Workgroup Report: National Toxicology Program Workshop on Hormonally Induced Reproductive Tumors - Relevance of Rodent Bioassays
Thayer, Kristina A., Foster, Paul M., Environmental Health Perspectives
The National Toxicology Program (NTP) hosted the workshop "Hormonally Induced Reproductive Tumors--Relevance of Rodent Bioassays" on 22-24 May 2006 in Raleigh, North Carolina, to discuss the adequacy of rodent models used in the 2-year bioassay by the NTP for detecting certain reproductive tumors. This workshop is the third in a series of activities associated with the NTP Roadmap to critically evaluate the NTP testing program and determine whether any refinements or new strategies are needed to maximize its impact on public health (NTP 2005c). More than 100 people from academia, industry, government, and nonprofit organizations attended the workshop, including an invited panel of 55 scientists with expertise in endocrinology, cancer biology, reproductive toxicology, statistics, and other related fields. The workshop opened with a series of presentations on each of the target tumors from the perspective of clinical and epidemiologic studies, mode(s) of action, and rodent models. To address the workshop's objectives, the invited panels also met in tumor site-specific breakout groups and summarized their discussions in a plenary session. The workshop agenda, presentations, background materials, roster of the invited panel and other attendees, and public comments can be found on the NTP website (http://ntp.niehs.nih.gov; see "Meetings & Workshops" or directly at http://ntp.niehs.nih.gov/go/18592).
The workshop's overall objective was to determine the utility and relevance to human disease outcome of experimental rodent models for evaluating four types of reproductive tumors (ovary, mammary gland, prostate, and testis) with known or presumed hormonal etiologies. The NTP is interested in these issues because of concern that current rodent models, including those used by the NTP, may not adequately detect carcinogens that act via the endocrine system or other nongenotoxic modes of action. Tumors of the mammary gland, ovary, prostate, and testis were selected for evaluation because of their significant human morbidity and mortality and concern that currently used models are not optimal for addressing these tumor types. In some cases, the modes of action thought to cause these types of tumors in rodents are not believed relevant to humans. For example, atrazineinduced mammary gland tumors in the Sprague-Dawley rat are attributed to precocious reproductive senescence that results from atrazine's effects on neuroendocrine function [U.S. Environmental Protection Agency (EPA) 2000]. In certain strains of rats, reproductive senescence in females is characterized by a state of persistent estrus and high levels of estrogen, which may increase the risk of tumor formation in estrogenresponsive tissues. This mode of action is generally not considered relevant to humans because circulating levels of endogenous estrogen decrease during menopause.
In other cases, the rodent species and strains currently used in NTP chronic bioassays (F344/N rat and B6C3F1/N mouse) are suspected to be poor models because they either do not develop a specific type of tumor or have a high spontaneous tumor incidence. Either situation can make detecting a chemically induced effect difficult. For example, in NTP chronic bioassays, prostate tumors are rarely observed in control animals and have never been clearly associated with a chemical exposure, suggesting that the current models are not sensitive for detecting potential human prostate carcinogens. However, this problem is not specific to the F344/N rat and B6C3F1/N mouse, as no conventional rodent models are considered ideal to assess chemically induced prostate cancer (NTP 2006e). At the other extreme are testicular Leydig cell tumors in the F344/N rat. The spontaneous incidence in control animals is so high (??70-90%) that it can be difficult to detect a statistically significant increase above background (NTP 2005a). The conventional rodent models may also be poor predictors of human response because of significant differences between rodents and humans in tumor prevalence, anatomy, or tumor histology. …