Linear versus nonlinear models for hemoglobin adduct formation by acrylamide and its metabolite glycidamide: implications for risk estimation.

Hemoglobin cysteine adduct levels formed by acrylamide (AA) and its epoxide metabolite glycidamide (GA) as previously determined (Bergmark et al., Toxicol. Appl. Pharmacol., 111: 352-363, 1991) in rats given single injections of AA were used to estimate tissue doses, D = integral of Cdt (area under the concentration curve in the blood compartment), of the two compounds. The data were adapted to linear or nonlinear kinetic models, where the latter model accounted for the Michaelis-Menten kinetics of the metabolic conversion of AA to GA. In the linear model, the first-order rates, k*, of elimination from all processes were estimated to be 0.50 and 0.48 h-1 for AA and GA, respectively. In the nonlinear model, the parametrical values Vmax = 19.1 h-1 and Km = 66 microM for the in vivo metabolic conversion of AA to GA, and k1 = 0.21 h-1 and k2 = 0.48 h-1 for the first-order rates of elimination from all other processes of AA and GA, respectively, were found to give the best fit to the exact dosimetric expressions [formula: see text] Using Equation B, it was estimated that the percentage of AA converted to GA approaches 58% when [AA]o, the initial concentration of AA, approaches zero. The implications for high-to-low-dose extrapolation of toxic effects of the derived mathematical relationships between administered dose and tissue dose are discussed.