Structure-permeability bioavailability properties

One of the key parameters for correlating molecular structure and chemical properties with bioavailability has been transcorneal flux or, alternatively, the corneal permeability coefficient. The epithelium has been modeled as a lipid barrier (possibly with a limited number of aqueous pores that, for this physical model, serve as the equivalent of the extracellular space in a more physiological description) and the stroma as an aqueous barrier (Fig. 11). The endothelium is very thin and porous compared with the epithelium [189] and often has been ignored in the analysis, although mathematically it can be included as part of the lipid barrier. Diffusion through bilayer membranes of various structures has been modeled for some time [202] and adapted to ophthalmic applications more recently [203,204]. For a series of molecules of similar size, it was shown that the permeability increases with octa-nol/water distribution (or partition) coefficient until a plateau is reached. Modeling of this type of data has led to the earlier statement that drugs need to be both... 

LC/MS/MS techniques with selective and sensitive detection methods make it possible to quantitatively analyze samples from Caco-2 cell and PAMPA buffer matrices. A high-throughput permeability screen with robust LC/MS technology can quickly generate information about structure-permeability relationships that are extremely valuable in the lead optimization phase for the selection of pre-clinical candidates with favorable oral bioavailability properties. 

Numerous publications have appeared in recent years describing structural and physical properties that are related directly or indirectly to permeability, oral absorption, or oral bioavailability. These have recently been reviewed " and here we summarize several contributions (see Table 10.5). 

Figure 8.1 shows that the properties that Abbot scientists predicted most frequently for their compounds, using their WWW portal, are structural alerts for toxicity and mutagenicity. " The second largest number of predictions is for log P followed by bioavailability scores (rule of five and internal Abbot score ), p.Ka, and solubility prediction as well as prediction of some other more complex properties, such as Blood Brain Barrier (BBB) permeability, binding energy, polar/non-polar surface area, and log D. 

A cascade method was proposed using recursive partitioning and descriptors generated with a program called Algorithm Builder were used with a 800-compound training set [35]. Their predictions are 2-class models with Fs less than and more than 30%, respectively. As parameters, they use a combination of solubility, pifractions ionized, human permeability, P-gp substrate specificity, physicochemical properties, and various structural descriptors. This is an attempt to model the components of bioavailability and then to integrate them into an overall prediction (see further in Section 16.4). 

---