part 3. statistical analyses of concordance and key characteristics

chapter 22. Analysis of key characteristics of human carcinogens Daniel Krewski, Mustafa Al-Zoughool, Michael Bird, Nicholas Birkett, Mélissa Billard, Brittany Milton, Jerry M. Rice, Vincent J. Cogliano, Mark A. Hill, Julian Little, and Jan M. Zielinski (deceased) in collaboration with other participants (see the Contributors list) in the Workshop on Tumour Site Concordance and Mechanisms of Carcinogenesis, which was convened by IARC in April and November 2012 in Lyon

Introduction evaluate and integrate the evidence limited evidence of carcinogenicity,

provided by human epidemiological inadequate evidence of carcinogen- PART 3

Since its establishment in the ear- studies, animal cancer bioassays, icity, or evidence suggesting lack of CHAPTER 22 ly 1970s, the IARC Monographs and information on possible biolog- carcinogenicity. The information on Programme has evaluated more ical mechanisms of action, to clas- biological mechanisms of action may than 1000 agents with evidence of sify agents into one of the following be evaluated as strong, moderate, or human exposure and for which some categories: carcinogenic to humans weak, and is taken into consideration suspicion exists of an increased (Group 1), probably carcinogenic to in the overall evaluation. cancer risk to humans. The IARC humans (Group 2A), possibly carci- The role of mechanistic informa- Monographs Programme has devel- nogenic to humans (Group 2B), not tion in evaluating carcinogenicity oped detailed criteria against which classifiable as to its carcinogenicity to has increased substantially during to evaluate the available scientific humans (Group 3), and probably not the history of the IARC Monographs evidence on the carcinogenic poten- carcinogenic to humans (Group 4). Programme. In 1991, IARC convened tial of such agents. These criteria, These evaluations involve classifying a Working Group on the Use of Data which are described in the Preamble the data from both the human and on Mechanisms of Carcinogenesis to the IARC Monographs (Cogliano the animal studies as providing suf- in Risk Identification, to explore how et al., 2004; IARC, 2006), are used to ficient evidence of carcinogenicity, mechanistic data could be used to

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 257 identify agents with the potential to processes at the cellular and mo- electrophile can react with cellu- cause cancer in humans. The con- lecular level and thought to be rele- lar macromolecules such as DNA, sensus report of the Working Group vant to carcinogenesis. Information RNA, and proteins to form adducts. documented several mechanisms on these 24 end-points was derived Some chemical carcinogens are di- that were considered to be relevant from human in vivo, human in vitro, rect-acting electrophiles (e.g. form- to human carcinogenesis at that animal in vivo, and animal in vitro aldehyde; sulfur mustard, and ethyl- time, including genotoxicity, cell studies (see Al-Zoughool et al., 2019). ene oxide), whereas others require proliferation, receptor mechanisms During the November 2012 meeting, biotransformation by enzymes in a in mitogenesis, alterations in DNA the Workshop participants identified process termed metabolic activation repair, intercellular communication, 10 broader key characteristics of car- (e.g. polycyclic aromatic hydrocar- and immune defects and immuno- cinogens (see Chapter 10, by Smith, bons and benzene) (Miller, 1970). suppression (Vainio et al., 1992). and Smith et al., 2016). Information Characteristic 2: Is genotoxic Toxicokinetic and other variables on these characteristics was extract- were also identified as factors affect- ed from the IARC Monographs and Genotoxicity is the ability to induce ing multistage carcinogenesis. Since used to develop a database of key DNA damage or other chromosomal 1991, IARC (2006) and other orga- characteristics for Group 1 agents alterations, as measured by three nizations – for example, the United (see Al-Zoughool et al., 2019). This associated toxicological end-points: States National Toxicology Program chapter focuses on the key charac- (i) DNA damage: an alteration in (National Toxicology Program, 2014) teristics of the Group 1 agents iden- the chemical structure or integrity of DNA, including a break in a DNA and the United States Environmental tified in the IARC Monographs up to strand, and/or chemical modifica- Protection Agency (EPA) (EPA, and including Volume 106, and pre- tions such as covalent binding to 2005) – have stressed the increas- sents the results of an exploratory the nucleotide bases (Hoeijmakers, ing importance of mechanistic infor- analysis of this database. 2009); (ii) gene mutations: changes mation in cancer risk assessment. in the normal nucleotide sequence of Related risk assessment practic- Methods cellular DNA that may have a central es concern mode of action (Meek Key characteristics role in human carcinogenesis (Ding et al., 2014) and pathways of toxic- et al., 2008); (iii) clastogenic effects ity (Krewski et al., 2014; Bourdon- As mentioned above, Chapter 10, reflect damage to chromosomes, in- Lacombe et al., 2015; Cote et al., by Smith, and Smith et al. (2016) cluding DNA breakage, or the rear- 2016), as well as dosimetric consid- describe 10 key characteristics of rangement, gain, or loss of chromo- erations (Gurusankar et al., 2017). human carcinogens, as listed in some fragments (Snyder, 2010). This chapter examines the avail- Table 22.2. The toxicological end- Characteristic 3: Alters DNA able data on mechanisms of action points initially considered by the repair or causes genomic of the Group 1 agents identified up Workshop participants and used instability to and including Volume 106 of the as indicators of these characteris- IARC Monographs (Table 22.1). The tics are also noted in Table 22.2. Alterations in DNA repair result in de- present analysis is based on a re- A brief summary of each of these fects in processes that monitor and view of human cancer mechanisms, characteristics and the associated correct DNA replication fidelity. Such conducted by the participants in the toxicological end-points is provided defects can confer strong mutator two-part Workshop on Tumour Site below. phenotypes that result in genomic Concordance and Mechanisms of instability. Characteristic 1: Is electrophilic Carcinogenesis, which was con- or can be metabolically Characteristic 4: Induces vened by IARC in April and Novem- activated to electrophiles epigenetic alterations ber 2012 in Lyon. This approach initially involved retrieval of informa- The first characteristic refers to Induced epigenetic alterations are tion from the IARC Monographs on agents that act as electrophiles stable changes in gene expression 24 toxicological end-points identi- themselves or that can be metab- and chromatin organization that are fied as likely indicators of biological olized to form electrophiles. An independent of mutation and that

258 Table 22.1. Number of Group 1 agents in Volumes 100–118 of the IARC Monographs, by type of agenta

Type of agent Volume Total

100 105 106 107 109 110 111 113 114 117 118

Pharmaceuticals 23 – – – – – – – – – – 23

Biological agents 11 – – – – – – – – – – 11

Arsenic, metals, 10 – – – – – 2b – – – – 12 fibres, and dusts

Radiation 18 – – – – – – – – – 1c 19

Personal habits and 12 – – – – – – – 1d – – 13 indoor combustions

Chemical agents and 33 1e 1f 2g 2h 1i – 1j – 1k 1l 43 related occupations

Total 107 1 1 2 2 1 2 1 1 1 2 121

a At the time that the present analysis was conducted, mechanistic information was available only for the 109 Group 1 agents evaluated in the IARC Monographs up to and including Volume 106. b Fluoro-edenite fibrous amphibole; occupational exposures associated with the Acheson process in the manufacture of silicon carbide fibres. c Ultraviolet radiation from welding. d Processed meat. e Diesel engine exhaust. f Trichloroethylene. g Polychlorinated biphenyls (PCBs); dioxin-like PCBs. h Outdoor air pollution; particulate matter in outdoor air pollution. i 1,2-Dichloropropane. j Lindane. k Pentachlorophenol (PCP). l Welding fumes. can be inherited through cell divi- reactive oxygen and detoxification of leading to the recruitment and acti- sion. Epigenetic phenomena include the radical species within cells and vation of inflammatory cells. Strong genomic imprinting, X-chromosome tissues. Reactive oxygen species links exist between inflammation PART 3 inactivation, global reconfiguration induce a cascade of events that can and the induction of oxidative stress CHAPTER 22 of the DNA methylome, changes in include DNA mutation and oxidative and genomic instability; this makes chromatin compaction states and DNA damage. Both are key events in it difficult to separate out the rel- histone modification patterns, and carcinogenesis (Klaunig et al., 2011). ative importance of each of these altered expression of microRNAs mechanisms. These linkages might Characteristic 6: Induces (miRNAs). These phenomena occur be the basis of the relationship be- chronic inflammation during organ development and con- tween chronic inflammation and can- tribute to the lineage-specific epi- Chronic inflammation can arise from cer (Multhoff and Radons, 2012). genome that is maintained over the persistent infection (e.g. with human Characteristic 7: Is lifetime of an organism. papillomavirus or with Helicobacter immunosuppressive pylori) as well as from exogenous ir- Characteristic 5: Induces ritants (e.g. silica or asbestos fibres). Immunosuppression is an induced oxidative stress Persistent infection and chronic in- reduction in the capacity of the Oxidative stress results from an flammation disrupt local tissue ho- immune system to respond effec- imbalance between formation of meostasis and alter cell signalling, tively to foreign antigens, including

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 259 Table 22.2. Key characteristics and toxicological end-points demonstrated by agents known to cause cancer in humans (adapted from Al-Zoughool et al., 2019)

Key characteristic Corresponding toxicological end-points

Is electrophilic or can be metabolically activated Metabolic activation to electrophiles Protein adducts ADME (differences in absorption, distribution, metabolism, and elimination)

Is genotoxic DNA damage Cytogenetic/clastogenic effects Gene mutations

Alters DNA repair or causes genomic instability DNA repair alteration, leading to genomic instability

Induces epigenetic alterations Epigenetic alterations (DNA methylation, histone modification, and altered expression of microRNAs)

Induces oxidative stress Oxidative stress

Induces chronic inflammation Chronic inflammation Chronic irritation

Is immunosuppressive Immune effects

Modulates receptor-mediated effects Receptor-mediated effects Hormonal effects

Causes immortalization Immortalization Alterations in telomere length

Alters cell proliferation, cell death, or nutrient Cell-cycle effects supply Bystander effects Alterations in cell signalling pathways Angiogenic effects Cell death Inhibition of gap-junctional intercellular communication

antigens on tumour cells. In contrast receptors, thereby inducing or mod- Characteristic 9: Causes to other key characteristics, immuno- ifying a plethora of signal transduc- immortalization suppression does not play a direct tion pathways that, among other re- Immortalization refers to a cell evad- part in transforming normal cells into sponses, stimulate cell proliferation. ing normal cellular senescence and tumour cells, but enables them to es- Receptor-mediated effects can in- proliferating indefinitely. In culture, cape immune surveillance. Among duce hormonal effects whereby ex- normal cells have a fixed number of other roles, the immune system also ternal agents can interfere with the replication cycles before they enter plays a major part in the inflammato- synthesis, secretion, transport, bind- cellular senescence and stop repli- ry response to injury. ing, action, or elimination of natural cating. Evasion of senescence is fre- quently associated with activation of Characteristic 8: Modulates hormones in the body. These exter- telomerase (Willeit et al., 2010) and receptor-mediated effects nal factors can also demonstrate re- plays a critical part in carcinogenesis activity similar to endogenously pro- Modulation of receptor-mediated ef- (Reddel, 2000). Carcinogenesis may fects can occur when agents mimic duced hormones, which can lead to involve activation of a telomerase the structure of endogenous ligands them mediating changes in homeo- that prevents loss of telomere length, that bind to cells and activate cell stasis, reproduction, development, leading to immortalization of cells surface receptors or intracellular or behaviour. (Willeit et al., 2010).

260 Characteristic 10: Alters cell (evaluated in Volume 106; Guha for 109 Group 1 agents identified in proliferation, cell death, or et al., 2012; IARC, 2014). Had these the IARC Monographs up to and in- nutrient supply two agents been evaluated within cluding Volume 106. The 86 Group 1 Volume 100, they would have been agents for which separate mech- Cell proliferation is affected by alter- included in Volume 100F; they have anistic summaries are provided in ations in the rates of cell growth with- therefore been listed with other the IARC Monographs up to and in a tissue. It may be a direct effect or chemical agents and related occupa- including Volume 106 are listed in a secondary regenerative effect after tions in Volume 100F*. Table 22.3, along with their relation- induction of cell death by cytotoxic Although additional Group 1 ship to the 111 distinct agents identi- agents. Two associated toxicological agents have since been identified fied up to and including Volume 109 end-points are (i) cell-cycle effects, (Table 22.1), the present analysis is used by Krewski et al. (Chapter 21) i.e. alterations in the functioning restricted to Group 1 agents identi- in a parallel analysis of overlap be- of the complex series of factors fied in the IARC Monographs up to tween tumours and tumour sites in that control the cell cycle and cell and including Volume 106, the most animals and humans. division, which have been associat- recent volume for which mechanis- ed with carcinogenesis (Diaz-Moralli tic information was available at the Database of mechanistic et al., 2013), and (ii) alterations in cell time of the present analysis. Group 1 characteristics signalling pathways, which relate to agents not included in the present A database of toxicological end- the ability of the agent to interfere analysis are (i) polychlorinated bi- points was assembled for the 86 with cell signalling pathways, leading phenyls (PCBs) and dioxin-like PCBs Group 1 agents identified up to and to expression of a carcinogenic trait (reviewed in Volume 107; Lauby- including Volume 106 of the IARC or phenotype in the cell. Secretan et al., 2013; IARC, 2016b), Monographs. The database in- For cell death, necrosis triggers (ii) outdoor air pollution and (iii) par- cludes information from in vivo and the invasion of cells such as mac- ticulate matter in outdoor air pollu- in vitro studies in humans and ani- rophages into the affected area, tion (both evaluated in Volume 109; mals. Information on the 24 toxico- and enhances the proliferation and Loomis et al., 2013; IARC, 2016a), logical end-points was retrieved from spread of cancer cells. Defects in (iv) 1,2-dichloropropane (reviewed in Section 4 (“Other relevant data”) of programmed cell death can cause Volume 110; Benbrahim-Tallaa et al., the IARC Monographs (Al-Zoughool cancer; evasion of apoptosis is a re- 2014; IARC, 2017a), (v) fluoro-eden- et al., 2019). quirement for both neoplastic trans- ite fibrous amphibole and (vi) occu- Recognizing that, among other formation and sustained growth of pational exposures associated with limitations, new data may have be- cancer cells. the Acheson process used in the come available since 2009, when Adequate cell nutrition is essen- manufacture of silicon carbide fi-

the various parts of Volume 100 were PART 3 tial to proliferating cancer cells, and bres (both evaluated in Volume 111; agents that promote or inhibit the Grosse et al., 2014; IARC, 2017b); compiled, PubMed searches were CHAPTER 22 growth of blood vessels (angiogene- (vii) lindane (Volume 113; Loomis conducted to identify evidence on sis) will affect tumour growth. et al., 2015), and (viii) processed any of the 24 toxicological end-points meat (Volume 114; Bouvard et al., linked to these agents that was not Group 1 agents included in 2015). recorded in the IARC Monographs the analysis In some cases, the discussion of (Birkett et al., 2019). The mechanistic Volume 100 of the IARC Monographs mechanisms of action in Section 4 database distinguishes information provided a review and update of the (“Other relevant data”) of the IARC derived from the Monographs from 107 Group 1 agents identified as Monographs is based on groups of that found in the PubMed search, of 2009. Since the publication of agents that act via the same mech- thereby permitting an assessment Volume 100, mechanistic informa- anism. For example, internalized of the extent to which Section 4 tion has become available on two radionuclides that emit α-particles (“Other relevant data”) of the IARC additional Group 1 agents: diesel are discussed in the Monographs as Monographs captured all relevant engine exhaust (reviewed in Volume a group with the same mechanism information on these end-points. 105; Benbrahim-Tallaa et al., 2012; of action. Birkett et al. (2019) re- The analyses in the present chapter IARC, 2013) and trichloroethylene viewed the mechanistic information are restricted to information taken

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 261 Table 22.3. Relationship between 86 agents used in the analysis of key characteristics of human carcinogens and 111 agents used in the analysis of concordance between tumours and tumour sites in humans and animals

Volumea Agent 86 agents used in the analysis of key 111 agents used in the analysis of number characteristics concordance between tumours and tumour sites in humans and animals

100A 1 Aristolochic acid Aristolochic acid Aristolochic acid, plants containing

100A 2 Azathioprine Azathioprine

100A 3 Busulfan Busulfan

100A 4 Chlorambucil Chlorambucil

100A 5 Chlornaphazine Chlornaphazine

100A 6 Cyclophosphamide Cyclophosphamide

100A 7 Ciclosporin Ciclosporin

100A 8 Diethylstilbestrol Diethylstilbestrol

100A 9 Estrogen-only menopausal therapy Estrogen-only menopausal therapy

100A 10 Estrogen–progestogen menopausal therapy Estrogen–progestogen menopausal therapy (combined) (combined)

100A 11 Estrogen–progestogen oral contraceptives Estrogen–progestogen oral contraceptives (combined) (combined)

100A 12 Etoposide in combination with cisplatin Etoposide (Group 2A) and bleomycin (Group 2B) Etoposide in combination with cisplatin and bleomycin

100A 13 Melphalan Melphalan

100A 14 PUVA (psoralen–UVA photochemotherapy) Methoxsalen in combination with UVA

100A 15 MOPP MOPP

100A 16 Phenacetin Phenacetin Phenacetin, analgesic mixtures containing

100A 17 1-(2-Chloroethyl)-3-(4-methylcyclohexyl)- 1-(2-Chloroethyl)-3-(4-methylcyclohexyl)- 1-nitrosourea (Methyl-CCNU) 1-nitrosourea (Methyl-CCNU)

100A 18 Tamoxifen Tamoxifen

100A 19 Thiotepa Thiotepa

100A 20 Treosulfan Treosulfan

100B 21 Opisthorchis viverrini and Clonorchis sinensis Clonorchis sinensis (infection with) Opisthorchis viverrini (infection with)

100B 22 Epstein–Barr virus Epstein–Barr virus

100B 23 Helicobacter pylori Helicobacter pylori (infection with)

100B 24 Hepatitis B virus Hepatitis B virus

100B 25 Hepatitis C virus Hepatitis C virus

100B 26 Human immunodeficiency virus type 1 Human immunodeficiency virus type 1

100B 27 Human papillomavirus Human papillomavirus

262 Table 22.3. Relationship between 86 agents used in the analysis of key characteristics of human carcinogens and 111 agents used in the analysis of concordance between tumours and tumour sites in humans and animals (continued)

Volumea Agent 86 agents used in the analysis of key 111 agents used in the analysis of number characteristics concordance between tumours and tumour sites in humans and animals

100B 28 Human T-cell lymphotropic virus type 1 Human T-cell lymphotropic virus type 1

100B 29 Kaposi sarcoma-associated herpesvirus Kaposi sarcoma-associated herpesvirus

100B 30 Schistosoma haematobium Schistosoma haematobium (infection with)

100C 31 Arsenic and inorganic arsenic compounds Arsenic and inorganic arsenic compounds

100C 32 Asbestos (all forms, including actinolite, amosite, Asbestos (all forms, including actinolite, amosite, anthophyllite, chrysotile, crocidolite, and anthophyllite, chrysotile, crocidolite, and tremolite) tremolite)

100C 33 Beryllium and beryllium compounds Beryllium and beryllium compounds

100C 34 Cadmium and cadmium compounds Cadmium and cadmium compounds

100C 35 Chromium(VI) compounds Chromium(VI) compounds

100C 36 Erionite Erionite

100C 37 Leather dust Leather dust

100C 38 Nickel and nickel compounds Nickel compounds

100C 39 Silica dust, crystalline, in the form of quartz or Silica dust, crystalline, in the form of quartz or cristobalite cristobalite

100C 40 Wood dust Wood dust

100D 41 Solar and UV radiation UV radiation (bandwidth 100–400 nm, encompassing UVC, UVB, and UVA) UV-emitting tanning devices Solar radiation

100D 42 X- and γ-radiation X- and γ-radiation Ionizing radiation (all types) PART 3 100D 43 Neutron radiation Neutron radiation CHAPTER 22 100D 44 Internalized radionuclides that emit α-particles Haematite mining with exposure to radon (underground) Plutonium-239 Internalized radionuclides that emit α-particles Thorium-232 (as Thorotrast) Radium-224 and its decay products Radium-226 and its decay products Radium-228 and its decay products Radon-222 and its decay products

100D 45 Internalized radionuclides that emit β-particles Fission products including Sr-90 Radioiodines, including iodine-131 Phosphorus-32, as phosphate Internalized radionuclides that emit β-particles

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 263 Table 22.3. Relationship between 86 agents used in the analysis of key characteristics of human carcinogens and 111 agents used in the analysis of concordance between tumours and tumour sites in humans and animals (continued)

Volumea Agent 86 agents used in the analysis of key 111 agents used in the analysis of number characteristics concordance between tumours and tumour sites in humans and animals

100E 46 Consumption of alcoholic beverages Acetaldehyde associated with consumption of alcoholic beverages Alcoholic beverages Ethanol in alcoholic beverages

100E 47 Betel quid and areca nut Areca nut Betel quid with tobacco Betel quid without tobacco

100E 48 Coal, indoor emissions from household Coal, indoor emissions from household combustion of combustion of

100E 49 N′-Nitrosonornicotine (NNN) and 4-(Methyl- N′-Nitrosonornicotine (NNN) and 4-(Methyl- nitrosamino)-1-(3-pyridyl)-1-butanone (NNK) nitrosamino)-1-(3-pyridyl)-1-butanone (NNK)

100E 50 Salted fish, Chinese-style Salted fish, Chinese-style

100E 51 Second-hand tobacco smoke Second-hand tobacco smoke

100E 52 Tobacco smoking Tobacco smoking

100E 53 Tobacco, smokeless Tobacco, smokeless

100F 54 Acid mists, strong inorganic Acid mists, strong inorganic

100F 55 Aflatoxins Aflatoxins

100F 56 Aluminium production Aluminium production

100F 57 4-Aminobiphenyl 4-Aminobiphenyl

100F 58 Auramine production Auramine production

100F 59 Benzene Benzene

100F 60 Benzidine Benzidine

100F 61 Benzidine, dyes metabolized to Benzidine, dyes metabolized to

100F 62 Benzo[a]pyrene Benzo[a]pyrene

100F 63 Bis(chloromethyl)ether; Chloromethyl methyl Bis(chloromethyl)ether; Chloromethyl methyl ether (technical grade) ether (technical grade)

100F 64 1,3-Butadiene 1,3-Butadiene

100F 65 Coal gasification Coal gasification

100F 66 Coal-tar distillation Coal-tar distillation

100F 67 Coal-tar pitch Coal-tar pitch

100F 68 Coke production Coke production

100F 69 Ethylene oxide Ethylene oxide

100F 70 Formaldehyde Formaldehyde

100F 71 Iron and steel founding, occupational exposure Iron and steel founding, occupational exposure during during

264 Table 22.3. Relationship between 86 agents used in the analysis of key characteristics of human carcinogens and 111 agents used in the analysis of concordance between tumours and tumour sites in humans and animals (continued)

Volumea Agent 86 agents used in the analysis of key 111 agents used in the analysis of number characteristics concordance between tumours and tumour sites in humans and animals

100F 72 Isopropyl alcohol manufacture using strong acids Isopropyl alcohol manufacture using strong acids

100F 73 Magenta production Magenta production

100F 74 4,4′-Methylenebis(2-chloroaniline) (MOCA) 4,4′-Methylenebis(2-chloroaniline) (MOCA)

100F 75 Mineral oils, untreated or mildly treated Mineral oils, untreated or mildly treated

100F 76 2-Naphthylamine 2-Naphthylamine

100F 77 ortho-Toluidine ortho-Toluidine

100F 78 Painter, occupational exposure as a Painter, occupational exposure as a

100F 79 2,3,7,8-Tetrachlorodibenzo-para-dioxin, 2,3,4,7,8-Pentachlorodibenzofuran 2,3,4,7,8-Pentachlorodibenzofuran, 2,3,7,8-Tetrachlorodibenzo-para-dioxin 3,3′,4,4′,5-Pentachlorobiphenyl 3,3′,4,4′,5-Pentachlorobiphenyl

100F 80 Rubber manufacturing industry, occupational Rubber manufacturing industry, occupational exposures in the exposures in the

100F 81 Shale oils Shale oils

100F 82 Soot (as found in occupational exposure of Soot (as found in occupational exposure of chimney sweeps) chimney sweeps)

100F 83 Sulfur mustard Sulfur mustard

100F 84 Vinyl chloride Vinyl chloride

105 85 Diesel and gasoline engine exhausts Engine exhaust, diesel

106 86 Trichloroethylene Trichloroethylene

107 Polychlorinated biphenylsb

109 Outdoor air pollutionb PART 3

b 109 Particulate matter in outdoor air pollution CHAPTER 22

UV, ultraviolet. a IARC Monographs Volumes 100A (IARC, 2012e), 100B (IARC, 2012b), 100C (IARC, 2012a), 100D (IARC, 2012f), 100E (IARC, 2012d), 100F (IARC, 2012c), 105 (IARC, 2013), 106 (IARC, 2014), 107 (IARC, 2016b), and 109 (IARC, 2016a). b Because the mechanistic sections for Monographs Volumes 107–109 were not available for review at the time that the present analysis was conducted, Group 1 agents in these volumes were not included in the present analysis.

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 265 Fig. 22.1. Key characteristics of 86 Group 1 agents. The number of agents is shown above each characteristic.

directly from the IARC Monographs: animal in vivo and animal in vitro sources of information noted above. Birkett et al. (2019) present the re- sources, it is possible to aggregate To evaluate the extent to which the sults of a sensitivity analysis incorpo- this information according to human Group 1 agents demonstrated more rating the additional information ob- and animal sources (by combining than one key characteristic, the num- tained through the PubMed search. across in vivo and in vitro sources) ber of agents demonstrating multiple After the collection of informa- or according to in vivo and in vitro characteristics was also tabulated. tion on the toxicological end-points sources (by combining across hu- A heat map showing the number man and animal sources). Of primary identified by the Workshop partici- (0, 1, 2, 3, or 4) of sources of infor- pants during the April 2012 meeting, interest here is aggregation across mation (human in vivo, human in vi- the database of key characteristics all four sources combined, to obtain tro, animal in vivo, and animal in vitro was then created by mapping the an overall indicator of whether any of studies) supporting a given charac- 24 toxicological end-points to the the key characteristics is associated teristic for a specified agent was pre- 10 characteristics as indicated in with each of the 86 Group 1 agents pared, to evaluate the consistency Table 22.2. As noted by Al-Zoughool of interest. of information provided by different et al. (2019), two of the toxicological Statistical analysis end-points – susceptibility and sources. A heat map showing the changes in gene expression – did Descriptive statistical methods were overlap between human and animal not link to any of the key character- used to explore the key characteris- sources of information (after combin- istics, and thus were not included in tics associated with the 86 Group 1 ing in vivo and in vitro sources in both the development of the database of agents, beginning with a tabulation cases) on the key characteristics was key characteristics. Because the da- of the number of agents demonstrat- also prepared, to evaluate the extent tabase includes information derived ing any of the 10 characteristics, to which there was overlap between from human in vivo, human in vitro, both overall and stratified by the four these two sources.

266 PART 3 CHAPTER 22 Sources of information characteristics on key of 86 Group 1 agents (sources are human in vivo, human in vitro, animal in vivo, and animal in vitro Human Human In Vivo Human In Vitro Animal In Vivo Animal In Vitro 22.2. Fig. studies).

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 267 Overall mechanistic data were epigenetic alterations”, and “induc- for each agent. As in Fig. 22.1, the also tabulated by type of agent es chronic inflammation”. Human in single most prominent characteristic (pharmaceuticals; biological agents; vitro studies provide the most infor- was genotoxicity: many agents (HIV- arsenic, metals, fibres and dusts; ra- mation on an additional three key 1 is a prominent exception) showed diation; personal habits and indoor characteristics: “alters DNA repair or a positive response for genotoxicity combustions; and chemical agents causes genomic instability”, “induces in at least one of the four sources and related occupations), to identify oxidative stress”, and “alters cell pro- of information, and for many agents possible differences in mechanistic liferation, cell death, or nutrient sup- more than one source provided ev- patterns by agent type. ply”, and equivalent information to idence of genotoxicity. For some animal in vivo studies on “modulates agents (e.g. all radiation sources, Results receptor-mediated effects”. some pharmaceutical agents, and The prominence of human stud- The key characteristics of the 86 some chemical agents), genotoxicity ies as sources of information on Group 1 agents considered here are was demonstrated in all four test sys- the key characteristics of human summarized in Fig. 22.1. The most tems, confirming that genotoxicity is carcinogens may be attributed to prevalent mechanistic characteristic central to the carcinogenic pathways the increasing use of molecular and was “is genotoxic”, followed by “al- of these agents. genetic markers in human studies. ters cell proliferation, cell death, or Fig. 22.4 also shows that most Epidemiological studies conducted nutrient supply”, “induces oxidative agents exhibit multiple key charac- in the occupational or general envi- stress”, “is electrophilic or can be teristics, with evidence drawn from ronment often analyse biomarkers metabolically activated to electro- more than one source of mechanis- of DNA adduct formation, clasto- philes”, and “induces chronic inflam- tic information. Radiation sources genic effects, and gene mutations, mation”. Nearly all agents demon- and tobacco smoke are associated all of which reflect DNA damage. strate genotoxicity as one of their with many of the key characteristics, Therefore, human in vivo studies mechanistic properties; a prominent suggesting that these agents act by are a major source of information on exception is human immunodeficien- multiple pathways. genotoxicity. cy virus type 1 (HIV-1). Evidence of Several Group 1 agents, including Fig. 22.3 shows the number of genotoxicity was provided by expres- several occupational exposures, are agents demonstrating multiple char- sion of the following toxicological complex mixtures of chemicals and acteristics as evidenced from stud- end-points: DNA damage, gene mu- other substances. Coal-tar pitch, tations, and cytogenetic/clastogenic ies in animals and in humans. The occupational exposure to soot, and effects (including chromosomal ab- 86 Group 1 agents considered here coke production have similar charac- errations, micronucleus formation, exhibit an average of approximately teristics, probably due to the strong and aneuploidy). four key characteristics; the modal presence in relevant workplaces of Fig. 22.2 shows the key charac- value is two characteristics, exhibit- polycyclic aromatic hydrocarbons, teristics exhibited by the 86 agents ed by 20 agents. All agents demon- although other factors such as the classified according to the source of strate at least one key characteristic, nature of inorganic substances and data (human in vivo, human in vitro, with two agents demonstrating nine their chemical composition could animal in vivo, and animal in vitro characteristics and 14 agents show- also have a role. Other occupation- studies) on these characteristics. ing six. No agent exhibited all 10 key ally relevant agents (e.g. exposures Information on all the mechanistic characteristics. during iron and steel founding and characteristics was available to dif- Fig. 22.4 presents a heat map aluminium production) demonstrate ferent degrees from all four sourc- indicating the strength of evidence only a single key characteristic, al- es. Information on genotoxicity was of the different characteristics for though this may reflect the difficulty available from each of the four sourc- the 86 individual Group 1 agents. of testing for other characteristics es for the majority of the agents. The intensity of the colour reflects in these occupational exposure Human in vivo studies contribute the the number of sources of informa- situations. most evidence on four of the 10 key tion (human in vivo, human in vitro, The degree of overlap between characteristics for these 86 agents, animal in vivo, and animal in vitro human and animal sources of infor- including “is genotoxic”, “induces studies) on each key characteristic mation on the 10 key characteristics

268 Fig. 22.3. Number of Group 1 agents demonstrating one or more key characteristics.

of human carcinogens is shown in icity. Ten agents – diethylstilbestrol, oxidative stress”), but Fig. 22.5 clar- the heat map in Fig. 22.5. This heat Kaposi sarcoma-associated herpes- ifies that there are both human and map, prepared by combining the in virus, arsenic and inorganic arsenic animal data for only five of these. vivo and in vitro sources of informa- compounds, cadmium and cadmium For chlornaphazine, Fig. 22.4 shows tion on the key characteristics for compounds, asbestos, crystalline two sources of information, for “is PART 3 humans and for animals, indicates silica, solar and ultraviolet radiation, electrophilic or can be metabolical- CHAPTER 22 whether information on the key char- sulfur mustard, diesel and gaso- ly activated to electrophiles” and “is acteristics for a given agent is de- line engine exhausts, and trichloro- genotoxic”, whereas the correspond- rived from both human and animal ethylene – demonstrate overlap be- ing data in Fig. 22.5 show overlap sources (reflecting concordance tween human and animal sources of between human and animal sources between humans and animals), from information for at least five of the key only for “is electrophilic or can be human sources alone, from animal characteristics. metabolically activated to electro- sources alone, or from neither of Comparisons between the results philes”, with human but not animal these. These results indicate overlap in Fig. 22.4 and Fig. 22.5 can pro- data on “is genotoxic”. between human and animal sourc- vide additional insights into the key Fig. 22.6 shows the key character- es of information for several agents. characteristics of the Group 1 agents istics of the six categories of Group 1 The concordance is particularly considered here. For example, in the agents considered in Volume 100: strong for genotoxicity: information case of diethylstilbestrol, Fig. 22.4 pharmaceuticals; biological agents; from both human and animal sourc- indicates that there is information arsenic, metals, fibres, and dusts; es is available for 63 of the 85 agents from 1, 2, or 3 sources on nine key radiation; personal habits and in- demonstrating evidence of genotox- characteristics (all except “induces door combustions; and chemical

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 269 Fig. 22.4. Heat map showing the strength of evidence for key characteristics of 86 Group 1 agents according to the number of information sources (sources are human in vivo, human in vitro, animal in vivo, and animal in vitro studies).

270 Fig. 22.5. Heat map showing the degree of concordance between human and animal sources of information on key characteristics of 86 Group 1 agents (after combining in vivo and in vitro sources of information for humans and for animals). PART 3 CHAPTER 22

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 271 Fig. 22.6. Key characteristics of 86 Group 1 agents by type of agent (expressed as a percentage of the number of agents of each type demonstrating each of the 10 mechanistic characteristics): (a) pharmaceuticals; (b) biological agents; (c) arsenic, metals, fibres, and dusts; (d) radiation; (e) personal habits and indoor combustions; and (f) chemical agents and related occupations. a

b

272 Fig. 22.6. Key characteristics of 86 Group 1 agents by type of agent (expressed as a percentage of the number of agents of each type demonstrating each of the 10 mechanistic characteristics): (a) pharmaceuticals; (b) biological agents; (c) arsenic, metals, fibres, and dusts; (d) radiation; (e) personal habits and indoor combustions; and (f) chemical agents and related occupations (continued). c

d PART 3 CHAPTER 22

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 273 Fig. 22.6. Key characteristics of 86 Group 1 agents by type of agent (expressed as a percentage of the number of agents of each type demonstrating each of the 10 mechanistic characteristics): (a) pharmaceuticals; (b) biological agents; (c) arsenic, metals, fibres, and dusts; (d) radiation; (e) personal habits and indoor combustions; and (f) chemical agents and related occupations (continued). e

f

274 agents and related occupations. al., 2019 and Birkett et al., 2019). Information drawn from the IARC Genotoxicity was the most preva- The profiles of key characteristics Monographs showed that the over- lent key characteristic demonstrat- of these agents show several inter- whelming majority of the agents ex- ed by agents in the categories of esting patterns. First, all but seven amined here induce one or more of pharmaceuticals; arsenic, metals, agents exhibited multiple character- these end-points. Even biological fibres, and dusts; personal habits istics, an observation consistent with agents such as viruses that act pri- and indoor combustions; and chem- previous findings on the complexity marily through non-genotoxic mech- ical agents and related occupations, and heterogeneity of carcinogenic anisms induce cytogenetic effects and genotoxicity was exhibited by pathways (Hanahan and Weinberg, and gene mutations as secondary all agents in the category of radia- 2011; Floor et al., 2012; Baker, 2014; events through chronic inflammation tion. Immortalization, genotoxicity, Pickup et al., 2014; Roessler et al., and oxidative stress. and altered cell proliferation, cell 2014). Agents in the categories of Another important observation death, or nutrient supply are promi- biological agents; arsenic, metals, fi- is that information on the key char- nent characteristics of the biological bres, and dusts; personal habits and acteristics of the 86 Group 1 agents agents. None of the biological agents indoor combustions; and radiation considered here is often derived from demonstrated modulation of recep- demonstrated a wide spectrum of bi- multiple sources (human in vivo, hu- tor-mediated effects, and none of the ological activity. Radiation has been man in vitro, animal in vivo, and ani- agents in the category of personal linked to many hallmarks of cancer mal in vitro studies); for many agents, habits and indoor combustions ap- (Boss et al., 2014): this mechanistic evidence is available from more than peared to act through modulation of profile, with multiple pathways in- one of these sources. Concordance receptor-mediated effects, through volved for most radiation agents, is between animal and human sources immunosuppression or through im- consistent with the broad spectrum of information was seen for sever- mortalisation. There are five agents of tumours associated with exposure al agents, particularly with respect in the category of radiation, all of to ionizing radiation (Chapter 21, by to genotoxicity, an observation that which demonstrate the following Krewski et al.). Viral oncogenesis lends additional support to the rele- key characteristics: is genotoxic; al- is also multifaceted, and the multi- vance of animal data for human can- ters DNA repair or causes genomic step nature of viral oncogenesis is cer risk assessment. instability; induces oxidative stress; thought to be influenced by host ge- Some caution must be used in causes immortalization; and alters netic variability (Mesri et al., 2014). interpreting the distribution of key cell proliferation, cell death, or nu- Genotoxicity was the most prev- characteristics across the Group 1 trient supply. The profiles of key alent mechanistic characteristic, agents considered here. It is possi- characteristics for pharmaceutical demonstrated by 85 of the 86 agents ble that the near universality of ge- agents and for chemical agents and considered, possibly reflecting the notoxicity as a carcinogenic mech- PART 3 related occupations are remarkably fact that the process of carcino- anism may be related to the way in CHAPTER 22 similar, possibly reflecting the fact genesis necessarily involves geno- which the IARC Monographs were that despite their different expo- mic changes. This finding is consis- compiled, with emphasis on the re- sure circumstances, some of these tent with an earlier evaluation of 180 porting of genotoxicity data. This chemical entities may act via similar Group 1, Group 2A, and Group 2B would have been partially mitigat- mechanisms. agents conducted by Bartsch and ed by the inclusion of mechanistic Malaveille (1989), who reported that information from outside the IARC Discussion 80–90% of the agents in these three Monographs in the preparation of categories demonstrated genotoxic the mechanistic database evaluated The present analysis of key charac- characteristics. In the present ana- separately by Birkett et al. (2019). It teristics of 86 agents classified as lyses, genotoxicity was considered should also be noted that the IARC carcinogenic to humans (Group 1) by to include the following end-points: Monographs have been published the IARC Monographs Programme DNA damage, cytogenetic effects over a long time span, extending was based on mechanistic infor- (including chromosomal aberra- from the early 1970s to the present mation retrieved from the IARC tions, micronucleus formation, and (Saracci and Wild, 2015). Studies of Monographs (see Al-Zoughool et aneuploidy), and gene mutations. agents in earlier Monographs would

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 275 have focused on changes such as information sources, of which the The epigenetic characteristics of DNA damage that could have been most notable was the identification the 74 Group 1 agents considered detected by the techniques available of evidence for six additional agents in Volumes 100A–E were assessed at that time. These agents may not that demonstrate modulation of re- by Herceg et al. (2013). As in the have been evaluated exhaustively ceptor-mediated effects, beyond present analysis, those authors used for pathways that have been iden- the seven agents noted in Fig. 22.1. DNA methylation, histone modifica- tified more recently, such as those Nonetheless, the overall findings are tion, and altered expression of miR- involving the multifactorial nature largely comparable with those pre- NAs as indicators of epigenetic alter- of carcinogenesis, or for the multi- sented without the additional data ations. They considered information plicity of pathways operating dur- search (for further details, see Birkett from both the IARC Monographs ing the process of agent-induced et al., 2019). and the general scientific literature, carcinogenesis. As the IARC Monographs Pro- and identified 22 of the 74 Group 1 A related limitation of the present gramme has evolved from its incep- agents (30%) as demonstrating epi- analysis is that it did not distinguish tion in the early 1970s until the pres- genetic effects. The present analy- direct genotoxicity of the agent or its ent time, the guidelines for carcin- sis, which examined Group 1 agents metabolites from genotoxicity that ogen identification as set out in the in Volumes 100A–F as well as occurs as a result of other respons- Preamble to the IARC Monographs Volumes 105 and 106, identified 33 es, because this information was (IARC, 2006) have been updated of the 86 Group 1 agents (38%) as not generally provided in the IARC from time to time, with increasing mediating epigenetic change. Monographs from which the mech- emphasis on the use of mechanis- In an earlier evaluation, Hernán- anistic data on the Group 1 agents tic information in the overall eval- dez et al. (2009) reported that 45 of the 371 agents (12%) in Groups 1, were abstracted. It is recommended uation in the most recent updates. 2A, and 2B at the time of their analy- that the distinction between primary Nonetheless, the identification of sis were not genotoxic. In their study, and secondary genotoxic effects be Group 1 agents continues to rest an agent was considered non-ge- noted in future Monographs, when- heavily on the availability of suffi- ever possible. notoxic if it showed negative results cient evidence of carcinogenicity in Another limitation of the present in the Salmonella mutagenicity as- epidemiological or clinical studies. results is that they are based on say (the Ames test) as well as in Of the 111 distinct Group 1 agents the information on mechanisms in the mouse lymphoma assay, the in identified up to and including Volume Section 4 (“Other relevant data”) of vitro chromosomal aberration test, 109, no less than 102 demonstrated the IARC Monographs, which fo- the in vitro micronucleus test, the sufficient evidence of carcinogeni- cused primarily on “established” and in vivo micronucleus test, and the city in humans, and the remaining “likely” mechanisms. A full systemat- in vivo chromosomal aberration test nine agents were placed in Group 1 ic review of the entire literature on bi- in bone marrow in rodents. These after reference to mechanistic data ological mechanisms for all Group 1 results support the role of non-ge- or other considerations (as “mech- agents was not undertaken, so the notoxic pathways in carcinogenesis, anistic upgrades” according to the database may not reflect all mech- an observation that is reinforced by anistic characteristics of the differ- evaluation criteria outlined in the the prevalence of multiple charac- ent agents. As a sensitivity analysis Preamble to the IARC Monographs; teristics of human carcinogens not to examine the extent to which the see Table 21.4, in Chapter 21, by associated with genotoxicity in the Monographs captured most of the Krewski et al.). Despite the inherent present analysis. relevant information in this regard, reliance on human epidemiological To ensure that all relevant evi- Birkett et al. (2019) conducted a data in identifying agents that may dence on the 10 key characteristics supplementary PubMed search to increase human cancer risk, Section of human carcinogens is taken into identify additional information on 4 (“Other relevant data”) of the IARC account in future Monographs eval- key characteristics not cited in the Monographs increasingly provides uations of agents that may cause Monographs, or published since detailed descriptions of the mecha- cancer in humans, a carefully de- 2009. Although this sensitivity anal- nisms by which agents under review signed systematic review of the sci- ysis was not based on an exhaus- may act, including agents not as- entific literature would be required tive search, it did identify additional signed to Group 1. in conjunction with each evaluation.

276 However, to conduct a series of com- The 10 key characteristics are and reproducible manner. The EPA prehensive systematic reviews of the features of carcinogens rather than is also currently supporting the de- key characteristics of all 86 agents mechanisms. The analysis presented velopment of software tools specifi- considered in the present analysis here does not address the sequence cally designed for systematic review would require a considerable ef- of events involved in carcinogenesis. of toxicological and epidemiological fort, and was not attempted as part For example, if the carcinogenic data (ICF, 2017). of the present project. The expert mechanism of action is being in- The strong evidence linking geno- opinion of future IARC Working vestigated for a genotoxic agent toxicity to carcinogenesis is consis- Groups charged with evaluating the that requires metabolic activation, tent with epidemiological data and mechanistic data on new agents the mechanism needs to consider experimental research. Genotoxic selected for evaluation by the IARC the entire metabolic pathway. If the effects include the formation of DNA Monographs would be of consider- agent is not metabolized to produce adducts or induction of single- and able value in this regard, but would an electrophile, DNA damage will double-strand DNA breaks. Several ideally be supported by a concomi- not occur. In such a case, biological lines of evidence from epidemio- tant systematic review of the relevant effects that occur after induction of logical studies and in experimental DNA damage also would not be ob- scientific literature on the key char- animals and model systems have served. This sequential relationship acteristics to ensure that the analysis shown that DNA adducts are strongly would be as complete as possible. is also apparent for characteristics associated with cancer (Kriek et al., Another issue that arises when such as chronic inflammation, which 1998; Phillips et al., 2015). Some discussing key characteristics of hu- acts through the production of oxida- genotoxic effects can lead to gene man carcinogens is whether indirect tive stress, release of cytokines, and mutation, an important event in the effects should be considered. Many stimulation of cell proliferation, which pathway towards carcinogenesis, agents have a direct carcinogenic ultimately produces DNA damage. especially if it involves oncogenes or effect, but in other cases the carci- The results of the present analysis tumour suppressor genes. Chromo- nogenic characteristic is the result of can provide a basis for future efforts somal aberrations are another type a secondary event along the mecha- to categorize mechanistic data for of genetic alteration that occurs fre- nistic pathway. For example, cell pro- carcinogens through a systemat- quently in many tumours, especial- liferation can arise either as a result of ic review process. A full systematic ly solid tumours. Most tumour cells a direct action of the agent on the cell review of all agents and all potential display aneuploidy, and for some tu- or indirectly, as a result of cytotoxici- carcinogenic mechanisms is an in- mours, characteristic chromosomal ty that stimulates cell proliferation to timidating prospect. However, such abnormalities have been identified replace cells, through alterations in a review would provide a more com- (e.g. the Philadelphia chromosome cell signalling without cytotoxicity, or prehensive examination of mech- via inhibition of cell proliferation that anisms, because it would include in chronic myeloid leukaemia). PART 3 then results in selection of an altered studies that failed to find effects. It The complexity of the pathways CHAPTER 22 clone of cells with a high prolifera- might also support a process that involved in carcinogenesis and the tion rate. Although the downstream involves a sequence of mechanistic fact that the cellular response to car- effect is the same (increased cell steps and mechanistic characteris- cinogen exposure is modulated by proliferation), the pathway leading to tics relevant to the development of host cell physiology, genetics, and that result can be different. A similar cancer in humans. other variables have prompted the issue arises with genotoxicity: many The importance of systematic re- development and application of sen- agents are not directly genotoxic but view in assembling all relevant evi- sitive assays that measure toxicity cause DNA damage by stimulating dence on a particular issue has been pathways and perturbations in the a chain of molecular changes (e.g. emphasized in the recent review of molecular functioning of the cell. The chronic inflammation). The current the EPA’s Integrated Risk Information newly proposed toxicological test- database does not contain the in- System (IRIS) (National Research ing paradigm (Krewski et al., 2014) formation needed to address these Council, 2014) and is currently being focuses on high-throughput screen- issues and cannot be used to draw implemented within the IRIS pro- ing to detect changes in the molec- conclusions about the detailed gramme as a way of summarizing ular pathways of the cell in response mechanism of action of an agent. all relevant data in a comprehensive to chemical exposure. This new

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 277 paradigm would be useful in com- four characteristics per agent, no work of Krewski et al. (Chapter 21) prehensive cancer risk assessment attempt was made to conduct a mul- that examined concordance be- and would be able to detect distinct tivariate analysis of these character- tween animals and humans for 39 key mechanistic pathways operating istics to determine whether similar tumour sites and 14 organ and tissue after carcinogen exposure. In a sim- agents tended to express similar systems, based on the database on ilar initiative, the Kyoto Encyclopedia characteristics. Recalling that phar- tumours and tumour sites in humans of Genes and Genomes (KEGG) maceuticals as a class demonstrat- and experimental animals developed website has compiled a comprehen- ed a mechanistic profile similar to by Grosse et al. (Annex 1). The pro- sive list of pathways associated with that of chemical agents and related files of key characteristics of agents specific diseases (see the KEGG occupations, it is possible that the associated with specific tumour sites pathway database at http://www.ge­ chemotherapeutic agents and some could be examined to obtain addi- nome.jp/kegg/pathway.html). KEGG of the chemical agents act via the tional insights into the mechanisms also identified major in vitro assays same carcinogenic mechanisms. by which specific tumours occur. It that can be used to detect targets of Cyclophosphamide and benzene would be of particular interest to an- these pathways. This attempt to un- (once used as a chemotherapeutic alyse whether certain tumour sites derstand the biological mechanisms agent) may have some commonality demonstrate signature profiles. of carcinogenesis is consistent with in this respect, as might treosulfan Extending the mechanistic data- current practice of using in vitro as- and butadiene through the formation base to include additional informa- says to detect changes in critical of the same diepoxide. Further study tion such as structural alerts rele- signalling and other molecular path- of these two groups, in terms of both vant to carcinogenesis could also ways in cancer development, as pro- mechanism of action and tumour site be informative. Although the present posed by Krewski et al. (2014). concordance, may provide insight version of the mechanisms database into tumours that result from long- includes the International Union of Further analyses term exposure to chemotherapeutic Pure and Applied Chemistry (IUPAC) The extensive database on key cha- agents. International Chemical Identifier racteristics of human carcinogens Searching for patterns within (InChI) for key chemical coding (Stein developed here offers considerable homogeneous classes of agents et al., 2003; Heller et al., 2015), this potential for further analysis. More would also be of future research in- information has not been taken into in-depth analyses are under way terest. For example, one could ex- account in the analyses completed to to explore the level of agreement amine mechanistic patterns within date. One possible source of auxilia- between mechanistic data derived subgroups of pharmaceuticals, in- ry information on toxicological end- from human sources on the one cluding antineoplastic agents, hor- points that may be related to the 10 hand and from animal sources on monal products, immunosuppres- key characteristics is EPA’s Toxicity the other, as well as from in vivo and sants, and analgesic mixtures. In a Forecaster (ToxCast) programme in vitro sources, issues that have similar vein, Shin et al. (2015) have (Judson et al., 2014; Knudsen et al., received only limited attention here. recently used bioactivity profiles for 2015), which now includes in vitro, An analysis of the key characteris- 38 agents derived from high-through- in vitro, and in silico data on diverse tics demonstrated by Group 1 agents put in vitro assays to investigate pat- toxicological end-points for more on a site-specific basis is also plan- terns of toxicity associated with dif- than 10 000 chemical substances, ned; if agents that cause tumours ferent scenarios of use. some of which overlap with the set at a specific site (e.g. the lung or Exposure to a single agent may of Group 1 agents considered in the liver) are shown to demonstrate result in the development of more this chapter. The ToxCast database similar characteristics, this could than one type of tumour, perhaps also includes information on several provide new insights into site-speci- through different pathways that in- hundred toxicological assays, which fic carcinogenesis. volve different mechanistic charac- could enrich the database of key Although the present anal- teristics. It would be of interest to characteristics used in the present ysis found that most Group 1 examine the key characteristics for analysis. agents demonstrated multiple key agents associated with specific tu- With the elaboration of the 10 characteristics, with an average of mour types. This would extend the key characteristics articulated by

278 Smith (Chapter 10) and Smith et al. There could be value in revisiting relevant mechanistic information. (2016), mechanistic evaluations of the present retrospective analysis Although this approach could expe- new agents undertaken within the of the 86 Group 1 agents identified dite identification of relevant articles, IARC Monographs Programme in the IARC Monographs up to and expert opinion and application of are beginning to make use of these including Volume 106, with respect weight-of-evidence criteria would still characteristics, including the use of to the conduct of a series of compre- have value in reducing the errors in formal methods of systematic review hensive systematic reviews on the 10 the assignment of key characteristics to identify relevant mechanistic infor- key characteristics of these agents, to specific agents. mation. This has been successfully followed by an in-depth evaluation attempted in recent evaluations of of the findings of the systematic re- Conclusions some organochlorine insecticides views by experts in relevant disci- and some chlorophenoxy herbicides plines. The development of criteria In considering the results present- (Loomis et al., 2015; Volume 113) for evaluating the weight of evidence ed in this chapter, it is important to and of red meat and processed meat for the key characteristics, similar to emphasize that these mechanistic (Bouvard et al., 2015; Volume 114). that included in the Preamble to the analyses are a first step in under- It is expected that the search IARC Monographs for human and standing the biological mechanisms strategies used in future mechanistic animal data (IARC, 2006), might be by which cancer may occur in hu- evaluations will be refined as expe- contemplated at that time. Group 1 mans. Although considerable effort rience with the key characteristics agents identified beyond Volume 106 was expended in developing the accumulates. In an earlier evaluation for which mechanistic information database of key characteristics and of evidence of epigenetic alterations had become available could also be their analyses in this chapter, these for 28 Group 1 agents, Chappell included in such an analysis. results should be viewed as prelim- et al. (2016) searched for evidence of Baker et al. (2016) have recently inary, to be refined through more DNA methylation, histone modifica- applied supervised machine learn- exhaustive systematic reviews of tion, and expression of non-coding ing techniques to classify PubMed the relevant scientific literature and/ miRNAs, as was done in the pres- literature according to the hallmarks or through discussion with a broad ent analysis, but with the addition of of cancer (Hanahan and Weinberg, panel of experts on the mechanisms several more detailed search terms, 2000, 2011). In a case study of basal of carcinogenesis. The 10 key char- specifically long non-coding RNA cell carcinoma and melanoma, only acteristics were endorsed by the (lncRNA), small RNA, chromatin, 46,727 of 121,488 abstracts from participants in the IARC Workshop and promotor methylation. Chappell their original systematic literature on Tumour Site Concordance and et al. (2016) noted that the great ma- search were classified as relevant, Mechanisms of Carcinogenesis, jority (89%) of the studies on lncR- which provided oversight for this NAs included in their review report- reflecting the potential time savings PART 3

project; additional experience with CHAPTER 22 ed alterations in miRNAs, leading that may be achieved through auto- the exploration of these characteris- to results largely consistent with the matic classification. An approach to tics in cancer research will serve to search terms used here: 43% (12 of extracting information on the 10 key 28) of the agents evaluated by these characteristics of human carcinogens define their utility more fully. Equally authors demonstrated evidence of would be to apply these machine important is to consider the nature epigenetic alterations, similar to the learning techniques and biomedical of the evidence needed to establish 38% (33 of 86) of agents included in text mining methods to identify, in an that specific mechanistic charac- the present analysis. Continued ex- automated fashion, articles associ- teristics are associated with human perience with the evaluation of the ating these key characteristics with carcinogens. The current database 10 key characteristics of human car- specific Group 1 agents. Because has relied on the demonstration of cinogens can be expected to further of the sheer size of a full systemat- certain toxicological end-points as refine the criteria used for their iden- ic review of mechanistic information evidence of these mechanistic char- tification, including both the toxico- on all Group 1 agents, the use of acteristics; further consideration of logical events associated with these automated search algorithms of this these and other possible markers key characteristics and the assays type could offer considerable effi- of the key characteristics of human used as evidence of these events. ciency gains in identifying potentially carcinogens is warranted.

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 279 Mechanistic considerations are extracted from the Monographs to Further in-depth analyses of this rich becoming increasingly prominent assemble a database on the basis database of characteristics of human in the IARC Monographs, thereby of 10 key characteristics of human carcinogens are expected to provide enriching the body of evidence on carcinogens. After some grouping of additional insights into the mecha- which future analyses of this type similar agents, the characteristic pro- nisms of human carcinogenesis. may be based. The authors are plan- files were examined for 86 Group 1 ning to update the mechanistic da- agents for which mechanistic infor- Acknowledgements tabase to include Monographs pub- mation was available in the IARC The authors are grateful to Kate Z. lished subsequent to Volume 106, a Monographs up to and including Guyton for many helpful comments, task that will be greatly facilitated by Volume 106, based on information which resulted in both improvements the documentation of key character- derived from human in vivo, human in the presentation of this chapter istics of agents evaluated in recent in vitro, animal in vivo, and animal and additional insights into the find- Monographs. in vitro studies. The most prevalent key characteristic was “is genotoxic”, ings in this chapter, and to the editors Summary followed by “alters cell proliferation, for their thorough review of several cell death, or nutrient supply” and “in- preliminary drafts. Mélissa Billard Since its inception in the early 1970s, duces oxidative stress”. Most agents was a Visiting Scientist at IARC the IARC Monographs Programme exhibited several of the 10 key char- during the summers of 2013 and has evaluated more than 1000 acteristics, with an average of four 2014, working under the direction of agents with respect to their carci- characteristics per agent, a finding Yann Grosse and Robert Baan on nogenic hazard; of these, up to and consistent with the notion that can- assembling the mechanistic data- including Volume 119 of the IARC cer development in humans involves base analysed in this chapter. Julian Monographs, 120 agents met the multiple pathways. Information on Little is the Canada Research Chair criteria for classification as carcino- the key characteristics was often in Human Genome Epidemiology genic to humans (Group 1). Volume available from multiple sources, with at the University of Ottawa. Daniel 100 of the IARC Monographs provid- many agents demonstrating concor- Krewski is the Natural Sciences and ed a review and update of Group 1 dance between human and animal Engineering Research Council of carcinogens. These agents were sources, particularly with respect Canada Chair in Risk Science at the divided into six broad categories: to genotoxicity. Although a detailed University of Ottawa. This chapter is pharmaceuticals; biological agents; comparison of the characteristics of dedicated to the memory of Jan M. arsenic, metals, fibres, and dusts; different types of agent was not at- Zielinski, who succumbed to can- radiation; personal habits and indoor tempted here, the overall characteris- cer in 2016; before becoming ill, Dr combustions; and chemical agents tic profiles for pharmaceutical agents Zielinski led the analytical team that and related occupations. Data on bi- and for chemical agents and relat- conducted the analyses reported ological mechanisms of action were ed occupations appeared similar. here.

280 References

Al-Zoughool M, Bird M, Rice JM, Billard Bouvard V, Loomis D, Guyton KZ, Grosse Guha N, Loomis D, Grosse Y, Lauby- M, Birkett N, Krewski D, et al. (2019). Y, Ghissassi FE, Benbrahim-Tallaa L, et al.; Secretan B, El Ghissassi F, Bouvard V, Development of a database of toxicological International Agency for Research on et al.; International Agency for Research end-points and key characteristics of agents Cancer Monograph Working Group (2015). on Cancer Monograph Working Group known to cause cancer in humans. J Toxicol Carcinogenicity of consumption of red and (2012). Carcinogenicity of trichloroethylene, Environ Health B Crit Rev. (forthcoming) processed meat. Lancet Oncol. 16(16):1599– tetrachloroethylene, some other chlorinated 600. http://dx.doi.org/10.1016/S1470- solvents, and their metabolites. Lancet Oncol. Baker S, Silins I, Guo Y, Ali I, Högberg J, 2045(15)00444-1 PMID:26514947 13(12):1192–3. http://dx.doi.org/10.1016/ Stenius U, et al. (2016). Automatic semantic S1470-2045(12)70485-0 PMID:23323277 classification of scientific literature according Chappell G, Pogribny IP, Guyton KZ, Rusyn to the hallmarks of cancer. Bioinformatics. I (2016). Epigenetic alterations induced by Gurusankar R, Yenugadhati N, Krishnan K, 32(3):432–40. http://dx.doi.org/10.1093/ genotoxic occupational and environmental Hays S, Haines D, Zidek A, et al. (2017). The bioinformatics/btv585 PMID:26454282 human chemical carcinogens: a systematic role of human biological monitoring in health literature review. Mutat Res Rev Mutat risk assessment. Int J Risk Assess Manag. Baker SG (2014). A cancer theory kerfuffle can Res. 768:27–45. http://dx.doi.org/10.1016/j. 20(1–3):136 – 97. http://dx.doi.org/10.1504/ lead to new lines of research. J Natl Cancer mrrev.2016.03.004 PMID:27234561 IJRAM.2017.082561 Inst. 107(2):dju405 http://dx.doi.org/10.1093/ jnci/dju405 PMID:25528755 Cogliano VJ, Baan RA, Straif K, Grosse Y, Hanahan D, Weinberg RA (2000). The Secretan MB, El Ghissassi F, et al. (2004). hallmarks of cancer. Cell. 100(1):57–70. http:// Bartsch H, Malaveille C (1989). Prevalence The science and practice of carcinogen dx.doi.org/10.1016/S0092-8674(00)81683-9 of genotoxic chemicals among animal and identification and evaluation. Environ Health PMID:10647931 human carcinogens evaluated in the IARC Perspect. 112(13):1269–74. http://dx.doi. Monograph Series. Cell Biol Toxicol. 5(2):115– org/10.1289/ehp.6950 PMID:15345338 Hanahan D, Weinberg RA (2011). Hallmarks of 27. http://dx.doi.org/10.1007/BF00122647 cancer: the next generation. Cell. 144(5):646– PMID:2766027 Cote I, Andersen ME, Ankley GT, Barone S, 74. http://dx.doi.org/10.1016/j.cell.2011.02.013 Birnbaum LS, Boekelheide K, et al. (2016). PMID:21376230 Benbrahim-Tallaa L, Baan RA, Grosse Y, The Next Generation of Risk Assessment Lauby-Secretan B, El Ghissassi F, Bouvard multi-year study – highlights of findings, Heller SR, McNaught A, Pletnev I, Stein V, et al.; International Agency for Research applications to risk assessment, and S, Tchekhovskoi D (2015). InChI, the on Cancer Monograph Working Group future directions. Environ Health Perspect. IUPAC International Chemical Identifier. J (2012). Carcinogenicity of diesel-engine 124(11):1671–82. http://dx.doi.org/10.1289/ Cheminform. 7:23. http://dx.doi.org/10.1186/ and gasoline-engine exhausts and some EHP233 PMID:27091369 s13321-015-0068-4 PMID:26136848 nitroarenes. Lancet Oncol. 13(7):663–4. http:// dx.doi.org/10.1016/S1470-2045(12)70280-2 Diaz-Moralli S, Tarrado-Castellarnau M, Herceg Z, Lambert MP, van Veldhoven K, PMID:22946126 Miranda A, Cascante M (2013). Targeting Demetriou C, Vineis P, Smith MT, et al. cell cycle regulation in cancer therapy. (2013). Towards incorporating epigenetic Benbrahim-Tallaa L, Lauby-Secretan B, Pharmacol Ther. 138(2):255–71. http:// mechanisms into carcinogen identification Loomis D, Guyton KZ, Grosse Y, El Ghissassi dx.doi.org/10.1016/j.pharmthera.2013.01.011 and evaluation. Carcinogenesis. 34(9):1955– F, et al.; International Agency for Research on PMID:23356980 67. http://dx.doi.org/10.1093/carcin/bgt212 Cancer Monograph Working Group (2014). PMID:23749751 Carcinogenicity of perfluorooctanoic acid, Ding L, Getz G, Wheeler DA, Mardis ER, tetrafluoroethylene, dichloromethane, 1,2-di- McLellan MD, Cibulskis K, et al. (2008). Hernández LG, van Steeg H, Luijten M, van chloropropane, and 1,3-propane sultone. Somatic mutations affect key pathways in lung Benthem J (2009). Mechanisms of non- PART 3 genotoxic carcinogens and importance of

Lancet Oncol. 15(9):924–5. http://dx.doi. adenocarcinoma. Nature. 455(7216):1069– CHAPTER 22 org/10.1016/S1470-2045(14)70316-X 75. http://dx.doi.org/10.1038/nature07423 a weight of evidence approach. Mutat Res. PMID:25225686 PMID:18948947 682(2–3):94–109. http://dx.doi.org/10.1016/j. mrrev.2009.07.002 PMID:19631282 Birkett N, Al-Zoughool M, Bird M, Zielinski JM, EPA (2005). Guidelines for carcinogen Krewski D (2019). Carcinogenic mechanisms risk assessment (EPA/630/P-03/001F). Hoeijmakers JH (2009). DNA damage, aging, for 109 human carcinogens. J Toxicol Environ Washington (DC), USA: U.S. Environmental and cancer. N Engl J Med. 361(15):1475–85. Health B Crit Rev. (forthcoming) Protection Agency. Available from: http:// http://dx.doi.org/10.1056/NEJMra0804615 www.epa.gov/cancerguidelines/. PMID:19812404 Boss MK, Bristow R, Dewhirst MW (2014). Linking the history of radiation biology to the Floor SL, Dumont JE, Maenhaut C, Raspe IARC (2006). Preamble to the IARC hallmarks of cancer. Radiat Res. 181(6):561– E (2012). Hallmarks of cancer: of all Monographs, amended January 2006. 77. http://dx.doi.org/10.1667/RR13675.1 cancer cells, all the time? Trends Mol Med. Available from: https://monographs.iarc.fr/ PMID:24811865 18(9):509–15. http://dx.doi.org/10.1016/j. previous-preamble/. molmed.2012.06.005 PMID:22795735 Bourdon-Lacombe JA, Moffat ID, Deveau IARC (2012a). Arsenic, metals, fibres, and M, Husain M, Auerbach S, Krewski D, et al. Grosse Y, Loomis D, Guyton KZ, Lauby- dusts. IARC Monogr Eval Carcinog Risks (2015). Technical guide for applications of Secretan B, El Ghissassi F, Bouvard V, Hum. 100C:1–499. Available from: http:// gene expression profiling in human health et al.; International Agency for Research publications.iarc.fr/120 PMID:23189751 risk assessment of environmental chemicals. on Cancer Monograph Working Group IARC (2012b). Biological agents. IARC Regul Toxicol Pharmacol. 72(2):292–309. (2014). Carcinogenicity of fluoro-edenite, Monogr Eval Carcinog Risks Hum. 100B:1– http://dx.doi.org/10.1016/j.yrtph.2015.04.010 silicon carbide fibres and whiskers, 441. Available from: http://publications.iarc. PMID:25944780 and carbon nanotubes. Lancet Oncol. fr/119 PMID:23189750 15(13):1427–8. http://dx.doi.org/10.1016/ S1470-2045(14)71109-X PMID:25499275

Part 3 • Chapter 22. Analysis of key characteristics of human carcinogens 281 IARC (2012c). Chemical agents and related Krewski D, Westphal M, Andersen ME, Phillips TD, Richardson M, Cheng YS, occupations. IARC Monogr Eval Carcinog Paoli GM, Chiu WA, Al-Zoughool M, et al. He L, McDonald TJ, Cizmas LH, et al. Risks Hum. 100F:1–599. Available from: (2014). A framework for the next generation (2015). Mechanistic relationships between http://publications.iarc.fr/123 PMID:23189753 of risk science. Environ Health Perspect. hepatic genotoxicity and carcinogenicity in 122(8):796–805. http://dx.doi.org/10.1289/ male B6C3F1 mice treated with polycyclic IARC (2012d). Personal habits and indoor ehp.1307260 PMID:24727499 aromatic hydrocarbon mixtures. Arch Toxicol. combustions. IARC Monogr Eval Carcinog 89(6):967–77. http://dx.doi.org/10.1007/ Risks Hum. 100E:1–575. Available from: Kriek E, Rojas M, Alexandrov K, Bartsch H s00204-014-1285-8 PMID:24888377 http://publications.iarc.fr/122 PMID:23193840 (1998). Polycyclic aromatic hydrocarbon-DNA adducts in humans: relevance as biomarkers Pickup MW, Mouw JK, Weaver VM (2014). The IARC (2012e). Pharmaceuticals. IARC for exposure and cancer risk. Mutat Res. extracellular matrix modulates the hallmarks Monogr Eval Carcinog Risks Hum. 100A:1– 400(1–2):215–31. http://dx.doi.org/10.1016/ of cancer. EMBO Rep. 15(12):1243–53. 437. Available from: http://publications.iarc. S0027-5107(98)00065-7 PMID:9685648 http://dx.doi.org/10.15252/embr.201439246 fr/118 PMID:23189749 PMID:25381661 Lauby-Secretan B, Loomis D, Grosse Y, El IARC (2012f). Radiation. IARC Monogr Ghissassi F, Bouvard V, Benbrahim-Tallaa L, Reddel RR (2000). The role of senescence Eval Carcinog Risks Hum. 100D:1–437. et al.; International Agency for Research on and immortalization in carcinogenesis. Available from: http://publications.iarc.fr/121 Cancer Monograph Working Group (2013). Carcinogenesis. 21(3):477–84. http://dx.doi. PMID:23189752 Carcinogenicity of polychlorinated biphenyls org/10.1093/carcin/21.3.477 PMID:10688868 and polybrominated biphenyls. Lancet Oncol. IARC (2013). Diesel and gasoline engine Roessler S, Budhu A, Wang XW (2014). 14(4):287–8. http://dx.doi.org/10.1016/S1470- exhausts and some nitroarenes. IARC Deciphering cancer heterogeneity: the 2045(13)70104-9 PMID:23499544 Monogr Eval Carcinog Risks Hum. 105:1–704. biological space. Front Cell Dev Biol. 2:12. Available from: http://publications.iarc.fr/129. Loomis D, Grosse Y, Lauby-Secretan B, El http://dx.doi.org/10.3389/fcell.2014.00012 PMID:26442290 Ghissassi F, Bouvard V, Benbrahim-Tallaa PMID:25364720 L, et al.; International Agency for Research IARC (2014). Trichloroethylene, tetrachlo- Saracci R, Wild CP (2015). International on Cancer Monograph Working Group roethylene, and some other chlorinated Agency for Research on Cancer: the first 50 (2013). The carcinogenicity of outdoor air agents. IARC Monogr Eval Carcinog Risks years, 1965–2015. Lyon, France: International pollution. Lancet Oncol. 14(13):1262–3. http:// Hum. 106:1–514. Available from: http:// Agency for Research on Cancer. Available dx.doi.org/10.1016/S1470-2045(13)70487-X publications.iarc.fr/130. PMID:26214861 from: http://publications.iarc.fr/449. PMID:25035875 IARC (2016a). Outdoor air pollution. IARC Shin H-M, Ernstoff A, Arnot JA, Wetmore BA, Loomis D, Guyton K, Grosse Y, El Ghissasi Monogr Eval Carcinog Risks Hum. 109:1– Csiszar SA, Fantke P, et al. (2015). Risk-based F, Bouvard V, Benbrahim-Tallaa L, et al.; 448. Available from: http://publications.iarc. high-throughput chemical screening and International Agency for Research on fr/538. prioritization using exposure models and in Cancer Monograph Working Group (2015). vitro bioactivity assays. Environ Sci Technol. IARC (2016b). Polychlorinated biphenyls and Carcinogenicity of lindane, DDT, and 49(11):6760 –71. http://dx.doi.org/10.1021/acs. polybrominated biphenyls. IARC Monogr Eval 2,4-dichlorophenoxyacetic acid. Lancet est.5b00498 PMID:25932772 Carcinog Risks Hum. 107:1–502. Available Oncol. 16(8):891–2. http://dx.doi.org/10.1016/ from: http://publications.iarc.fr/131. S1470-2045(15)00081-9 PMID:26111929 Smith MT, Guyton KZ, Gibbons CF, Fritz JM, Portier CJ, Rusyn I, et al. (2016). Key IARC (2017a). Some chemicals used as Meek ME, Boobis A, Cote I, Dellarco V, characteristics of carcinogens as a basis solvents and in polymer manufacture. IARC Fotakis G, Munn S, et al. (2014). New for organizing data on mechanisms of Monogr Eval Carcinog Risks Hum. 110:1–276. developments in the evolution and application carcinogenesis. Environ Health Perspect. Available from: http://publications.iarc.fr/547. of the WHO/IPCS framework on mode of 124(6):713–21. http://dx.doi.org/10.1289/ action/species concordance analysis. J Appl ehp.1509912 PMID:26600562 IARC (2017b). Some nanomaterials and Toxicol. 34(1):1–18. http://dx.doi.org/10.1002/ some fibres. IARC Monogr Eval Carcinog jat.2949 PMID:24166207 Risks Hum. 111:1–316. Available from: http:// Snyder RD (2010). Possible structural and functional determinants contributing to the publications.iarc.fr/552. Mesri EA, Feitelson MA, Munger K (2014). clastogenicity of pharmaceuticals. Environ Human viral oncogenesis: a cancer hallmarks Mol Mutagen. 51(8–9):800–14. http://dx.doi. ICF (2017). DRAGON: an online tool for analysis. Cell Host Microbe. 15(3):266–82. org/10.1002/em.20626 PMID:20872827 systematic review. Fairfax (VA), USA: http://dx.doi.org/10.1016/j.chom.2014.02.011 ICF. Available from: https://www.icf.com/ PMID:24629334 resources/solutions-and-apps/dragon-online- Stein SE, Heller SR, Tchekhovskoi DV (2003). An open standard for chemical tool-systematic-review. Miller JA (1970). Carcinogenesis by structure representation: the IUPAC Chemical chemicals: an overview – G. H. A. Clowes Identifier. In: Proceedings of the International Judson R, Houck K, Martin M, Knudsen T, Memorial Lecture. Cancer Res. 30(3):559–76. Chemical Information Conference, Nimes, Thomas RS, Sipes N, et al. (2014). In vitro PMID:4915745 and modelling approaches to risk assessment France, 19–22 October 2003. Tetbury, United from the U.S. Environmental Protection Multhoff G, Radons J (2012). Radiation, Kingdom: Infonortics; pp. 131–43. Agency ToxCast programme. Basic Clin inflammation, and immune responses Vainio H, Magee PN, McGregor DB, Pharmacol Toxicol. 115(1):69–76. http://dx.doi. in cancer. Front Oncol. 2:58. http:// McMichael AJ, editors (1992). Mechanisms org/10.1111/bcpt.12239 PMID:24684691 dx.doi.org/10.3389/fonc.2012.00058 of carcinogenesis in risk identification. Lyon, PMID:22675673 Klaunig JE, Wang Z, Pu X, Zhou S (2011). France: International Agency for Research on Oxidative stress and oxidative damage National Research Council (2014). Review Cancer (IARC Scientific Publication No. 116). in chemical carcinogenesis. Toxicol. of EPA’s Integrated Risk Information System Willeit P, Willeit J, Mayr A, Weger S, Appl Pharmacol. 254(2):86–99. http:// (IRIS) Process. Washington (DC), USA: Oberhollenzer F, Brandstätter A, et al. (2010). dx.doi.org/10.1016/j.taap.2009.11.028 National Academies Press. http://dx.doi. Telomere length and risk of incident cancer PMID:21296097 org/10.17226/18764 and cancer mortality. JAMA. 304(1):69–75. Knudsen TB, Keller DA, Sander M, Carney National Toxicology Program (2014). Report http://dx.doi.org/10.1001/jama.2010.897 EW, Doerrer NG, Eaton DL, et al. (2015). on carcinogens. 13th ed. Research Triangle PMID:20606151 FutureTox II: in vitro data and in silico Park (NC), USA: Department of Health and models for predictive toxicology. Toxicol Sci. Human Services, Public Health Service. 143(2):256 – 67. http://dx.doi.org/10.1093/ toxsci/kfu234 PMID:25628403

282