Quantitative risk approaches for formaldehyde.
Formaldehyde presents unique data that highlight significant issues in the extrapolation of animal studies to human risk assessments. Formaldehyde causes rare nasal cancer in rats at 15 ppm, but not at lower levels of 2 ppm and 0.5 ppm in the range of human exposures. Mice and hamster studies even at high levels have results similar to low dose rats. Higginson et al. reviewed the human epidemiology studies and concluded that no excess cancer risk was observed; and if a risk exists, it is very low. Formaldehyde is a natural metabolite--the human body turns over 51 g/day. Cells, therefore, have detoxification and other defence mechanisms to formaldehyde. Recent CIIT biomechanism results elucidate these factors. These data raise two issues: First, the appropriateness of linear quantitative risk methodology given the non-linear nature of the biological data. Either a non-linear (threshold) statistical model or NOEL approach are appropriate risk assessment techniques for formaldehyde. Second, the rare nasal cancer observed in rats also occurs in control animals. A comparison of relative risk between background and low formaldehyde exposures has been calculated for both groups. Non-linear (MLE) 5 stage multistage models estimate 0 per million risk from both background and 1 ppm of formaldehyde exposure. Linear or upperbound (95%) estimates are 7,200/million from background and 5,000/million from 1 ppm formaldehyde. These estimates have a significant impact on formaldehyde regulatory programs for warning labels and "safe" exposure levels.