Do rats comply with EPA policy on cancer risk assessment for formaldehyde?

Formaldehyde has been shown to cause nasal squamous cell carcinomas in the rat following 2-year inhalation exposure. The incidence of this tumor in a historical data base of 16,794 rats was nil, indicating that it is a rare spontaneous tumor. Five different mathematical extrapolation models were applied to the rat nasal tumor data to produce estimates at 10(-4) risk (the size of the historical data base) of between 3.232 and 0.003 ppm formaldehyde depending on the model and choice of maximum likelihood estimate or lower confidence limit values. Assuming that an ambient level of 0.07 ppm formaldehyde exists in a rat house, the multistage linear model did not predict correctly within the observed data. The EPA policy model (multistage third order) was not inconsistent with the observed data (P = 0.259). However, the unit risk, derived from this form of modeling, shows considerable inconsistency, at ambient levels of formaldehyde, when compared to observed incidences of nasal tumors in the general human population. It is proposed that the multistage models are inappropriate, and that caution should be exercised in the extrapolation of highly nonlinear animal tumor data.