Three sources of variability in tumor incidence were investigated. To assess the impact of random variability, we evaluated the operating characteristics of several statistical procedures for comparing tumor incidences. We found that those techniques that use pre-determined levels of significance (e.g., p<0.01) for all comparisons may produce unacceptably high false positive rates. We concluded that a procedure derived by Westfall and Soper, if used properly, seems to provide a reasonable balance between controlling the overall false positive rate and maintaining power for detecting carcinogenic effects. To investigate the impact of body weight differences on tumor incidence, we evaluated data from the NTP Dietary Restriction Study. Our investigation suggests that a dietary restriction strategy that focuses on achieving similar body weights in dosed and control groups may produce false positive outcomes if substantially more food restriction is required for control groups than for dosed animals. We also found that equal food restriction to dosed and control animals may produce false negative outcomes if the resulting body weights are substantially different in dosed and control groups. Alternative methods of reducing body weight are being studied, including dietary modifiation and requiring lighter animals from the breeder. Finally, we evaluated the impact Helicobacter hepaticus infection on the incidence of liver neoplasms. Control B6C3F1 male mice from nine NTP studies infected with Helicobacter hepaticus showed increased (p<0.05) liver tumor incidences compared with controls from 26 uninfected bioassays. The increased incidences of hepatocellular neoplasms were seen primarily in those males exhibiting H. hepaticus associated hepatitis. We concluded that interpretation of carcinogenic effects in the liver of B6C3F1 mice may be confounded if there is H. hepaticus associated hepatitis.
Showing the most recent 10 out of 13 publications
