Probabilistic estimates of heterocyclic amines and polycyclic aromatic hydrocarbons concentrations in meats and breads applicable to exposure assessments

Random effect meta-regressions were constructed to estimate concentrations of two heterocyclic amines (HCA) and eight polycyclic aromatic hydrocarbons (PAH) in meat and breads. Eighteen HCA studies and nine PAH studies of food concentration were assembled. Concentration was computed for beef, poultry, pork, and seafood, and bread. Fixed effect predictors included cooking time, form of the food, cooking method, interaction between form and cooking method, temperature at which the food was cooked, fuel of the flame source, percentage of fat, and other elements. Meat type was significant to all HCAs but only three of the PAHs. Cooking method or an interaction between cooking method and food form was significant in all the overall models for each compound, and 80% of models created for stratifications of the data based on meat type. Improvement on compilations such as the Computerized Heterocyclic Amines Resource for Research in Epidemiology of Disease (CHARRED) database comes from inclusion of additional studies, PAH compounds, more generalizable food categories, more cooking methods (such as smoking), and addition of seafood. Meta-regression allows parameters to be estimated with separation of between-study heterogeneity, and extrapolation of exposures to more foods. Resulting uncertainty estimates are useful in a probabilistic exposure assessment.