Development of a database on tumors and tumor sites in humans and in experimental animals for 'Group 1 agents identified through volume 109 of the IARC Monographs

ABSTRACT Volume 100 in the series of IARC Monographs on the Evaluation of Carcinogenic Risks to Humans comprises an update and review of relevant information on all agents determined to induce cancer in humans. These Group 1 agents are categorized in 6 Monographs (Volumes 100A-F) published in 2012. This paper describes the methodology and stringent criteria used in the creation of a comprehensive database on tumors noted in animals and humans for the carcinogens reviewed in Volume 100, and for additional Group 1 agents that were identified in subsequent Monographs through Volume 109. The development of this database involved the systematic collection of relevant data on tumors detected in humans and experimental animals identified by the Working Groups that conducted evaluations reported in the IARC Monographs. The database includes all human tumor sites identified by the Working Groups, along with all tumor sites for which there was sufficient evidence in experimental animals. This database provides a basis for assessing the degree of concordance between tumor sites observed in humans and experimental animals for Group 1 agents identified through Volume 109.


Introduction
Since the early 1970s, the Monographs Programme of the International Agency for Research on Cancer (IARC) has conducted hazard evaluations of agents that may increase the risk for cancer in humans. The reviews of the relevant literature and the evaluations were conducted by international Working Groups of expert scientists according to a well-established and rigorous protocol that is described in the Preamble to the IARC Monographs (IARC 2006). The reviews and evaluations are published as IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (IARC 2015a).
For ease of reference, these five agents are included in an expanded group of 'chemical agents and related occupations' denoted by Vol 100F*. Although additional Group 1 agents were identified since, the current dataset extends only through Vol 109, the last Monograph for which final results were available at the time the present paper was completed.
The reviews and updates in Monograph Volumes 100A-F specifically focused on the identification of tumors and tumor types, both in humans and in experimental animals, resulting from exposure to each of the IARC Group 1 agents. In addition, the organ site(s) where tumors were reported to originate were documented where possible. The availability of this information on the more than 100 human carcinogens in IARC's Group 1 prompted us to investigate the level of concordance that may exist between humans and experimental animals with respect to tumors and tumor sites. To this end, pertinent information in Volume 100 is captured in Supplemental Table 1 that might then serve as a basis to develop a database on tumors and tumor sites in animals and humans. The creation of such a databasedesigned to be amenable to bio-statistical analysis (Krewski et al. 2019, this Volume)was motivated by the interest in a statistical assessment of the degree of concordance between animal and human tumors and tumor sites. This important scientific question is dependent upon the extent to which animal cancer data collected here may be extrapolated to humans. It is anticipated that the database may also find other applications including the development of human tumor profiles to assist in the identification of additional Group 1 agents.
It should be noted that for agents in IARC's Groups 2A/2B (probably/possibly carcinogenic to humans), the information on cancer in humans may often be lacking or not be sufficiently strong for reliable, adequate interspecies comparisons to be conducted. For this reason, the concordance analysis (Krewski et al. 2019) was focused on agents in Group 1. In addition, it was decided that sufficient evidence of carcinogenicity in humans and sufficient evidence of carcinogenicity in experimental animals were required for an agent to be included in the statistical concordance analysis. With less than sufficient evidence of carcinogenicity, in humans or animals, the definition of a target site in either species might become less reliable or not be possible. Therefore, although the dataset given in Supplemental Table 1 provides information on all Group 1 agents, the actual database of human carcinogens eligible for concordance analysis is appreciably smaller (Krewski et al. 2019).

Methods
In making an evaluation of the evidence of carcinogenicity to humans, an IARC Monographs Working Group is generally asked to identify target sites in humans for which there is sufficient evidence of carcinogenicity of the agent under study. However, the Working Group is not required to identify target sites for carcinogenicity in experimental animals at the time of the evaluation, but simply to assess the overall weight of the evidence in experimental animals. Consequently, for the purpose of this IARC Scientific Publication, species-specific tumor target sites in experimental animals needed to be identified for each Group 1 agent prior to proceeding to explore concordance between animal and human cancers.
During the six meetings for Volumes 100 A-F, the respective Working Groups identified studies in experimental animals that provided results on species-specific tumor sites. This was based upon criteria adapted from the Preamble to the IARC Monographs. It was considered that there was sufficient evidence for identifying a species-specific tumor site in experimental animals under any one of the three following conditions: • An increased incidence of malignant neoplasms or an appropriate combination of benign and malignant neoplasms originating from the same organ (or tissue) is identified in two or more independent studies in one species carried out at different times or in different labs or under different protocols. • An increased incidence of malignant neoplasms or an appropriate combination of benign and malignant neoplasms originating from the same organ (or tissue) is identified in both genders of one species in one wellconducted study, ideally performed under Good Laboratory Practice (GLP).  Rat Urinary bladder Transitional cell carcinoma Wei et al. (1999Wei et al. ( , 2002 Gillett et al. (1992), MF, beagle, i.v.; White et al. (1993), MF, beagle, p.o.; Gillett et al. (1987), MF, beagle, inh.
(Continued ) • A single study in one species and gender might be considered to provide sufficient evidence for a specific target site when malignant neoplasms occur to an unusual degree with regard to incidence, type of tumor, or age at onset.
Confirmation of the tumors and tumor sites identified in Volume 100 was performed by one member each from the IARC secretariat and from the University of Ottawa (UO) project team, who systematically consulted the original publications describing the studies cited in the Volume-100 reviews. It was decided by the Workshop Participants that extraction of the following information was required for each study: species, strain, gender, route of exposure; targeted tumor site including histology. Further information would be recorded as 'study details', e.g., dose, number of test animals, number of control animals, age at the start of exposure, duration of exposure, duration of follow-up, and statistical analyses. The two team members independently captured the information and any disagreements were resolved following a group discussion. Ultimately, more than 2,000 investigations were reviewed, and more than 1,000 of these contributed to the identification of species-specific tumor sites in experimental animals. Studies were not considered if any one of the following exclusion descriptors was applicable: • Initiation-promotion studies • Co-carcinogenicity investigations • Studies in genetically modified animals • Investigations with precancerous lesions as the outcome • Carcinogenicity of metabolites and derivatives • Non-laboratory animals (livestock; companion animals) • Analogous agents (similar structure or similar virus type) demonstrate sufficient evidence of carcinogenicity in experimental animals, as indicated above. Strain, gender and route of exposure reported for each animal study are also presented. Comments are provided as appropriate. For example, no human tumor site is specified for 'aristolochic acid', as this agent was placed in Group 1 on the basis of classification of "plants containing aristolochic acid" as a Group 1 agent, supported by mechanistic data on genotoxicity (IARC 2012a). As with other 'mechanistic upgrades', this agent is listed in the complete dataset, but not included in the statistical analysis of concordance (Krewski et al. 2019). All the information on tumors and tumor sites in humans and experimental animals from IARC Monograph Volumes 100-109 is provided in Supplemental Table 1.

Observations
For some Group-1 agents, there were only a few studies that contributed to the identification of a target site in experimental animals, and frequently the investigations did not enable defining a target site as a result of inadequate reporting. There were many instances where the reported tumor incidences were uninformative, possibly as a result of the small number of animals tested. In other cases, studies noted an increased incidence of tumors but without indication of malignancy or proper description of histopathological details. Further, some reports did not specify the purity of the administered agent. In these cases, it was necessary to consider the possibility of confounding, as the existence of other agents in the administered sample may have contributed to the outcome. In some experiments, animals were observed only for short periods of time, especially the studies investigating acute adverse effects, which precluded observation of carcinogenic outcomes that may take longer to develop.

Conclusions
The dataset developed here to define target sites for carcinogenicity in humans and experimental animals summarizes all available data on Group 1 agents identified in Volumes 100-109 of the IARC Monographs. At the time of completion of Monograph Volume 109, a total of 111 Group 1 agents were identified, and these are included in the list presented in the Supplemental Material. This comprehensive set of data constitutes a unique basis for addressing the important scientific question such as to which extent these animal cancer data are comparable with human cancer results. The value of this dataset is demonstrated by the initial concordance analyses that were conducted with the database derived from it (Krewski et al. 1983 this Volume).