1. Age-Related Differences in Susceptibility to Carcinogenesis—Toward an Improved Analysis of Data on Age-Related Differences in Cancer Sensitivity in the EPA Children’s Cancer Risk Guidance Document Researchers: Dale Hattis, Principal Investigator Rob Goble, Research Professor Abel Russ, Research Associate Jen Ericson and Jill Mailloux, Student Research Assistants Margaret Chu, EPA Project Monitor Opinions are mine and do not necessarily reflect EPA policy

2. INNOVATIVE ASPECTS OF THE ANALYSIS --Paper #1 Compare measures of potency, rather than uncorrected cancer incidence, among groups. Where dosage spans multiple age groups, use dummy variables to represent the observed tumor risk as the sum of cancer contributions from dosing in different periods: –The periods are: fetal (gd 12-19), pre-weaning (1-21 d); weaning - 2 mo; adult (2 mo - 2 yr). –Where continuous dosing occurs in only a fraction of a period that fraction is used as the corresponding “dummy” rather than 1. Use likelihood methods to first derive appropriate statistical weighting of the different observations, and to avoid bias from excluding “0” points. Express dosage for animals of different weights on a metabolically consistent basis (either concentration in air or food, or per unit body weight to the three quarters power).

3. Paper #2--Monte Carlo Analysis of Uncertainties for Application to Human Risk Assessments Uncertainties in the central estimates of the sensitivities of each life stage per dose in mg/kg^3/4, relative to adults Uncertainties from chemical-to-chemical differences in life-stage related sensitivities Uncertainties in the mapping of comparative ages/times between rodents and humans Bottom line:--Overall expected increment to lifetime tumor risks from full lifetime constant exposure per mg/kg^3/4

4. The Poisson One-hit Transformation--From the Fraction of Animals with at Least One Tumor to The Number of Tumors Per Animal

5. Effect of the One-Hit Transformation for Various Observations of % Tumors in Animal Groups

6. Use of Part-Period Dummy Variables in Combination To Represent Different Exposure Patterns--Maltoni Vinyl Chloride Experiments

7. Detailed Model for Statistical Fitting

8. Summary Results of Analyses for Paper #1 Central estimate results: 5-60 fold increased carcinogenic sensitivity in the birth-weaning period per dose/(body weight 3/4 -day) for mutagenic carcinogens--no detectable increase for nonmutagens Somewhat smaller increase—about 5 fold—for radiation carcinogenesis per Gray Greater increase for mutagens for continuous, rather than discrete dosing protocols Greater increase in males than females Similar increased sensitivity in the fetal period for direct-acting nitrosoureas, but no such increased fetal sensitivity for carcinogens requiring metabolic activation Greater increase in early life sensitivity in liver, and less in lung, than for other tumor sites.

9. Overview of the Data Base

10. Media Concentration or Dose/BW^3/4 Dosimetry

11. Overall Results--Continuous vs Discrete Dosing Protocols (Caveat: Continuous dosing results include 4/9 nonmutagens)

12. Overall Results--Radiation Exposures

13. Age-Related Pharmacodynamic Sensitivity for Carcinogenesis-- Mutagens vs Non-Mutagens--Continuous Dosing Protocols

14. Different Results for Mutagens by Sex-- Continuous + Discrete Dosing Data Combined

15. Females--Lognormal Plots of Likelihood-Based Uncertainty Distributions for Cancer Transformations Per Daily Dose for Various Life Stages for Mutagenic Chemicals (Relative to Comparable Exposures of Adults)--Discrete + Continuous Dosing Experiments

16. Males--Lognormal Plots of Likelihood-Based Uncertainty Distributions for Cancer Transformations Per Daily Dose for Various Life Stages for Mutagenic Chemicals (Relative to Comparable Exposures of Adults)--Discrete + Continuous Dosing Experiments

17. Direct-Acting vs Metabolically-Activated Mutagens-- Standard Age Periods, Discrete Dosing Experiments

18. Direct-Acting vs Metabolically-Activated Mutagens-- Narrower Age Periods, Discrete Dosing Experiments

19. Lactational vs Direct Birth-Weaning Exposures (Continuous + Discrete Dosing)

20. Uncertainty From Chemical-to-Chemical Differences in Life-Stage-Specific Sensitivities for Carcinogenesis Data are inadequate for separate estimation of 5 model parameters for individual chemicals and sites. Approach: –Calculate log differences between observed and model- predicted cancer transformations per mg/kg^3/4 -day for cases where exposure was confined to a single life stage. –Average these life-stage-specific log differences within chemicals for each sex. –Analyze the distribution of average differences among chemicals.

21. Chemical-to-Chemical Mean Log Predicted vs Observed Differences

22. Uncertainties in the mapping of comparative ages/times between rodents and humans Find comparable developmental benchmark for calibrating similar ages across species--current approach based on times of sexual maturity Find interspecies comparisons of other ages based on the fraction of body weights attained relative to the weight at the time of sexual maturity. Derive distributional treatment of uncertainty in interspecies time mapping from separate results for rat/human and mouse/human projections.

23. Population-weighted differences in mean height for NHANES III subjects of different ages (2-90 years)

24. Population- weighted differences in Log(Mean Weight in kg) for NHANES III subjects of different ages (2-90 years)

25. Post-natal growth of Sprague-Dawley Rats, based on data compiled for EPA

26. Post-natal growth of ICR/Jcl mice, based on data of Nomura (1976)

27. Species Differences in Times of Sexual Maturity

28. Inferences of Corresponding Human Ages from Weight-Based Comparisons Relative to the Times of Sexual Maturity--Female Mice

29. Inferences of Corresponding Human Ages from Weight-Based Comparisons Relative to the Times of Sexual Maturity--Female Rats

30. Estimated Lengths of Various Life Stages in Humans Inferred from the Ages of Sexual Maturity in Mice, Rats, and Humans, and Patterns of Growth of Body Weight for Rodents Through 60 Days of Age, and for Humans Through Age 16

31. Monte Carlo Simulations All done in Microsoft Excel 3 Uncertainty distributions are lognormal, except that the length of the human equivalents to the birth- weaning and weaning-60 day periods are limited to 15 years in females and 16 years in males Distributional results are the means at each percentile for three simulations of 5000 trials each

32. Detailed Results by Life Stage For Males--Uncertainty Distributions of Risks for Full Lifetime Exposures to a Generic Mutagenic Carcinogen at a Constant Dose Rate Per Kg of Body Weight 3/4 (The numbers represent the increment to lifetime relative risk/dose where the risk from treatment for the adult period is 1)

33. Detailed Results by Life Stage For Females--Uncertainty Distributions of Risks for Full Lifetime Exposures to a Generic Mutagenic Carcinogen at a Constant Dose Rate Per Kg of Body Weight 3/4 (The numbers represent the increment to lifetime relative risk/dose where the risk for the full adult period is 1)

34. Overall Bottom Line -- Population Expected Risks from Lifetime Constant Exposure to a Mutagenic Carcinogen per Body Weight^3/4 Relative to Adult-Only Exposure

35. Take-Home Conclusions From the Analysis of the Current Data Base Improved life-stage specific analyses of risks from mutagenic carcinogens are possible using current information. These involve appreciable uncertainties, particularly in the mapping of rodent exposure periods to human equivalents. The current analysis suggests that early-life exposure could make important contributions to full-life cancer risks. The mean estimate is a 3.5 fold increment to the risks for full life exposure per body weight^3/4 relative to adult-only exposure, with 5%-95% confidence limits of 1.7 - 7.4 fold. The increments will be somewhat less for constant daily dosage expressed on a mg/kg body weight basis.