Introduction

Design of successful drug development candidates requires balancing a number of different characteristics simultaneously including intrinsic activity, biopharmaceutical properties, synthesis, stability, and many others. Independently, each of these can be considered a barrier to drug performance or development success (Kerns and Di, 2003). Among the most important of these processes determining in vivo performance are absorption, distribution, metabolism and excretion, collectively referred to as ADME. The structure and physicochemical characteristics of the drug are important determinants of these processes, as are the characteristics of the physiological mechanisms. Further complicating the issue is the interrelationship of many of these processes, frequently in antagonistic ways. Increasing solute lipophilicity, for example, can decrease aqueous solubility, which frequently compromises oral absorption. Also, it can increase metabolic clearance, thus making difficult sustaining the pharmacologically relevant systemic exposure. In contrast, permeability, in many cases, increases with increasing lipophilicity, favoring absorption (Conradi et al., 1996; Hansch et al., 2004). The actual result will be determined by the relative contributions of these two competing phenomena.

Permeability is also an important determinant of distribution, metabolism and excretion. Given these multiple dependencies upon specific drug structural characteristics, it can be seen that simultaneously optimizing all of these processes can present a formidable challenge.

Occasionally, one specific, clearly dominant problem, can be identified in a particular drug discovery program. In this case, focusing on the specific property can result in successful solution of the problem and identification of a promising drug candidate. The recent interest in development of computational structure property relationships has resulted in the discovery of promising tools to aid in these problem solving and property optimization objectives (Ekins et al., 2000; Lombardo et al., 2003; Wilson et al., 2003).

More generally, the interrelationship of multiple solute properties must be considered simultaneously in order to achieve the desired outcome. A simple example of this is the interrelationship of solute solubility and permeability in determining oral absorption potential (Hilgers et al., 2003). Briefly, bi-directional Caco-2 cell permeability data and aqueous solubility data were measured for a series of relatively homologous antimicrobial agents at Pharmacia. All compounds were dosed both intravenously and orally in rats in order to calculate total clearance and oral bioavailability. The clearance data supported the expectation that these compounds had minimal first pass metabolism such that bioavailability was directly related to the fraction of solute absorbed (Hilgers et al., 2003). In attempting to predict absorption of these compounds, it was found that no significant correlation existed between bioavailability (fraction absorbed) and either solubility or permeability alone.

While high permeability solutes did seem to be well absorbed, so were many of the low permeability solutes. What was common among these poorly permeable, well absorbed compounds was high aqueous solubility, suggesting that this property may help to compensate for the poor in vitro permeability.

One simple model for interrelating solubility and permeability quantitative data to predict absorption is the Maximum Absorbable Dose (MAD) model (Johnson and Swindell, 1996; Curatolo, 1998). MAD is a hypothetical value that estimates the amount of an infinite dose that could be absorbed, given a combination of solubility and absorption rate (calculated from permeability coefficient). In the present case, since a finite dose was actually given, the fraction of the dose absorbed can be estimated by dividing the MAD value by that actually dosed. Since MAD calculates a theoretical mass of solute assuming an infinite dose, it is possible to predict masses absorbed greater than than atually dosed. In this case, the predicted fraction absorbed is assigned a value of 1. In general, the correlation between predicted fraction dose absorbed and observed bioavailability was good except for the most highly permeable, lowest solubility compounds (Hilgers et al., 2003). On the assumption that for such solutes aqueous solubility may underestimate the intestinal solubility arising from bile salts and other surfactants present in the lumen, solubility of these compounds were measured in simulated intestinal fluid (SIF) and predicted fraction absorbed recalculated. These results are shown in Figure 1. The lesson from this example is that solubility and permeability must be considered together in prioritizing these candidates with respect to absorption potential and, for highly permeable solutes, solubility in biologically relevant media may need to be considered.

Figure 1. Oral absorption of oxazolidinone antibiotics in the rat. Rats were administered 25 mg/kg suspensions of the compounds as previously described (Hilgers, et al., 2003). Predicted fraction absorbed was calculated from Caco-2 cell absorptive permeability coefficients and solubility in simulated intestinal fluid (SIF) according to the Maximum Absorbable Dose (MAD) model.

Figure 1. Oral absorption of oxazolidinone antibiotics in the rat. Rats were administered 25 mg/kg suspensions of the compounds as previously described (Hilgers, et al., 2003). Predicted fraction absorbed was calculated from Caco-2 cell absorptive permeability coefficients and solubility in simulated intestinal fluid (SIF) according to the Maximum Absorbable Dose (MAD) model.

Predicted fraction absorbed

0 0

Post a comment