the outcome and the
toxicokinetics of the chemical. Thus, when dose-response is lacking or unclear, the
synthesis will also consider the potential influence of such factors on the response
pattern.
Coherence
Description: Examines the extent to which findings are cohesive across different
endpoints that are related to, or dependent on, one another (e.g., based on known
biology of the organ system or disease, or mechanistic understanding such as
toxicokinetic/dynamic understanding of the chemical or related chemicals). In some
instances, additional analyses of mechanistic evidence from research on the chemical
under review or related chemicals that evaluate linkages between endpoints or
organ-specific effects may be needed to interpret the evidence. These analyses may
require additional literature search strategies.
Application: Syntheses will consider potentially related findings, both within and across
studies, particularly when relationships are observed within a cohort or within a
narrowly defined category (e.g., occupation; strain or sex; life stage of exposure).
Syntheses will emphasize evidence indicative of a progression of effects, such as
temporal or dose-dependent increases in the severity of the type of endpoint
observed. If an expected coherence between findings is not observed, possible
explanations should be explored including the biology of the effects as well as the
sensitivity and specificity of the measures used.
Mechanistic evidence
related to biological
plausibility
Description: There are multiple uses for mechanistic information (see Section 9.2) and
this consideration overlaps with "coherence." This examines the biological support (or
lack thereof) for findings from the human and animal health effect studies and
becomes more impactful on the hazard conclusions when notable uncertainties in the
strength of those sets of studies exist. These analyses can also improve understanding
of dose- or duration-related development of the health effect. In the absence of
human or animal evidence of apical health endpoints, the synthesis of mechanistic
information may drive evidence integration conclusions (when such information is
available).
Application: Syntheses can evaluate evidence on precursors, biomarkers, or other
molecular or cellular changes related to the health effect(s) of interest to describe the
likelihood that the observed effects result from exposure. This will be an analysis of
existing evidence, and not simply whether a theoretical pathway can be postulated.
This analysis may not be limited to evidence relevant to the PECO but may also include
evaluations of biological pathways (e.g., for the health effect; established for other,
possibly related, chemicals). The synthesis will consider the sensitivity of the
mechanistic changes and the potential contribution of alternate or previously
unidentified mechanisms of toxicity.
This document is a draft for review purposes only and does not constitute Agency policy.
75	DRAFT-DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Consideration
Description and synthesis methods
Natural experiments
Description: Specific to epidemiology studies and rarely available, this examines effects
in populations that have experienced well-described, pronounced changes in chemical
exposure (e.g., lead exposures before and after banning lead in gasoline).
Application: Compared to other observational designs, natural experiments have the
benefit of dividing people into exposed and unexposed groups without them
influencing their own exposure status. During synthesis, associations in medium- and
/?/g/?-confidence natural experiments can substantially reduce concerns about residual
confounding.
Table 16. Individual and social factors that may increase susceptibility to
exposure-related health effects
Factor
Examples
Demographic
Gender, age, race/ethnicity, education, income, occupation, geography
Genetic variability
Polymorphisms in genes regulating cell cycle, DNA repair, cell division, cell
signaling, cell structure, gene expression, apoptosis, and metabolism
Life stage
In utero, childhood, puberty, pregnancy, women of childbearing age, old age
Health status
Pre-existing conditions or disease such as psychosocial stress, body mass
index, frailty, nutritional status, chronic disease
Behaviors or practices
Diet, mouthing, smoking, alcohol consumption, pica, subsistence, or
recreational hunting and fishing
Social determinants
Income, socioeconomic status, neighborhood factors, health care access, and
social, economic, and political inequality
9.1. SYNTHESES OF HUMAN AND ANIMAL HEALTH EFFECTS EVIDENCE
The syntheses of the human and animal health effect evidence will focus on describing
aspects of the evidence that best inform causal interpretations, including the exposure context
examined in the sets of studies. These syntheses (or the lack of data within these lines of evidence)
help determine the approach to be taken in synthesizing the available mechanistic evidence (see
Section 9.2). In this way, the mechanistic synthesis might range from a high-level summary of
potential mechanisms of action to specific, focused questions needed to fill data gaps identified
from the human and animal syntheses and integration (e.g., shape of dose-response at low doses,
applicability of the animal evidence to humans, addressing susceptible populations).
Evidence synthesis will be based primarily on studies of high and medium confidence.
Low-confidence studies may be used, if few or no studies with higher confidence are available, to
help evaluate consistency, or if the study designs of the /ow-confidence studies address notable
uncertainties in the set of high- or med/um-confidence studies on a given health effect. If
This document is a draft for review purposes only and does not constitute Agency policy.
76	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 Hexavalent Chromium IRIS Assessment
/ow-confidence studies are used, then a careful examination of risk of bias and sensitivity with
potential impacts on the evidence synthesis conclusions will be included in the narrative.
As previously described, these syntheses will articulate the strengths and the weaknesses of
the available evidence organized around the considerations described in Table 15, as well as issues
that stem from the evaluation of individual studies (e.g., concerns about bias or sensitivity). If
possible, results across studies will be compared using graphs and charts or other data
visualization strategies. The analysis will typically include an examination of results stratified by
any or all of the following: study confidence classification (or specific issues within confidence
evaluation domains), population or species, exposures (e.g., level, patterns [intermittent or
continuous], duration, intensity), sensitivity (e.g., low vs. high), and other factors that may have
been identified in the refined evaluation plan. The number of studies and the differences
encompassed by the studies will determine the extent to which specific types of factors can be
examined to stratify study results. Additionally, for both the human and animal evidence syntheses,
if supported by the available data, additional analyses across studies (such as meta-analysis) may
also be conducted.
9.2. MECHANISTIC INFORMATION
Mechanistic information includes any experimental measurement related to a health
outcome that informs the biological or chemical events associated with phenotypic effects; these
measurements can improve understanding of the biological effects following exposure to a
chemical but are not generally considered by themselves adverse outcomes. Mechanistic data are
reported in a diverse array of observational and experimental studies across species, model
systems, and exposure paradigms, including in vitro, in vivo (by various routes of exposure),
ex vivo, and in silico studies. The evidence available to describe mechanistic events or MOAs (U.S.
EPA. 2005a) is typically aggregated from numerous studies, often involving a diverse range of
exposure paradigms and models, as well as a wide spectrum of diverse endpoints. In addition, a
chemical may operate through multiple mechanistic pathways fU.S. EPA. 2005al. Similarly,
multiple mechanistic pathways might interact to cause a single, adverse effect In contrast to the
defined scope of the evaluation and syntheses of PECO-specific human or animal health effect
studies, the potential utility and interpretation of mechanistic information can be quite broad and
difficult to define. Thus, to be pragmatic and provide clear and transparent syntheses of the most
useful information, the mechanistic syntheses for most health outcomes will focus on a subset of
the most relevant mechanistic studies. It should be stressed, however, that the process of
evaluating mechanistic information differs fundamentally from evaluations of the other evidence
streams. More specifically, the mechanistic analysis for any specific substance will depend on
evaluating the confidence that the relevant data are consistent with a plausible biological
understanding of how a chemical exposure might generate an adverse outcome, rather than
focusing on evaluations of individual studies.
This document is a draft for review purposes only and does not constitute Agency policy.
11	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 Hexavalent Chromium IRIS Assessment
The synthesis of mechanistic information is used to inform the integration of health effect
evidence for both hazard identification (i.e., biological plausibility or coherence of the available
human or animal evidence, inferences regarding human relevance, or the identification of
susceptible populations and life stages across the human and animal evidence) and dose-response
evaluation. Therefore, the synthesis of the mechanistic data will focus on the evidence most likely
to be useful for augmenting the human or animal health effect evidence. Based on the identified
gaps in understanding, the mechanistic synthesis may focus on providing information on precursor
events, a biological understanding of how effects develop or are related, the human relevance of
animal results, or identifying likely susceptible populations and life stages. This means that, for
example, if extensive /j/g/j-confidence human or animal evidence is available, the need to synthesize
all available mechanistic evidence will be diminished. In these cases, the synthesis will focus on the
analysis and interpretation of smaller sets of mechanistic studies that specifically address
controversial issues to resolve, such as those related to applicability of animal evidence to humans
when the human evidence is weak or the shape of the dose-response at low exposure levels when
this understanding is highly uncertain and data informing this uncertainty exist.
To identify the focused set(s) of studies for use in analyses of critical mechanistic questions,
the synthesis will apply a phased approach that progressively focuses the scope of the mechanistic
information to be considered. This stepwise focusing, which began during the literature search and
screening steps based on problem formulation decisions, depends primarily on the potential hazard
signals that arise from the human and/or animal health effect studies, or from mechanistic studies
that signal potential hazards that have not been examined in health effect studies. Cr(VI)
mechanistic information will be collected and inventoried (i.e., capturing details relating to
exposure characteristics, model system, and assays tested to allow for sorting and retrieval to
address critical mechanistic questions) for all health outcomes meeting PECO criteria, including
cancer and effects on the GI, respiratory, reproductive, developmental, immune, and hematological
systems. Other mechanistic information (e.g., relevant to non-PECO health outcomes) will be
reviewed and sorted to facilitate later decisions, including identification of areas of research
unexamined in the available human or animal health effect studies.
For cancer, it is acknowledged that the issue of whether Cr(VI) causes cancer by the oral
route of exposure via a mutagenic mode of action is critical to address (see Section 2.3); therefore, a
specific and thorough analysis integrating the evidence for potential mechanisms of cancer relevant
to the oral route of exposure will be conducted. Given the focus of the lung cancer assessment on
dose-response analysis, the mechanistic information relevant to cancer via the inhalation route will
be investigated to identify and synthesize those studies that could influence the dose-response
assessment for lung cancer, if available. It is not anticipated that other mechanistic analyses
relevant to cancer will be conducted in the assessment; however, if other cancer types are identified
that require a focused mechanistic analysis, these will be documented in the updated protocol
released with the draft assessment To facilitate the two primary mechanistic evaluations for
This document is a draft for review purposes only and does not constitute Agency policy.
78	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 Hexavalent Chromium IRIS Assessment
cancer, an inventory of the available mechanistic studies was developed. As shown in Table 8,
mechanistic studies investigating genotoxicity, oxidative stress, alterations in cell proliferation and
cell death, electrophilicity, receptor-mediated effects, altered DNA repair, immortalization, chronic
inflammation, and epigenetic alterations have been identified in the mechanistic studies database
relevant to cancer. Mechanistic events relevant to these characteristics will be investigated, and
any areas lacking evidence will be identified. The identification of mechanistic evidence that may
indicate potentially relevant susceptible subpopulations or life stages will be particularly
important
The information collected as described above (e.g., in sortable inventories) will be used to
identify studies available for consideration in addressing the specific gaps in understanding
identified as critical to address from the evaluation and synthesis of the human and animal lines of
evidence, including postulated mechanistic pathways or MOAs that may be involved in the toxicity
of the chemical. Subsequently, from the studies available to potentially address the identified gaps
in understanding, the synthesis will focus on those considered most impactful to the specified
evaluation based on study design characteristics (which may or may not encompass all studies
considered relevant for a particular question), with the rationale for any focusing transparently
documented. As the potential influence of the information provided by these studies can vary
depending on the hazard question(s) or the associated mechanistic events or pathways, the level of
rigor will also depend on their potential impact of increased understanding to hazard identification
or dose-response decisions, and may range from overviews of potential mechanisms or cursory
insights drawn from sets of unanalyzed results to detailed evaluations of a subset of the most
relevant mechanistic studies.
Although the application of this approach cannot be predefined, for the small subsets of
studies that best address the key mechanistic questions, the synthesis will first prioritize studies
based on their toxicological relevance to answering the specific question (e.g., model system,
specificity of the assay for the effect of interest). For example, mechanistic information from in vivo
studies will be analyzed first, with primary consideration given to endpoint-relevant routes.
Analysis of ex vivo and in vitro studies will then be prioritized by those most informative to
evaluating the mechanistic events indicated by the in vivo data, including studies conducted under
conditions most relevant to human exposures and in model systems best replicating in vivo human
biology. The path for focusing the mechanistic database will be documented in the updated
protocol released with the draft assessment
More rigorous analyses will be particularly important when the set(s) of studies available to
inform influential mechanistic conclusions are inconsistent and potentially conflicting, or when the
studies include experiments that directly challenge the necessity of proposed mechanistic
relationships between exposure and an apical effect (e.g., altering a receptor-mediated pathway
through chemical intervention or using knock-out animals). More detailed analyses may also be
useful when it is apparent that study design aspects in the available human and animal health effect
This document is a draft for review purposes only and does not constitute Agency policy.
79	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 Hexavalent Chromium IRIS Assessment
studies are likely to have significant flaws or introduce important uncertainties (e.g., potential
shortcomings identified during the evaluation of exposure methods may be clarified using
mechanistic studies).
For the more rigorous mechanistic analyses, the review will be facilitated using
pathway-based organizational methods and established evidence evaluation frameworks. These
approaches provide transparency and objectivity to the integration and interpretation of
mechanistic events and pathways anchored to the specific questions that have been identified (e.g.,
anchored to a specific health effect) across diverse sets of relevant data (e.g., human, animal, and in
vitro studies). The key characteristics of carcinogens have been used to organize the large
mechanistic database relevant to cancer for Cr(VI) exposure (see Table 8) and will serve to
organize the mechanistic analysis and help identify key events that will be evaluated using the MOA
analysis framework described in EPA's cancer guidelines (U.S. EPA. 2005a). Similar approaches
(e.g., identification of key characteristics or mechanistic events anchored to a specific health effect)
will be used to organize mechanistic databases for noncancer health effects. The mechanistic
analyses will inform the evidence integration across lines of evidence, as well as the dose-response
analyses, that are described in Sections 11 and 12.
This document is a draft for review purposes only and does not constitute Agency policy.
80	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 Hexavalent Chromium IRIS Assessment
10. INTEGRATION ACROSS LINES OF EVIDENCE
For the analysis of human health outcomes that might result from chemical exposure, IRIS
assessments draw integrated conclusions across human, animal, and mechanistic evidence (see
Section 9). During evidence integration, a two-step, sequential process will be used as follows (and
depicted in Figure 4):
•	First, judgments regarding the strength of the evidence from the available human and
animal studies are made in parallel. These judgments incorporate mechanistic evidence (or
MOA understanding) in exposed humans and animals, respectively, that informs the
biological plausibility and coherence of the available human or animal health effect studies.
Note that at this stage, the animal evidence judgment does not yet consider the human
relevance of that evidence.
•	Second, the animal and human evidence judgments are combined to draw an overall
conclusion(s) that incorporates inferences drawn based on information on the human
relevance of the animal evidence (i.e., based on default assumptions or empirical evidence),
coherence across the human and animal evidence streams, and susceptibility.
STEP 1: INTEGRATION OF HEALTH EFFECT	STEP 2: OVERALL INTEGRATION OF EVIDENCE
AND MECHANISTIC EVIDENCE IN HUMANS OR	FOR HAZARD ID
ANIMALS
HUMAN EVIDENCE JUDGMENT
The synthesis of evidence about health effects
and mechanisms from human studies is
combined (integrated) to make a judgment
about health effects in human studies
The judgments regarding the human and
animal evidence are integrated in light of
evidence on the human relevance of the
findings in animals, susceptibility, and the
coherence of the findings across evidence
streams to draw a conclusion about the
evidence for health effects in humans.
EVIDENCE INTEGRATION CONCLUSION
The synthesis of evidence about health effects
and mechanisms from animal studies is
combined (integrated) to make a judgment
about health effects in animals.
ANIMAL EVIDENCE JUDGMENT
Figure 4. Process for evidence integration.
This document is a draft for review purposes only and does not constitute Agency policy.
81	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 Hexavalent Chromium IRIS Assessment
The decision points within the structured two-step evidence integration process will be
summarized in an evidence profile table for each hazard (see Table 17 for a template) in support of
the evidence integration narrative. Human and animal evidence judgments from Step 1 and the
overall evidence integration conclusion from Step 2 are reached using decision frameworks (see
Sections 10.1 and 10.2 for details) that are based on considerations originally described by Austin
Bradford Hill (Hill. 1965). This process is similar to that used by the Grading of Recommendations
Assessment, Development, and Evaluation [GRADE; (Morgan etal.. 2016: Guvattetal.. 2011:
Schiinemann et al.. 2 0111]. which arrives at an overall level of confidence conclusion based on
considering the body of evidence. As described in Section 9, the human, animal, and mechanistic
syntheses serve as inputs providing a foundation for the evidence integration decisions; thus, the
major conclusions from these syntheses will be summarized in the evidence profile table (see
Table 17) supporting the evidence integration narrative. The evidence profile table summarizes the
judgments and their evidence basis for each step of the structured evidence integration process.
Separate sections are included for human and animal evidence judgments, inference across
streams, and the overall evidence integration conclusion. The table presents the key information
from the evidence that informed each judgment.
10.1. INTEGRATION WITHIN THE HUMAN AND ANIMAL EVIDENCE
As summarized above, before drawing overall evidence integration conclusions about
whether a chemical is likely to cause particular health effect(s) in humans given relevant exposure
circumstances, judgments are drawn regarding the strength of evidence for the available human
and animal evidence, separately. If relevant mechanistic evidence in exposed humans and animals
(or their cells) was synthesized, this line of evidence will be integrated with the evidence from
health effects studies. The considerations outlined in Table 15 (see Section 9) are evaluated in the
context of how they impact the strength of evidence (see Table 18), and the judgments are reached
using the structured frameworks explained in Tables 19 and 20 (for human and animal evidence,
respectively). These judgments are summarized in tabular format using the template in Table 17 to
transparently convey expert judgments made throughout the evidence synthesis and integration
processes. The evidence profile table allows for consistent documentation of the supporting
rationale for each decision.
This document is a draft for review purposes only and does not constitute Agency policy.
82	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 17. Evidence profile table template
Studies and
interpretation
Factors that
increase strength
Factors that
decrease strength
Summary of findings
Human and animal
evidence
judgments
Inference across lines
of evidence
Overall evidence
integration
conclusion
[Health effect or outcome grouping]
Evidence from human studies [route]
•	Human relevance of
findings in animals
•	Coherence across
lines of evidence
(i.e., for both health
effect-specific and
mechanistic data)
•	Other inferences
o Information on
susceptibility
o MOA analysis
inferences (e.g.,
cross-species
inferences of
toxicokinetics, or
quantitative
implications)
o Relevant
information from
other sources
(e.g., read across;
other potentially
related health
hazards)
Describe
conclusion(s) and
primary basis for
the integration of
all available
evidence (across
human, animal,
and mechanistic):
• ©©©
Evidence
demonstrates
•	®©o
Evidence
indicates
•	©oo
Evidence
suggests
•	ooo
Evidence
inadequate
•	— — —
Strong
evidence
supports no
effect
•	References
•	Study
confidence
(based on
evaluation of
risk of bias
and
sensitivity)
•	Study design
description
•	Consistency or
replication
•	Dose-response
gradient
•	Coherence of
observed
effects (apical
studies)
•	Effect size
(magnitude,
severity)
•	Mechanistic
evidence
providing
plausibility
•	Medium- or
/?/g/?-confidence
studies3
•	Unexplained
inconsistency
•	Imprecision
•	/.ow-confidence
studies3 or
other concerns
about methods
or design across
studies
•	Other (e.g.,
single/few
studies)
•	Evidence
demonstrating
implausibility
(e.g.,
mechanistic)
•	Results information
(general endpoints
affected/unaffected)
across studies
•	Human mechanistic
evidence informing
biological plausibility:
discuss how data
influenced the human
evidence judgment
(e.g., evidence of
precursors in exposed
humans)
Could be multiple rows (e.g.,
grouped by study confidence
or population) if this informs
heterogeneity of results
Describe the strength
of the evidence from
human studies, and
primary basis for
judgment:
•	©0©
Robust
•	®©o
Moderate
•	©oo
Slight
•	ooo
Indeterminate
Compelling
evidence of no
effect
This document is a draft for review purposes only and does not constitute Agency policy.
83	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Studies and
interpretation
Factors that
increase strength
Factors that
decrease strength
Summary of findings
Human and animal
evidence
judgments
Inference across lines
of evidence
Overall evidence
integration
conclusion
Evidence for an effect in animals [route]
References
• Consistency or
• Unexplained
Study
replication
inconsistency
confidence
• Dose-response
• Imprecision
(based on
gradient
• /.ow-confidence
evaluation of
• Coherence of
studies3 or
risk of bias
observed
other concerns
and
effects (apical
about methods
sensitivity)
studies)
or design across
Study design
• Effect size
studies
description
(magnitude,
• Other (e.g.,

severity)
single/few

• Mechanistic
studies)

evidence
• Evidence

providing
demonstrating

plausibility
implausibility

• Medium- or
(e.g.,

/?/g/?-confidence
mechanistic)

studies3

•	Results information
(general endpoints
affected/unaffected)
across studies
•	Animal mechanistic
evidence informing
biological plausibility:
discuss how
mechanistic data
influenced the animal
evidence judgment
(e.g., evidence of
coherent molecular
changes in animal
studies)
Could be multiple rows (e.g.,
grouped by study confidence
or population) if this informs
heterogeneity of results
Describe the strength
of the evidence from
animal studies, and
primary basis for
judgment:
•	©0©
Robust
•	©©O
Moderate
•	©oo
Slight
•	ooo
Indeterminate
•	— — —
Compelling
evidence of no
effect
aStudy confidence, based on evaluation of risk of bias and study sensitivity (see Section 6), should be considered when evaluating each of the other factors that increase
or decrease strength (e.g., consistency). Notably, lack of findings in studies deemed insensitive neither increases nor decreases strength.
This document is a draft for review purposes only and does not constitute Agency policy.
84	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 18. Considerations that inform judgments regarding the strength of the human and animal evidence
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
The structured categories and criteria in Tables 19 and 20 will guide the application of strength of evidence judgments for an outcome or health effect.
Evidence synthesis scenarios that do not warrant an increase or decrease in evidence strength will be considered "neutral" and do not need to be described in
Table 17.
Risk of bias; sensitivity
(across studies)
• An evidence base of high- or medium-confidence
studies increases strength.
•	An evidence base of mostly /ow-confidence studies decreases
strength. An exception to this is an evidence base of studies
where the primary issues resulting in low confidence are related
to insensitivity. This may increase evidence strength in cases
where an association is identified because the expected impact
of study insensitivity is towards the null.
•	Decisions to increase strength for other considerations in this
table should generally not be made if there are serious concerns
for risk of bias.
Consistency
• Similarity of findings for a given outcome (e.g., of a
similar magnitude, direction) across independent
studies or experiments increases strength, particularly
when consistency is observed across populations (e.g.,
location) or exposure scenarios in human studies, and
across laboratories, populations (e.g., species), or
exposure scenarios (e.g., duration; route; timing) in
animal studies.
• Unexplained inconsistency (conflicting evidence) decreases
strength. Generally, strength should not be decreased if
discrepant findings can be reasonably explained by study
confidence conclusions; variation in population or species, sex,
or life stage; exposure patterns (e.g., intermittent or
continuous); levels (low or high); or duration or intensity.
Strength (effect
magnitude) and
precision
•	Evidence of a large magnitude effect (considered
either within or across studies) can increase strength.
Effects of a concerning rarity or severity can also
increase strength, even if they are of a small
magnitude.
•	Precise results from individual studies or across the set
of studies increases strength, noting that biological
significance is prioritized over statistical significance.
• Strength may be decreased if effect sizes that are small in
magnitude are concluded not to be biologically significant, or if
there are only a few studies with imprecise results.
This document is a draft for review purposes only and does not constitute Agency policy.
85	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
Biological
gradient/dose-response
•	Evidence of dose-response increases strength.
Dose-response may be demonstrated across studies or
within studies and it can be dose- or
duration-dependent. It also may not be a monotonic
dose-response (monotonicity should not necessarily be
expected, e.g., different outcomes may be expected at
low vs. high doses due to activation of different
mechanistic pathways or induction of systemic toxicity
at very high doses).
•	Decreases in a response after cessation of exposure
(e.g., symptoms of current asthma) also may increase
strength by increasing certainty in a relationship
between exposure and outcome (this is most
applicable to epidemiology studies because of their
observational nature).
•	A lack of dose-response when expected based on biological
understanding and having a wide-range of doses/exposures
evaluated in the evidence base can decrease strength.
•	In experimental studies, strength may be decreased when
effects resolve under certain experimental conditions (e.g., rapid
reversibility after removal of exposure). However, many
reversible effects are of high concern. Deciding between these
situations is informed by factors such as the toxicokinetics of the
chemical and the conditions of exposure fsee U.S. EPA (1998a)l,
endpoint severity, judgments regarding the potential for delayed
or secondary effects, as well as the exposure context focus of the
assessment (e.g., addressing intermittent or short-term
exposures).
•	In rare cases, and typically only in toxicology studies, the
magnitude of effects at a given exposure level might decrease
with longer exposures (e.g., due to tolerance or acclimation).
Like the discussion of reversibility above, a decision about
whether this decreases evidence strength depends on the
exposure context focus of the assessment and other factors.
•	If the data are not adequate to evaluate a dose-response
pattern, then strength is neither increased or decreased.
Coherence
• Biologically related findings within an organ system, or
across populations (e.g., sex) increase strength,
particularly when a temporal- or dose-dependent
progression of related effects is observed within or
across studies, or when related findings of increasing
severity are observed with increasing exposure.
• An observed lack of expected coherent changes (e.g.,
well-established biological relationships) will typically decrease
evidence strength. However, the biological relationships
between the endpoints being compared and the sensitivity and
specificity of the measures used need to be carefully examined.
The decision to decrease depends on the availability of evidence
across multiple related endpoints for which changes would be
anticipated, and it considers factors (e.g., dose and duration of
exposure, strength of expected relationship) across the studies
of related changes.
This document is a draft for review purposes only and does not constitute Agency policy.
86	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
Mechanistic evidence
related to biological
plausibility
•	Mechanistic evidence of precursors or health effect
biomarkers in well-conducted studies of exposed
humans or animals, in appropriately exposed human or
animal cells, or other relevant human or animal models
increases strength, particularly when this evidence is
observed in the same cohort/population exhibiting the
health outcome.
•	Evidence of changes in biological pathways or that
provides support for a proposed MOA in models also
increases strength, particularly when support is
provided for rate-limiting or key events or conserved
across multiple components of the pathway or MOA.
•	Mechanistic understanding is not a prerequisite for drawing a
conclusion that a chemical causes a given health effect; thus,
absence of knowledge should not be used a basis for decreasing
strength (NTP, 2015; NRC, 2014).
•	Mechanistic evidence in well-conducted studies that
demonstrates that the health effect(s) are unlikely to occur, or
only likely to occur under certain scenarios (e.g., above certain
exposure levels), can decrease evidence strength. A decision to
decrease depends on an evaluation of the strength of the
mechanistic evidence supporting vs. opposing biological
plausibility, as well as the strength of the health effect-specific
findings (e.g., stronger health effect data require more certainty
in mechanistic evidence opposing plausibility).
This document is a draft for review purposes only and does not constitute Agency policy.
87	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 Hexavalent Chromium IRIS Assessment
For human and animal evidence, the analyses of each consideration in Table 18 will be used
to develop a strength-of-evidence judgment. Tables 19 and 20 provide the judgments for each
category and the criteria that will guide how to apply the judgments. Briefly, the terms robust and
moderate are standardized characterizations for judgments on the extent of support provided by
human or animal studies that the health effect(s) result from chemical exposure. Repeated
observations of effects by independent studies examining various aspects of exposure or response
(e.g., different exposure settings, dose levels or patterns, populations or species, and related
endpoints) will result in a stronger strength of evidence judgment These terms are applied to
human and animal evidence separately and are differentiated by the quantity and quality of
information available to rule out alternative explanations for the results. The term slight indicates
situations in which there is some evidence indicating an association within the evidence stream, but
substantial uncertainties in the data exist to prevent stronger judgments from being drawn.
Indeterminate reflects evidence stream judgments when no studies are available, or situations in
which the evidence is inconsistent and/or primarily of low confidence. Compelling evidence of no
effect represents a situation in which extensive evidence across a range of populations and
exposures has identified no effects/associations. This scenario is seldom used because it requires a
high degree of confidence in the conduct of individual studies, including consideration of study
sensitivity and comprehensive assessments of health outcomes and life stages of exposure.
Publication bias has the potential to result in strength-of-evidence judgments that are stronger than
would be merited if the entire body of research were available. However, the existence of
publication bias can be difficult to determine and is not a component of the strength-of-evidence
framework for human or animal studies. If potential publication bias is evaluated for an outcome, it
may inform the level of certainty regarding the completeness of the assessment database for that
outcome.
This document is a draft for review purposes only and does not constitute Agency policy.
88	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 19. Framework for evidence judgments from studies in humans
Within-stream
strength-of-
evidence
judgment
Description
Robust
(©©©)
...evidence in
human studies
(strong signal of
effect with little
residual
uncertainty)
A set of high- or medium-confidence independent studies reporting an association between the
exposure and the health outcome, with reasonable confidence that alternative explanations,
including chance, bias, and confounding, can be ruled out across studies. The set of studies is
primarily consistent, with reasonable explanations when results differ; and an exposure
response gradient is demonstrated. Additional supporting evidence, such as associations with
biologically related endpoints in human studies (coherence) or large estimates of risk or
severity of the response, may increase confidence but are not required.
Mechanistic evidence from exposed humans, if available, may add support informing
considerations such as exposure response, temporality, coherence, and MOA, thus, raising the
level of certainty to robust for a set of studies that otherwise would be described as moderate.
Moderate
(©©O)
...evidence in
human studies
(signal of effect
with some
uncertainty)
A smaller number of studies (at least one high- or medium-confidence study with supporting
evidence), or with some heterogeneous results, that do not reach the degree of confidence
required for robust. For multiple studies, there is primarily consistent evidence of an
association, but there may be some uncertainty due to potential chance, bias, or confounding.
For a single study, there is a large magnitude or severity of the effect, or a dose-response
gradient, or other supporting evidence, and there are no serious residual methodological
uncertainties. Supporting evidence could include associations with related endpoints, including
mechanistic evidence from exposed humans, if available, based on considerations such as
exposure response, temporality, coherence, and MOA.
Slight
(©OO)
...evidence in
human studies
(signal of effect
with large
amount of
uncertainty)
One or more studies reporting an association between exposure and the health outcome,
where considerable uncertainty exists. In general, the evidence is limited to a set of consistent
/ow-confidence studies, or higher confidence studies with unexplained heterogeneity.
Supporting coherent evidence is sparse. Biological support from mechanistic evidence in
exposed humans may also be independently interpreted as slight. This also includes scenarios
where there are serious residual uncertainties across studies (these uncertainties typically
relate to exposure characterization or outcome ascertainment, including temporality) in a set of
largely consistent medium- or /?/g/?-confidence studies. This category serves primarily to
encourage additional study where evidence does exist that might provide some support for an
association, but for which the evidence does not reach the degree of confidence required for
moderate.
Indeterminate
(OOO)
...evidence in
human studies
(signal cannot be
determined for
or against an
effect)
No studies available in humans or situations when the evidence is highly inconsistent and
primarily of low confidence. In addition, this may include situations where higher confidence
studies exist, but unexplained heterogeneity exists, and there are additional outstanding
concerns such as effect estimates of low magnitude, uninterpretable patterns with respect to
exposure levels, or uncertainties or methodological limitations that result in an inability to
discern effects from exposure. A set of largely null studies could be concluded to be
indeterminate if the evidence does not reach the level required for compelling evidence of no
effect.
This document is a draft for review purposes only and does not constitute Agency policy.
89	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Within-stream
strength-of-
evidence
judgment
Description
Compelling
evidence of no
effect
(...)
...in human
studies
(strong signal for
lack of an effect
with little
uncertainty)
Several /?/g/?-confidence studies showing null results (for example, an odds ratio of 1.0), ruling
out alternative explanations including chance, bias, and confounding with reasonable
confidence. Each of the studies should have used an optimal outcome and exposure
assessment and adequate sample size (specifically for higher exposure groups and for
susceptible populations). The set as a whole should include the full range of levels of exposures
that human beings are known to encounter, an evaluation of an exposure-response gradient,
and an examination of at-risk populations and life stages.
This document is a draft for review purposes only and does not constitute Agency policy.
90	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 20. Framework for evidence judgments from studies in animals
Within-stream
strength-of-
evidence
judgment
Description
Robust
(©©©)
...evidence in
animals
(strong signal of
effect with little
residual
uncertainty)
The set of high- or medium-confidence experiments includes consistent findings of adverse or
toxicologically significant effects across multiple laboratories, exposure routes, experimental
designs (e.g., a subchronic study and a two-generation study), or species, and the experiments
can reasonably rule out the potential for nonspecific effects (e.g., resulting from toxicity) to
have resulted in the findings. Any inconsistent evidence (evidence that cannot be reasonably
explained by the respective study design or differences in animal model) is from a set of
experiments of lower confidence. At least two of the following additional factors in the set of
experiments support a causal association: coherent effects across multiple related endpoints
(may include mechanistic endpoints); an unusual magnitude of effect, rarity, age at onset, or
severity; a strong dose-response relationship; or consistent observations across animal life
stages, sexes, or strains. Alternatively, mechanistic data in animals (in vivo or in vitro) that
address the above considerations or that provide experimental support for an MOA that
defines a causal relationship with reasonable confidence may raise the level of certainty to
robust for evidence that otherwise would be described as moderate or, exceptionally, slight or
indeterminate.
Moderate
(©©O)
...evidence
in animals
(signal of effect
with some
uncertainty)
A set of evidence that does not reach the degree of certainty required for robust, but which
includes at least one high- or medium-confidence study and information strengthening the
likelihood of a causal association. Although the results are largely consistent, notable
uncertainties remain. However, while inconsistent evidence and/or evidence indicating
nonspecific effects (e.g., maternal toxicity at doses causing developmental effects) may exist, it
is not sufficient to reduce or discount the level of concern regarding the positive findings from
the supportive experiments or it is from a set of experiments of lower confidence. The set of
experiments supporting the effect provide additional information supporting a causal
association, such as consistent effects across laboratories or species; coherent effects across
multiple related endpoints (may include mechanistic endpoints); an unusual magnitude of
effect, rarity, age at onset, or severity; a strong dose-response relationship; and/or consistent
observations across exposure scenarios (e.g., route, timing, duration), sexes, or animal strains.
Mechanistic data in animals (in vivo or in vitro) that address the above considerations or that
provide information supporting an association between exposure and effect with reasonable
confidence may raise the level of certainty to moderate for evidence that otherwise would be
described as slight.
This document is a draft for review purposes only and does not constitute Agency policy.
91	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Within-stream
strength-of-
evidence
judgment
Description
Slight
(©OO)
...evidence in
animals
(signal of effect
with large
amount of
uncertainty)
Scenarios in which there is a signal of a possible effect, but the evidence is conflicting or weak.
Most commonly, this includes situations where only /ow-confidence experiments are available
and supporting coherent evidence is sparse. It also applies when one medium- or
/?/g/?-confidence experiment is available without additional information strengthening the
likelihood of a causal association (e.g., corroboration within the same study or from other
studies). Lastly, this includes scenarios in which there is evidence that would typically be
characterized as moderate, but inconsistent evidence (evidence that cannot be reasonably
explained by the respective study design or differences in animal model) from a set of
experiments of higher confidence (may include mechanistic evidence) exists. Strong biological
support from mechanistic studies in exposed animals or animal cells may also be independently
interpreted as slight. Notably, to encourage additional research, it is important to describe
situations for which evidence does exist that might provide some support for an association but
is insufficient for a conclusion of moderate.
Indeterminate
(OOO)
...evidence of the
effect under
review in
animals
(signal cannot be
determined for
or against an
effect)
No animal studies were available, the available endpoints are not informative to the hazard
question under evaluation, or the evidence is highly inconsistent and primarily of low
confidence. In addition, this may include situations where higher confidence studies exist, but
there is unexplained heterogeneity and additional concerns such as small effect sizes (given
what is known about the endpoint) or a lack of dose-dependence. A set of largely null studies
could be concluded to be indeterminate if the evidence does not reach the level required for
compelling evidence of no effect.
Compelling
evidence of no
effect
(...)
...in animals
(strong signal for
lack of an effect
with little
uncertainty)
A set of /?/g/?-confidence experiments examining a reasonable spectrum of endpoints relevant
to a type of toxicity that demonstrate a lack of biologically significant effects across multiple
species, both sexes, and a broad range of exposure levels. The data are compelling in that the
experiments have examined the range of scenarios across which health effects in animals could
be observed, and an alternative explanation (e.g., inadequately controlled features of the
studies' experimental designs; inadequate sample sizes) for the observed lack of effects is not
available. The experiments were designed to specifically test for effects of interest, including
suitable exposure timing and duration, post exposure latency, and endpoint evaluation
procedures, and to address potentially susceptible populations and life stages. Mechanistic
data in animals (in vivo or in vitro) that address the above considerations or that provide
information supporting the lack of an association between exposure and effect with reasonable
confidence may provide additional support to this judgment.
10.2. OVERALL EVIDENCE INTEGRATION CONCLUSIONS
1	The second stage of evidence integration combines animal and human evidence judgments
2	while also considering mechanistic information on the human relevance of the animal evidence,
3	relevance of the mechanistic evidence to humans (especially in cases where animal evidence is
This document is a draft for review purposes only and does not constitute Agency policy.
92	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 Hexavalent Chromium IRIS Assessment
lacking), coherence across lines of evidence, and information on susceptible populations. Based on
the integration across lines of evidence, this stage culminates in an evidence integration narrative
as described at the beginning of this chapter that summarizes the conclusions regarding each
potential health effect (i.e., each noncancer health effect and specific type of cancer, or broader
grouping of related outcomes as defined in the evaluation plan). For each health effect, this
narrative will include a summary of the strength of the evidence and an overall conclusion across
the lines of evidence, with exposure context provided. The first sentence of the evidence
integration narrative should include the summary conclusion, and for evaluations of
carcinogenicity, include the cancer descriptor fU.S. EPA. 2005al Table 21 describes the five
evidence integration conclusion levels, the integration conclusion language associated with each
level, and the types of evidence that fit each level. The five integration conclusion levels reflect the
differences in the amount and quality of the data that inform the evaluation of whether exposure
may cause the health effect(s) under specified exposure conditions.
For evaluations of carcinogenicity, consistent with EPA's cancer guidelines (U.S. EPA.
2005a), one of EPA's standardized cancer descriptors will be used as a shorthand characterization
of the evidence integration narrative, describing the overall potential for carcinogenicity. These
are: (1) carcinogenic to humans, (2) likely to be carcinogenic to humans, (3) suggestive evidence of
carcinogenic potential, (4) inadequate information to assess carcinogenic potential, or (5) not likely
to be carcinogenic to humans. More than one descriptor can be used when a chemical's effects differ
by dose or exposure route fU.S. EPA. 2005al In some cases, mutagenicity will also be evaluated
(e.g., when there is evidence of carcinogenicity) because it influences the approach to
dose-response assessment and subsequent application of adjustment factors for exposures early in
life riJ.S. EPA. 2005a. b).
For each cancer subtype, an evidence integration narrative will be provided as described
above, and an appropriate descriptor will be selected as described in the EPA's cancer guidelines
(U.S. EPA. 2005a). If a systematic review of more than one cancer type was conducted, then the
conclusion for the cancer type(s) with the highest confidence will be used as the basis for the
standardized cancer descriptor. When considering evidence on carcinogenicity across human and
animal evidence streams, consistent with EPA guidance fU.S. EPA. 2005al. site concordance is not
required. The cancer descriptor and evidence integration narrative (including application of the
MOA framework) will also consider the conditions of carcinogenicity, including exposure
(e.g., route; dose) and susceptibility (e.g., genetics; life stage), as the data allow (Farland. 2005: U.S.
EPA. 2005a. bl.
This document is a draft for review purposes only and does not constitute Agency policy.
93	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 21. Conclusions for the evidence integration narrative
Evidence integration
conclusion3 in narrative
Evidence
integration
conclusion level
Explanation and example scenarios'3
The currently available
evidence demonstrates
that (chemical) causes
(health effect) in humans0
under relevant exposure
circumstances. This
conclusion is based on
studies of (humans or
animals) that assessed
(exposure or dose) levels of
(range of concentrations or
specific cutoff-level
concentration01).
Evidence
demonstrates
A strong evidence base demonstrating that (chemical) exposure
causes (health effect) in humans.
•	This conclusion level is used if there is robust human
evidence supporting an effect.
•	This conclusion level could also be used with moderate
human evidence and robust animal evidence if there is
strong mechanistic evidence that MOAs and key precursors
identified in animals are anticipated to occur and progress
in humans.
The currently available
evidence indicates that
(chemical) likely causes
(health effect) in humans
under relevant exposure
circumstances. This
conclusion is based on
studies of (humans or
animals) that assessed
(exposure or dose) levels of
(range of concentrations or
specific cutoff-level
concentration).
Evidence
indicates (likely0)
An evidence base that indicates that (chemical) exposure likely
causes (health effect) in humans, although outstanding
questions or limitations may remain, and the evidence is
insufficient for the higher conclusion level.
•	This conclusion level is used if there is robust animal
evidence supporting an effect and slight to indeterminate
human evidence, or with moderate human evidence when
strong mechanistic evidence is lacking.
•	This conclusion level could also be used with moderate
human evidence supporting an effect and slight or
indeterminate animal evidence, or with moderate animal
evidence supporting an effect and slight or indeterminate
human evidence. In these scenarios, any uncertainties in
the moderate evidence are not sufficient to substantially
reduce confidence in the reliability of the evidence, or
mechanistic evidence in the slight or indeterminate
evidence base (e.g., precursors) exists to increase
confidence in the reliability of the moderate evidence.
This document is a draft for review purposes only and does not constitute Agency policy.
94	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Evidence integration
conclusion3 in narrative
Evidence
integration
conclusion level
Explanation and example scenarios'3
The currently available
evidence suggests that
(chemical) may cause
(health effect) in humans
under relevant exposure
circumstances. This
conclusion is based on
studies of (humans or
animals) that assessed
(exposure or dose) levels of
(range of concentrations or
specific cutoff-level
concentration).
Evidence suggests
An evidence base that suggests that (chemical) exposure may
cause (health effect) in humans, but there are very few studies
that have contributed to the evaluation, the evidence is very
weak or conflicting, and/or the methodological conduct of the
studies is poor.
•	This conclusion level is used if there is slight human
evidence and indeterminate to slight animal evidence.
•	This conclusion level is also used with slight animal
evidence and indeterminate to slight human evidence.
•	This conclusion level could also be used with moderate
human evidence and slight or indeterminate animal
evidence, or with moderate animal evidence and slight or
indeterminate human evidence. In these scenarios,
outstanding issues regarding the moderate evidence have
substantially reduced confidence in the reliability of the
evidence, or mechanistic evidence in the slight or
indeterminate evidence base (e.g., null results in
well-conducted evaluations of precursors) exists to
decrease confidence in the reliability of the moderate
evidence.
Exceptionally, when there is general scientific understanding of
mechanistic events that result in a health effect, this conclusion
level could also be used if there is strong mechanistic evidence
that is sufficient to highlight potential human toxicity'—in the
absence of informative conventional studies in humans or in
animals (i.e., indeterminate evidence in both).
The currently available
evidence is inadequate to
assess whether (chemical)
may cause (health effect) in
humans under relevant
exposure circumstances.
Evidence
inadequate
This conveys either a lack of information or an inability to
interpret the available evidence for (health effect). On an
assessment-specific basis, a single use of this "inadequate"
conclusion level might be used to characterize the evidence for
multiple health effect categories (i.e., all health effects that were
examined and did not support other conclusion levels).8
•	This conclusion level is used if there is indeterminate
human and animal evidence.
•	This conclusion level is also used with slight animal
evidence and compelling evidence of no effect human
evidence.
•	This conclusion level could also be used with sliaht to
robust animal evidence and indeterminate human evidence
if strong mechanistic information indicated that the animal
evidence is unlikely to be relevant to humans.
A conclusion of inadequate is not a determination that the agent
does not cause the indicated health effect(s). It simply indicates
that the available evidence is insufficient to reach conclusions.
This document is a draft for review purposes only and does not constitute Agency policy.
95	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Evidence integration
conclusion3 in narrative
Evidence
integration
conclusion level
Explanation and example scenarios'3
Strong evidence supports
no effect in humans under
relevant exposure
circumstances. This
conclusion is based on
studies of (humans or
animals) that assessed
(exposure or dose) levels of
(range of concentrations).
Strong evidence
supports no
effect
This represents a situation in which extensive evidence across a
range of populations and exposure levels has identified no
effects/associations. This scenario requires a high degree of
confidence in the conduct of individual studies, including
consideration of study sensitivity, and comprehensive
assessments of the endpoints and life stages of exposure
relevant to the heath effect of interest.
•	This conclusion level is used if there is compelling evidence
of no effect in human studies and compelling evidence of no
effect to indeterminate effect in animals.
•	This conclusion level is also used if there is indeterminate
human evidence and compelling evidence of no effect
animal evidence in models concluded to be relevant to
humans.
This conclusion level could also be used with compellina
evidence of no effect in human studies and moderate to robust
animal evidence if strong mechanistic information indicated that
the animal evidence is unlikely to be relevant to humans.
aEvidence integration conclusions are typically developed at the level of the health effect when there are sufficient
studies on the topic to evaluate the evidence at that level; this should always be the case for "evidence
demonstrates" and "strong evidence supports no effect," and typically for "evidence indicates (likely)." However,
some databases only allow for evaluations at the category of health effects examined; this will more frequently
be the case for conclusion levels of "evidence suggests" and "evidence is inadequate."
terminology of "is" refers to the default option; terminology of "could also be" refers to situational options
dependent on mechanistic understanding.
cln some assessments, these conclusions might be based on data specific to a particular life stage of exposure, sex,
or population (or another specific group). In such cases, this would be specified in the narrative conclusion, with
additional detail provided in the narrative text. This applies to all conclusion levels.
dlf concentrations cannot be estimated, an alternative expression of exposure level such as "occupational
exposure levels," will be provided. This applies to all conclusion levels.
eFor some applications, such as benefit-cost analysis, to better differentiate the categories of "evidence
demonstrates" and "evidence indicates," the latter category should be interpreted as evidence that supports an
exposure-effect linkage that is likely to be causal.
'Scientific understanding of adverse outcome pathway (AOPs) and of the human implications of new toxicity
testing methods (e.g., from high-throughput screening, from short-term in vivo testing of alternative species, or
from new in vitro testing) will continue to increase. This may make possible the development of hazard
conclusions when there are mechanistic or other relevant data that can be interpreted with a similar level of
confidence to positive animal results in the absence of conventional studies in humans or in animals.
Specific narratives for each of these health effects may also be deemed unnecessary.
10.3. HAZARD CONSIDERATIONS FOR DOSE-RESPONSE
This section provides a transition from hazard identification to the dose-response section,
highlighting (1) information that will inform the selection of outcomes or broader health effect
This document is a draft for review purposes only and does not constitute Agency policy.
96	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 Hexavalent Chromium IRIS Assessment
categories for which toxicity values will be derived, (2) whether toxicity values can be derived to
protect specific populations or life stages, (3) how dose-response modeling will be informed by
toxicokinetic information, and (4) identification of biologically based benchmark response (BMR)
levels. The pool of outcomes and study-specific endpoints will be discussed to identify which
categories of effects and study designs are considered the strongest and most appropriate for
quantitative assessment of a given health effect. Health effects that were analyzed in relation to
exposure levels within or closer to the range of exposures encountered in the environment are
particularly informative. When there are multiple endpoints for an organ/system, considerations
for characterizing the overall impact on this organ/system will be discussed. For example, if there
are multiple histopathological alterations relevant to liver function changes, liver necrosis may be
selected as the most representative endpoint to consider for dose-response analysis. This section
may review or clarify which endpoints or combination of endpoints in each organ/system
characterize the overall effect for dose-response analysis. For cancer types, consideration will be
given to the overall risk of multiple types of tumors. Multiple tumor types (if applicable) will be
discussed, and a rationale given for any grouping.
Biological considerations that are important for dose-response analysis (e.g., that could help
with selection of a BMR) will be discussed. The impact of route of exposure on toxicity to different
organs/systems will be examined. The existence and validity of PBPK models or toxicokinetic
information that may allow the estimation of internal dose will be presented. In addition,
mechanistic evidence presented in Section 9 that will influence the dose-response analyses will be
highlighted, for example, evidence related to susceptibility or potential shape of the dose-response
curve (i.e., linear, nonlinear, or threshold model). Mode(s) of action will be summarized, including
any interactions between them relevant to understanding overall risk. Some biological
considerations relevant to dose-response for cancer are:
•	Is there evidence for direct mutagenicity?
•	Does tumor latency decrease with increasing exposure?
•	If there are multiple tumor types, which cancers have a longer latency period?
•	Is incidence data available (incidence data are preferred to mortality data)?
•	Were there different background incidences in different (geographic) populations?
•	While benign and malignant tumors of the same cell of origin are generally evaluated
together, was there an increase only in malignant tumors?
This section will draw from Sections 9 and 10 to describe the evidence (i.e., human, animal,
mechanistic) regarding populations and life stages susceptible to the hazards identified and factors
This document is a draft for review purposes only and does not constitute Agency policy.
97	DRAFT-DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
that increase risk of the hazards. This section should include a discussion of the populations that
may be, in general, susceptible to the health effects identified to be hazards of exposure to the
assessed chemical, even if there are no specific data on effects of exposure to that chemical in the
potentially susceptible population. Background information about biological mechanisms or ADME,
as well as biochemical and physiological differences among life stages may be used to guide the
selection of populations and life stages to consider. At a minimum, particular consideration will be
given to infants and children, pregnant women, and women of childbearing age. Evidence on
factors that contribute to some population groups having increased responses to chemical exposure
and/or factors that contribute to increases in exposure or dose will be summarized and evaluated
with respect to patterns across studies pertinent to consistency, coherence, and the magnitude and
direction of effect measures. Relevant factors may include intrinsic factors (e.g., age, sex, genetics,
health status, behaviors), extrinsic factors (e.g., socioeconomic, access to health care), and
differential exposure levels or frequency (e.g., occupation-related exposure, residential proximity to
locations with greater exposure intensity).
The section will consider options for using data related to susceptible populations to impact
dose-response analysis. In particular, an attempt will be made to highlight where it might be
possible to develop separate risk estimates for a specific population or life stage or determine
whether evidence is available to select a data-derived uncertainty factor (UF).
This document is a draft for review purposes only and does not constitute Agency policy.
98	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
11. DOSE-RESPONSE ASSESSMENT: SELECTING
STUDIES AND QUANTITATIVE ANALYSIS
The previous sections of this protocol describe how systematic review principles are
applied to evaluate study quality (potential bias and sensitivity) and reach evidence synthesis and
integration conclusions on health outcomes (or hazard identification) associated with exposure to
the chemical of interest Selection of specific data sets for dose-response assessment and
performance of the dose-response assessment is conducted after hazard identification is complete
and involves database and chemical-specific biological judgments. A number of EPA guidance and
support documents detail data requirements and other considerations for dose-response modeling
especially EPA's Benchmark Dose Technical Guidance fU.S. EPA. 2012bl. EPA's Review of the
Reference Dose and Reference Concentration Processes (U.S. EPA. 2005a. 2002). Guidelines for
Carcinogen Risk Assessment fU.S. EPA. 2005al. and Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA. 2005b). 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.
For IRIS assessments, dose-response assessments are typically performed for both
noncancer and cancer hazards, and for both oral and inhalation routes of exposure following
chronic exposure10 to the chemical of interest, if supported by existing data. For noncancer
hazards, an oral reference dose (RfD) and/or an inhalation reference concentration (RfC) are
usually derived. An RfD or an RfC 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 fU.S. EPA. 2002. §4.21.
These health effects may also include cancer effects [e.g., in a case where a nonlinear MOA is
concluded that indicates a key precursor event necessary for carcinogenicity does not occur below
a specific exposure level fU.S. EPA. 2005a. §3.3.41: see Section 11.2.3], Reference values are not
predictive risk values; that is, they provide no information about risks at higher or lower exposure
levels.
When low-dose linear extrapolation for cancer effects is supported, particularly for
chemicals with direct mutagenic activity or those for which the data indicate a linear component
below the POD, an oral slope factor (OSF) and/or an inhalation unit risk (IUR) are used to estimate
human cancer risks. An OSF is a plausible upper-bound lifetime cancer risk from chronic ingestion
10Dose-response assessments may also be conducted for shorter durations, particularly if the evidence base
for a chemical indicates risks associated with shorter exposures to the chemical (U.S. EPA. 20021.
This document is a draft for review purposes only and does not constitute Agency policy.
99	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 Hexavalent Chromium IRIS Assessment
of a chemical per unit of mass consumed per unit body weight, per day (mg/kg-day). An IUR is a
plausible upper-bound lifetime cancer risk from chronic inhalation of a chemical per unit of air
concentration (expressed as ppm or ng/m3). In contrast with reference values (RfVs), an OSF or
IUR can be used in conjunction with exposure information to predict cancer risk at a given dose.
As discussed in Section 2 ("Scoping and Initial Problem Formulation Summary") of this
assessment, IRIS will conduct the assessment with a goal of developing an RfD and RfC for the
noncancer effects of Cr(VI) and quantitative cancer assessments for inhaled and ingested Cr(VI)
consistent with the available mechanistic evidence.
The derivation of cancer risk estimates may also depend on the nature of the hazard
conclusion. Specifically, EPA generally conducts dose-response assessments and derives cancer
values for chemicals that are classified as carcinogenic or likely to be carcinogenic to humans. When
there is suggestive evidence of carcinogenicity to humans, EPA generally would not conduct a
dose-response assessment and derive a cancer value. Similarly, for noncancer outcomes, EPA will
make decisions on whether to conduct dose-response assessments based on the strength of the
evidence of a hazard. However, when the evidence includes a well-conducted study, quantitative
analyses may be useful for some purposes, for example, providing a sense of the magnitude and
uncertainty of potential risks, ranking potential hazards, or setting research priorities (U.S. EPA.
2005a).
11.1. SELECTING STUDIES FOR DOSE-RESPONSE ASSESSMENT
The dose-response assessment begins with a review of the important health effects
highlighted in the hazard identification step (see Section 10), particularly among the studies of
highest quality and that exemplify the study attributes summarized in Table 22. This review also
considers whether there are opportunities for quantitative evidence integration. Examples of
quantitative integration, from simplest to more complex, include (1) combining results for an
outcome across sex (within a study); (2) characterizing overall toxicity, as in combining effects that
comprise a syndrome, or occur on a continuum (e.g., precursors and eventual overt toxicity, benign
tumors that progress to malignant tumors); and (3) conducting a meta-analysis or meta-regression
of all studies addressing a category of important health effects.
Some studies that are used qualitatively for hazard identification may or may not be useful
quantitatively for dose-response assessment due to such factors as the lack of quantitative
measures of exposure or lack of variability measures for response data. If the needed information
cannot be located (see Section 7), semiquantitative analysis may be feasible (e.g., via
NOAEL/lowest-observed-adverse-effect level [LOAEL]). 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.
This document is a draft for review purposes only and does not constitute Agency policy.
100	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table 22. Attributes used to evaluate studies for derivation of toxicity values


Considerations
Study attributes
Human studies
Animal studies
Study confidence
High- or medium-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).
Animal studies provide supporting evidence when adequate human
studies are available, and are considered principal studies when
adequate human studies are not available. For some hazards, studies
of animal species known to respond similarly to humans would be
preferred over studies of other species.
Relevance of
exposure
paradigm
Exposure
route
Studies involving human environmental exposures (oral,
inhalation).
Studies by a route of administration relevant to human
environmental exposure are preferred. A validated toxicokinetic
model can also be used to extrapolate across exposure routes.

Exposure
durations
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
levels
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, §2.1.1) 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 (e.g., matching procedures, blocking) 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.
This document is a draft for review purposes only and does not constitute Agency policy.
101	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Study attributes
Considerations
Human studies
Animal studies
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.
Studies providing actual measurements of exposure (e.g., analytical
inhalation concentrations vs. target concentrations) are preferred.
Relevant internal dose measures may facilitate extrapolation to
humans, as would availability of a suitable animal PBPK model in
conjunction with an animal study reported in terms of administered
exposure.
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.
102	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 Hexavalent Chromium IRIS Assessment
Among the studies that support the hazard conclusions, those that are 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. 2012b). 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 (e.g., IUR or RfC) 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 (U.S.
EPA. 2012bl.
11.2. 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. 2012b: 2005a. §31:
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 11.2.1 for more details.
2)	Derivation of cancer risk estimates or reference values nearly always involves extrapolation
to exposures lower than the POD and is described in more detail in Sections 11.2.2 and
11.2.3., respectively.
When sufficient and appropriate human data and laboratory animal data are both available
for the same outcome, human data are generally preferred for the dose-response assessment
because their use eliminates the need to perform interspecies extrapolations.
For reference values, IRIS assessments typically derive a candidate value from each suitable
data set, whether for human or animal (see Section 11.1). Evaluating these candidate values
grouped within a particular organ/system yields a single organ/system-specific value for each
organ/system under consideration. Next, evaluation of these organ/system-specific values results
in the selection of a single overall reference value to cover all health outcomes across all
organs/systems. While this overall reference value is the focus of the assessment, the
organ/system-specific values can be useful for subsequent cumulative risk assessments that
consider the combined effect of multiple agents acting at a common organ/system.
For cancer, if there are multiple tumor sites in a study population (human or animal), final
cancer risk estimates will typically address overall cancer risk.
This document is a draft for review purposes only and does not constitute Agency policy.
103	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 Hexavalent Chromium IRIS Assessment
11.2.1. 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 MOA and quantitatively support model
parameters that represent rates and other quantities associated with the key precursor events of
the MOA. 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 fhttp://www.epa.gov/ncea/bmds)
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, and reporting modeling results [see the EPA Benchmark Dose
Technical Guidance fU.S. EPA. 2012bl], 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. For linear extrapolation of cancer risk, the POD is used to calculate an OSF or IUR,
and for nonlinear extrapolation, the POD is used in calculating an RfD or RfC.
The response level at which the POD is calculated is guided by the severity of the endpoint
If linear extrapolation is used, selection of a response level corresponding to the point of departure
is not highly influential, so standard values near the low end of the observable range are generally
used (for example, 10% extra risk for cancer bioassay data, 1% for epidemiologic data, lower for
rare cancers). Nonlinear approaches account for both statistical and biologic considerations. 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 biologic significance. In the absence of such definition, one control
standard deviation from the control mean is often used for minimally adverse effects, one-half
This document is a draft for review purposes only and does not constitute Agency policy.
104	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 Hexavalent Chromium IRIS Assessment
standard deviation for more severe effects. The point of departure is the 95% lower bound on the
dose associated with the selected response level.
EPA has developed standard approaches for determining the relevant dose to be used in the
dose-response modeling in the absence of appropriate toxicokinetic modeling. These standard
approaches also facilitate comparison across exposure patterns and species:
•	Intermittent study exposures are standardized to a daily average over the duration of
exposure. For chronic effects, daily exposures are averaged over the life span. Exposures
during a critical period, however, are not averaged over a longer duration fU.S. EPA. 2005a.
33.1.1: 1991. §3.21.
•	Doses are standardized to equivalent human terms to facilitate comparison of results from
different species. Oral doses are scaled allometrically using mg/kg3/4-day as the equivalent
dose metric across species. Allometric scaling pertains to equivalence across species, not
across life stages, and is not used to scale doses from adult humans or mature animals to
infants or children fU.S. EPA. 2011a: 2005a. §3.1.31 Inhalation exposures are scaled using
dosimetry models that apply species-specific physiologic and anatomic factors and consider
whether the effect occurs at the site of first contact or after systemic circulation (U.S. EPA.
2012a: 1994. §31.
•	It can be informative to convert doses across exposure routes. If this is done, the
assessment describes the underlying data, algorithms, and assumptions fU.S. EPA. 2005a.
§3.1.41.
•	In the absence of study-specific data on, for example, intake rates or body weight, the EPA
has developed recommended values for use in dose-response analysis fU.S. EPA. 19881.
11.2.2. Extrapolation: Slope Factors and Unit Risks
An OSF or IUR facilitates estimation of human cancer risks when low-dose linear
extrapolation for cancer effects is supported, particularly for chemicals with direct mutagenic
activity or those for which the data indicate a linear component below the POD. Low-dose linear
extrapolation is also used as a default when the data are insufficient to establish the MOA fU.S. EPA.
2005a)- If data are sufficient to ascertain one or more MOAs consistent with low-dose nonlinearity,
or to support their biological plausibility, low-dose extrapolation may use the reference-value
approach when suitable data are available fU.S. EPA. 2005al: see Section 11.2.3 below.
Differences in susceptibility may warrant derivation of multiple slope factors or unit risks,
with separate estimates for susceptible populations and life stages (U.S. EPA. 2005a. b). If
appropriate chemical-specific data on susceptibility from early life exposures are available, then
these data are used to develop cancer risk values that specifically address any potential for
differential potency in early life stages fU.S. EPA. 2005a. b). If such data are not available, the
evidence synthesis and integration analyses support a mutagenic MOA for carcinogenicity, and the
This document is a draft for review purposes only and does not constitute Agency policy.
105	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
1	extrapolation approach is linear, the dose-response assessment should indicate that in the
2	development of risk estimates, the default age-dependent adjustment factors should be used with
3	the cancer slope factor or unit risk and age-specific estimates of exposure (U.S. EPA. 2005a. b).
4	The derivation of an OSF and IUR for Cr(VI) conducted as part of the current assessment
5	will be performed consistent with EPA guidance. For the oral assessment, both linear and
6	nonlinear approaches will be presented for Cr(VI) carcinogenicity (U.S. EPA. 2005a) to provide
7	insights into uncertainties related to model choice and mechanisms.
11.2.3. Extrapolation: Reference Values
8	Reference value derivation is EPA's most frequently used type of nonlinear extrapolation
9	method and is most commonly used for noncancer effects. This approach is also used for cancer
10	effects if there are sufficient data to ascertain the MOA and conclude that it is not linear at low
11	doses. For these cases, reference values for each relevant route of exposure are developed
12	following EPA's established practices fU.S. EPA. 2005a. §3.3.41: in general, the reference value is
13	based not on tumor incidence, but on a key precursor event in the MOA that is necessary for tumor
14	formation.
15	For each data set selected for reference value derivation, reference values are estimated by
16	applying relevant adjustments to the PODs to account for the conditions of the reference value
17	definition—for human variation, extrapolation from animals to humans, extrapolation to chronic
18	exposure duration, and extrapolation to a minimal level of risk (if not observed in the data set).
19	Extrapolation between routes of exposure will not be performed for Cr(VI) (see Sections 3.1 and
20	6.4). Increasingly, data-based adjustments fU.S. EPA. 2014al and Bayesian methods for
21	characterizing population variability (NRC. 2014) are feasible and may be distinguished from the
22	UF considerations outlined below. The assessment will discuss the scientific bases for estimating
23	these data-based adjustments and UFs:
24	• Animal-to-human extrapolation: If animal results are used to make inferences about
25	humans, the reference value derivation incorporates the potential for cross-species
26	differences, which may arise from differences in toxicokinetics or toxicodynamics. If
27	available, a biologically based model that adjusts fully for toxicokinetic and toxicodynamic
28	differences across species may be used. Otherwise, the POD is standardized to equivalent
29	human terms or is based on toxicokinetic or dosimetry modeling that may range from
30	detailed chemical-specific to default approaches (U.S. EPA. 2014a. 2011a). and a factor of
31	101/2 (rounded to 3) is applied to account for the remaining uncertainty involving
32	toxicokinetic and toxicodynamic differences.
33	• Human variation: The assessment accounts for variation in susceptibility across the human
34	population and the possibility that the available data may not represent individuals who are
35	most susceptible to the effect, by using a data-based adjustment, UF, or a combination of the
36	two. Where appropriate data or models for the effect or for characterizing the internal dose
37	are available, the potential for data-based adjustments for toxicodynamics or toxicokinetics
This document is a draft for review purposes only and does not constitute Agency policy.
106	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
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
is considered (U.S. EPA. 2014a. 2002).12 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; (U.S. EPA. 2002. §4.4.5:1998a. §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, 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. 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 fU.S. EPA. 2002.1998a. 1996.1994.19911.
•	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 (U.S. EPA. 2002.
1998a. 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. 2002.1998a. 1996.1994.19911.
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 101/2
(i.e., 3) if either one or the other is missing (U.S. EPA. 2002. §4.4.51.
The POD for an RfV is divided by the product of these factors. U.S. EPA f2002. §4.4.51
recommends that any composite factor that exceeds 3,000 represents excessive uncertainty and
recommends against relying on the associated RfV.
The derivation of an RfD and RfC for the noncancer effects of Cr(VI) will be conducted
consistent with EPA guidance summarized above.
"Examples of adjusting the toxicokinetic portion of interhuman variability include the IRIS boron
assessment's use of non-chemical-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. 2011bl.
12Note 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.
107	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
12. PROTOCOL HISTORY
1	Release date: March 15, 2019
This document is a draft for review purposes only and does not constitute Agency policy.
108	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
REFERENCES
Agerstrand. M: Kuester. A: Bachmann. T: Breitholtz. M: Ebert. I: Rechenberg. B: Ruden. C. (2011).
Reporting and evaluation criteria as means towards a transparent use of ecotoxicity data for
environmental risk assessment of pharmaceuticals. Environ Pollut 159: 2487-2492.
http://dx.doi.Org/10.1016/i.envpol.2011.06.023.
Anger. G: Halstenberg. 1: Hochgeschwender. K: Scherhag. C: Korallus. U: Knopf. H: Schmidt. P:
Ohlinger, M. (2005). Chromium compounds. In B Elvers (Ed.), Ullmann's Encyclopedia of
Industrial Chemistry (7th ed., pp. 1-36). Hoboken, NJ: WileyVCH Verlag GmbH & Co.
http://dx.doi.org/10.1002/143560Q7.a07 067.
ATS (American Thoracic Society). (1987). Standardization of spirometry-1987 update. Am J Respir
CritCare Med 136: 1285-1298.
ATS (American Thoracic Society). (1995). Standardization of spirometry 1994 update. American
Thoracic Society. Am J Respir CritCare Med 152: 1107-1136.
http://dx.doi.Org/10.1164/airccm.152.3.7663792.
ATSDR (Agency for Toxic Substances and Disease Registry). (2012). Toxicological profile for
chromium. Atlanta, GA: US Department of Health and Human Services, Public Health
Service. http://www.atsdr.cdc.gov/ToxProfiles/tp.asp?id=62&tid=17.
Bahadori. T: Thayer. K. (2018). National Academy of Sciences committee to review advances made
to the IRIS process. Presentation presented at National Academy of Sciences Review of
Advances Made to the IRIS Process: A Workshop, February 1-2, 2018, Washington, D.C.
Balashazv. I: Hofmann. W: Heistracher. T. (2003). Local particle deposition patterns may play a key
role in the development of lung cancer. J Appl Physiol (1985) 94: 1719-1725.
http://dx.doi.org/10.1152/iapplphysiol.00527.2002.
Barnhart. 1. (1997). Chromium chemistry and implications for environmental fate and toxicity.
Journal of Soil Contamination 6: 561-568.
Bedinger. GM. (2015). 2014 Minerals yearbook. Arsenic [advance release], Reston, VA: U.S.
Geological Survey.
Beronius. A: Molander. L: Ruden. C: Hanberg. A. (2014). Facilitating the use of non-standard in vivo
studies in health risk assessment of chemicals: a proposal to improve evaluation criteria
and reporting. J Appl Toxicol 34: 607-617. http://dx.doi.org/10.1002/iat.2991.
Beronius. A: Molander. L: Zilliacus. I: Ruden. C: Hanberg. A. (2018). Testing and refining the Science
in Risk Assessment and Policy (SciRAP) web-based platform for evaluating the reliability
and relevance of in vivo toxicity studies. J Appl Toxicol 38: 1460-1470.
http://dx.doi.org/10.1002/iat3648.
Bourdon-Lacombe. TA: Moffat. ID: Deveau. M: Husain. M: Auerbach. S: Krewski. D: Thomas. RS:
Bushel. PR: Williams. A: Yauk. CL. (2015). Technical guide for applications of gene
expression profiling in human health risk assessment of environmental chemicals. Regul
Toxicol Pharmacol 72: 292-309. http://dx.doi.org/10.1016/i.vrtph.2015.04.010.
This document is a draft for review purposes only and does not constitute Agency policy.
109	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Brazma. A: Hingamp. P: Ouackenbush. T: Sherlock. G: Spellman. P: Stoeckert. C: Aach. T: Ansorge. W:
Ball. CA: Causton. HC: Gaasterland. T: Glenisson. P: Holstege. FC: Kim. IF: Markowitz. V:
Matese. 1C: Parkinson. H: Robinson. A: Sarkans. U: Schulze-Kremer. S: Stewart. 1: Taylor. R:
Vilo. 1: Vingron. M. (2001). Minimum information about a microarray experiment (MIAME)-
toward standards for microarray data. Nat Genet 29: 365-371.
http://dx.doi.org/10.1038/ngl2Ql-365.
CA State Water Board (California State Water Quality Control Board). (2017). Frequently asked
questions about hexavalent chromium in public water systems [Fact Sheet], Sacramento,
CA.
https://www.waterboards.ca.gov/drinking water/certlic/drinkingwater/documents/chro
mium6/chrome 6 faqs.pdf.
CalEPA (California Environmental Protection Agency). (2003). Ambient air monitoring for
hexavalent chromium and metals in Barrio Logan: May 2001 through May 2002. (Report:
2003-10-14). Sacramento, CA.
CalEPA (California Environmental Protection Agency). (2004). Barrio Logan report: A compilation
of air quality studies in Barrio Logan. November 2004. (Report: 2005-11-00). Sacramento,
CA.
CalEPA (California Environmental Protection Agency). (2011). Public health goal for Hexavalent
Chromium (Cr VI) in drinking water. Sacramento, CA: Pesticide and Environmental
Toxicology Branch, Office of Environmental Health Hazard Assessment
http://oehha.ca.gOv/media/downloads/water/public-health-goal/cr6phg072911.pdf.
CDPH (California Department of Public Health). (2013). Notice of proposed rulemaking: Title 22,
California Code of Regulations subject: Hexavalent chromium mcl (DPH-11-005). Available
online at (accessed
Collins. BT: Stout. MP: Levine. KE: Kissling. GE: Melnick. RL: Fennell. TR: Walden. R: Abdo. K:
Pritchard. IB: Fernando. RA: Burka. LT: Hooth. MI. (2010). Exposure to hexavalent
chromium resulted in significantly higher tissue chromium burden compared with trivalent
chromium following similar oral doses to male F344/N rats and female B6C3F1 mice.
Toxicol Sci 118: 368-379. http://dx.doi.org/10.1093/toxsci/kfq263.
Crissman. TW: Goodman. DG: Hildebrandt. PK: Maronpot. RR: Prater. DA: Riley. TH: Seaman. WT:
Thake. DC. (2004). Bestpractices guideline: Toxicologic histopathology. Toxicol Pathol 32:
126-131. http://dx.doi.Org/10.1080/01926230490268756.
Crump. C: Crump. K: Hack. E: Luippold. R: Mundt. K: Liebig. E: Panko. I: Paustenbach. D: Proctor. D.
(2003). Dose-response and risk assessment of airborne hexavalent chromium and lung
cancer mortality. Risk Anal 23: 1147-1163. http://dx.doi.0rg/lO.llll /i.0272-
4332.2003.00388.x.
Culver. BH: Graham. BL: Coates. AL: Wanger. I: Berry. CE: Clarke. PK: Hallstrand. TS: Hankinson. TL:
Kaminskv. DA: Macintvre. NR: Mccormack. MC: Rosenfeld. M: Stanoievic. S: Weiner. DT:
Laboratories. ACoPSfPF. (2017). Recommendations for a Standardized Pulmonary Function
Report An Official American Thoracic Society Technical Statement Am J Respir Crit Care
Med 196: 1463-1472. http://dx.doi.org/10.1164/rccm.201710-1981ST.
Cummings. DE: Fendorf. S: Singh. N: Sani. RK: Peyton. BM: Magnuson. TS. (2007). Reduction of
Cr(VI) under acidic conditions by the facultative Fe(lll)-reducing bacterium Acidiphilium
cryptum. Environ Sci Technol 41: 146-152. http://dx.doi.org/10.1021/esQ61333k.
This document is a draft for review purposes only and does not constitute Agency policy.
110	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
De Flora. S: Camoirano. A: Bagnasco. M: Bennicelli. C: Corbett. GE: Kerger. BP. (1997). Estimates of
the chromium(VI) reducing capacity in human body compartments as a mechanism for
attenuating its potential toxicity and carcinogenicity. Carcinogenesis 18: 531-537.
EFSA (European Food Safety Authority). (2017). Guidance on the use of the weight of evidence
approach in scientific assessments. EFSA J15: 4971. http://dx.doi.org/10.2903 /
j.efsa.2017.4971.
Farland. WH. (2005). Implementation of the cancer guidelines and accompanying supplemental
guidance - Science Policy Council Cancer Guidelines Implementation Workgroup
communication I: Application of the mode of action framework in mutagenicity
determinations for carcinogenicity. Available online at
https://www.epa.gov/sites/production/files/2015-
01/documents/cgiwgcommuniation i.pdf (accessed
FDA. US. (2013). Beverages: Bottle water. Code of Federal Regulations: 21 CFR 165.110.
http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm.
Federal Register. (2010). Federal Register: March 29 2010 National Primary Drinking Water
Regulations Announcement of the Results of EPAs Review of Existing Drinking Water
Standards and Request for Public Comment and/or Information on Related Issues Notice.
(815Z10001). http://nepis.epa.gov/exe/ZyPURL.cgi?Dockey=Pl007TNX.txt.
Fendorf. S: Wielinga. BW: Hansel. CM. (2000). Chromium transformations in natural environments:
The role of biological and abiological. processes in chromium(VI) reduction. International
Geology Review 42: 691-701. http://dx.doi.Org/10.1080/00206810009465107.
Fendorf. SE. (1995). Surface reactions of chromium in soils and waters. Geoderma 67: 55-71.
http://dx.doi.org/10.1016/0016-706ir94100062-F.
Gibb. H: Lees. P: Pinskv. P: Roonev. B. (2000a). Lung cancer among workers in chromium chemical
production. Am J Ind Med 38: 115-126.
Gibb. HI: Lees. PS: Pinskv. PF: Roonev. BC. (2000b). Clinical findings of irritation among chromium
chemical production workers. Am J Ind Med 38: 127-131.
Gibb. HI: Lees. PS: Wang. 1: Grace O'Learv. K. (2015). Extended followup of a cohort of chromium
production workers. Am J Ind Med 58: 905-913. http://dx.doi.org/10.10Q2/aiim.22479.
Glaser. U: Hochrainer. D: Kloeppel. H: Kuhnen. H. (1985). Low level chromium (VI) inhalation
effects on alveolar macrophages and immune functions in Wistar rats. Arch Toxicol 57: 250-
256. http://dx.doi.org/10.1007/BF00324787.
Glaser. U: Hochrainer. D: Steinhoff. D. (1990). Investigation of irritating properties of inhaled Cr(VI)
with possible influence on its carcinogenic action. In NH Seemayer; W Hadnagy (Eds.),
Environmental hygiene II (pp. 239-245). Berlin, Germany: Springer-Verlag.
Guvatt. G: Oxman. AD: Akl. EA: Kunz. R: Vist. G: Brozek. 1: Norris. S: Falck-Ytter. Y: Glasziou. P:
deBeer. H: Taeschke. R: Rind. D: Meerpohl. I: Dahm. P: Schiinemann. HI. (2011). GRADE
guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin
Epidemiol 64: 383-394. http://dx.doi.Org/10.1016/i.iclinepi.2010.04.026.
Hausladen. DM: Fendorf. S. (2017). Hexavalent chromium generation within naturally structured
soils and sediments. Environ Sci Technol51: 2058-2067.
http://dx.doi.org/10.1021/acs.est.6b04039.
This document is a draft for review purposes only and does not constitute Agency policy.
Ill	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Health Canada. (2016). Guidelines for Canadian drinking water quality: Guideline technical
document: Chromium. In Guidelines for Canadian Drinking Water Quality. Ottawa, Ontario:
Minister of Health.
Higgins. TPT: Green. S. (2011). Cochrane handbook for systematic reviews of interventions, version
5.1.0 (Updated March 2011): The Cochrane Collaboration. Retrieved from
http://handbook.cochrane.org/
Hill. AB. (1965). The environment and disease: Association or causation? Proc R Soc Med 58: 295-
300.
Hirst. TA: Howick. 1: Aronson. IK: Roberts. N: Perera. R: Koshiaris. C: Heneghan. C. (2014). The need
for randomization in animal trials: an overview of systematic reviews [Review], PLoS ONE
9: e98856. http://dx.doi.org/10.1371/iournal.pone.0098856.
Hoenig. TM: Heisev. DM. (2001). The abuse of power: The pervasive fallacy of power calculations for
data analysis. Am Stat 55: 19-24.
Hruz. T: Laule. 0: Szabo. G: Wessendorp. F: Bleuler. S: Oertle. L: Widmaver. P: Gruissem. W:
Zimmermann. P. (2008). Genevestigator v3: A reference expression database for the meta-
analysis of transcriptomes. 2008: 420747. http://dx.doi.org/10.1155/2008/420747.
Huang. L: Yu. CH: Hopke. PK: Liov. PI: Buckley. BT: Shin. TY: Fan. ZT. (2014). Measurement of soluble
and total hexavalent chromium in the ambient airborne particles in New Jersey. Aerosol Air
Qual Res 14: 1939-1949. http://dx.doi.org/10.4209/aaqr.2013.10.0312.
IARC (International Agency for Research on Cancer). (2012). A review of human carcinogens. Part
C: Arsenic, metals, fibres, and dusts [IARC Monograph], Lyon, France: World Health
Organization. http://monographs.iarc.fr/ENG/Monographs/vollOOC/monolOOC.pdf.
ICF Consulting. (2018). DRAGON: An online tool for systematic review. Available online at
https://www.icf.com/resources/solutions-and-apps/dragon-online-tool-systematic-review
(accessed
Illumina. Inc... (2018). BaseSpace correlation engine [Computer Program], San Diego, CA. Retrieved
from https://www.illumina.com/products/by-type/informatics-products/basespace-
correlation-engine.html
IPCS (International Programme on Chemical Safety). (2013). Inorganic chromium(VI) compounds.
(78). Geneva, Switzerland: World Health Organization.
http://www.who.int/ipcs/publications/cicad/cicad 78.pdf.
IRIS (Integrated Risk Information System). (2015). Advancing Systematic Review Workshop:
Agenda.Advancing Systematic Review Workshop, December 16 - 17, 2015, Arlington, VA.
Ishikawa. Y: Nakagawa. K: Satoh. Y: Kitagawa. T: Sugano. H: Hirano. T: Tsuchiva. E. (1994a).
Characteristics of chromate workers' cancers, chromium lung deposition and precancerous
bronchial lesions: an autopsy study. Br J Cancer 70: 160-166.
Ishikawa. Y: Nakagawa. K: Satoh. Y: Kitagawa. T: Sugano. H: Hirano. T: Tsuchiva. E. (1994b). Hot
spots of chromium accumulation at bifurcations of chromate workers' bronchi. Cancer Res
54: 2342-2346.
Tackson. MA: Yang. L: Lea. I: Rashid. A: Kuo. B: Williams. A: Lvn Yauk. C: Fostel. 1. (2017). The TGx-
28.65 biomarker online application for analysis of transcriptomics data to identify DNA
damage-inducing chemicals in human cell cultures. Environ Mol Mutagen 58: 529-535.
http://dx.doi.org/10.1002/em.22114.
This document is a draft for review purposes only and does not constitute Agency policy.
112	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Tardine. PM: Mehlhorn. TL: Bailey. WB: Brooks. SC: Fendorf. S: Gentry. RW: Phelps. TT: Saiers. IE.
(2011).	Geochemical processes governing the fate and transport of chromium(III) and
chromium(VI) in soils. Vadose Zone J 10: 1058-1070.
http://dx.doi.org/10.2136/vzi2010.0102.
Kim. 1G: Dixon. IB. (20021. Oxidation and fate of chromium in soils. Soil Sci Plant Nutr 48: 483-490.
http://dx.doi.Org/10.1080/00380768.2002.10409230.
Kirman. CR: Avlward. LL: Suh. M: Harris. MA: Thompson. CM: Haws. LC: Proctor. DM: Lin. SS: Parker.
W: Hays. SM. (2013). Physiologically based pharmacokinetic model for humans orally
exposed to chromium. Chem Biol Interact 204: 13-27.
http://dx.doi.Org/10.1016/i.cbi.2013.04.003.
Kirman. CR: Hays. SM: Avlward. LL: Suh. M: Harris. MA: Thompson. CM: Haws. LC: Proctor. DM.
(2012).	Physiologically based pharmacokinetic model for rats and mice orally exposed to
chromium. Chem Biol Interact 200: 45-64. http://dx.doi.Org/10.1016/i.cbi.2012.08.016.
Kirman. CR: Suh. M: Hays. SM: Gurleytik. H: Gerads. R: De Flora. S: Parker. W: Lin. S: Haws. LC:
Harris. MA: Proctor. DM. (2016). Reduction of hexavalent chromium by fasted and fed
human gastric fluid. II. Ex vivo gastric reduction modeling. Toxicol Appl Pharmacol 306:
120-133. http://dx.doi.Org/10.1016/i.taap.2016.07.002.
Kirman. CR: Suh. M: Proctor. DM: Hays. SM. (2017). Improved physiologically based
pharmacokinetic model for oral exposures to chromium in mice, rats, and humans to
address temporal variation and sensitive populations. Toxicol Appl Pharmacol.
http://dx.doi.Org/10.1016/i.taap.2017.03.023.
Kondo. K: Takahashi. Y: Ishikawa. S: Uchihara. H: Hirose. Y: Yoshizawa. K: Tsuvuguchi. M: Takizawa.
H: Mivoshi. T: Sakivama. S: Monden. Y. (2003). Microscopic analysis of chromium
accumulation in the bronchi and lung of chromate workers. Cancer 98: 2420-2429.
http://dx.doi.org/10.1002/cncr.11818.
Krauth. D: Woodruff. TT: Bero. L. (2013). Instruments for assessing risk of bias and other
methodological criteria of published animal studies: a systematic review [Review], Environ
Health Perspect 121: 985-992. http://dx.doi.org/10.1289/ehp.1206389.
Krawic. A: Luczak. M: Zhitkovich. A. (2017). Variation in extracellular detoxification is a link to
different carcinogenicity among chromates in rodent and human lungs. Chem Res Toxicol
30: 1720-1729. http://dx.doi.org/10.1021/acs.chemrestox.7b00172.
Kuo. HW: Lai. IS: Lin. TI. (1997). Nasal septum lesions and lung function in workers exposed to
chromic acid in electroplating factories. Int Arch Occup Environ Health 70: 272-276.
Kupershmidt. I: Su. 01: Grewal. A: Sundaresh. S: Halperin. I: Flvnn. I: Shekar. M: Wang. H: Park. 1: Cui.
W: Wall. GD: Wisotzkev. R: Alag. S: Akhtari. S: Ronaghi. M. (2010). Ontology-based meta-
analysis of global collections of high-throughput public data. PLoS ONE 5.
http://dx.doi.org/10.1371/iournal.pone.0013066.
Lindberg. E: Hedenstierna. G. (1983). Chrome plating: Symptoms, findings in the upper airways, and
effects on lung function. Arch Environ Occup Health 38: 367-374.
Losi. ME: Amrhein. C: Frankenberger. WT. (1994). Environmental biochemistry of chromium
[Review], Rev Environ Contam Toxicol 136: 91-121.
Luippold. RS: Mundt. KA: Austin. RP: Liebig. E: Panko. 1: Crump. C: Crump. K: Proctor. D. (2003).
Lung cancer mortality among chromate production workers. Occup Environ Med 60: 451-
457.
This document is a draft for review purposes only and does not constitute Agency policy.
113	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
MacKenzie. RD: Bverrum. RU: Decker. CF: Hoppert. CA: Langham. RF. (1958). Chronic toxicity
studies. II. Hexavalent and trivalent chromium administered in drinking water to rats. AMA
Arch Ind Health 18: 232-234.
Macleod. MR. (2013). Systematic reviews of experimental animal studies. Presentation presented at
Workshop on weight of evidence; US National Research Council Committee to review the
Integrated Risk Information System (IRIS) process, March 27-28, 2013, Washington, D.C.
Malsch. PA: Proctor. DM: Finlev. BL. (1994). Estimation of a chromium inhalation reference
concentration using the benchmark dose method: a case study [Review], Regul Toxicol
Pharmacol 20: 58-82. http://dx.doi.org/10.1006/rtph.1994.1036.
Mancuso. TF. (1975). Consideration of chromium as an industrial carcinogen. In International
conference on heavy metals in the environment: v III, health; October. Toronto, ON, Canada:
Institute for Environmental Studies.
Mancuso. TF. (1997). Chromium as an industrial carcinogen: Parti. Am J Ind Med 31: 129-139.
http://dx.doi. org/10.1002 /CSTCIll 097-0274C199702131:2<129::ATD-ATTM1 >3.0.CQ:2-V.
McClain. CN: Fendorf. S: Webb. SM: Maher. K. (2017). Quantifying Cr(VI) production and export
from serpentine soil of the California coast range. Environ Sci Technol 51: 141-149.
http://dx.doi.org/10.1021/acs.est.6b03484.
McConnell. ER: Bell. SM: Cote. I: Wang. RL: Perkins. El: Garcia-Revero. N: Gong. P: Burgoon. LP.
(2014). Systematic Omics Analysis Review (SOAR) tool to support risk assessment PLoS
ONE 9: ell0379. http://dx.doi.org/10.1371/iournal.pone.0110379.
Miksche. LW: Lewalter. 1. (1997). Health surveillance and biological effect monitoring for
chromium-exposed workers [Review], Regul Toxicol Pharmacol 26: S94-S99.
http://dx.doi.org/10.1006/rtph.1997.1146.
Miller. MR: Hankinson. I: Brusasco. V: Burgos. F: Casaburi. R: Coates. A: Crapo. R: Enright. P: van Per
Grinten. CP: Gustafsson. P: Tensen. R: Tohnson. PC: Macintvre. N: Mckav. R: Navaias. P:
Pedersen. OF: Pellegrino. R: Viegi. G: Wanger. 1: Force. AET. (2005). Standardisation of
spirometry. EurRespir J 26: 319-338. http://dx.doi.org/10.1183 /09031936.05.00034805.
Molander. L: Agerstrand. M: Beronius. A: Hanberg. A: Ruden. C. (2015). Science in Risk Assessment
and Policy (SciRAP): An online resource for evaluating and reporting in vivo (eco) toxicity
studies. Hum Ecol Risk Assess 21: 753-762.
Morgan. RL: Thayer. KA: Bero. L: Bruce. N: Falck-Ytter. Y: Ghersi. P: Guvatt. G: Hooiimans. C:
Langendam. M: Mandrioli. P: Mustafa. RA: Rehfuess. EA: Roonev. AA: Shea. B: Silbergeld. EK:
Sutton. P: Wolfe. MS: Woodruff. TT: Verbeek. TH: Hollowav. AC: Santesso. N: Schiinemann. HI.
(2016). GRAPE: Assessing the quality of evidence in environmental and occupational health.
Environ Int 92-93: 611-616. http://dx.doi.Org/10.1016/i.envint.2016.01.004.
NASEM (National Academies of Sciences, Engineering, and Medicine). (2018). Progress toward
transforming the Integrated Risk Information System (IRIS) Program. A 2018 evaluation
(2018). Washington, P.C.: The National Academies Press.
http://dx.doi.org/10.17226/25086.
Ng. AW: Bidani. A: Heming. TA. (2004). Innate host defense of the lung: effects of lung-lining fluid
pH [Review], Lung 182: 297-317. http://dx.doi.org/10.1007/s00408-004-2511 -6.
NIOSH (National Institute for Occupational Safety and Health). (2013a). Criteria for a recommended
standard: Occupational exposure to hexavalent chromium. (PHHS (NIOSH) Publication
This document is a draft for review purposes only and does not constitute Agency policy.
114	PRAFT-PO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Number 2013-128). Centers for Disease Control and Prevention.
https: / /www.cdc. gov/niosh /docs /2 013-128/.
NIOSH (National Institute for Occupational Safety and Health). (2013b). Occupational exposure to
hexavalent chromium. (DHHS (NIOSH) Publication No. 2013128). Department of Health and
Human Services, Centers for Disease Control and Prevention.
http: / /www. cdc. gov /niosh /docs /2 013-12 8/pdfs/2 013 12 8.pdf.
NT PEP (New Jersey Department of Environmental Protection). (2009). Derivation of ingestion-
based soil remediation criterion for Cr+6 based on the NTP chronic bioassay data for
Sodium Dichromate Dihydrate. http://www.state.ni.us/dep/dsr/chromium/soil-cleanup-
derivation.pdf.
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.
NTP (National Toxicology Program). (2008). Toxicology and carcinogenesis studies of sodium
dichromate dihydrate (Cas No. 7789-12-0) in F344/N rats and B6C3F1 mice (drinking
water studies) (pp. 1-192). (NTP TR 546; NIH Publication No. 08-5887).
NTP (National Toxicology Program). (2011). Report on carcinogens: Twelfth edition (12th ed.).
Research Triangle Park, NC: U.S. Department of Health and Human Services, Public Health
Service, National Institutes of Health, http: //ntp.niehs.nih.gov/ntp/roc/twelfth/rocl2.pdf.
NTP (National Toxicology Program). (2015). Handbook for conducting a literature-based health
assessment using 0HAT approach for systematic review and evidence integration. U.S. Dept
of Health and Human Services, National Toxicology Program.
https://ntp.niehs.nih.gov/ntp/ohat/pubs/handbookian2015 508.pdf.
O'Flahertv. El. (1993). A pharmacokinetic model for chromium. Toxicol Lett 68: 145-158.
http: / /dx.doi.org/10.1016/0378-4274C93190127-1.
O'Flahertv. EI. (1996). A physiologically based model of chromium kinetics in the rat. Toxicol Appl
Pharmacol 138: 54-64. http: //dx.doi.org/10.1006/taap.1996.0Q97.
O'Flahertv. EI: Kerger. BP: Hays. SM: Paustenbach. DT. (2001). A physiologically based model for the
ingestion of chromium(III) and chromium(VI) by humans. Toxicol Sci 60: 196-213.
O'Flahertv. El: Radike. MI. (1991). Pharmacokinetic modeling of trivalent and hexavalent chromium
based on ingestion and inhalation of soluble chromium compounds. (AL-TR-1991-0162).
Wright-Patterson AFB, Ohio: Armstrong Laboratory.
https: //ntrl. ntis.gov/NTRL/dash board/sea rchResults.xhtml?searchOuerv=AD A2 5 62 3 8.
0A0PS (EPA Office of Air Quality Planning and Standards). (2012). Residual risk assessment for the
chromic acid anodizing, decorative chromium electroplating, and hard chromium
electroplating source categories. Research Triangle Park, NC: U.S. Environmental Protection
Agency.
OEHHA (California Office of Environmental Health Hazard Assessment). (2008). TSP for noncancer
RELs - Appendix P.3 Chronic RELs and toxicity summaries using the previous version of the
Hot Spots Risk Assessment guidelines (OEHHA 1999). Sacramento, CA: California
Environmental Protection Agency.
http://oehha.ca.gov/media/downloads/crnr/appendixd3final.pdf.
Oregon PEO (Oregon Pepartment of Environmental Quality). (2016). Portland Metals air
monitoring data. Portland, OR.
This document is a draft for review purposes only and does not constitute Agency policy.
115	PRAFT-PO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
OSHA (Occupational Safety & Health Administration). (2006). Occupational exposure to hexavalent
chromium. Final rule. Fed Reg 71: 10099-10385.
Oze. C: Bird. DK: Fendorf. S. (2007). Genesis of hexavalent chromium from natural sources in soil
and groundwater. Proc Natl Acad Sci USA 104: 6544-6549.
http://dx.doi.Org/10.1073/pnas.0701085104.
Oze. C: Fendorf. S: Bird. DK: Coleman. RG. (2004). Chrominium geochemistry of serpentine soils.
International Geology Review 46: 97-126. http://dx.doi.Org/10.2747/0020-6814.46.2.97.
Park. RM: Bena. IF: Stavner. LT: Smith. RT: Gibb. HI: Lees. PS. (2004). Hexavalent chromium and lung
cancer in the chromate industry: a quantitative risk assessment. Risk Anal 24: 1099-1108.
http://dx.doi.Org/10.llll/i.0272-4332.2004.00512.x.
Proctor. DM: Suh. M: Avlward. LL: Kirman. CR: Harris. MA: Thompson. CM: Gurleytik. H: Gerads. R:
Haws. LC: Hays. SM. (2012). Hexavalent chromium reduction kinetics in rodent stomach
contents. Chemosphere 89: 487-493.
http://dx.doi.Org/10.1016/i.chemosphere.2012.04.065.
Proctor. DM: Suh. M: Mittal. L. iz: Hirsch. S: Salgado. RV: Bartlett. C: Van Landingham. C: Rohr. A:
Crump. K. (2016). Inhalation cancer risk assessment of hexavalent chromium based on
updated mortality for Painesville chromate production workers. J Expo Sci Environ
Epidemiol 26: 224-231. http://dx.doi.org/10.1038/ies.2015.77.
Oiagen (Qiagen Bioinformatics). (2018). Ingenuity Pathway Analysis [Computer Program],
Redwoord City, CA. Retrieved from
https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/
RIVM (National Institute for Public Health and the Environment (Netherlands)). (2001). Re-
evaluation of human-toxicological maximum permissible risk levels (pp. 58-61). (RIVM
report 711701 025). Bilthoven, the Netherlands: Rijksinstituutvoor Volksgezondheid en
Milieu RIVM. http://www.rivm.nl/bibliotheek/rapporten/711701025.html.
Sasso. AF: Schlosser. PM. (2015). An evaluation of in vivo models for toxicokinetics of hexavalent
chromium in the stomach. Toxicol Appl Pharmacol 287: 293-298.
http://dx.doi.Org/10.1016/i.taap.2015.06.016.
Schlosser. PM: Sasso. AF. (2014). A revised model of ex-vivo reduction of hexavalent chromium in
human and rodent gastric juices. Toxicol Appl Pharmacol 280: 352-361.
http://dx.doi.Org/10.1016/i.taap.2014.08.010.
Schiinemann. H: Hill. S: Guvatt. G: Akl. EA: Ahmed. F. (2011). The GRADE approach and Bradford
Hill's criteria for causation. J Epidemiol Community Health 65: 392-395.
http://dx.doi.org/10.1136/iech.2010.119933.
Sciome (Sciome, LLC.). (2018). BMDExpress 2 software (Version 44) [Computer Program],
Research Triangle Park, NC: National Toxicology Program. Retrieved from
https://www.sciome.com/bmdexpress/
Shaw Environmental (Shaw Environmental, Inc). (2006). Final report Industry profile, exposure
profile, technological feasibility evaluation, and environmental impact for industries
affected by a revised OSHA standard for hexavalent chromium (General industry,
construction, and maritime sectors). Washington, DC: U.S. Department of Labor,
Occupational Safety and Health Administration.
Smith. MT: Guvton. KZ: Gibbons. CF: Fritz. TM: Portier. CI: Rusvn. I: DeMarini. DM: Caldwell. TC:
Kavlock. RT: Lambert. P: Hecht. SS: Bucher. TR: Stewart. BW: Baan. R: Cogliano. VI: Straif. K.
This document is a draft for review purposes only and does not constitute Agency policy.
116	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
(2016). Key characteristics of carcinogens as a basis for organizing data on mechanisms of
carcinogenesis [Review], Environ Health Perspectl24: 713-721.
http://dx.doi.org/10.1289/ehp.1509912.
Sterne. T: Higgins. T: Reeves. B. (2016). R0BINS-I: a tool for assessing risk of bias in non-randomized
studies of interventions, Version 7 March 2016. Br Med J 355: i4919.
Subramanian. A: Tamavo. P: Mootha. VK: Mukheriee. S: Ebert. BL: Gillette. MA: Paulovich. A:
Pomerov. SL: Golub. TR: Lander. ES: Mesirov. TP. (2005). Gene set enrichment analysis: a
knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl
Acad Sci USA 102: 15545-15550. http://dx.doi.org/10.1073/pnas.05065801Q2.
SWRCB. Title 22. Division 4. Chapter 14. Article 3: Amendments. (2014).
TCEO (Texas Commission on Environmental Quality). (2014). Development support document for
hexavalent chromium (Particulate compounds). Austin, TX.
https://www.tceq.texas.gov/assets/public/implementation/tox/dsd/final/august2014/he
xavalent chromium.pdf.
TCEO (Texas Commission on Environmental Quality). (2016). Development support document for
hexavalent chromium oral reference dose. Austin, TX.
https://www.tceq.texas.gov/assets/public/implementation/tox/dsd/final/chromium ord.
pdf.
U.S. EPA (U.S. Environmental Protection Agency). (1988). Recommendations for and documentation
of biological values for use in risk assessment (pp. 1-395). (EPA/600/6-87/008). Cincinnati,
OH: U.S. Environmental Protection Agency, Office of Research and Development, Office of
Health and Environmental Assessment
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=34855.
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=23162.
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). (1998a). 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
This document is a draft for review purposes only and does not constitute Agency policy.
117	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
U.S. EPA (U.S. Environmental Protection Agency). (1998b). Toxicological review of hexavalent
chromium [EPA Report] (pp. 80). (630R98009). Washington, DC.
http://www.epa.gov/ncea/iris/toxreviews/0144-tr.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2002). A review of the reference dose and
reference concentration processes (pp. 1-192). (EPA/630/P-02/002F). Washington, DC:
U.S. Environmental Protection Agency, Risk Assessment Forum.
http://www.epa.gov/osa/review-reference-dose-and-reference-concentration-processes.
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.
U.S. EPA (U.S. Environmental Protection Agency). (2005a). Guidelines for carcinogen risk
assessment [EPAReport] (pp. 1-166). (EPA/630/P-03/001F). Washington, DC: U.S.
Environmental Protection Agency, Risk Assessment Forum.
http://www2.epa.gov/osa/guidelines-carcinogen-risk-assessment.
U.S. EPA (U.S. Environmental Protection Agency). (2005b). Supplemental guidance for assessing
susceptibility from early-life exposure to carcinogens [EPAReport], (EPA/630/R-03/003F).
Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
https://www3.epa.gov/airtoxics/childrens supplement final.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2008a). Cr (VI) quantitative risk assessment Q*1
based on mouse carcinogenicity studies. Washington, DC: Health Effects Division, Office of
Pesticide Programs. http://www.regulations.gov/#!documentDetail:D=EPA-HO-OPP-20Q3-
0250-0088.
U.S. EPA (U.S. Environmental Protection Agency). (2008b). Evaluation of the carcinogenic potential
of inorganic hexavalent chromium (Cr(VI)). Washington, DC: Health Effects Division, Office
of Pesticide Programs. http: //www.regulations.gov/#!documentDetail:D=EPA-HO-OPP-
2003-0250-0089.
U.S. EPA (U.S. Environmental Protection Agency). (2008c). Reregistration eligibility decision for
chromated arsenicals. (EPA 739-R-08-006). Office of Prevention, Pesticides And Toxic
Substances; U.S. Environmental Protection Agency.
U.S. EPA (U.S. Environmental Protection Agency). (2010). Toxicological review of hexavalent
chromium (external review draft). (EPA/635/R-10/004A). Washington, DC.
http://cfpub.epa.gov/ncea/iris drafts/recordisplay.cfm?deid=221433.
U.S. EPA (U.S. Environmental Protection Agency). (2011a). Recommended use of body weight 3/4
as the default method in derivation of the oral reference dose (pp. 1-50).
(EPA/100/R11/0001). Washington, DC: U.S. Environmental Protection Agency, Risk
Assessment Forum, Office of the Science Advisor.
https://www.epa.gov/risk/recommended-use-body-weight-34-default-method-derivation-
oral-reference-dose.
U.S. EPA (U.S. Environmental Protection Agency). (2011b). Toxicological review of
trichloroethylene (CASRN 79-01-6) in support of summary information on the Integrated
Risk Information System (IRIS) [EPAReport], (EPA/635/R-09/011F). Washington, DC.
https://cfpub.epa.gov/ncea/iris/iris documents/documents/toxreviews/0106tr.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2012a). Advances in inhalation gas dosimetry for
derivation of a reference concentration (RfC) and use in risk assessment (pp. 1-140).
This document is a draft for review purposes only and does not constitute Agency policy.
118	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
(EPA/600/R-12/044). Washington, DC.
https://cfpub.epa. gov/ncea/risk/recordisplay.cfm?deid=244650&.CFID=50524762&.CFTOK
EN=17139189.
U.S. EPA (U.S. Environmental Protection Agency). (2012b). 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). (2013). Scientific workshop: Factors affecting the
reduction and absorption of hexavalent chromium in the gastrointestinal (GI) tract:
Potential impact on evaluating the carcinogenicity of ingested hexavalent chromium. Paper
presented at Hexavalent Chromium Workshop, September 19 & 24, 2018, Crystal City, VA.
U.S. EPA (U.S. Environmental Protection Agency). (2014a). 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). (2014b). Preliminary materials for the Integrated
Risk Information System (IRIS) toxicological review of hexavalent chromium part 1:
Experimental animal studies [CASRN) 18540-29-9] [EPA Report], (EPA/635/R-14/094).
Research Triangle Park, NC. https: //www.epa.gov/iris/iris-bimonthly-public-meeting-iun-
2014.
U.S. EPA (U.S. Environmental Protection Agency). (2014c). Preliminary materials for the Integrated
Risk Information System (IRIS) toxicological review of hexavalent chromium part 2: Human,
toxicokinetic, and mechanistic Studies [CASRN) 18540-29-9] [EPA Report], (EPA/635/R-
14/218). Research Triangle Park, NC. https: //www.epa.gov/iris/iris-bimonthly-public-
meeting-oct-2014.
U.S. EPA (U.S. Environmental Protection Agency). (2014d). The third Unregulated Contaminant
Monitoring Rule (UCMR 3) occurrence data. Retrieved from
http://water.epa.gOv/lawsregs/rulesregs/sdwa/ucmr/data.cfm#ucmr2013
U.S. EPA (U.S. Environmental Protection Agency). (2016). US EPAs report on the environment
(ROE). Available online at https://cfpub.epa.gov/roe/index.cfm (accessed
U.S. EPA (U.S. Environmental Protection Agency). (2017). Guidance to assist interested persons in
developing and submitting draft risk evaluations under the Toxic Substances Control Act
(EPA/740/R17/001). Washington, DC: Office of Chemical Safety and Pollution Prevention.
https://www.epa.gov/sites/production/files/2017-
06/documents/tsca ra guidance final.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2018a). Application of systematic review in TSCA
risk evaluations. (740-P1-8001). Washington, D.C.: U.S. Environmental Protection Agency,
Office of Chemical Safety and Pollution Prevention.
https://www.epa.gov/sites/production/files/2018-
06/documents/final application of sr in tsca 05-31-18.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2018b). An umbrella Quality Assurance Project
Plan (QAPP) for PBPK Models [EPA Report], (ORD QAPP ID No: B-0030740-QP-1-1).
Research Triangle Park.
USGS (U.S. Geological Survey). (1995). Chromium life cycle study. (IC 9411). Washington, D.C.: U.S.
Department of the Interior, Bureau of Mines.
This document is a draft for review purposes only and does not constitute Agency policy.
119	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Varghese. A: Cawlev. M: Hong. T. (2017). Supervised clustering for automated document
classification and prioritization: A case study using toxicological abstracts. Environ Syst
Decis. http://dx.doi.Org/10.1007/sl0669-017-9670-5.
Vesterinen. HM: Sena. ES: Egan. KT: Hirst. TC: Churolov. L: Currie. GL: Antonic. A: Howells. DW:
Macleod. MR. (2014). Meta-analysis of data from animal studies: a practical guide. J
Neurosci Methods 221: 92-102. http://dx.doi.org/10.1016/i.ineumeth.2013.09.010.
WHO (World Health Organization). (1999). WHO laboratory manual for the examination of human
semen and sperm-cervical mucus interaction. In WHO laboratory manual for the
examination of human semen and sperm-cervical mucus interaction (4th ed.). Cambridge,
UK: Cambridge University Press.
WHO (World Health Organization). (2000). Air quality guidelines for Europe (2nd ed.). Copenhagen,
Denmark: World Health Organization, Regional Office for Europe.
http://www.euro.who.int/en/health-topics/environment-and-health/air-
qualitv/publications/pre2009/air-quality-guidelines-for-europe.
WHO (World Health Organization). (2003). Chromium in drinking water.
(WHO/SDE/WSH/03.04/04). Geneva, Switzerland.
http://www.who.int/water sanitation health/dwq/chemicals/chromium.pdf.
WHO (World Health Organization). (2010). WHO laboratory manual for the examination and
processing of human semen [WHO EHC], In WHO laboratory manual for the examination
and processing of human semen (5 th ed.). Geneva, Switzerland.
This document is a draft for review purposes only and does not constitute Agency policy.
120	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
APPENDICES
APPENDIX A. ELECTRONIC DATABASE SEARCH STRATEGIES
Table A-l. Literature search query strings for computerized databases
Database
search date
Terms
PubMed
(1/29/2013)
(7/19/2013)
(2/5/2014)
(4/1/2015)
(4/1/2016)
(5/24/2017)
(5/24/2018)
hexavalent chromium OR (hexavalent AND chromium) OR CRVI OR CR VI OR Chromium VI OR
"Chromic acid" OR "Calcium chromate" OR "Potassium dichromate" OR "Potassium chromate"
OR "Sodium chromate" OR "lead chromate" OR "zinc chromate" OR "strontium chromate" OR
"ammonium dichromate" OR 13765-19-0[RN] OR 1333-82-0[RN] OR 7789-00-6[RN] OR
7778-50-9[RN] OR 7775-ll-3[RN] OR 7789-12-0[RN] OR 13530-65-9[RN] OR 7738-94-5[rn] OR
18540-29-9[rn] OR 7758-97-6[RN] OR 11119-70-3[rn] OR 11103-86-9[rn] OR 13530-65-9[rn] OR
7788-98-9[rn] OR 77898-09-5[rn] OR 7789-06-2[rn]
Web of
Science
(1/29/2013)
(7/19/2013)
(2/5/2014)
(4/1/2015)
(4/1/2016)
(5/24/2017)
(5/24/2018)
Topic = (hexavalent chromium OR (hexavalent AND chromium) Chromium VI OR CrVI OR Cr VI OR
"Chromic acid" OR "Calcium chromate" OR "Chromic trioxide" OR "Potassium dichromate" OR
"Potassium chromate" OR "Sodium chromate" OR "Sodium dichromate dehydrate" OR "lead
chromate" OR "zinc chromate" OR "strontium chromate" OR "ammonium dichromate" OR
"ammonium chromate" OR 13765-19-0 OR 1333-82-0 OR 7789-00-6 OR 7778-50-9 OR 7775-11-3
OR 7789-12-0 OR 13530-65-9 OR 7738-94-5 OR 18540-29-9 OR 7758-97-6 OR 11119-70-3 OR
11103-86-9 OR 13530-65-9 OR 7788-98-9 OR 77898-09-5 OR 7789-06-2)
AND
Research Areas = Toxicology, Biochemistry molecular biology, Public environmental occupational
health, Dermatology, Cell biology, Oncology, Life sciences biomedicine other topics, Allergy,
Veterinary sciences, Developmental biology, Immunology, Reproductive biology, Pathology,
Physiology, Urology nephrology, Hematology, Neurosciences neurology, Respiratory system,
Cardiovascular system cardiology, Obstetrics gynecology, Infectious diseases, Gastroenterology
hepatology, Microscopy
This document is a draft for review purposes only and does not constitute Agency policy.
121	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Database
search date
Terms
Web of
Science
(1/29/2013)
(7/19/2013)
(2/5/2014)
(4/1/2015)
(12/1/2017)
(5/24/2017)
(5/24/2018)
Topic = (hexavalent chromium OR (hexavalent AND chromium) Chromium VI OR CrVI OR Cr VI OR
"Chromic acid" OR "Calcium chromate" OR "Chromic trioxide" OR "Potassium dichromate" OR
"Potassium chromate" OR "Sodium chromate" OR "Sodium dichromate dehydrate" OR "lead
chromate" OR "zinc chromate" OR "strontium chromate" OR "ammonium dichromate" OR
"ammonium chromate" OR 13765-19-0 OR 1333-82-0 OR 7789-00-6 OR 7778-50-9 OR 7775-11-3
OR 7789-12-0 OR 13530-65-9 OR 7738-94-5 OR 18540-29-9 OR 7758-97-6 OR 11119-70-3 OR
11103-86-9 OR 13530-65-9 OR 7788-98-9 OR 77898-09-5 OR 7789-06-2)
AND
Research Areas = Chemistry, Environmental sciences ecology, Spectroscopy, Pharmacology
pharmacy, Water resources, Genetics heredity, Science technology other topics, Biophysics, Food
sciences technology, Endocrinology metabolism, Research experimental medicine, Nutrition
dietetics, Zoology, General internal medicine, Construction building technology, Parasitology,
Medical laboratory technology, Education educational research, Otorhinolaryngology,
Rheumatology, Anatomy morphology, Emergency medicine, Mycology, Sport sciences, Psychiatry
AND
cancer* OR carcinogen* OR chronic OR subchronic OR genotox* OR inhalation absorption OR
oral absorption OR mice OR mouse OR Mutagenicity OR pharmacokinetic
OR rat OR rats OR toxic* NOT (fish OR bacteria* OR microorganism* OR plant*) OR tumor*
Toxline
(includes
TSCATS)
(1/29/2013)
(7/19/2013)
(2/5/2014)
(4/1/2015)
(4/1/2016)
(5/24/2017)
(5/24/2018)
18540-29-9 OR 7789-09-5 OR 13765-19-0 OR 1333-82-0 OR 7758-97-6 OR 7789-00-6 OR
7778-50-9 OR 7775-11-3 OR 7789-12-0 OR 7789-06-2 OR 13530-65-9 OR 7788-98-9 OR 7738-94-5
OR 13530-68-2
TSCATS2
(1/29/2013)
(7/19/2013)
(2/5/2014)
(4/1/2015)
(4/1/2016)
(5/24/2017)
(5/24/2018)
18540-29-9
This document is a draft for review purposes only and does not constitute Agency policy.
122	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Database
search date
Terms
Combined
reference set
(duplicates eliminated through electronic screen)
Toxline = Toxicology Literature Online; TSCATS2 = Toxic Substances Control Act Test Submissions 2.0.
This document is a draft for review purposes only and does not constitute Agency policy.
123	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Table A-2. Processes used to augment the search of core computerized
databases for Cr(VI)
System
used
Selected key reference(s) or sources
Date
Additional
references
identified
Manual
search of
citations
from health
assessment
documents
ATSDR (Agencv for Toxic Substances and Disease Registrv). (2012).
Toxicological profile for chromium. Atlanta, GA: U.S. Department of Health
and Human Services, Public Health Service.
httD://www.atsdr.cdc.gov/ToxProfiles/tD.asD?id=62&tid=17.
1/2013
40 citations
added

U.S. EPA (U.S. Environmental Protection Agencv). (2010). Toxicological
review of hexavalent chromium (external review draft). (EPA/635/R-
10/004A). Washington, DC.
http://cfpub.eoa.gov/ncea/iris drafts/recordisplav.cfm?deid=221433.
1/2013
59 citations
added

OSHA (Occupational Safetv & Health Administration). (2006). Occupational
exposure to hexavalent chromium. Final rule. Fed Reg 71: 10099-10385.
5/2014
3 citations
added
IPCS (International Programme on Chemical Safetv). (2013). Inorganic
chromium (VI) compounds. (78). Geneva, Switzerland: World Health
Organization, http://www.who.int/ipcs/publications/cicad/cicad_78.pdf.
5/2014
5 citations
added
NIOSH (National Institute for Occupational Safetv and Health). (2013b).
Occupational exposure to hexavalent chromium. (DHHS [NIOSH]
Publication No. 2013128). Department of Health and Human Services,
Centers for Disease Control and Prevention.
http://www.cdc.gov/niosh/docs/2013-128/pdfs/2013_128.pdf.
5/2014
1 citation
added
References
obtained
during the
assessment
process
Snowball search
1/2013,
Ongoing

Search of
online
chemical
assessment-
related
websites
Combination of Chemical Abstracts Service registry number (CASRN) and
synonyms searched on the following websites:
•	American Conference of Governmental Industrial Hygienists (ACGIH)
(http://www.acgih.org)
•	American Industrial Hygiene Association Workplace Environmental
Exposure Levels (AIHA WEELs)
(http://www.tera.org/OARS/WEELhtml)
•	Agency for Toxic Substances and Disease Registry (ATSDR)
(http://www.atsdr.cdc.gov/substances/index.asp)
•	California Environmental Protection Agency (CalEPA) Office of
Environmental Health Hazard Assessment (OEHHA)
(http://www.oehha.ca.gov/risk.html)
o OEHHA Toxicity Criteria Database
(http://www.oehha.ca.gov/tcdb/index.asp)


This document is a draft for review purposes only and does not constitute Agency policy.
124	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
System
used
Selected key reference(s) or sources
Date
Additional
references
identified

o Biomonitoring California-Priority Chemicals
(httDs://biomonitoring. ca.gov/chemicals/Drioritv-chemicals)
o Biomonitoring California-Designated Chemicals
(httDs://biomonitoring. ca.gov/chemicals/designated-chemicals)
o Cal/Ecotox Database
(https://oehha.ca.gov/ecotoxicologv/general-info/calecotox-
database)
o OEHHA fact sheets
(http://www.oehha.ca.gov/Dublic info/facts/index.html)
o Noncancer health effects table (reference exposure levels [RELs]:
http://www.oehha.ca.gov/air/allrels.html)
o Cancer Potency Factors (see Appendix A and Appendix B;
httD://www.oehha.ca.gov/air/hot SDOts/tsd052909.html)
• CalEPA Drinking Water Notification Levels
(http://www.swrcb.ca.gov/drinking water/certlic/drinkingwater/Not


ification Levels, shtml)
•	Chemical Risk Information Platform (CHRIP)
(http://www.safe.nite.go.ip/english/db.html)
•	Consumer Product Safety Commission (CPSC) (http://www.cpsc.gov)
•	European Centre for Ecotoxicology and Toxicology of Chemicals
(ECETOC) publications (http://www.ecetoc.org/publications)
•	European Chemicals Agency (ECHA); general site
(http://echa.europa.eu/information-on-chemicals)
•	ECHA info on Registered Substances
(http://echa.europa.eu/information-on-chemicals/registered-
substances)
•	ECHA Information from the Existing Substances Regulation (ESR)
(http://echa.europa.eu/information-on-chemicals/information-from-
existing-substances-regulation)
• eChemPortal [participating databases: Aggregated Computational
Toxicology Resource (ACToR), AGRITOX, Canadian Categorization
Results (CCR), CCR DATA, Canada's Existing Substances Assessment
Repository (CESAR), CHRIP, ECHA CHEM, Data Bank of Environmental
Properties of Chemicals (EnviChem), European chemical Substances
Information System (ESIS), Globally Harmonized System-Japan (GHS-
J), High Production Volume Information System (HPVIS), Hazardous
Substances Data Bank (HSDB), Hazardous Substances and New
Organisms Chemical Classification Information Database (HSNO
CCID), INCHEM, Japan CHEmicals Collaborative Knowledge (J-CHECK),
JECDB, NICNAS PEC, OECD HPV, OECD SIDS IUCLID, UNEP SIDS, United
Kingdom (UK) Coordinated Chemicals Risk Management Programme
Publications (CCRMP) Outputs, US EPA IRIS, US EPA Substance
Registry Services (SRS)
This document is a draft for review purposes only and does not constitute Agency policy.
125	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
System
used
Selected key reference(s) or sources
Date
Additional
references
identified

(http://www. echemDortal.org/echemDortal/DarticiDant/Dage. action?


oagelD=9)l
•	Environment Canada—search entire site
(httD://www.ec.gc.ca/default.asD?lang=En&n=ECD35C36) if not
found below:
o Toxic substances managed under Canadian Environmental
Protection Act (CEPA) (htto://www.ec.gc.ca/toxiques-
toxics/Default.asD?lang=En&n=98E80CC6-l) search results
o Final assessments (httoV/www.ec.gc.ca/lcoe-
ceDa/default.asD?lang=En&xml=09F567A7-BlEE-lFEE-73DB-
8AE6C1EB7658)
o Draft assessments (httoV/www.ec.gc.ca/lcoe-
ceoa/default.asD?lang=En&xml=6892C255-5597-C162-95FC-
4B905320F8C9)
•	EPA Chemical Data Access Tool (CDAT)
(httD://iava.eDa.gov/oDDt chemical search/)
•	EPA Acute Exposure Guideline Levels
(httD://www. eDa.gov/oDDt/aegl/Dubs/chemlist. htm)
•	EPA National Service Center for Environmental Publications (NSCEP)
(htto://www. eoa.gov/nceDihom/)
•	EPA Office of Pesticide Programs (OPP)
(httD://iasDub.eDa.gov/aDex/Desticides/f?D=chemicalsearch:l)
•	EPA Science Inventory (htto://cfoub.eoa.gov/si/)
•	Emergency Response Planning Guidelines (ERPGs)
(httos://www.aiha.org/get-
involved/AIHAGuidelineFoundation/EmergencvResoonsePlanningGui
delines/Pages/default.asox)
•	Food and Drug Administration (FDA) (htto://www.fda.gov/)
•	Federal Docket (www.regulations.gov)
•	Health Canada—search entire site (httD://www.hc-sc.gc.ca/index-
eng.Dho)
•	Health Canada Drinking Water Documents (htto://www.hc-
sc.gc.ca/ewh-semt/Dubs/water-eau/index-eng.DhD#tech doc)
•	Health Canada First Priority List Assessments (htto://www.hc-
sc.gc.ca/ewh-semt/Dubs/contaminants/osll-lsDl/index-eng.DhD)
•	Health Canada Second Priority List Assessments (htto://www.hc-
sc.gc.ca/ewh-semt/Dubs/contaminants/Dsl2-lsD2/index-eng.DhD)
•	International Agency for Research on Cancer (IARC) Monographs:
(httDs://monograohs. iarc.fr/agents-classified-bv-the-iarc)
•	IRISTrack/new assessments and reviews
(httD://cfoub. eoa.gov/ncea/iris/search/)
This document is a draft for review purposes only and does not constitute Agency policy.
126	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
System
used
Selected key reference(s) or sources
Date
Additional
references
identified

•	Japan Existing Chemical Data Base (JECDB)
(http://dra4.nihs.go.ip/mhlw data/isp/SearchPageENG.isp)
•	National Academies Press (NAP)—search site (http://www.nao.edu/)
•	National Cancer Institute (NCI) (http://www.cancer.gov)
•	National Center for Toxicological Research (NCTR)
(http://www.fda.gov/AboutFDA/CentersOffices/OC/OfficeofScientific


and Medical Programs/NCTR/defau It. htm)
•	National Industrial Chemicals Notification and Assessment Scheme
(NICNAS); priority existing chemical (PEC) only covered by
eChemPortal (http://www.nicnas.gov.au)
•	National Institute for Environmental Health Sciences (NIEHS)
(http://www.niehs.nih.gov/)
•	National Institute for Occupational Safety and Health (NIOSH)
(http://www.cdc.gov/niosh/topics/)
•	National Institute for Occupational Safety and Health Technical
Information Center (NIOSHTIC) 2 (http://www2a.cdc.gov/nioshtic-2/)
•	National Toxicology Program (NTP)—Report on Carcinogens (RoC),
status, results, and management reports
o RoC (12th-14th): (https://ntp.niehs.nih.gov/pubhealth/roc/index-
l.html)
o NTP site search:
(http://ntpsearch.niehs.nih.gov/texis/search/?auerv=arsenic&pr=
ntp web entire site all&mu=Entire+NTP+Site)
•	Organisation for Economic Cooperation and Development (OECD)
high production volume (HPV)/Screening Information Data Set
(SIDS)/lnternational Uniform Chemical Information Database (IUCLID)
(cross-check with eChem;
http://webnet.oecd.org/hpv/ui/Search.aspx)
•	Occupational Safety and Health Administration (OSHA)
(http://www.osha.gov/dts/chemicalsampling/toc/toc chemsamp.ht
ml)
•	Registry of Toxic Effects of Chemical Substances (RTECS)
(http://www.ccohs.ca/search.html)
•	United Nations Environment Programme (UNEP) SIDS (through 2007;
http://www.inchem.org/pages/sids.html)
This document is a draft for review purposes only and does not constitute Agency policy.
127	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
APPENDIX B. TYPICAL DATA EXTRACTION FIELDS
Table B-l. Key data extraction elements to summarize study design,
experimental model, methodology, and results
Field label
Data extraction elements
HUMAN
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, etc.)

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 and CAS number

Exposure assessment (e.g., blood, urine, hair, air, drinking water, job classification, residence,
administered treatment in controlled study, etc.)

Methodological details for exposure assessment (e.g., HPLC-MS/MS, limit of detection)

Statistical methods
This document is a draft for review purposes only and does not constitute Agency policy.
128	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Field label
Data extraction elements
Results
Exposure levels (e.g., mean, median, measures of variance as presented in paper, such as
standard deviation (SD), standard error of the mean (SEM), 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, etc.) or description of qualitative results. When
possible, convert measures of effect to a common metric with associated 95% confidence
intervals (CI). 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 (RR, 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.
ANIMAL
Funding
Funding source(s)

Reporting of conflict of interest by authors
Animal model
Sex

Species

Strain

Source of animals

Age or life stage at start of dosing and at health outcome assessment

Diet and husbandry information (e.g., diet name/source)
Treatment
Chemical name and CAS number

Source of chemical

Purity of chemical

Dose levels or concentration (as presented and converted to mg/kg BW-day when possible)

Other dose-related details, such as whether administered dose level was verified by
measurement, information on internal dosimetry

Vehicle used for exposed animals

Route of administration (e.g., oral, inhalation, dermal, injection)

Duration and frequency of dosing (e.g., hours, days, weeks when administration was ended,
days per week)
This document is a draft for review purposes only and does not constitute Agency policy.
129	DRAFT-DO NOT CITE OR QUOTE

-------
Systematic Review Protocol for the Hexavalent Chromium IRIS Assessment
Field label
Data extraction elements
Methods
Study design (e.g., single treatment, acute, subchronic [e.g., 90 days in a rodent], chronic,
multigenerational, developmental, other)

Guideline compliance (i.e., use of EPA, OECD, NTP, or another guideline for study design,
conducted under GLP guideline conditions, non-GLP but consistent with guideline study,
nonguideline peer reviewed publication)

Number of animals per group (and dams per group in developmental studies)

Randomization procedure, allocation concealment, blinding during outcome assessment

Method to control for litter effects in developmental studies

Use of negative controls and whether controls were untreated, vehicle-treated, or both

Report on data from positive controls—was expected response observed?

Endpoint health category (e.g., reproductive)

Endpoint (e.g., infertility)

Diagnostic or method to measure endpoint

Statistical methods
Results
Measures of effect at each dose or concentration level (e.g., mean, median, frequency, and
measures of precision or variance) 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 will be expressed as mean difference,
standardized mean difference, and percentage control response. Categorical data will be
expressed as relative risk (RR, also called risk ratio).

NOEL, LOEL, BMD analysis, statistical significance of other dose levels, or other estimates of
effect presented in paper.
Note: The NOEL and LOEL are highly influenced by study design, do not give any quantitative
information about the relationship between dose and response, and can be subject to author's
interpretation (e.g., a statistically significant effect may not be considered biologically
important). Also, a NOEL does not necessarily mean zero response. Ideally, the response rate
at specific dose levels is used as the primary measure to characterize the response.

Observations on dose-response (e.g., trend analysis, description of whether dose-response
shape appears to be monotonic, nonmonotonic)

Data on internal concentration, toxicokinetics, or toxicodynamics (when reported)
Other
Documentation of author queries, use of digital rulers to estimate data values from figures,
exposure unit, and statistical result conversions, etc.
BMD = benchmark dose; CAS = Chemical Abstract Service; GLP = good laboratory practice; HPLC-MS/MS = high-
performance liquid chromatography-tandem mass spectrometry; LOEL = lowest-observed-effect level;
NOEL = no-observed-effect level; NTP = National Toxicology Program; OECD = Organisation for Economic
Cooperation and Development.
This document is a draft for review purposes only and does not constitute Agency policy.
130	DRAFT-DO NOT CITE OR QUOTE

-------