Biologically motivated computational modeling: contribution to risk assessment.

The article highlighted in this issue is ‘‘Human Respiratory Tract Cancer Risks of Inhaled Formaldehyde: Dose Response Predictions Derived from Biologically Motivated ComputationalModelingofaCombinedRodent andHuman Dataset,’’ by Rory Conolly, Julia Kimbell, Derek Janszen, Paul Schlosser, Darin Kalisak, Julian Preston, and Frederick Miller. In the featured article, Conolly et al. describe the development of the human component of a biologically motivated computational model to predict exposure response at levels of formaldehyde less than those associated with squamous cell carcinomas (SCC) observed in Fischer 344 rats exposed by inhalation. The article addresses extension of the computational model to the entire respiratory tract of humans, complementing a previous description, which presented modeling for the nasal airways of rats (Conolly et al., 2003). Extension to the entire respiratory tract is relevant for prediction of risk associated with oronasal breathing of humans, as occurs at higher exertion levels characteristic of those likely in the occupational environment. The computational model incorporates regenerative cell proliferation as a required step in the induction of tumors by formaldehyde and a contribution from mutagenicity that has greatest impact at low exposures through modeling of complex functional relationships for cancer due to actions of formaldehyde on mutation, cell replication, and exponential clonal expansion. Species variations in dosimetry are taken into account by computational fluid dynamics modeling of formaldehyde flux in various regions of the nose and a single path model for the lower respiratory tract of humans. Specifically, the animal model includes an anatomically realistic three-dimensional computational fluid dynamics model (CFD) of rat nasal airflow and site-specific flux of formaldehyde into the tissue in which the nasal SCC developed. Flux is the relevant dose metric for the two relevant noncancer effects in the tissues: formation of DNA-protein crosslinks (DPX) and cytolethality/regenerative cellular proliferation (CRCP). A twostage clonal growth model links the modes of action with mutation accumulation and tumor formation. In the human component of the model described in this issue, predictions of regional dosimetry are based on human versions of the CFD model and a linked typical path model for the lower respiratory tract. Regional formation of DPX driven by the CFDpredicted flux of formaldehyde into tissue is predicted by a human DPX model based on scale up from rat and rhesus monkey DPX models. CRCP data were those for the rat. Baseline parameter values for the human clonal growth model were calibrated against human lung cancer incidence data (Peto et al., 1992; SEER, 2003). The body of chemical specific and more generic biological information on which the model is based is extensive. Model structure is also consistent with the outcome of consideration of weight of evidence for mode of action of tumor induction in a formal framework (Liteplo and Meek, 2003; Meek et al., 2003). The database on which this weight of evidence analysis and the computational exposure response model on formaldehyde draws is impressive, including a specifically designed inhalation bioassay in which extensive dose-response data on intermediate endpoints were collected (Monticello et al., 1991, 1996). The clonal expansion model also draws upon an impressive more generic body of work on relative roles of mutation and cell proliferation in cancer induction dating back over 20 years. The three-dimensional computational fluid dynamics models contribute considerably not only to formaldehyde-specific but generic understanding of site specificity as a function of airflow resulting from the complex anatomy of the nasal passages and lung of rats and humans. The authors clearly describe uncertainties of the model structure and parameter values which include the lack of direct correlation between DPX and mutation (dose-response data for other potentially mutagenic lesions associated with formaldehyde not being available). Formal sensitivity analysis and Monte-Carlo based analysis of variability and uncertainty were also not conducted precluding identification of model 1 For correspondence via fax: (613) 954-2486. E-mail: Bette_Meek@ hc-sc.gc.ca.