Document 1QR7Yb4pEabq9Nz8y8wGOxDNm

Benchm ark Doses for Liver Tumors in Sprague Dawley Rats fed Perfluorooctane Sulfonic Acid Potassium Salt (PFOS) David W. Gaylor, Ph.D. Sciences International, Inc. January 24,2002 Introduction The carcinogen risk assessment guidelines proposed by the U.S. Environmental Protection Agency (1999) recommend the use of a benchmark dose (BMD) approach for low dose cancer risk assessment. Unless stipulated otherwise, the BMD is the dose at which the excess lifetime tumor incidence is 10%, denoted by BMDio. A value o f 10% was selected as this is about the lowest incidence that can be estimated with adequate precision from typical chronic bioassays in rodents. Further, a lower 95% confidence limit is calculated for the benchmark dose (BMDLio) to account for the experimental variation o f the bioassay. The BMDLio is then used as a point-of-departure for low dose cancer risk assessment. When a nonlinear dose response curve is expected in the low dose range, a margin of exposure between the BMDLio and anticipated human exposure levels is considered. Otherwise, linear extrapolation from the BMDLio to zero is used for low dbse cancer risk estimation. In either case, the BMDLio serves as the point-ofdeparture. Bioassay Data The data used for calculation of the BMDLio were collected in the 104-Week Dietary Chronic Toxicity and Carcinogenicity Study with Perfluorooctane Sulfonic Acid Potassium Salt (PFOS; T-6295) in Rats. The BMDL is calculated for hepatocellular adenomas and carcinomas combined for males and females. All tumors were adenomas except for one carcinoma in the high dose females. In order to calculate lifetime incidence rates for each dose group, it is necessary to calculate the number o f animals at risk. Clearly, animals that were removed from the study for interim sacrifices or that died before the terminal sacrifice were not at risk for a lifetime. The Pdy-3 approach developed by the National Toxicology Program (Bailer and Portier, 1988) is used here to calculate the effective number o f animals at risk. Obviously, an animal that survives for the lifetime o f the study until the terminal sacrifice counts as a whole lifetime exposure. Also, any animal that is removed from the study with a hepatocellular adenoma/carcinoma prior to the terminal sacrifice lived long enough to develop the tumor is counted as a lifetime exposure. All other animals are given a weight o f (t/T)3, where t is the week that an animal was removed from the study without a hepatocellular adenoma/carcinoma and terminal sacrifices began at week T=105.. Relatively little weight is given to an animal removed early in a study. For example, the animals removed at an interim sacrifice halfivay through the study 1 at 53 weeks receive a weight o f (53/105)3= 0.13 o f a lifetime, whereas an animal that died on week 96 receives a weight o f (963/105) = 0.76 o f a lifetime. The weights are summed for each dose group to obtain the effective number o f animals at risk for each group. The number o f animals with hepatocellular adenoma/carcinoma, effective number of animals at risk, and average serum levels o f PFOS at 14 weeks for each dose group are displayed in Table 1. Table 1. Results from the 104-week carcinogenicity study in SD rats fed PFOS. Dose (ppm) 14-wk Serum (ug/ml) Number o f animals with liver tumors* Effective number o f animals 0 .0.05 0.5 4.04 2.0 17.1 5.0 43.9 20.0 148 0 2.67 0.5 6.96 2.0 27.3 5.0 64.4 20.0 223 Males 0 3 3 1 7 Females 0 1 1 1 6 33 32 37 38 38 39 35 29 37 41 *Hepatocellular adenomas except one hepatocellular carcinoma in the high dose females. 2 As noted before, the effective numbers of animals at risk reflect the lower survival in the controls and low dose males and the females fed 2 ppm and the higher survival in the high dose females. Benchmark Dose Calculations The numbers o f animals with hepatocellular adenoma/carcinoma and the effective number o f animals at risk were entered into the U.S. Environmental Protection Agency benchmark dose software program (BMDS). Estimates o f the benchmark dose were obtained using the multistage model P = 1 - exp[-( q0+qid + q2d2+ q3 d3 +q4 d4)] where P represents the proportion o f animals with tumors, d is the dietary dose or serum level, and the q 's are estimated from the experimental dose response data. Goodness-of-fit p-values for the multistage model are above 0.1 indicating an adequate fit o f the model. Small p-values would indicate a statistically significant deviance from the multistage model. The goodness-of-fit p-values, BMDio, and BMDLio in terms o f dietary concentration and 14-week serum levels o f PFOS for males and females are displayed in Table 2. Table 2. Goodness-of-fit p-values for the multistage model, BMDio, and BMDLio values obtained using the US EPA benchmark dose software program (BMDS). Sex Male Female Male Female p-value BMDio BMDLio Dietary Concentration 0.24 18.2 ppm 0.S4 16.7 ppm 14-Week Serum Level 0.23 135 ug/ml 0.54 193 ug/ml 7.9 ppm 8.0 ppm 62 ug/ml 92 ug/ml 3 References Bailer, A.J. and Portier, C.J. Effects of treatment-induced mortality and tumor-induced mortality on tests for carcinogenicity in small samples. Biometrics 44:417-431 (1988). U.S. Environmental Protection Agency. Guidelinesfo r Carcinogen R isk Assessment. NCEA-F-0644, Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, DC. July, 1999. 4