Can the time from synthesis design to validated chemistry be shortened?

The traditional way of identifying a potential new drug is to synthesize an d test one candidate at a time. Design and optimization of reaction and wor kup conditions for each new molecule are accomplished through experimentati on varying one-variable-at-a-time (OVAT). While this approach of drug disco very has been extremely successful, market pressure to discover and bring n ew therapies to the customer in half the time is forcing pharmaceutical org anizations to look for new ways to find active compounds and optimize serie s leads. Companies are now using combinatorial chemistry to rapidly synthes ize and screen hundreds to thousands of compounds to identify lead candidat es and synthesizing hundreds more variations of the lead structure to optim ize activity. What has not changed is the need for reliable chemical transf ormations that will perform for a wide range of compounds. Considerable tim e is still being expended designing and validating these transformations be fore the parallel syntheses can begin. The challenge still being faced is r educing the time between synthesis design and validated chemistry. The goal of validated chemistry is achieved when sufficient experimental informatio n is obtained to permit the identification of reaction conditions or variab les that have significant influence on yield and purity of the chemical tra nsformation. Reaching this level of understanding may be shortened consider ably by using experiment designs that can take advantage of the parallel ex perimentation capabilities that the combinatorial chemistry field has suppl ied. Experiment designs that are more suitable for parallel experimentation and provide more information than OVAT experiments are the factorial desig ns. These designs involve the variation of all of the studied variables in a systematic manner. The outcome of these experiments are quantitative esti mates of the influence of each variable, the identification of variable int eractions (synergy), the estimation of experimental noise (error estimates) , and polynomial models that can be used to optimize the chemical transform ation. Because of the structure of these experiment designs, additional exp eriments run in the future can be added to the original design to extract a dditional information from the combined set. This last feature removes the need to commit to a large number of runs before sufficient knowledge about the chemistry is known. A recent example from the combinatorial chemistry l iterature is used to illustrate the features of factorial and fractional fa ctorial designs, and to demonstrate the benefits of using these types of ex periments. Graphical analysis of the data is used to illustrate that a form al training in statistics is not needed to take advantage of these designs. (C) 1999 John Wiley & Sons, Inc.