In this paper we propose a new rule for drug design when optimizing drug oral bioavailability or fraction of a drug that reaches the peripheral circulation after oral administration. Defining the molecular properties that play a role in oral bioavailability is a critical, but poorly understood aspect of new drug discovery.

We have discovered that molecular rigidity plays an unexpectedly important role in determining the oral bioavailability of drug candidates. This observation could be widely applicable to improving the rate of new drug discovery. It now appears that primary emphasis in drug design should be on reduced number of rotatable bonds rather than the previously emphasized, but correlated property of reduced molecular weight.

Oral bioavailability is a critical issue in the discovery of many new drug candidates. Lipinski’s "rule of five" has been a key guide to suggesting the molecular properties that should be altered to improve the so-called "drug-like" properties that include oral bioavailability. This rule was developed using a drug database that allows one to pinpoint common properties of known drugs, but not to specifically identify preferred properties. In general there has not been an oral bioavailability database of sufficient size and molecular diversity to allow a rigorous analysis of how molecular properties influence drug oral bioavailability. We have analyzed a recently developed database of rat oral bioavailability involving a diverse compound collection at the former SmithKline Beecham Laboratories and found this property to be highly dependent on molecular rigidity as measured by the number of freely rotatable bonds. Molecular weight is seen to be a determinant of oral bioavailability only to the degree that it correlates with the number of rotatable bonds. A new rule for predicting an improved likelihood of high or drug-like oral bioavailability for new drug candidates has been proposed.

Much of my career in Medicinal Chemistry has involved efforts to improve the pharmacokinetics of lead structures, often in compound classes that generally fail the rule of five, such as peptides or peptidomimetics. The lack of a good oral bioavailability database to guide design has been frustrating and made this largely a matter of guesswork. In recent years many projects at SmithKline Beecham (now GlaxoSmithKline) placed a high priority on obtaining consistent oral bioavailability data in rats suitable for a systematic analysis. It has been rewarding to finally have an opportunity to analyze this data in a way that reveals new priorities for drug design.