To appreciate why most small molecule discovery efforts fail, it is helpful to understand the small molecule discovery process. Researchers begin with a Target Product Profile (TPP), a set of minimum characteristics a new drug must possess to warrant development and use in people.
Researchers must possess a deep biological understanding of the disease. They must also have a “biochemical pathway” or protein or macromolecule “target” through which the new drug is expected to exert its function. The hypothesis that modulation of this “target” will affect disease goes through a continuous process of “validation” at all stages of the path to the patient. Suitable chemical compound libraries are “screened” through a battery of in vitro assays to identify a small number of compounds with some hint of activity against the target. Whether the candidate drugs are first identified in vitro by high-throughput screening (HTS) or by structure-based drug design, any compound advancing as a “lead” candidate must then be co-optimized by subsequent chemical structure modifications for as many as 20 or 30 properties that contribute to the TPP. Lead finding and optimization is an extremely complex process in which failure is more frequent than success.
The screening and design processes yield families of compounds related by their structure-activity relationships (SAR). At this point a team of skilled medicinal chemists synthesizes a designed set of new but related compounds. Biologists evaluate these compounds for target binding and function in vitro, providing data that help the chemists sharpen the SAR. Through further rounds of refinement, chemists generate compounds with improved performance against the TPP.
Even with powerful computational algorithms that attempt to predict problems with particular compound structures and compound families, iterative cycles of design, synthesis and screening remain today the only viable method to refine favourable properties and eliminate the undesired ones in search of small molecule drug candidates for in vivo characterization. It is not uncommon for drug discovery teams to sift through hundreds or thousands of compounds in careful in vitro assays—or screen millions in high-throughput mode—to find a very limited number of lead compounds in two or three classes related by SAR.
If oral delivery is crucial to the TPP, the degree of difficulty in identifying lead compounds rises dramatically. The human body has multiple layers of defence to resist intrusion by foreign molecules. The body foils drug developers by a host of protective mechanisms. For example, the body may simply not absorb the drug from the stomach or gastrointestinal tract. Or the body may metabolize the drug rapidly. If the drug needs to reach the central nervous system, then the blood-brain barrier poses an additional hurdle.
Achieving sufficient and consistent oral bioavailability is a high hurdle for passage to the next stage. The few compound classes nominated for the in vitro medicinal chemistry optimization cycle are now iteratively re-optimized (or more likely, eliminated) by this requirement for an orally bioavailable molecule. Only when orally bioavailable candidates are identified can in vivo proof-of-concept studies to validate the biological linkage of the target and the disease be initiated in animals.
The few compounds that survive all of these hurdles are thoroughly evaluated in animals for their efficacy and possible toxicity. From the potential candidates that advance through these in vivo studies, usually only a single compound and a backup are nominated for further highly regimented, FDA-mandated IND-enabling work—with no guarantee that any compound will be qualified for subsequent clinical development.
The time and manpower required to advance a compound to candidate status is substantial. HTS and lead identification each require two to four full-time scientists working for six to 12 months. Lead optimization is significantly more demanding. A typical program has about 15 to 20 scientists and can last for two to three years, with a 10 to 20 percent chance to find an IND candidate out of the thousands or millions of chemicals evaluated throughout the process.
To find a potent, safe, and effective drug candidate for clinical development, drug discovery as practised today results in a steep “de-selection funnel.” That is, during serial assessment the undesirable properties of proposed hit and lead compounds rapidly reduce the number of candidates under evaluation. Thus, compound attrition rates are very high. Nevertheless, these steps, combined with scientific rigour and luck, transform simple organic chemicals into valuable drugs for the safe and effective treatment of human disease.