Lyon, France
Amended January 2019
IARC Monographs on the Identification of
Carcinogenic Hazards to Humans
PREAMBLE
IARC Monographs Preamble
Contents
A. GENERAL PRINCIPLES AND PROCEDURES ...................................... 1
1. Background ............................................................................................................. 1
2. Objective
and scope ................................................................................................ 2
3. Selectio
n of agents for review ................................................................................. 3
4. The
Working Group and other meeting participants .............................................. 4
5.
Working procedures ................................................................................................ 6
6. Overview of the scientific review and evaluation process ...................................... 8
7. Respon
sibilities of the Working Group ................................................................ 10
B. SCIENTIFIC REVIEW AND EVALUATION ......................................... 12
1. Exposure characterization ..................................................................................... 12
(a) Identif
ication of the agent ................................................................................ 12
(b) Detection and analysis...................................................................................... 13
(c) Production and use ........................................................................................... 13
(d) Exposure ........................................................................................................... 14
(e) Regulations and guidelines ............................................................................... 14
(f) Critical review of exposure assessment in key epidemiological studies .......... 15
2. Studies
of cancer in humans .................................................................................. 16
(a) Types of study considered ................................................................................ 16
(b) Identification of eligible studies of cancer in humans ..................................... 17
(c) Assessment of study quality and informativeness ........................................... 17
(d) Meta-analyses and pooled analyses ................................................................. 20
(e) Considerations in assessing the body of epidemiological evidence................. 20
3. Studies of cancer in experimental animals ............................................................ 22
(a) T
ypes of studies considered ............................................................................. 23
(b) Study evaluation ............................................................................................... 23
(c) Outcomes and statistical analyses .................................................................... 24
4. Mechani
stic evidence ............................................................................................ 25
(a) Absorpt
ion, distribution, metabolism, and excretion ....................................... 26
(b) Evidence relevant to key characteristics of carcinogens .................................. 26
IARC Monographs Preamble
(c) Other relevant evidence .................................................................................... 28
(d) Study quality and importance to the evaluation ............................................... 28
5. Summary of data reported ..................................................................................... 29
(a) Exposure character
ization ................................................................................ 29
(b) Cancer in humans ............................................................................................. 29
(c) Cancer in experimental animals ....................................................................... 30
(d) Mechanistic evidence ....................................................................................... 30
6. Evaluation and rationale ........................................................................................ 30
(a) Carcinogenicity in humans ............................................................................... 31
(b) Carcinogenicity in experimental animals ......................................................... 32
(c) Mechanistic evidence ....................................................................................... 33
(d) Overall evaluation ............................................................................................ 35
(e) Rationale ........................................................................................................... 36
References ................................................................................................................. 38
IARC Monographs Preamble
1
The Preamble to the IARC Monographs describes the objective and scope of 1
the programme, general principles and procedures, and scientific review and 2
evaluations. The IARC Monographs embody principles of scientific rigour, 3
impartial evaluation, transparency, and consistency. The Preamble should be 4
consulted when reading a Monograph or a summary of a Monograph’s 5
evaluations. Separate Instructions for Authors describe the operational 6
procedures for the preparation and publication of a volume of the Monographs. 7
A. GENERAL PRINCIPLES AND PROCEDURES 8
1. Background 9
Soon after the International Agency for Research on Cancer (IARC) was 10
established in 1965, it started to receive frequent requests for advice on the 11
carcinogenicity of chemicals, including requests for lists of established and 12
suspected human carcinogens. In 1970, an IARC Advisory Committee on 13
Environmental Carcinogenesis recommended “that a compendium on carcinogenic 14
chemicals be prepared by experts. The biological activity and evaluation of 15
practical importance to public health should be referenced and documented.” The 16
next year, the IARC Governing Council adopted a resolution that IARC should 17
prepare “monographs on the evaluation of carcinogenic risk of chemicals to man”, 18
which became the initial title of the series. 19
In succeeding years, the scope of the programme broadened as Monographs 20
were developed for complex mixtures, occupational exposures, physical agents, 21
biological organisms, pharmaceuticals, and other exposures. In 1988,of 22
chemicals” was dropped from the title, and in 2019, “evaluation of carcinogenic 23
risks” became “identification of carcinogenic hazards”, in line with the objective of 24
the programme. 25
Identifying the causes of human cancer is the first step in cancer prevention. 26
The identification of a cancer hazard may have broad and profound implications. 27
National and international authorities and organizations can and do use information 28
on causes of cancer in support of actions to reduce exposure to carcinogens in the 29
workplace, in the environment, and elsewhere. Cancer prevention is needed as 30
much today as it was when IARC was established, because the global burden of 31
cancer is high and continues to increase as a result of population growth and ageing 32
and upward trends in some exposures, especially in low- and middle-income 33
countries (http://publications.iarc.fr/Non-Series-Publications/World-Cancer-34
Reports). 35
IARC’s process for developing Monographs, which has evolved over several 36
decades, involves the engagement of international, interdisciplinary Working 37
Groups of expert scientists, the transparent synthesis of different streams of 38
evidence (exposure characterization, cancer in humans, cancer in experimental 39
animals, and mechanisms of carcinogenesis), and the integration of these streams 40
IARC Monographs Preamble
2
of evidence into an overall evaluation and classification according to criteria 1
developed and refined by IARC. Since the Monographs programme was 2
established, the understanding of carcinogenesis has greatly deepened. Scientific 3
advances are incorporated into the evaluation methodology. In particular, strong 4
mechanistic evidence has had an increasing role in the overall evaluations since 5
1991. 6
The Preamble is primarily a statement of the general principles and procedures 7
used in developing a Monograph, to promote transparency and consistency across 8
Monographs evaluations. In addition, IARC provides Instructions for Authors 9
(https://monographs.iarc.fr/instructions-for-authors/), which specify more detailed 10
working procedures. IARC routinely updates these Instructions for Authors to 11
reflect advances in methods for cancer hazard identification and accumulated 12
experience, including input from experts. 13
2. Objective and scope 14
The objective of the programme is to prepare, with the engagement of 15
international, interdisciplinary Working Groups of experts, scientific reviews and 16
evaluations of evidence on the carcinogenicity of a wide range of agents. 17
The Monographs assess the strength of the available evidence that an agent can 18
cause cancer in humans, based on three streams of evidence: on cancer in humans 19
(see Part B, Section 2), on cancer in experimental animals (see Part B, Section 3), 20
and on mechanistic evidence (see Part B, Section 4). In addition, the exposure to 21
each agent is characterized (see Part B, Section 1). In this Preamble, the term 22
“agent” refers to any chemical, physical, or biological entity or exposure 23
circumstance (e.g. occupation as a painter) for which evidence on the 24
carcinogenicity is evaluated. 25
A cancer hazard is an agent that is capable of causing cancer, whereas a cancer 26
risk is an estimate of the probability that cancer will occur given some level of 27
exposure to a cancer hazard. The Monographs assess the strength of evidence that 28
an agent is a cancer hazard. The distinction between hazard and risk is 29
fundamental. The Monographs identify cancer hazards even when risks appear to 30
be low in some exposure scenarios. This is because the exposure may be 31
widespread at low levels, and because exposure levels in many populations are not 32
known or documented. 33
Although the Monographs programme has focused on hazard identification, 34
some epidemiological studies used to identify a cancer hazard are also used to 35
estimate an exposureresponse relationship within the range of the available data. 36
However, extrapolating exposureresponse relationships beyond the available data 37
(e.g. to lower exposures, or from experimental animals to humans) is outside the 38
scope of Monographs Working Groups (IARC, 2014). In addition, the 39
Monographs programme does not review quantitative risk characterizations 40
developed by other health agencies. 41
IARC Monographs Preamble
3
The identification of a cancer hazard should trigger some action to protect 1
public health, either directly as a result of the hazard identification or through the 2
conduct of a risk assessment. Although such actions are outside the scope of the 3
programme, the Monographs are used by national and international authorities and 4
organizations to inform risk assessments, formulate decisions about preventive 5
measures, motivate effective cancer control programmes, and choose among 6
options for public health decisions. Monographs evaluations are only one part of 7
the body of information on which decisions to control exposure to carcinogens 8
may be based. Options to prevent cancer vary from one situation to another and 9
across geographical regions and take many factors into account, including different 10
national priorities. Therefore, no recommendations are given in the Monographs 11
with regard to regulation, legislation, or other policy approaches, which are the 12
responsibility of individual governments or organizations. The Monographs 13
programme also does not make research recommendations. However, it is 14
important to note that Monographs contribute significantly to the science of 15
carcinogenesis by synthesizing and integrating streams of evidence about 16
carcinogenicity and pointing to critical gaps in knowledge. 17
3. Selection of agents for review 18
Since 1984, about every five years IARC convenes an international, 19
interdisciplinary Advisory Group to recommend agents for review by the 20
Monographs programme. IARC selects Advisory Group members who are 21
knowledgeable about current research on carcinogens and public health priorities. 22
Before an Advisory Group meets, IARC solicits nominations of agents from 23
scientists and government agencies worldwide. Since 2003, IARC also invites 24
nominations from the public. IARC charges each Advisory Group with reviewing 25
nominations, evaluating exposure and hazard potential, and preparing a report that 26
documents the Advisory Group’s process for these activities and its rationale for 27
the recommendations. 28
For each new volume of the Monographs, IARC selects the agents for review 29
from those recommended by the most recent Advisory Group, considering the 30
availability of pertinent research studies and current public health priorities. On 31
occasion, IARC may select other agents if there is a need to rapidly evaluate an 32
emerging carcinogenic hazard or an urgent need to re-evaluate a previous 33
classification. All evaluations consider the full body of available evidence, not just 34
information published after a previous review. 35
A Monograph may review: 36
(a) An agent not reviewed in a previous Monograph, if there is potential human 37
exposure and there is evidence for assessing its carcinogenicity. A group of 38
related agents (e.g. metal compounds) may be reviewed together if there is 39
evidence for assessing carcinogenicity for one or more members of the 40
group. 41
IARC Monographs Preamble
4
(b) An agent reviewed in a previous Monograph, if there is new evidence of 1
cancer in humans or in experimental animals, or mechanistic evidence to 2
warrant re-evaluation of the classification. In the interests of efficiency, the 3
literature searches may build on previous comprehensive searches. 4
(c) An agent that has been established to be carcinogenic to humans and has 5
been reviewed in a previous Monograph, if there is new evidence of cancer 6
in humans that indicates new tumour sites where there might be a causal 7
association. In the interests of efficiency, the review may focus on these new 8
tumour sites. 9
4. The Working Group and other meeting participants 10
Five categories of participants can be present at Monographs meetings: 11
(i) Working Group members are responsible for all scientific reviews and 12
evaluations developed in the volume of the Monographs. The Working 13
Group is interdisciplinary and comprises subgroups of experts in the fields of 14
(a) exposure characterization, (b) cancer in humans, (c) cancer in 15
experimental animals, and (d) mechanistic evidence. IARC selects Working 16
Group members on the basis of expertise related to the subject matter and 17
relevant methodologies, and absence of conflicts of interest. Consideration is 18
also given to diversity in scientific approaches and views, as well as 19
demographic composition. Working Group members generally have 20
published research related to the exposure or carcinogenicity of the agents 21
being reviewed, and IARC uses literature searches to identify most experts. 22
Since 2006, IARC also has encouraged public nominations through its Call 23
for Experts. IARC’s reliance on experts with knowledge of the subject matter 24
and/or expertise in methodological assessment is confirmed by decades of 25
experience documenting that there is value in specialized expertise and that 26
the overwhelming majority of Working Group members are committed to the 27
objective evaluation of scientific evidence and not to the narrow advancement 28
of their own research results or a pre-determined outcome (Wild & Cogliano, 29
2011). Working Group members are expected to serve the public health 30
mission of IARC, and should refrain from consulting and other activities for 31
financial gain that are related to the agents under review, or the use of inside 32
information from the meeting, until the full volume of the Monographs is 33
published. 34
IARC identifies, from among Working Group members, individuals to serve 35
as Meeting Chair and Subgroup Chairs. At the opening of the meeting, the 36
Working Group is asked to endorse the selection of the Meeting Chair, with 37
the opportunity to propose alternatives. The Meeting Chair and Subgroup 38
Chairs take a leading role at all stages of the review process (see Part A, 39
Section 7), promote open scientific discussions that involve all Working 40
IARC Monographs Preamble
5
Group members in accordance with normal committee procedures, and 1
ensure adherence to the Preamble. 2
(ii) Invited Specialists are experts who have critical knowledge and experience 3
but who also have a conflict of interest that warrants exclusion from 4
developing or influencing the evaluations of carcinogenicity. Invited 5
Specialists do not draft any section of the Monograph that pertains to the 6
description or interpretation of cancer data, and they do not participate in the 7
evaluations. These experts are invited in limited numbers when necessary to 8
assist the Working Group by contributing their unique knowledge and 9
experience to the discussions. 10
(iii) Representatives of national and international health agencies may attend 11
because their agencies are interested in the subject of the meeting. They do 12
not draft any section of the Monograph or participate in the evaluations. 13
(iv) Observers with relevant scientific credentials may be admitted in limited 14
numbers. Attention is given to the balance of Observers from constituencies 15
with differing perspectives. Observers are invited to observe the meeting and 16
should not attempt to influence it, and they agree to respect the Guidelines 17
for Observers at IARC Monographs meetings. Observers do not draft any 18
section of the Monograph or participate in the evaluations. 19
(v) The IARC Secretariat consists of scientists who are designated by IARC and 20
who have relevant expertise. The IARC Secretariat coordinates and 21
facilitates all aspects of the evaluation and ensures adherence to the 22
Preamble throughout development of the scientific reviews and 23
classifications (see Part A, Sections 5 and 6). The IARC Secretariat 24
organizes and announces the meeting, identifies and recruits the Working 25
Group members, and assesses the declared interests of all meeting 26
participants. The IARC Secretariat supports the activities of the Working 27
Group (see Part A, Section 7) by searching the literature and performing title 28
and abstract screening, organizing conference calls to coordinate the 29
development of pre-meeting drafts and discuss cross-cutting issues, and 30
reviewing drafts before and during the meeting. Members of the IARC 31
Secretariat serve as meeting rapporteurs, assist the Meeting Chair and 32
Subgroup Chairs in facilitating all discussions, and may draft text or tables 33
when designated by the Meeting Chair and Subgroup Chairs. Their 34
participation in the evaluations is restricted to the role of clarifying or 35
interpreting the Preamble. 36
All participants are listed, with their principal affiliations, in the front matter of 37
the published volume of the Monographs. Working Group members and Invited 38
Specialists serve as individual scientists and not as representatives of any 39
organization, government, or industry (Cogliano et al., 2004). 40
The roles of the meeting participants are summarized in Table 1. 41
IARC Monographs Preamble
6
Table 1. Roles of participants at IARC Monographs meetings
Category of participant
Role
Prepare
text,
tables, and
analyses
Participate
in
discussions
Participate
in
evaluations
Eligible to
serve as
Chair
Working Group members
Invited Specialists
a
Representatives of health
agencies
b
Observers
b
IARC Secretariat
c
d
a
Only for the section on exposure characterization
b
Only at times designated by the Meeting Chair and Subgroup Chairs
c
When needed or requested by the Meeting Chair and Subgroup Chairs
d
Only for clarifying or interpreting the Preamble
5. Working procedures 1
A separate Working Group is responsible for developing each volume of the 2
Monographs. A volume contains one or more Monographs, which can cover either 3
a single agent or several related agents. Approximately one year before the meeting 4
of a Working Group, a preliminary list of agents to be reviewed, together with a 5
Call for Data and a Call for Experts, is announced on the Monographs programme 6
website (http://monographs.iarc.fr). 7
Before a meeting invitation is extended, each potential participant, including the 8
IARC Secretariat, completes the WHO Declaration of Interests form to report 9
financial interests, employment and consulting (including remuneration for serving 10
as an expert witness), individual and institutional research support, and non-11
financial interests such as public statements and positions related to the subject of 12
the meeting. IARC assesses the declared interests to determine whether there is a 13
conflict that warrants any limitation on participation (see Table 2). 14
Approximately two months before a Monographs meeting, IARC publishes the 15
names and affiliations of all meeting participants together with a summary of 16
declared interests, in the interests of transparency and to provide an opportunity for 17
undeclared conflicts of interest to be brought to IARC’s attention. It is not 18
IARC Monographs Preamble
7
acceptable for Observers or third parties to contact other participants before a 1
meeting or to lobby them at any time. Meeting participants are asked to report all 2
such contacts to IARC (Cogliano et al., 2005). 3
The Working Group meets at IARC for approximately eight days to discuss and 4
finalize the scientific review and to develop summaries and evaluations. At the 5
opening of the meeting, all participants update their Declaration of Interests forms, 6
which are then reviewed by IARC. Declared interests related to the subject of the 7
meeting are disclosed to the meeting participants during the meeting and in the 8
published volume (Cogliano et al., 2004). The objectives of the meeting are peer 9
review and consensus. During the first part of the meeting, subgroup sessions 10
(covering exposure characterization, cancer in humans, cancer in experimental 11
animals, and mechanistic evidence) review the pre-meeting drafts, develop a joint 12
subgroup draft, and draft subgroup summaries. During the last part of the meeting, 13
the Working Group meets in plenary session to review the subgroup drafts and 14
summaries and to develop the consensus evaluations. As a result, the entire volume 15
is the joint product of the Working Group, and there are no individually authored 16
sections. After the meeting, the master copy is verified by the IARC Secretariat 17
and is then edited and prepared for publication. The aim is to publish the volume 18
within approximately nine months of the Working Group meeting. A summary of 19
the evaluations and key supporting evidence is prepared for publication in a 20
scientific journal or is made available on the Monographs programme website 21
soon after the meeting. 22
In the interests of transparency, IARC engages with the public throughout the 23
process, as summarized in Table 2. 24
IARC Monographs Preamble
8
Table 2. Public engagement during Monographs development
Approximate timeframe Engagement
Every 5 years
IARC convenes an Advisory Group to recommend
high-priority agents for future review
~1 year before a Monographs
meeting
IARC selects agents for review in a new volume of the
Monographs
IARC posts on its website:
Preliminary List of Agents to be reviewed
Call for Data and Call for Experts
Request for Observer Status
WHO Declaration of Interests form
~8 months before a
Monographs meeting
Call for Experts closes
~4 months before a
Monographs meeting
Request for Observer Status closes
~2 months before a
Monographs meeting
IARC posts the names of all meeting participants
together with a summary of declared interests, and a
statement discouraging contact of the Working Group
by interested parties
~1 month before a
Monographs meeting
Call for Data closes
~2–4 weeks after a
Monographs meeting
IARC publishes a summary of evaluations and key
supporting evidence
~9 months after a
Monographs meeting
IARC Secretariat publishes the verified and edited
master copy of plenary drafts as a Monographs
volume
6. Overview of the scientific review and evaluation process 1
The Working Group considers all pertinent epidemiological studies, cancer 2
bioassays in experimental animals, and mechanistic evidence, as well as pertinent 3
information on exposure in humans. In general, for cancer in humans, cancer in 4
experimental animals, and mechanistic evidence, only studies that have been 5
published or accepted for publication in the openly available scientific literature are 6
reviewed. Under some circumstances, materials that are publicly available and 7
IARC Monographs Preamble
9
whose content is final may be reviewed if there is sufficient information to permit 1
an evaluation of the quality of the methods and results of the studies (see Step 1, 2
below). Such materials may include reports and databases publicly available from 3
government agencies, as well as doctoral theses. The reliance on published and 4
publicly available studies promotes transparency and protects against citation of 5
premature information. 6
The principles of systematic review are applied to the identification, screening, 7
synthesis, and evaluation of the evidence related to cancer in humans, cancer in 8
experimental animals, and mechanistic evidence (as described in Part B, 9
Sections 2–4 and as detailed in the Instructions for Authors). Each Monograph 10
specifies or references information on the conduct of the literature searches, 11
including search terms and inclusion/exclusion criteria that were used for each 12
stream of evidence. 13
In brief, the steps of the review process are as follows: 14
Step 1. Comprehensive and transparent identification of the relevant 15
information: The IARC Secretariat identifies relevant studies through 16
initial comprehensive searches of literature contained in authoritative 17
biomedical databases (e.g. PubMed, PubChem) and through a Call for 18
Data. These literature searches, designed in consultation with a librarian 19
and other technical experts, address whether the agent causes cancer in 20
humans, causes cancer in experimental systems, and/or exhibits key 21
characteristics of established human carcinogens (in humans or in 22
experimental systems). The Working Group provides input and advice to 23
IARC to refine the search strategies, and identifies literature through 24
other searches (e.g. from reference lists of past Monographs, retrieved 25
articles, and other authoritative reviews). 26
For certain types of agents (e.g. regulated pesticides and pharmaceuticals), 27
IARC also provides an opportunity to relevant regulatory authorities, and 28
regulated parties through such authorities, to make pertinent unpublished 29
studies publicly available by the date specified in the Call for Data. 30
Consideration of such studies by the Working Group is dependent on the 31
public availability of sufficient information to permit an independent 32
evaluation of (a) whether there has been selective reporting (e.g. on 33
outcomes, or from a larger set of conducted studies), (b) study quality 34
(e.g. design, methodology, and reporting of results), and (c) study results. 35
Step 2. Screening, selection, and organization of the studies: The IARC 36
Secretariat screens the retrieved literature for inclusion based on title and 37
abstract review, according to pre-defined exclusion criteria. For instance, 38
studies may be excluded if they were not about the agent (or a metabolite 39
of the agent), or if they reported no original data on epidemiological or 40
toxicological end-points (e.g. review articles). The Working Group 41
IARC Monographs Preamble
10
reviews the title and abstract screening done by IARC, and performs full-1
text review. Any reasons for exclusion are recorded, and included studies 2
are organized according to factors pertinent to the considerations 3
described in Part B, Sections 2–4 (e.g. design, species, and end-point). 4
Inclusion of a study does not imply acceptance of the adequacy of the 5
study design or of the analysis and interpretation of the results. 6
Step 3. Evaluation of study quality: The Working Group evaluates the quality 7
of the included studies based on the considerations (e.g. design, 8
methodology, and reporting of results) described in Part B, Sections 2–4. 9
Based on these considerations, the Working Group may accord greater 10
weight to some of the included studies. Interpretation of the results and 11
the strengths and limitations of a study are clearly outlined in square 12
brackets at the end of study descriptions (see Part B). 13
Step 4. Report characteristics of included studies, including assessment of 14
study quality: Pertinent characteristics and results of included studies are 15
reviewed and succinctly described, as detailed in Part B, Sections 1–4. 16
Tabulation of data may facilitate this reporting. This step may be iterative 17
with Step 3. 18
Step 5. Synthesis and evaluation of strength of evidence: The Working Group 19
summarizes the overall strengths and limitations of the evidence from the 20
individual streams of evidence (cancer in humans, cancer in experimental 21
animals, and mechanistic evidence; see Part B, Section 5). The Working 22
Group then evaluates the strength of evidence from each stream of 23
evidence by using the transparent methods and defined descriptive terms 24
given in Part B, Sections 6ac. The Working Group then develops, and 25
describes the rationale for, the consensus classification of carcinogenicity 26
that integrates the conclusions about the strength of evidence from studies 27
of cancer in humans, studies of cancer in experimental animals, and 28
mechanistic evidence (see Part B, Section 6d). 29
7. Responsibilities of the Working Group 30
The Working Group is responsible for identifying and evaluating the relevant 31
studies and developing the scientific reviews and evaluations for a volume of the 32
Monographs. The IARC Secretariat supports these activities of the Working Group 33
(see Part A, Section 4). Briefly, the Working Group’s tasks in developing the 34
evaluation are, in sequence: 35
(i) Before the meeting, the Working Group ascertains that all appropriate 36
studies have been identified and selected, and assesses the methods and quality of 37
each individual study, as outlined above (see Part A, Section 6). The Working 38
Group members prepare pre-meeting working drafts that present accurate tabular 39
or textual summaries of informative studies by extracting key elements of the study 40
IARC Monographs Preamble
11
design and results, and highlighting notable strengths and limitations. They 1
participate in conference calls organized by IARC to coordinate the development 2
of working drafts and to discuss cross-cutting issues. Pre-meeting reviews of all 3
working drafts are generally performed by two or more subgroup members who 4
did not participate in study identification, data extraction, or study review for the 5
draft. Each study summary is written or reviewed by someone who is not 6
associated with the study. 7
(ii) At the meeting, within subgroups, the Working Group members critically 8
review, discuss, and revise the pre-meeting drafts and adopt the revised versions as 9
consensus subgroup drafts. Subgroup Chairs ensure that someone who is not 10
associated with the study leads the discussion of each study summary. A proposed 11
classification of the strength of the evidence reviewed in the subgroup using the 12
IARC Monographs criteria (see Part B, Sections 6ac) is then developed from the 13
consensus subgroup drafts of the evidence summaries (see Part B, Section 5). 14
(iii) During the plenary session, each subgroup presents its drafts for scientific 15
review and discussion to the other Working Group members, who did not 16
participate in study identification, data extraction, or study review for the drafts. 17
Subgroup Chairs ensure that someone who is not associated with the study leads 18
the discussion of each study summary. After review, discussion, and revisions as 19
needed, the subgroup drafts are adopted as a consensus Working Group product. 20
The summaries and classifications of the strength of the evidence, developed in the 21
subgroup in line with the IARC Monographs criteria (see Part B, Sections 6ac), 22
are considered, revised as needed, and adopted by the full Working Group. The 23
Meeting Chair proposes an overall evaluation using the guidance provided in 24
Part B, Section 6d. 25
The Working Group strives to achieve consensus evaluations. Consensus 26
reflects broad agreement among the Working Group, but not necessarily 27
unanimity. The Meeting Chair may poll the Working Group to determine the 28
diversity of scientific opinion on issues where consensus is not apparent. 29
Only the final product of the plenary session represents the views and expert 30
opinions of the Working Group. The entire Monographs volume is the joint 31
product of the Working Group and represents an extensive and thorough peer 32
review of the body of evidence (individual studies, synthesis, and evaluation) by an 33
interdisciplinary expert group. Initial working papers and subsequent revisions are 34
not released, because they would give an incomplete and possibly misleading 35
impression of the consensus developed by the Working Group over a full week of 36
deliberation. 37
IARC Monographs Preamble
12
B. SCIENTIFIC REVIEW AND EVALUATION 1
2
This part of the Preamble discusses the types of evidence that are considered 3
and summarized in each section of a Monograph, followed by the scientific criteria 4
that guide the evaluations. In addition, a section of General Remarks at the front of 5
the volume discusses the reasons the agents were scheduled for evaluation and any 6
key issues encountered during the meeting. 7
1. Exposure characterization 8
This section identifies the agent and describes its occurrence, main uses, and 9
production locations and volumes, where relevant. It also summarizes the 10
prevalence, concentrations in relevant studies, and relevant routes of exposure in 11
humans worldwide. Methods of exposure measurement and analysis are described, 12
and methods of exposure assessment used in key epidemiological studies reviewed 13
by the Working Group are described and evaluated. 14
Over the course of the Monographs programme, concepts of exposure and dose 15
have evolved substantially with deepening understanding of the interactions of 16
agents and biological systems. The concept of exposure has broadened and become 17
more holistic, extending beyond chemical, physical, and biological agents to 18
stressors as construed generally, including psychosocial stressors (National 19
Research Council, 2012; National Academies of Sciences, Engineering, and 20
Medicine, 2017). Overall, this broader conceptualization supports greater 21
integration between exposure characterization and other sections of the 22
Monographs. Concepts of absorption, distribution, metabolism, and excretion are 23
considered in the first subsection of mechanistic evidence (see Part B, Section 4a), 24
whereas validated biomarkers of internal exposure or metabolites that are routinely 25
used for exposure assessment are reported on in this section (see Part B, 26
Section 1b). 27
(a) Identification of the agent 28
The agent being evaluated is unambiguously identified. Details will vary 29
depending on the type of agent but will generally include physical and chemical 30
properties relevant to the agent’s identification, occurrence, and biological activity. 31
If the material that has been tested in experimental animals or in vitro systems is 32
different from that to which humans are exposed, these differences are noted. 33
For chemical agents, the Chemical Abstracts Service Registry Number is 34
provided, as well as the latest primary name and other names in common use, 35
including important trade names, along with available information on the 36
composition of common mixtures or products containing the agent, and potentially 37
toxic and/or carcinogenic impurities. Physical properties relevant to understanding 38
the potential for human exposure and measures of exposure used in studies in 39
IARC Monographs Preamble
13
humans are summarized. These might include physical state, volatility, aqueous 1
and fat solubility, and half-life in the environment and/or in human tissues. 2
For biological agents, taxonomy and structure are described. Mode of 3
replication, life-cycle, target cells, persistence, latency, and host responses, 4
including morbidity and mortality through pathologies other than cancer, are also 5
presented. 6
For foreign bodies, fibres and particles, composition, size range, relative 7
dimensions, and accumulation, persistence, and clearance in target organs are 8
summarized. Physical agents that are forms of radiation are described in terms of 9
frequency spectrum and energy transmission. 10
Exposures may result from, or be influenced by, a diverse range of social and 11
environmental factors, including components of diet, sleep, and physical activity 12
patterns. In these instances, this section will include a description of the agent, its 13
variability across human populations, and its composition or characteristics 14
relevant to understanding its potential carcinogenic hazard to humans and to 15
evaluating exposure assessments in epidemiological studies. 16
(b) Detection and analysis 17
Key methods of detection and quantification of the agent are presented, with an 18
emphasis on those used most widely in surveillance, regulation, and 19
epidemiological studies. Measurement methods for sample matrices that are 20
deemed important sources of human exposure (e.g. air, drinking-water, food, 21
residential dust) and for validated exposure biomarkers (e.g. the agent or its 22
metabolites in human blood, urine, or saliva) are described. Information on 23
detection and quantification limits is provided when it is available and is useful for 24
interpreting studies in humans and in experimental animals. This is not an 25
exhaustive treatise but is meant to help readers understand the strengths and 26
limitations of the available exposure data and of the epidemiological studies that 27
rely on these measurements. 28
(c) Production and use 29
Historical and geographical patterns and trends in production and use are 30
included when they are available, to help readers understand the contexts in which 31
exposures may occur, both within key epidemiological studies reviewed by the 32
Working Group and in human populations generally. Industries that produce, use, 33
or dispose of the agent are described, including their global distribution, when 34
available. National or international listing as a high-production-volume chemical or 35
similar classification may be included. Production processes with significant 36
potential for occupational exposure or environmental pollution are indicated. 37
Trends in global production volumes, technologies, and other data relevant to 38
understanding exposure potential are summarized. Minor or historical uses with 39
IARC Monographs Preamble
14
significant exposure potential or with particular relevance to key epidemiological 1
studies are included. Particular effort may be directed towards finding data on 2
production in low- and middle-income countries, where rapid economic 3
development may lead to higher exposures than those in high-income countries. 4
(d) Exposure 5
A concise overview of quantitative information on sources, prevalence, and 6
levels of exposure in humans is provided. Representative data from research 7
studies, government reports and websites, online databases, and other citable, 8
publicly available sources are tabulated. Data from low- and middle-income 9
countries are sought and included to the extent feasible; information gaps for key 10
regions are noted. Naturally occurring sources of exposure, if any, are noted. 11
Primary exposure routes (e.g. inhalation, ingestion, skin uptake) and other 12
considerations relevant to understanding the potential for cancer hazard from 13
exposure to the agent are reported. 14
For occupational settings, information on exposure prevalence and levels (e.g. 15
in air or human tissues) is reported by industry, occupation, region, and other 16
characteristics (e.g. process, task) where feasible. Information on historical 17
exposure trends, protection measures to limit exposure, and potential co-exposures 18
to other carcinogenic agents in workplaces is provided when available. 19
For non-occupational settings, the occurrence of the agent is described with 20
environmental monitoring or surveillance data. Information on exposure 21
prevalence and levels (e.g. concentrations in human tissues) as well as exposure 22
from and/or concentrations in food and beverages, consumer products, 23
consumption practices, and personal microenvironments is reported by region and 24
other relevant characteristics. Particular importance is placed on describing 25
exposures in life stages or in states of disease or nutrition that may involve greater 26
exposure or susceptibility. 27
Current exposures are of primary interest; however, information on historical 28
exposure trends is provided when available. Historical exposures may be relevant 29
for interpreting epidemiological studies, and when agents are persistent or have 30
long-term effects. Information gaps for important time periods are noted. Exposure 31
data that are not deemed to have high relevance to human exposure are generally 32
not considered. 33
(e) Regulations and guidelines 34
Regulations or guidelines that have been established for the agent (e.g. 35
occupational exposure limits, maximum permitted levels in foods and water, 36
pesticide registrations) are described in brief to provide context about government 37
efforts to limit exposure; these may be tabulated if they are informative for the 38
interpretation of existing or historical exposure levels. Information on applicable 39
IARC Monographs Preamble
15
populations, specific agents concerned, basis for regulation (e.g. human health risk, 1
environmental considerations), and timing of implementation may be noted. 2
National and international bans on production, use, and trade are also indicated. 3
This section aims to include major or illustrative regulations and may not be 4
comprehensive, because of the complexity and range of regulatory processes 5
worldwide. An absence of information on regulatory status should not be taken to 6
imply that a given country or region lacks exposure to, or regulations on exposure 7
to, the agent. 8
(f) Critical review of exposure assessment in key epidemiological studies 9
Epidemiological studies evaluate cancer hazard by comparing outcomes across 10
differently exposed groups. Therefore, the type and quality of the exposure 11
assessment methods used are key considerations when interpreting study findings 12
for hazard identification. This section summarizes and critically reviews the 13
exposure assessment methods used in the individual epidemiological studies that 14
contribute data relevant to the Monographs evaluation. 15
Although there is no standard set of criteria for evaluating the quality of 16
exposure assessment methods across all possible agents, some concepts are 17
universally relevant. Regardless of the agent, all exposures have two principal 18
dimensions: intensity (sometimes defined as concentration or dose) and time. Time 19
considerations include duration (time from first to last exposure), pattern or 20
frequency (whether continuous or intermittent), and windows of susceptibility. 21
This section considers how each of the key epidemiological studies characterizes 22
these dimensions. Interpretation of exposure information may also be informed by 23
consideration of mechanistic evidence (e.g. as described in Part B, Section 4a), 24
including the processes of absorption, distribution, metabolism, and excretion. 25
Exposure intensity and time in epidemiological studies can be characterized by 26
using environmental or biological monitoring data, records from workplaces or 27
other sources, expert assessments, modelled exposures, job-exposure matrices, and 28
subject or proxy reports via questionnaires or interviews. Investigators use these 29
data sources and methods individually or in combination to assign levels or values 30
of an exposure metric (which may be quantitative, semi-quantitative, or qualitative) 31
to members of the population under study. 32
In collaboration with the Working Group members reviewing human studies (of 33
cancer and of mechanisms), key epidemiological studies are identified. For each 34
selected study, the exposure assessment approach, along with its strengths and 35
limitations, is summarized using text and tables. Working Group members identify 36
concerns about exposure assessment methods and their impacts on overall quality 37
for each study reviewed (see Part B, Sections 2d and 4d). In situations where the 38
information provided in the study is inadequate to properly consider the exposure 39
assessment, this is indicated. When adequate information is available, the likely 40
IARC Monographs Preamble
16
direction of bias due to error in exposure measurement, including misclassification 1
(overestimated effects, underestimated effects, or unknown) is discussed. 2
2. Studies of cancer in humans 3
This section includes all pertinent epidemiological studies (see Part B, 4
Section 2b) that include cancer as an outcome. These studies encompass certain 5
types of biomarker studies, for example, studies with biomarkers as exposure 6
metrics (see Part B, Section 2) or those evaluating histological or tumour subtypes 7
and molecular signatures in tumours consistent with a given exposure (Alexandrov 8
et al., 2016). Studies that evaluate early biological effect biomarkers are reviewed 9
in Part B, Section 4. 10
(a) Types of study considered 11
Several types of epidemiological studies contribute to the assessment of 12
carcinogenicity in humans; they typically include cohort studies (including variants 13
such as casecohort and nested casecontrol studies), casecontrol studies, 14
ecological studies, and intervention studies. Rarely, results from randomized trials 15
may be available. Exceptionally, case reports and case series of cancer in humans 16
may also be reviewed. In addition to these designs, innovations in epidemiology 17
allow for many other variants that may be considered in any given Monographs 18
evaluation. 19
Cohort and casecontrol studies typically have the capacity to relate individual 20
exposures under study to the occurrence of cancer in individuals, and provide an 21
estimate of effect (such as relative risk) as the main measure of association. Well-22
conducted cohort and casecontrol studies provide most of the evidence of cancer 23
in humans evaluated by Working Groups. Intervention studies are much less 24
common, but when available can provide strong evidence for making causal 25
inferences. 26
In ecological studies, the units of investigation are usually whole populations 27
(e.g. in particular geographical areas or at particular times), and cancer frequency is 28
related to a summary measure of the exposure in the population under study. In 29
ecological studies, data on individual exposure and outcome are not available, 30
which renders this type of study more prone to confounding and exposure 31
misclassification. In some circumstances, however, ecological studies may be 32
informative, especially when the unit of exposure is most accurately measured at 33
the population level (see, for example, the Monograph on arsenic in drinking-34
water; IARC, 2004). 35
Exceptionally, case reports and case series may provide compelling evidence 36
about the carcinogenicity of an agent. In fact, many of the early discoveries of 37
occupational cancer hazards came about because of observations by workers and 38
their clinicians, who noted a high frequency of cancer in workers who share a 39
common occupation or exposure. Such observations may be the starting point for 40
IARC Monographs Preamble
17
more structured investigations, but in exceptional circumstances, when the risk is 1
high enough, the case series may in itself provide compelling evidence. This would 2
be especially warranted in situations where the exposure circumstance is fairly 3
unusual, as it was in the example of plants containing aristolochic acid (IARC, 4
2012a). 5
The uncertainties that surround the interpretation of case reports, case series, 6
and ecological studies typically make them inadequate, except in rare instances 7
as described above, to form the sole basis for inferring a causal relationship. 8
However, when considered together with cohort and casecontrol studies, these 9
types of study may support the judgement that a causal relationship exists. 10
Epidemiological studies of benign neoplasms, pre-neoplastic lesions, 11
malignant precursors, and other end-points are also reviewed when they relate 12
to the agents reviewed. On occasion they can strengthen inferences drawn from 13
studies of cancer itself. For example, benign brain tumours may share common 14
risk factors with those that are malignant, and benign neoplasms (or those of 15
uncertain behaviour) may be part of the causal path to malignancies (e.g. 16
myelodysplastic syndromes, which may progress to acute myeloid leukaemia). 17
(b) Identification of eligible studies of cancer in humans 18
Relevant studies of cancer in humans are identified by using systematic review 19
principles as described in Part A, further elaborated in the Instructions for Authors, 20
and as detailed below. Eligible studies include all studies in humans of exposure to 21
the agent of interest with cancer as an outcome. Multiple publications on the same 22
study population are identified so that the number of independent studies is 23
accurately represented. Multiple publications may result, for example, from 24
successive follow-ups of a single cohort, from analyses focused on different 25
aspects of an exposuredisease association, or from inclusion of overlapping 26
populations. Usually in such situations, only the most recent, most comprehensive, 27
or most informative report is reviewed in detail. 28
(c) Assessment of study quality and informativeness 29
Epidemiological studies are potentially susceptible to several different sources 30
of error, summarized briefly below. Qualities of individual studies that address 31
these issues are also described below. 32
Study quality is assessed as part of the structured expert review process 33
undertaken by the Working Group. A key aspect of quality assessment is 34
consideration of the possible roles of chance and bias in the interpretation of 35
epidemiological studies. Chance, which is also called random variation, can 36
produce misleading study results. This variability in study results is strongly 37
influenced by the sample size: smaller studies are more likely than larger studies to 38
have effect estimates that are imprecise. Confidence intervals around a study’s 39
IARC Monographs Preamble
18
point estimate of effect are used routinely to indicate the range of values of the 1
estimate that could easily be produced by chance alone. 2
Bias is the effect of factors in study design or conduct that lead an association to 3
erroneously appear stronger or weaker than the association that really exists 4
between the agent and the disease. Biases that require consideration are varied but 5
are usually categorized as selection bias, information bias (e.g. error in 6
measurement of exposure and diseases), and confounding (or confounding bias), 7
(Rothman et al., 2008). Selection bias in an epidemiological study occurs when 8
inclusion of participants from the eligible population or their follow-up in the study 9
is influenced by their exposure or their outcome (usually disease occurrence). 10
Under these conditions, the measure of association found in the study will not 11
accurately reflect the association that would otherwise have been found in the 12
eligible population (Hernán et al., 2004). Information bias results from inaccuracy 13
in exposure or outcome measurement. Both can cause an association between 14
hypothesized cause and effect to appear stronger or weaker than it really is. 15
Confounding is a mixing of extraneous effects with the effects of interest 16
(Rothman et al., 2008). An association between the purported causal factor and 17
another factor that is associated with an increase or decrease in incidence of disease 18
can lead to a spurious association or absence of a real association of the presumed 19
causal factor with the disease. When either of these occurs, confounding is present. 20
In assessing study quality, the Working Group consistently considers the 21
following aspects: 22
Study description: Clarity in describing the study design and its 23
implementation, and the completeness of reporting of all other key 24
information about the study and its results. 25
Study population: Whether the study population was appropriate for 26
evaluating the association between the agent and cancer. Whether the 27
study was designed and carried out to minimize selection bias. Cancer 28
cases in the study population must have been identified in a way that was 29
independent of the exposure of interest, and exposure assessed in a way 30
that was not related to disease (outcome) status. In these respects, 31
completeness of recruitment into the study from the population of interest 32
and completeness of follow-up for the outcome are essential measures. 33
Outcome measurement: The appropriateness of the cancer outcome 34
measure (e.g. mortality vs incidence) for the agent and cancer type under 35
consideration, outcome ascertainment methodology, and the extent to 36
which outcome misclassification may have led to bias in the measure(s) 37
of association. 38
Exposure measurement: The adequacy of the methods used to assess 39
exposure to the agent, and the likelihood (and direction) of bias in the 40
IARC Monographs Preamble
19
measure(s) of association due to error in exposure measurement, 1
including misclassification (as described in Part B, Section 1f). 2
Assessment of potential confounding: To what extent the authors took 3
into account in the study design and analysis other variables (including 4
co-exposures, as described in Part B, Section 1d) that can influence the 5
risk of disease and may have been related to the exposure of interest. 6
Important sources of potential confounding by such variables should have 7
been addressed either in the design of the study, such as by matching or 8
restriction, or in the analysis, by statistical adjustment. In some instances, 9
where direct information on confounders is unavailable, use of indirect 10
methods to evaluate the potential impact of confounding on exposure11
disease associations is appropriate (e.g. Axelson & Steenland, 1988; 12
Richardson et al., 2014). 13
Other potential sources of bias: Each epidemiological study is unique in 14
its study population, its design, its data collection, and, consequently, its 15
potential biases. All possible sources of bias are considered for their 16
possible impact on the results. The possibility of reporting bias (i.e. 17
selective reporting of some results and the suppression of others) should 18
be explored. 19
Statistical methodology: Adequacy of the statistical methods used and 20
their ability to obtain unbiased estimates of exposureoutcome 21
associations, confidence intervals, and test statistics for the significance 22
of measures of association. Appropriateness of methods used to 23
investigate confounding, including adjusting for matching when 24
necessary and avoiding treatment of probable mediating variables as 25
confounders. Detailed analyses of cancer risks in relation to summary 26
measures of exposure such as cumulative exposure, or temporal variables 27
such as age at first exposure or time since first exposure, are reviewed 28
and summarized when available. 29
For the sake of economy and simplicity, in this Preamble the list of possible 30
sources of error is referred to with the phrase “chance, bias, and confounding”, but 31
it should be recognized that this phrase encompasses a comprehensive set of 32
concerns pertaining to study quality. 33
These sources of error do not constitute and should not be used as a formal 34
checklist of indicators of study quality. The judgement of experienced experts is 35
critical in determining how much weight to assign to different issues in considering 36
how all of these potential sources of error should be integrated and how to rate the 37
potential for error related to each of these considerations. 38
The informativeness of a study is its ability to show a true association, if there is 39
one, between the agent and cancer, and the lack of an association, if no association 40
exists. Key determinants of informativeness include: having a study population of 41
IARC Monographs Preamble
20
sufficient size to obtain precise estimates of effect; sufficient elapsed time from 1
exposure to measurement of outcome for an effect, if present, to be observable; 2
presence of an adequate exposure contrast (intensity, frequency, and/or duration); 3
biologically relevant definitions of exposure; and relevant and well-defined time 4
windows for exposure and outcome. 5
(d) Meta-analyses and pooled analyses 6
Independent epidemiological studies of the same agent may lead to inconsistent 7
results that are difficult to interpret or reconcile. Combined analyses of data from 8
multiple studies may be conducted as a means to address this ambiguity. There are 9
two types of combined analysis. The first involves combining summary statistics 10
such as relative risks from individual studies (meta-analysis), and the second 11
involves a pooled analysis of the raw data from the individual studies (pooled 12
analysis) (Greenland & O’Rourke, 2008). 13
The strengths of combined analyses are increased precision because of 14
increased sample size and, in the case of pooled analyses, the opportunity to better 15
control for potential confounders and to explore in more detail interactions and 16
modifying effects that may explain heterogeneity among studies. A disadvantage 17
of combined analyses is the possible lack of comparability of data from various 18
studies, because of differences in population characteristics, subject recruitment, 19
procedures of data collection, methods of measurement, and effects of unmeasured 20
covariates that may differ among studies. These differences in study methods and 21
quality can influence results of either meta-analyses or pooled analyses. If 22
published meta-analyses are to be considered by the Working Group, their 23
adequacy needs to be carefully evaluated, including the methods used to identify 24
eligible studies and the accuracy of data extracted from the individual studies. 25
The Working Group may conduct ad hoc meta-analyses during the course of a 26
Monographs meeting, when there are sufficient studies of an exposureoutcome 27
association to contribute to the Working Group’s assessment of the association. 28
The results of such unpublished original calculations, which would be specified in 29
the text by presentation in square brackets, might involve updates of previously 30
conducted analyses that incorporate the results of more recent studies, or de novo 31
analyses. 32
Irrespective of the source of data for the meta-analyses and pooled analyses, the 33
following key considerations apply: the same criteria for data quality must be 34
applied as for individual studies; sources of heterogeneity among studies must be 35
carefully considered; and the possibility of publication bias should be explored. 36
(e) Considerations in assessing the body of epidemiological evidence 37
The ability of the body of epidemiological evidence to inform the Working 38
Group about the carcinogenicity of the agent is related to both the quantity and the 39
IARC Monographs Preamble
21
quality of the evidence. There is no formulaic answer to the question of how many 1
studies of cancer in humans are needed from which to draw inferences about 2
causality, although more than a single study in a single population will almost 3
always be needed. The number will depend on the considerations relating to 4
evidence described below. 5
After the quality of individual epidemiological studies of cancer has been 6
assessed and the informativeness of the various studies on the association between 7
the agent and cancer has been evaluated, a judgement is made about the strength of 8
evidence that the agent in question is carcinogenic to humans. In making its 9
judgement, the Working Group considers several aspects of the body of evidence 10
(e.g. Hill, 1965; Rothman et al., 2008; Vandenbroucke et al., 2016). 11
A strong association (e.g. a large relative risk) is more likely to indicate 12
causality than is a weak association, because it is more difficult for confounding to 13
falsely create a strong association. However, it is recognized that estimates of 14
effect of small magnitude do not imply lack of causality and may have impact on 15
public health if the disease or exposure is common. Estimates of effect of small 16
magnitude could also contribute useful information to the assessment of causality 17
if level of risk is commensurate with level of exposure when compared with risk 18
estimates from populations with higher exposure (e.g. as seen in residential radon 19
studies compared with studies of radon from uranium mining). 20
Associations that are consistently observed in several studies of the same 21
design, or in studies that use different epidemiological approaches, or under 22
different circumstances of exposure are more likely to indicate a causal 23
relationship than are isolated observations from single studies. If there are 24
inconsistent results among investigations, possible reasons are sought (e.g. 25
differences in study informativeness because of latency, exposure levels, or 26
assessment methods). Results of studies that are judged to be of high quality and 27
informativeness are given more weight than those of studies judged to be 28
methodologically less sound or less informative. 29
Temporality of the association is an essential consideration: that is, the exposure 30
must precede the outcome. 31
An observation that cancer risk increases with increasing exposure is 32
considered to be a strong indication of causality, although the absence of a graded 33
response is not necessarily evidence against a causal relationship, and there are 34
several reasons why the shape of the exposureresponse association may be non-35
monotonic (e.g. Stayner et al., 2003). The demonstration of a decline in risk after 36
cessation of or reduction in exposure in individuals or in whole populations also 37
supports a causal interpretation of the findings. 38
Confidence in a causal interpretation of the evidence from studies of cancer in 39
humans is enhanced if it is coherent with physiological and biological knowledge, 40
including information about exposure to the target organ, latency and timing of the 41
exposure, and characteristics of tumour subtypes. 42
IARC Monographs Preamble
22
The Working Group considers whether there are subpopulations with 1
increased susceptibility to cancer from the agent. For example, molecular 2
epidemiology studies that identify associations between genetic polymorphisms 3
and inter-individual differences in cancer susceptibility to the agent(s) being 4
evaluated may contribute to the identification of carcinogenic hazards to 5
humans. Such studies may be particularly informative if polymorphisms are 6
found to be modifiers of the exposureresponse association, because evaluation 7
of polymorphisms may increase the ability to detect an effect in susceptible 8
subpopulations. 9
When, in the process of evaluating the studies of cancer in humans, the 10
Working Group identifies several high-quality, informative epidemiological 11
studies that clearly show either no positive association or an inverse association 12
between an exposure and a specific type of cancer, a judgement may be made that, 13
in the aggregate, they suggest evidence of lack of carcinogenicity for that cancer 14
type. Such a judgement requires, first, that the studies strictly meet the standards of 15
design and analysis described above. Specifically, the possibility that bias, 16
confounding, or misclassification of exposure or outcome could explain the 17
observed results should be considered and ruled out with reasonable confidence. In 18
addition, all studies that are judged to be methodologically sound should (a) be 19
consistent with an estimate of relative effect of unity (or below unity) for any 20
observed level of exposure, (b) when considered together, provide a combined 21
estimate of relative risk that is at or below unity, and (c) have a narrow confidence 22
interval. Moreover, neither any individual well-designed and well-conducted study 23
nor the pooled results of all the studies should show any consistent tendency that 24
the relative risk of cancer increases with increasing level of exposure. It must be 25
noted that evidence of lack of carcinogenicity obtained from several 26
epidemiological studies can apply only to the type(s) of cancer studied, to the 27
exposure levels reported and the timing and route of exposure studied, to the 28
intervals between first exposure and disease onset observed in these studies, and to 29
the general population(s) studied (i.e. there may be susceptible subpopulations or 30
life stages). Experience from studies of cancer in humans indicates that the period 31
from first exposure to the development of clinical cancer is sometimes longer than 32
20 years; therefore, latency periods substantially shorter than about 30 years cannot 33
provide evidence of lack of carcinogenicity. Furthermore, there may be critical 34
windows of exposure, for example, as with diethylstilboestrol and clear cell 35
adenocarcinoma of the cervix and vagina (IARC, 2012a). 36
3. Studies of cancer in experimental animals 37
Most human carcinogens that have been studied adequately for carcinogenicity 38
in experimental animals have produced positive results in one or more animal 39
species. For some agents, carcinogenicity in experimental animals was 40
demonstrated before epidemiological studies identified their carcinogenicity in 41
IARC Monographs Preamble
23
humans. Although this observation cannot establish that all agents that cause 1
cancer in experimental animals also cause cancer in humans, it is biologically 2
plausible that agents for which there is sufficient evidence of carcinogenicity in 3
experimental animals (see Part B, Section 6b) present a carcinogenic hazard to 4
humans. Accordingly, in the absence of additional scientific information, such as 5
strong evidence that a given agent causes cancer in experimental animals through a 6
species-specific mechanism that does not operate in humans (see Part B, Sections 4 7
and 6; Capen et al., 1999; IARC, 2003), these agents are considered to pose a 8
potential carcinogenic hazard to humans. The inference of potential carcinogenic 9
hazard to humans does not imply tumour site concordance across species (Baan et 10
al., 2019). 11
(a) Types of studies considered 12
Relevant studies of cancer in experimental animals are identified by using 13
systematic review principles as described in Part A, further elaborated in the 14
Instructions for Authors, and as detailed below. Consideration is given to all 15
available long-term studies of cancer in experimental animals with the agent under 16
review (or possibly metabolites or derivatives of the agent) (see Part A, Section 7) 17
after a thorough evaluation of the study features (see Part B, Section 3b). Those 18
studies that are judged to be irrelevant to the evaluation or judged to be inadequate 19
(e.g. too short a duration, too few animals, poor survival; see below) may be 20
omitted. Guidelines for conducting long-term carcinogenicity experiments have 21
been published (e.g. OECD, 2018). 22
In addition to conventional long-term bioassays, alternative studies (e.g. in 23
genetically engineered mouse models) may be considered in assessing 24
carcinogenicity in experimental animals, also after a critical evaluation of the study 25
features. For studies of certain exposures, such as viruses that typically only infect 26
humans, use of such specialized experimental animal models may be particularly 27
important; models include genetically engineered mice with targeted expression of 28
viral genes to tissues from which human cancers arise, as well as humanized mice 29
implanted with the human cells usually infected by the virus. 30
Other types of studies can provide supportive evidence. These include: 31
experiments in which the agent was administered in the presence of factors that 32
modify carcinogenic effects (e.g. initiationpromotion studies); studies in which 33
the end-point was not cancer but a defined precancerous lesion; and studies of 34
cancer in non-laboratory animals (e.g. companion animals) exposed to the agent. 35
(b) Study evaluation 36
Considerations of importance in the interpretation and evaluation of a particular 37
study include: (i) whether the agent was clearly characterized, including the nature 38
and extent of impurities and contaminants and the stability of the agent, and, in the 39
IARC Monographs Preamble
24
case of mixtures, whether the sample characterization was adequately reported; 1
(ii) whether the dose was monitored adequately, particularly in inhalation 2
experiments; (iii) whether the doses, duration and frequency of treatment, duration 3
of observation, and route of exposure were appropriate; (iv) whether appropriate 4
experimental animal species and strains were evaluated; (v) whether there were 5
adequate numbers of animals per group; (vi) whether animals were allocated 6
randomly to groups; (vii) whether the body weight, food and water consumption, 7
and survival of treated animals were affected by any factors other than the test 8
agent; (viii) whether the histopathology review was adequate; and (ix) whether the 9
data were reported and analysed adequately. 10
(c) Outcomes and statistical analyses 11
An assessment of findings of carcinogenicity in experimental animals involves 12
consideration of (i) study features such as route, doses, schedule and duration of 13
exposure, species, strain (including genetic background where applicable), sex, 14
age, and duration of follow-up; (ii) the spectrum of neoplastic response, from pre-15
neoplastic lesions and benign tumours to malignant neoplasms; (iii) the incidence, 16
latency, severity, and multiplicity of neoplasms and pre-neoplastic lesions; (iv) the 17
consistency of the results for a specific target organ or organs across studies of 18
similar design; and (v) the possible role of modifying factors (e.g. diet, infection, 19
stress). 20
Key factors for statistical analysis include: (i) number of animals studied and 21
number examined histologically, (ii) number of animals with a given tumour type 22
or lesion, and (iii) duration of survival. 23
Benign tumours may be combined with malignant tumours in the assessment of 24
tumour incidence when (a) they occur together with and originate from the same 25
cell type as malignant tumours in an organ or tissue in a particular study and 26
(b) they appear to represent a stage in the progression to malignancy (Huff et al., 27
1989). The occurrence of lesions presumed to be pre-neoplastic may in certain 28
instances aid in assessing the biological plausibility of any neoplastic response 29
observed. 30
Evidence of an increased incidence of neoplasms with increasing level of 31
exposure strengthens the inference of a causal association between the exposure 32
and the development of neoplasms. The form of the doseresponse relationship 33
can vary widely, including non-linearity, depending on the particular agent under 34
study and the target organ. The doseresponse relationship can also be affected by 35
differences in survival among the treatment groups. 36
The statistical methods used should be clearly stated and should be the 37
generally accepted techniques refined for this purpose (Peto et al., 1980; Gart et al., 38
1986; Portier & Bailer, 1989; Bieler & Williams, 1993). The choice of the most 39
appropriate statistical method requires consideration of whether there are 40
differences in survival among the treatment groups; for example, reduced survival 41
IARC Monographs Preamble
25
because of non-tumour-related mortality can preclude the occurrence of tumours 1
later in life and a survival-adjusted analysis would be warranted. When detailed 2
information on survival is not available, comparisons of the proportions of tumour-3
bearing animals among the effective number of animals (alive at the time that the 4
first tumour was discovered) can be useful when significant differences in survival 5
occur before tumours appear. The lethality of the tumour also requires 6
consideration: for rapidly fatal tumours, the time of death provides an indication of 7
the time of tumour onset and can be assessed using life-table methods; non-fatal or 8
incidental tumours that do not affect survival can be assessed using methods such 9
as the MantelHaenszel test for changes in tumour prevalence. Because tumour 10
lethality is often difficult to determine, methods such as the poly-k test that do not 11
require such information can also be used. When results are available on the 12
number and size of tumours seen in experimental animals (e.g. papillomas on 13
mouse skin, liver tumours observed through nuclear magnetic resonance 14
tomography), other, more complicated statistical procedures may be needed 15
(Sherman et al., 1994; Dunson et al., 2003). 16
The concurrent control group is generally the most appropriate comparison 17
group for statistical analysis; however, for uncommon tumours, the analysis may 18
be improved by considering historical control data, particularly when between-19
study variability is low. Historical controls should be selected to resemble the 20
concurrent controls as closely as possible with respect to species, sex, and strain, as 21
well as other factors, such as basal diet and general laboratory environment, which 22
may affect tumour response rates in control animals (Haseman et al., 1984; Fung et 23
al., 1996; Greim et al., 2003). It is generally not appropriate to discount a tumour 24
response that is significantly increased compared with concurrent controls by 25
arguing that it falls within the range of historical controls. 26
Meta-analyses and pooled analyses may be appropriate when the experimental 27
protocols are sufficiently similar. 28
4. Mechanistic evidence 29
Mechanistic data may provide evidence of carcinogenicity and may also help in 30
assessing the relevance and importance of findings of cancer in experimental 31
animals and in humans (Guyton et al., 2009; Parkkinen et al., 2018) (see Part B, 32
Section 6). Mechanistic studies have gained in prominence, increasing in their 33
volume, diversity, and relevance to cancer hazard evaluation, whereas studies 34
pertinent to other streams of evidence evaluated in the Monographs (i.e. studies of 35
cancer in humans and lifetime cancer bioassays in rodents) may only be available 36
for a fraction of agents to which humans are currently exposed (Guyton et al., 37
2009, 2018). Mechanistic studies and data are identified, screened, and evaluated 38
for quality and importance to the evaluation by using systematic review principles 39
as described in Part A, further elaborated in the Instructions for Authors, and as 40
detailed below. 41
IARC Monographs Preamble
26
The Working Group’s synthesis reflects the extent of available evidence, 1
summarizing groups of included studies with an emphasis on characterizing 2
consistencies or differences in results within and across experimental designs. 3
Greater emphasis is given to informative mechanistic evidence from human-related 4
studies than to that from other experimental test systems, and gaps are identified. 5
Tabulation of data may facilitate this review. The specific topics addressed in the 6
evidence synthesis are described below. 7
(a) Absorption, distribution, metabolism, and excretion 8
Studies of absorption, distribution, metabolism, and excretion in mammalian 9
species are addressed in a summary fashion; exposure characterization is addressed 10
in Part B, Section 1. The Working Group describes the metabolic fate of the agent 11
in mammalian species, noting the metabolites that have been identified and their 12
chemical reactivity. A metabolic schema may indicate the relevant metabolic 13
pathways and products and whether supporting evidence is from studies in humans 14
and/or studies in experimental animals. Evidence on other adverse effects that 15
indirectly confirm absorption, distribution, and/or metabolism at tumour sites is 16
briefly summarized when direct evidence is sparse. 17
(b) Evidence relevant to key characteristics of carcinogens 18
A review of Group 1 human carcinogens classified up to and including IARC 19
Monographs Volume 100 revealed several issues relevant to improving the 20
evaluation of mechanistic evidence for cancer hazard identification (Smith et al., 21
2016). First, it was noted that human carcinogens often share one or more 22
characteristics that are related to the multiple mechanisms by which agents cause 23
cancer. Second, different human carcinogens may exhibit a different spectrum of 24
these key characteristics and operate through distinct mechanisms. Third, for many 25
carcinogens evaluated before Volume 100, few data were available on some 26
mechanisms of recognized importance in carcinogenesis, such as epigenetic 27
alterations (Herceg et al., 2013). Fourth, there was no widely accepted method to 28
search systematically for relevant mechanistic evidence, resulting in a lack of 29
uniformity in the scope of mechanistic topics addressed across IARC Monographs 30
evaluations. 31
To address these challenges, the key characteristics of human carcinogens were 32
introduced to facilitate systematic consideration of mechanistic evidence in IARC 33
Monographs evaluations (Smith et al., 2016; Guyton et al., 2018). The key 34
characteristics described by Smith et al. (2016) (see Table 3), such as “is 35
genotoxic”, “is immunosuppressive”, or “modulates receptor-mediated effects”, 36
are based on empirical observations of the chemical and biological properties 37
associated with the human carcinogens identified by the IARC Monographs 38
programme up to and including Volume 100. The list of key characteristics and 39
IARC Monographs Preamble
27
associated end-points may evolve, based on the experience of their application and 1
as new human carcinogens are identified. Key characteristics are distinct from the 2
“hallmarks of cancer”, which relate to the properties of cancer cells (Hanahan & 3
Weinberg, 2000, 2011). Key characteristics are also distinct from hypothesized 4
mechanistic pathways, which describe a sequence of biological events postulated 5
to occur during carcinogenesis. As such, the evaluation approach based on key 6
characteristics, outlined below, “avoids a narrow focus on specific pathways and 7
hypotheses and provides for a broad, holistic consideration of the mechanistic 8
evidence” (National Academies of Sciences, Engineering, and Medicine, 2017). 9
Table 3. The key characteristics of carcinogens described by Smith et al.
(2016)
Ten key characteristics of carcinogens
1. Is electrophilic or can be metabolically activated to an electrophile
2. Is genotoxic
3. Alters DNA repair or causes genomic instability
4. Induces epigenetic alterations
5. Induces oxidative stress
6. Induces chronic inflammation
7. Is immunosuppressive
8. Modulates receptor-mediated effects
9. Causes immortalization
10. Alters cell proliferation, cell death, or nutrient supply
Studies in exposed humans and in human primary cells or tissues that 10
incorporate end-points relevant to key characteristics of carcinogens are 11
emphasized when available. For each key characteristic with adequate evidence for 12
evaluation, studies are grouped according to whether they involve (a) humans or 13
human primary cells or tissues or (b) experimental systems; further organization 14
(as appropriate) is by end-point (e.g. DNA damage), duration, species, sex, strain, 15
and target organ as well as strength of study design. Studies investigating 16
susceptibility related to key characteristics of carcinogens (e.g. of genetic 17
polymorphisms, or in genetically engineered animals) can be highlighted and may 18
provide additional support for conclusions on the strength of evidence. Findings 19
relevant to a specific tumour type may be noted. 20
IARC Monographs Preamble
28
(c) Other relevant evidence 1
Other informative evidence may be described when it is judged by the Working 2
Group to be relevant to an evaluation of carcinogenicity and to be of sufficient 3
importance to affect the overall evaluation. Quantitative structureactivity 4
information, such as on specific chemical and/or biological features or activities 5
(e.g. electrophilicity, molecular docking with receptors), may be informative. In 6
addition, evidence that falls outside of the recognized key characteristics of 7
carcinogens, reflecting emerging knowledge or important novel scientific 8
developments on carcinogen mechanisms, may also be included. Available 9
evidence relevant to criteria provided in authoritative publications (e.g. Capen et 10
al., 1999; IARC, 2003) on thyroid, kidney, urinary bladder, or other tumours in 11
experimental animals induced by mechanisms that do not operate in humans is also 12
described. 13
(d) Study quality and importance to the evaluation 14
Based on formal considerations of the quality of the studies (e.g. design, 15
methodology, and reporting of results), the Working Group may give greater 16
weight to some included studies. 17
For observational and other studies in humans, the quality of study design, 18
exposure assessment, and assay accuracy and precision are considered, in 19
collaboration with the Working Group members reviewing exposure 20
characterization and studies of cancer in humans, as are other important factors, 21
including those described above for evaluation of epidemiological evidence 22
(García-Closas et al., 2006, 2011; Vermeulen et al., 2018) (Part B, Sections 1 23
and 2). 24
In general, in experimental systems, studies of repeated doses and of chronic 25
exposures are accorded greater importance than are studies of a single dose or time 26
point. Consideration is also given to factors such as the suitability of the dosing 27
range, the extent of concurrent toxicity observed, and the completeness of 28
reporting of the study (e.g. the source and purity of the agent, the analytical 29
methods, and the results). Route of exposure is generally considered to be a less 30
important factor in the evaluation of experimental studies, recognizing that the 31
exposures and target tissues may vary across experimental models and in exposed 32
human populations. Non-mammalian studies can be synthetically summarized 33
when they are considered to be supportive of evidence in humans or higher 34
organisms. 35
In vitro test systems can provide mechanistic insights, but important 36
considerations include the limitations of the test system (e.g. in metabolic 37
capabilities) as well as the suitability of a particular test article (i.e. because of 38
physical and chemical characteristics) (Hopkins et al., 2004). For studies on some 39
end-points, such as for traditional studies of mutations in bacteria and in 40
IARC Monographs Preamble
29
mammalian cells, formal guidelines, including those from the Organisation for 1
Economic Co-operation and Development, may be informative in conducting the 2
quality review (OECD, 1997, 2016a, b). However, existing guidelines will not 3
generally cover all relevant assays, even for genotoxicity. Possible considerations 4
when evaluating the quality of in vitro studies encompass the methodology and 5
design (e.g. the end-point and test method, the number of replicate samples, the 6
suitability of the concentration range, the inclusion of positive and negative 7
controls, and the assessment of cytotoxicity) as well as reporting (e.g. of the source 8
and purity of the agent, and of the analytical methods and results). High-content 9
and high-throughput in vitro data can serve as an additional or supportive source of 10
mechanistic evidence (Chiu et al., 2018; Guyton et al., 2018), although large-scale 11
screening programmes measuring a variety of end-points were designed to evaluate 12
large chemical libraries in order to prioritize chemicals for additional toxicity 13
testing rather than to identify the hazard of a specific chemical or chemical group. 14
The synthesis is focused on the evidence that is most informative for the overall 15
evaluation. In this regard, it is of note that some human carcinogens exhibit a 16
single or primary key characteristic, evidence of which has been influential in their 17
cancer hazard classifications. For instance, ethylene oxide is genotoxic (IARC, 18
1994), 2,3,7,8-tetrachlorodibenzo-para-dioxin modulates receptor-mediated effects 19
(IARC, 1997), and etoposide alters DNA repair (IARC, 2012a). Similarly, 20
oncogenic viruses cause immortalization, and certain drugs are, by design, 21
immunosuppressive (IARC, 2012a, b). Because non-carcinogens can also induce 22
oxidative stress, this key characteristic should be interpreted with caution unless it 23
is found in combination with other key characteristics (Guyton et al., 2018). 24
Evidence for a group of key characteristics can strengthen mechanistic conclusions 25
(e.g. “induces oxidative stress” together with “is electrophilic or can be 26
metabolically activated to an electrophile”, “induces chronic inflammation”, and 27
“is immunosuppressive”); see, for example, 1-bromopropane (IARC, 2018). 28
5. Summary of data reported 29
(a) Exposure characterization 30
Exposure data are summarized to identify the agent and describe its production, 31
use, and occurrence. Information on exposure prevalence and intensity in different 32
settings, including geographical patterns and time trends, may be included. 33
Exposure assessment methods used in key epidemiological studies reviewed by the 34
Working Group are described and evaluated. 35
(b) Cancer in humans 36
Results of epidemiological studies pertinent to an evaluation of carcinogenicity 37
in humans are summarized. The overall strengths and limitations of the 38
epidemiological evidence base are highlighted to indicate how the evaluation was 39
reached. The target organ(s) or tissue(s) in which a positive association between 40
IARC Monographs Preamble
30
the agent and cancer was observed are identified. Exposureresponse and other 1
quantitative data may be summarized when available. When the available 2
epidemiological studies pertain to a mixed exposure, process, occupation, or 3
industry, the Working Group seeks to identify the specific agent considered to be 4
most likely to be responsible for any excess risk. The evaluation is focused as 5
narrowly as the available data permit. 6
(c) Cancer in experimental animals 7
Results pertinent to an evaluation of carcinogenicity in experimental animals 8
are summarized to indicate how the evaluation was reached. For each animal 9
species, study design, and route of administration, there is a statement about 10
whether an increased incidence, reduced latency, or increased severity or 11
multiplicity of neoplasms or pre-neoplastic lesions was observed, and the tumour 12
sites are indicated. Special conditions resulting in tumours, such as prenatal 13
exposure or single-dose experiments, are mentioned. Negative findings, inverse 14
relationships, doseresponse patterns, and other quantitative data are also 15
summarized. 16
(d) Mechanistic evidence 17
Results pertinent to an evaluation of the mechanistic evidence on 18
carcinogenicity are summarized to indicate how the evaluation was reached. The 19
summary encompasses the informative studies on absorption, distribution, 20
metabolism, and excretion; on the key characteristics with adequate evidence for 21
evaluation; and on any other aspects of sufficient importance to affect the overall 22
evaluation, including on whether the agent belongs to a class of agents for which 23
one or more members have been classified as carcinogenic or probably 24
carcinogenic to humans, and on criteria with respect to tumours in experimental 25
animals induced by mechanisms that do not operate in humans. For each topic 26
addressed, the main supporting findings are highlighted from exposed humans, 27
human cells or tissues, experimental animals, or in vitro systems. When 28
mechanistic studies are available in exposed humans, the tumour type or target 29
tissue studied may be specified. Gaps in the evidence are indicated (i.e. if no 30
studies were available in exposed humans, in in vivo systems, etc.). Consistency or 31
differences of effects across different experimental systems are emphasized. 32
6. Evaluation and rationale 33
Consensus evaluations of the strength of the evidence of cancer in humans, the 34
evidence of cancer in experimental animals, and the mechanistic evidence are 35
made using transparent criteria and defined descriptive terms. The Working Group 36
then develops a consensus overall evaluation of the strength of the evidence of 37
carcinogenicity for each agent under review. 38
IARC Monographs Preamble
31
An evaluation of the strength of the evidence is limited to the agents under 1
review. When multiple agents being evaluated are considered by the Working 2
Group to be sufficiently closely related, they may be grouped together for the 3
purpose of a single and unified evaluation of the strength of the evidence. 4
The framework for these evaluations, described below, may not encompass all 5
factors relevant to a particular evaluation of carcinogenicity. After considering all 6
relevant scientific findings, the Working Group may exceptionally assign the agent 7
to a different category than a strict application of the framework would indicate, 8
while providing a clear rationale for the overall evaluation. 9
When there are substantial differences of scientific interpretation among the 10
Working Group members, the overall evaluation will be based on the consensus of 11
the Working Group. A summary of the alternative interpretations may be provided, 12
together with their scientific rationale and an indication of the relative degree of 13
support for each alternative. 14
The categories of the classification refer to the strength of the evidence that an 15
exposure is carcinogenic and not to the risk of cancer from particular exposures. 16
The terms probably carcinogenic and possibly carcinogenic have no quantitative 17
significance and are used as descriptors of different strengths of evidence of 18
carcinogenicity in humans; probably carcinogenic signifies a greater strength of 19
evidence than possibly carcinogenic. 20
(a) Carcinogenicity in humans 21
Based on the principles outlined in Part B, Section 2, the evidence relevant to 22
carcinogenicity from studies in humans is classified into one of the following 23
categories: 24
Sufficient evidence of carcinogenicity: A causal association between 25
exposure to the agent and human cancer has been established. That is, a 26
positive association has been observed in the body of evidence on 27
exposure to the agent and cancer in studies in which chance, bias, and 28
confounding were ruled out with reasonable confidence. 29
Limited evidence of carcinogenicity: A causal interpretation of the positive 30
association observed in the body of evidence on exposure to the agent and 31
cancer is credible, but chance, bias, or confounding could not be ruled out 32
with reasonable confidence. 33
Inadequate evidence regarding carcinogenicity: The available studies are of 34
insufficient quality, consistency, or statistical precision to permit a 35
conclusion to be drawn about the presence or the absence of a causal 36
association between exposure and cancer, or no data on cancer in humans 37
are available. Common findings that lead to a determination of 38
inadequate evidence of carcinogenicity include: (a) there are no data 39
available in humans; (b) there are data available in humans, but they are 40
IARC Monographs Preamble
32
of poor quality or informativeness; and (c) there are studies of sufficient 1
quality available in humans, but their results are inconsistent or otherwise 2
inconclusive. 3
Evidence suggesting lack of carcinogenicity: There are several high-quality 4
studies covering the full range of levels of exposure that humans are 5
known to encounter, which are mutually consistent in not showing a 6
positive association between exposure to the agent and the studied 7
cancers at any observed level of exposure. The results from these studies 8
alone or combined should have narrow confidence intervals with an upper 9
limit below or close to the null value (e.g. a relative risk of unity). Bias 10
and confounding were ruled out with reasonable confidence, and the 11
studies were considered informative. A conclusion of evidence suggesting 12
lack of carcinogenicity is limited to the cancer sites, populations and life 13
stages, conditions and levels of exposure, and length of observation 14
covered by the available studies. In addition, the possibility of a very 15
small risk at the levels of exposure studied can never be excluded. 16
When there is sufficient evidence, a separate sentence identifies the 17
target organ(s) or tissue(s) for which a causal interpretation has been 18
established. When there is limited evidence, a separate sentence identifies 19
the target organ(s) or tissue(s) for which a positive association between 20
exposure to the agent and the cancer(s) was observed in humans. When 21
there is evidence suggesting lack of carcinogenicity, a separate sentence 22
identifies the target organ(s) or tissue(s) where evidence of lack of 23
carcinogenicity was observed in humans. Identification of a specific 24
target organ or tissue as having sufficient evidence or limited evidence or 25
evidence suggesting lack of carcinogenicity does not preclude the 26
possibility that the agent may cause cancer at other sites. 27
(b) Carcinogenicity in experimental animals 28
The evidence relevant to carcinogenicity from studies in experimental animals 29
is classified into one of the following categories: 30
Sufficient evidence of carcinogenicity: A causal relationship has been 31
established between exposure to the agent and cancer in experimental 32
animals based on an increased incidence of malignant neoplasms or of an 33
appropriate combination of benign and malignant neoplasms in (a) two or 34
more species of animals or (b) two or more independent studies in one 35
species carried out at different times or in different laboratories and/or 36
under different protocols. An increased incidence of malignant neoplasms 37
or of an appropriate combination of benign and malignant neoplasms in 38
both sexes of a single species in a well-conducted study, ideally 39
IARC Monographs Preamble
33
conducted under Good Laboratory Practices (GLP), can also provide 1
sufficient evidence. 2
Exceptionally, a single study in one species and sex may be considered to 3
provide sufficient evidence of carcinogenicity when malignant neoplasms 4
occur to an unusual degree with regard to incidence, site, type of tumour, 5
or age at onset, or when there are marked findings of tumours at multiple 6
sites. 7
Limited evidence of carcinogenicity: The data suggest a carcinogenic effect 8
but are limited for making a definitive evaluation because, for example, 9
(a) the evidence of carcinogenicity is restricted to a single experiment and 10
does not meet the criteria for sufficient evidence; (b) the agent increases 11
the incidence only of benign neoplasms or lesions of uncertain neoplastic 12
potential; (c) the agent increases tumour multiplicity or decreases tumour 13
latency but does not increase tumour incidence; (d) the evidence of 14
carcinogenicity is restricted to initiationpromotion studies; (e) the 15
evidence of carcinogenicity is restricted to observational studies in non-16
laboratory animals (e.g. companion animals); or (f) there are unresolved 17
questions about the adequacy of the design, conduct, or interpretation of 18
the available studies. 19
Inadequate evidence regarding carcinogenicity: The studies cannot be 20
interpreted as showing either the presence or the absence of a 21
carcinogenic effect because of major qualitative or quantitative 22
limitations, or no data are available on cancer in experimental animals. 23
Evidence suggesting lack of carcinogenicity: Well-conducted studies (e.g. 24
conducted under GLP) involving both sexes of at least two species are 25
available showing that, within the limits of the tests used, the agent was 26
not carcinogenic. The conclusion of evidence suggesting lack of 27
carcinogenicity is limited to the species, tumour sites, age at exposure, 28
and conditions and levels of exposure covered by the available studies. 29
(c) Mechanistic evidence 30
Based on the principles outlined in Part B, Section 4, the mechanistic evidence 31
is classified into one of the following categories: 32
Strong mechanistic evidence: Results in several different experimental systems 33
are consistent, and the overall mechanistic database is coherent. Further 34
support can be provided by studies that demonstrate experimentally that the 35
suppression of key mechanistic processes leads to the suppression of tumour 36
development. Typically, a substantial number of studies on a range of 37
relevant end-points are available in one or more mammalian species. 38
Quantitative structureactivity considerations, in vitro tests in non-human 39
IARC Monographs Preamble
34
mammalian cells, and experiments in non-mammalian species may provide 1
corroborating evidence but typically do not in themselves provide strong 2
evidence. However, consistent findings across a number of different test 3
systems in different species may provide strong evidence. 4
Of note, “strong” relates not to potency but to strength of evidence. The 5
classification applies to three distinct topics: 6
(a) Strong evidence that the agent belongs, based on mechanistic 7
considerations, to a class of agents for which one or more members have 8
been classified as carcinogenic or probably carcinogenic to humans. The 9
considerations can go beyond quantitative structureactivity relationships to 10
incorporate similarities in biological activity relevant to common key 11
characteristics across dissimilar chemicals (e.g. based on molecular docking, 12
omics data). 13
(b) Strong evidence that the agent exhibits key characteristics of carcinogens. 14
In this case, three descriptors are possible: 15
(1) The strong evidence is in exposed humans. Findings relevant to a 16
specific tumour type may be informative in this determination. 17
(2) The strong evidence is in human primary cells or tissues. Specifically, 18
the strong findings are from biological specimens obtained from 19
humans (e.g. ex vivo exposure), from human primary cells, and/or, in 20
some cases, from other humanized systems (e.g. a human receptor or 21
enzyme). 22
(3) The strong evidence is in experimental systems. This may include one 23
or a few studies in human primary cells and tissues. 24
(c) Strong evidence that the mechanism of carcinogenicity in experimental 25
animals does not operate in humans. Certain results in experimental animals 26
(see Part B, Section 6b) would be discounted, according to relevant criteria 27
and considerations in authoritative publications (e.g. Capen et al., 1999; 28
IARC, 2003). Typically, this classification would not apply when there is 29
strong mechanistic evidence that the agent exhibits key characteristics of 30
carcinogens. 31
Limited mechanistic evidence: The evidence is suggestive, but, for example, 32
(a) the studies cover a narrow range of experiments, relevant end-points, 33
and/or species; (b) there are unexplained inconsistencies in the studies of 34
similar design; and/or (c) there is unexplained incoherence across studies of 35
different end-points or in different experimental systems. 36
Inadequate mechanistic evidence: Common findings that lead to a 37
determination of inadequate mechanistic evidence include: (a) few or no 38
data are available; (b) there are unresolved questions about the adequacy of 39
IARC Monographs Preamble
35
the design, conduct, or interpretation of the studies; (c) the available results 1
are negative. 2
(d) Overall evaluation 3
Finally, the bodies of evidence included within each stream of evidence are 4
considered as a whole, in order to reach an overall evaluation of the 5
carcinogenicity of the agent to humans. The three streams of evidence are 6
integrated and the agent is classified into one of the following categories (see 7
Table 4), indicating that the Working Group has established that: 8
The agent is carcinogenic to humans (Group 1) 9
This category applies whenever there is sufficient evidence of carcinogenicity in 10
humans. 11
In addition, this category may apply when there is both strong evidence in 12
exposed humans that the agent exhibits key characteristics of carcinogens and 13
sufficient evidence of carcinogenicity in experimental animals. 14
The agent is probably carcinogenic to humans (Group 2A) 15
This category generally applies when the Working Group has made at least 16
two of the following evaluations, including at least one that involves either 17
exposed humans or human cells or tissues: 18
Limited evidence of carcinogenicity in humans, 19
Sufficient evidence of carcinogenicity in experimental animals, 20
Strong evidence that the agent exhibits key characteristics of 21
carcinogens. 22
If there is inadequate evidence regarding carcinogenicity in humans, there 23
should be strong evidence in human cells or tissues that the agent exhibits key 24
characteristics of carcinogens. If there is limited evidence of carcinogenicity in 25
humans, then the second individual evaluation may be from experimental systems 26
(i.e. sufficient evidence of carcinogenicity in experimental animals or strong 27
evidence in experimental systems that the agent exhibits key characteristics of 28
carcinogens). 29
Additional considerations apply when there is strong evidence that the 30
mechanism of carcinogenicity in experimental animals does not operate in humans 31
for one or more tumour sites. Specifically, the remaining tumour sites should still 32
support an evaluation of sufficient evidence in experimental animals in order for 33
this evaluation to be used to support an overall classification in Group 2A. 34
Separately, this category generally applies if there is strong evidence that the 35
agent belongs, based on mechanistic considerations, to a class of agents for which 36
one or more members have been classified in Group 1 or Group 2A. 37
IARC Monographs Preamble
36
The agent is possibly carcinogenic to humans (Group 2B) 1
This category generally applies when only one of the following evaluations 2
has been made by the Working Group: 3
Limited evidence of carcinogenicity in humans, 4
Sufficient evidence of carcinogenicity in experimental animals, 5
Strong evidence that the agent exhibits key characteristics of 6
carcinogens. 7
Because this category can be based on evidence from studies in experimental 8
animals alone, there is no requirement that the strong mechanistic evidence be 9
in exposed humans or in human cells or tissues. This category may be based on 10
strong evidence in experimental systems that the agent exhibits key 11
characteristics of carcinogens. 12
As with Group 2A, additional considerations apply when there is strong 13
evidence that the mechanism of carcinogenicity in experimental animals does 14
not operate in humans for one or more tumour sites. Specifically, the remaining 15
tumour sites should still support an evaluation of sufficient evidence in 16
experimental animals in order for this evaluation to be used to support an 17
overall classification in Group 2B. 18
The agent is not classifiable as to its carcinogenicity to humans (Group 3) 19
Agents that do not fall into any other group are generally placed in this 20
category. 21
This includes the case when there is strong evidence that the mechanism of 22
carcinogenicity in experimental animals does not operate in humans for one or 23
more tumour sites in experimental animals, the remaining tumour sites do not 24
support an evaluation of sufficient evidence in experimental animals, and other 25
categories are not supported by data from studies in humans and mechanistic 26
studies. 27
An evaluation in Group 3 is not a determination of non-carcinogenicity or 28
overall safety. It often means that the agent is of unknown carcinogenic potential 29
and that there are significant gaps in research. 30
If the evidence suggests that the agent exhibits no carcinogenic activity, either 31
through evidence suggesting lack of carcinogenicity in both humans and 32
experimental animals, or through evidence suggesting lack of carcinogenicity in 33
experimental animals complemented by strong negative mechanistic evidence in 34
assays relevant to human cancer, then the Working Group may add a sentence to 35
the evaluation to characterize the agent as well-studied and without evidence of 36
carcinogenic activity. 37
(e) Rationale 38
The reasoning that the Working Group used to reach its evaluation is 39
summarized so that the basis for the evaluation offered is transparent. This section 40
IARC Monographs Preamble
37
integrates the major findings from studies of cancer in humans, cancer in 1
experimental animals, and mechanistic evidence. It includes concise statements of 2
the principal line(s) of argument that emerged in the deliberations of the Working 3
Group, the conclusions of the Working Group on the strength of the evidence for 4
each stream of evidence, an indication of the body of evidence that was pivotal to 5
these conclusions, and an explanation of the reasoning of the Working Group in 6
making its evaluation. 7
Table 4. Integration of streams of evidence in reaching overall
classifications (the evidence in bold italic represents the basis of the overall
evaluation)
Stream of evidence Classification based on
strength of evidence
Evidence of
cancer in
humans
a
Evidence of
cancer in
experimental
animals
Mechanistic
evidence
Sufficient Not necessary Not necessary
Carcinogenic to humans
(Group 1)
Limited or
Sufficient
Strong (b)(1)
(exposed humans)
Limited Sufficient
Strong (b)(2–3),
Limited, or
Inadequate
Probably carcinogenic to
humans (Group 2A)
Inadequate Sufficient
Strong (b)(2)
(human cells or
tissues)
Limited
Less than
Sufficient
Strong (b)(1–3)
Limited or
Not necessary
Strong (a)
(mechanistic class)
Limited
Less than
Sufficient
Limited or
Inadequate
Possibly carcinogenic to
humans (Group 2B)
Inadequate Sufficient
Strong (b)(3),
Limited, or
Inadequate
Inadequate
Less than
Sufficient
Strong b(1–3)
Limited Sufficient
Strong (c) (does not
operate in humans)
b
Inadequate Sufficient
Strong (c) (does not
operate in humans)
b
Not classifiable as to its
carcinogenicity to humans
(Group 3)
All other situations not listed above
a
Human cancer(s) with highest evaluation
b
The strong evidence that the mechanism of carcinogenicity in experimental animals does not
operate in humans must specifically be for the tumour sites supporting the classification of sufficient
evidence in experimental animals.
IARC Monographs Preamble
38
References 1
Alexandrov LB, Ju YS, Haase K, Van Loo P, Martincorena I, Nik-Zainal S, et al. (2016). 2
Mutational signatures associated with tobacco smoking in human cancer. Science. 3
354(6312):61822. https://doi.org/10.1126/science.aag0299 PMID:27811275 4
Axelson O, Steenland K (1988). Indirect methods of assessing the effects of tobacco use in 5
occupational studies. Am J Ind Med. 13(1):10518. https://doi.org/10.1002/ajim.4700130107 6
PMID:3344750 7
Baan RA, Stewart BW, Straif K, editors (2019). Tumour site concordance and mechanisms of 8
carcinogenesis (IARC Scientific Publication No. 165). Lyon, France: International Agency 9
for Research on Cancer. Available from: http://publications.iarc.fr/578. 10
Bieler GS, Williams RL (1993). Ratio estimates, the delta method, and quantal response tests for 11
increased carcinogenicity. Biometrics. 49(3):793801. https://doi.org/10.2307/2532200 12
PMID:8241374 13
Capen CC, Dybing E, Rice JM, Wilbourn JD, editors (1999). Species differences in thyroid, 14
kidney and urinary bladder carcinogenesis (IARC Scientific Publication No. 147). Lyon, 15
France: IARC Press. Available from: http://publications.iarc.fr/302 PMID:10627184 16
Chiu WA, Guyton KZ, Martin MT, Reif DM, Rusyn I (2018). Use of high-throughput in vitro 17
toxicity screening data in cancer hazard evaluations by IARC Monograph Working Groups. 18
ALTEX. 35(1):5164. https://doi.org/10.14573/altex.1703231 PMID:28738424 19
Cogliano V, Baan R, Straif K, Grosse Y, Secretan B, El Ghissassi F, et al. (2005). Transparency 20
in IARC Monographs. Lancet Oncol. 6(10):747. https://doi.org/10.1016/S1470-21
2045(05)70380-6 22
Cogliano VJ, Baan RA, Straif K, Grosse Y, Secretan MB, El Ghissassi F, et al. (2004). The 23
science and practice of carcinogen identification and evaluation. Environ Health Perspect. 24
112(13):126974. https://doi.org/10.1289/ehp.6950 PMID:15345338 25
Dunson DB, Chen Z, Harry J (2003). A Bayesian approach for joint modeling of cluster size and 26
subunit-specific outcomes. Biometrics. 59(3):52130. https://doi.org/10.1111/1541-27
0420.00062 PMID:14601753 28
Fung KY, Krewski D, Smythe RT (1996). A comparison of tests for trend with historical controls 29
in carcinogen bioassay. Can J Stat. 24(4):43154. https://doi.org/10.2307/3315326 30
Garcia-Closas M, Vermeulen R, Cox D, Lan Q, Caporaso N, Hunter D, et al. (2011). Population-31
based study designs in molecular epidemiology. In: Rothman N, Hainaut P, Schulte P, Smith 32
M, Boffetta P, Perera FP, editors. Molecular epidemiology: principles and practices. Lyon, 33
France: International Agency for Research on Cancer; pp. 24159. 34
García-Closas M, Vermeulen R, Sherman ME, Moore LE, Smith MT, Rothman N (2006). 35
Application of biomarkers in cancer epidemiology. In: Schottenfeld D, Fraumeni JF Jr, 36
editors. Cancer epidemiology and prevention. 3rd ed. New York (NY), USA: Oxford 37
University Press; pp. 7088. https://doi.org/10.1093/acprof:oso/9780195149616.003.0006 38
Gart JJ, Krewski D, Lee PN, Tarone RE, Wahrendorf J (1986). Statistical methods in cancer 39
research, Volume III: The design and analysis of long-term animal experiments (IARC 40
Scientific Publication No. 79). Lyon, France: International Agency for Research on Cancer. 41
Greenland S, O’Rourke K (2008). Meta-analysis. In: Rothman KJ, Greenland S, Lash TL, 42
editors. Modern epidemiology. 3rd ed. Philadelphia (PA), USA: Wolters Kluwer 43
Health/Lippincott Williams & Wilkins; pp. 652682. 44
Greim H, Gelbke H-P, Reuter U, Thielmann HW, Edler L (2003). Evaluation of historical control 45
data in carcinogenicity studies. Hum Exp Toxicol. 22(10):541–9. 46
https://doi.org/10.1191/0960327103ht394oa PMID:14655720 47
IARC Monographs Preamble
39
Guyton KZ, Kyle AD, Aubrecht J, Cogliano VJ, Eastmond DA, Jackson M, et al. (2009). 1
Improving prediction of chemical carcinogenicity by considering multiple mechanisms and 2
applying toxicogenomic approaches. Mutat Res. 681(23):23040. 3
https://doi.org/10.1016/j.mrrev.2008.10.001 PMID:19010444 4
Guyton KZ, Rusyn I, Chiu WA, Corpet DE, van den Berg M, Ross MK, et al. (2018). 5
Application of the key characteristics of carcinogens in cancer hazard identification. 6
Carcinogenesis. 39(4):61422. https://doi.org/10.1093/carcin/bgy031 PMID:29562322 7
Hanahan D, Weinberg RA (2000). The hallmarks of cancer. Cell. 100(1):5770. 8
https://doi.org/10.1016/S0092-8674(00)81683-9 PMID:10647931 9
Hanahan D, Weinberg RA (2011). Hallmarks of cancer: the next generation. Cell. 144(5):64610
74. https://doi.org/10.1016/j.cell.2011.02.013 PMID:21376230 11
Haseman JK, Huff J, Boorman GA (1984). Use of historical control data in carcinogenicity 12
studies in rodents. Toxicol Pathol. 12(2):12635. 13
https://doi.org/10.1177/019262338401200203 PMID:11478313 14
Herceg Z, Lambert MP, van Veldhoven K, Demetriou C, Vineis P, Smith MT, et al. (2013). 15
Towards incorporating epigenetic mechanisms into carcinogen identification and evaluation. 16
Carcinogenesis. 34(9):195567. https://doi.org/10.1093/carcin/bgt212 PMID:23749751 17
Hernán MA, Hernández-Díaz S, Robins JM (2004). A structural approach to selection bias. 18
Epidemiology. 15(5):61525. https://doi.org/10.1097/01.ede.0000135174.63482.43 19
PMID:15308962 20
Hill AB (1965). The environment and disease: association or causation? Proc R Soc Med. 21
58:295300. PMID:14283879 22
Hopkins AL, Groom CR, Alex A (2004). Ligand efficiency: a useful metric for lead selection. 23
Drug Discov Today. 9(10):4301. https://doi.org/10.1016/S1359-6446(04)03069-7 24
PMID:15109945 25
Huff JE, Eustis SL, Haseman JK (1989). Occurrence and relevance of chemically induced benign 26
neoplasms in long-term carcinogenicity studies. Cancer Metastasis Rev. 8(1):122. 27
https://doi.org/10.1007/BF00047055 PMID:2667783 28
IARC (1994). Some industrial chemicals. IARC Monogr Eval Carcinog Risks Hum. 60:1560. 29
Available from: http://publications.iarc.fr/78 PMID:7869568 30
IARC (1997). Polychlorinated dibenzo-para-dioxins and polychlorinated dibenzofurans. IARC 31
Monogr Eval Carcinog Risks Hum. 69:1631. Available from: http://publications.iarc.fr/87 32
PMID:9379504 33
IARC (2003). Predictive value of rodent forestomach and gastric neuroendocrine tumours in 34
evaluating carcinogenic risks to humans (IARC Technical Publication No. 39). Lyon, France: 35
IARC Press. 36
IARC (2004). Some drinking-water disinfectants and contaminants, including arsenic. IARC 37
Monogr Eval Carcinog Risks Hum. 84:1477. Available from: http://publications.iarc.fr/102 38
PMID:15645577 39
IARC (2012a). Pharmaceuticals. IARC Monogr Eval Carcinog Risks Hum. 100A:1437. 40
Available from: http://publications.iarc.fr/118 PMID:23189749 41
IARC (2012b). Biological agents. IARC Monogr Eval Carcinog Risks Hum. 100B:1441. 42
Available from: http://publications.iarc.fr/119 PMID:23189750 43
IARC (2014). Report of the IARC Advisory Group to Recommend on Quantitative Risk 44
Characterization (IARC Internal Report 14/001). Available from: 45
https://monographs.iarc.fr/wp-content/uploads/2018/06/14-001.pdf. 46
IARC (2018). Some industrial chemicals. IARC Monogr Eval Carcinog Risks Hum. 115:1292. 47
Available from: http://publications.iarc.fr/563 48
IARC Monographs Preamble
40
National Academies of Sciences, Engineering, and Medicine (2017). Using 21st century science 1
to improve risk-related evaluations. Washington (DC), USA: National Academies Press. 2
https://doi.org/10.17226/24635 PMID:28267305 3
National Research Council (2012). Exposure science in the 21st century: a vision and a strategy. 4
Washington (DC), USA: National Academies Press. https://doi.org/10.17226/13507 5
PMID:24901193 6
OECD (1997). Test no. 471: Bacterial reverse mutation test, OECD Guidelines for the Testing of 7
Chemicals, Section 4. Paris, France: Organisation for Economic Co-operation and 8
Development. https://doi.org/10.1787/9789264071247-en 9
OECD (2016a). Test no. 476: In vitro mammalian cell gene mutation tests using the Hprt and 10
xprt genes, OECD Guidelines for the Testing of Chemicals, Section 4. Paris, France: 11
Organisation for Economic Co-operation and Development. 12
https://dx.doi.org/10.1787/9789264264809-en 13
OECD (2016b). Test no. 490: In vitro mammalian cell gene mutation tests using the thymidine 14
kinase gene, OECD Guidelines for the Testing of Chemicals, Section 4. Paris, France: 15
Organisation for Economic Co-operation and Development. 16
https://doi.org/10.1787/9789264264908-en 17
OECD (2018). Test no. 451: Carcinogenicity studies, OECD Guidelines for the Testing of 18
Chemicals, Section 4: Health effects. Paris, France: Organisation for Economic Co-operation 19
and Development. https://doi.org/10.1787/9789264071186-en 20
Parkkinen VP, Wallmann C, Wilde M, Clarke B, Illari P, Kelly MP, et al. (2018). Evaluating 21
evidence of mechanisms in medicine: principles and procedures. Cham, Switzerland: 22
Springer. https://doi.org/10.1007/978-3-319-94610-8 23
Peto R, Pike MC, Day NE, Gray RG, Lee PN, Parish S, et al. (1980). Guidelines for simple, 24
sensitive significance tests for carcinogenic effects in long-term animal experiments. In: 25
Long-term and short-term screening assays for carcinogens: a critical appraisal. IARC 26
Monogr Eval Carcinog Risk Chem Hum Suppl. 2 Suppl. Lyon, France: International Agency 27
for Research on Cancer; pp. 311426. Available from: http://publications.iarc.fr/134 28
PMID:6935176 29
Portier CJ, Bailer AJ (1989). Testing for increased carcinogenicity using a survival-adjusted 30
quantal response test. Fundam Appl Toxicol. 12(4):7317. https://doi.org/10.1016/0272-31
0590(89)90004-3 PMID:2744275 32
Richardson DB, Laurier D, Schubauer-Berigan MK, Tchetgen Tchetgen E, Cole SR (2014). 33
Assessment and indirect adjustment for confounding by smoking in cohort studies using 34
relative hazards models. Am J Epidemiol. 180(9):93340. https://doi.org/10.1093/aje/kwu211 35
PMID:25245043 36
Rothman KJ, Greenland S, Poole C, Lash TL (2008). Causation and causal inference. In: 37
Rothman KJ, Greenland S, Lash TL, editors. Modern epidemiology. 3rd ed. Philadelphia 38
(PA), USA: Wolters Kluwer Health/Lippincott Williams & Wilkins; pp. 531. 39
Sherman CD, Portier CJ, Kopp-Schneider A (1994). Multistage models of carcinogenesis: an 40
approximation for the size and number distribution of late-stage clones. Risk Anal. 41
14(6):103948. https://doi.org/10.1111/j.1539-6924.1994.tb00074.x PMID:7846311 42
Smith MT, Guyton KZ, Gibbons CF, Fritz JM, Portier CJ, Rusyn I, et al. (2016). Key 43
characteristics of carcinogens as a basis for organizing data on mechanisms of carcinogenesis. 44
Environ Health Perspect. 124(6):71321. https://doi.org/10.1289/ehp.1509912 45
PMID:26600562 46
Stayner L, Steenland K, Dosemeci M, Hertz-Picciotto I (2003). Attenuation of exposure-response 47
curves in occupational cohort studies at high exposure levels. Scand J Work Environ Health. 48
29(4):31724. https://doi.org/10.5271/sjweh.737 PMID:12934726 49
IARC Monographs Preamble
41
Vandenbroucke JP, Broadbent A, Pearce N (2016). Causality and causal inference in 1
epidemiology: the need for a pluralistic approach. Int J Epidemiol. 45(6):177686. 2
https://doi.org/10.1093/ije/dyv341 PMID:26800751 3
Vermeulen R, Bell DA, Jones DP, Garcia-Closas M, Spira A, Wang TW, et al. (2018). 4
Application of biomarkers in cancer epidemiology. In: Thun MJ, Linet MS, Cerhan JR, 5
Haiman CA, Schottenfeld D, editors. Schottenfeld and Fraumeni cancer epidemiology and 6
prevention. 4th ed. New York (NY), USA: Oxford University Press; pp. 7796. 7
Wild CP, Cogliano VJ (2011). A plea on behalf of expert evaluation and the experts involved. Int 8
J Epidemiol. 40(1):25361. https://doi.org/10.1093/ije/dyq038 PMID:20223796 9