Gerhard Eisenbrand

• The weight of evidence pro/contra classifying the process-related food contaminant (PRC) acrylamide (AA) as a genotoxic carcinogen is reviewed. Current dietary AA exposure estimates reflect margins of exposure (MOEs) < 500. Several arguments support the view that AA may not act as a genotoxic carcinogen, especially not at consumer-relevant exposure levels: Biotransformation of AA into genotoxic glycidamide (GA) in primary rat hepatocytes is markedly slower than detoxifying coupling to glutathione (GS). Repeated feeding of rats with AA containing foods, bringing about uptake of 100 µg/kg/day of AA, resulted in dose x time-related buildup of AA-hemoglobin (Hb) adducts, whereas GA-Hb adducts remained within the background. Since hepatic oxidative biotransformation of AA into GA was proven by simultaneous urinary mercapturic acid monitoring it can be concluded that at this nutritional intake level any GA formed in the liver from AA is quantitatively coupled to GS to be excreted as mercapturic acid in urine. In an oral single dose–response study in rats, AA induced DNA N7-GA-Gua adducts dose-dependently in the high dose range (> 100 µg/kg b w). At variance, in the dose range below 100 µg/kg b.w. down to levels of average consumers exposure, DNA N7 -Gua lesions were found only sporadically, without dose dependence, and at levels close to the lower bound of similar human background DNA N7-Gua lesions. No DNA damage was detected by the comet assay within this low dose range. GA is a very weak mutagen, known to predominantly induce DNA N7-GA-Gua adducts, especially in the lower dose range. There is consensus that DNA N7-GA-Gua adducts exhibit rather low mutagenic potency. The low mutagenic potential of GA has further been evidenced by comparison to preactivated forms of other process-related contaminants, such as N-Nitroso compounds or polycyclic aromatic hydrocarbons, potent food borne mutagens/carcinogens. Toxicogenomic studies provide no evidence supporting a genotoxic mode of action (MOA), rather indicate effects on calcium signalling and cytoskeletal functions in rodent target organs. Rodent carcinogenicity studies show induction of strain- and species-specific neoplasms, with MOAs not considered likely predictive for human cancer risk. In summary, the overall evidence clearly argues for a nongenotoxic/nonmutagenic MOA underlying the neoplastic effects of AA in rodents. In consequence, a tolerable intake level (TDI) may be defined, guided by mechanistic elucidation of key adverse effects and supported by biomarker-based dosimetry in experimental systems and humans.