Predicting blood:air partition coefficients using basic physicochemical properties

Quantitative Property Property Relationships (QPPRs) for human and rat blood:air partition coefficients (PBAs) have been derived, based on vapour pressure (Log(VP)), the octanol:water partition coefficient (Log(K(OW))) and molecular weight (MW), using partial least squares multilinear modelling. These parameters are all included in the standard data to be submitted under REACH. The chemical dataset consisted of volatile organic chemicals, principally aliphatic hydrocarbons, benzene derivatives with one aromatic ring, and ethers, with and without halogen atoms. Other chemicals represented were cyclic hydrocarbons and carbonic acid esters. Separate rat and human models were derived, as well as mixed ones. Log(VP) and Log(K(OW)) contributed most to the prediction of Log(PBA) in the three-parameter model, while the contribution of MW was relatively small. Still, the three-parameter model differed significantly from the two-parameter model and performed better. Its performance was comparable to that of models published in public literature, which are based on more complex molecular parameters or on measured olive:oil air and saline/water:air partition coefficients. Since, based on the available data for humans, rats, mice, dogs and rabbits, existence of interspecies differences of PBAs cannot be clearly excluded, the use of separate models for each species is advisable. Concluding, the three-parameter human model Log(PBA)=6.96-1.04 Log(VP)-0.533 Log(K(OW))-0.00495MW and the three-parameter rat model 6.16-0.888 Log(VP)-0.521 Log(K(OW))-0.00201MW provide robust and reliable models for predicting PBA values of volatile organic chemicals using commonly available chemical properties of molecules.