Draft Recommendations on Classification of Rodent Neoplasms for Peto Analysis

The Executive Committee of the Society of Toxicologic Pathology appointed an ad hoc working group to review the current use of the Peto model for statistical analysis of rodent carcinogenicity study data and to provide recommendations for pathologists regarding appropriate and consistent classifi cation of neoplasms for analysis by the Peto model. In the Peto model, neoplasms that are detected as in-life observations are classifi ed as Mortality Independent and the incidences of these neoplasms are analyzed by the onset-rate method. For neoplasms that are not detected in-life in animals that come to necropsy before the end of the treatment period, the original Peto model requires pathologists to classify neoplasms as Fatal or Incidental ( Non-fatal ). Each Fatal neoplasm is modeled statistically as rapidly fatal regardless of how long the neoplasm has been present in the animal, and the date of death is used as a surrogate for date of onset of Fatal neoplasms. Fatal neoplasms are analyzed by the death-rate (life-table analysis) method, while Incidental neoplasms are analyzed by the prevalence method. In practice, most neoplasms that cause death are not rapidly fatal, e.g., a slowly progressive pituitary adenoma in a rat. The Peto model does not appropriately model slowly growing or slowly progressive neoplasms that cause death. Recommendations: 1. For animals that die or are killed prior to scheduled sacrifi ce, all neoplasms should be classifi ed into one of four categories on an animal-by-animal basis: Observed In Life (Mortality-Independent)—Neoplasms that were observed during in-life observations and for which the day of onset (fi rst observation) was recorded; Incidental—Neoplasms that were not observed prior to necropsy and likely did not contribute to death of the animal or bringing the animal to necropsy; Rapidly Fatal —Neoplasms that were not observed prior to necropsy, likely contributed to death or bringing an animal to necropsy, and appeared to grow and develop rapidly. For these neoplasms, the day of death would serve as a surrogate for the day of onset of the neoplasm; Not Rapidly Fatal—Neoplasms that were not observed prior to necropsy, likely contributed to death or bringing an animal to necropsy, and did not appear to grow or develop rapidly. 2. Other guidelines for pathologists interpreting carcinogenicity studies include: More than one neoplasm may be Rapidly Fatal, Not Rapidly Fatal , and/or Incidental in a single animal. An individual animal may have 2 or more Rapidly Fatal neoplasms if the pathologist believes that each neoplasm had a rapid onset and likely contributed to bringing the animal to necropsy; Each pathologist should use scientifi c judgment when differentiating Rapidly Fatal and Not Rapidly Fatal neoplasms. This scientifi c judgment is similar to that used in determining whether neoplasms contribute to the cause of death. Criteria for malignancy (tissue invasion, metastasis, necrosis, high mitotic index, anaplasia, etc.) can be helpful as guides for classifying a neoplasm as Rapidly Fatal. 3. Industry and regulatory statisticians should continue to explore alternatives to the Peto method that do not require classifi cation of neoplasms according to their impact on the death of the animal, and should use these methods if they offer equivalent or superior analytical capabilities.