Few, if any, products undergo more safety testing and scrutiny than a pharmaceutical. This, of course, is for good reason as the most critical issue with regard to pharmaceuticals is the risk benefit ratio. Whilst patient benefits can be quite clear in most cases, the risk of undergoing a treatment regimen must be as carefully and broadly quantified as possible. Often this characterisation results in findings that result in discontinuation or withdrawal. Thus as drug withdrawals and experimental therapeutics that do not fulfil safety criteria during clinical trials are regarded as ‘failures’, they are actually successes for patient safety and for the science base of drug discovery. The lessons learned and the data gathered during this sometimes painful process is of critical importance in buttressing future efforts against similar problems.
The failure of a drug or clinical candidate is invariably associated with a massive amount of basic science that goes into problem-solving activities. This effort, if captured and integrated into the Discovery process, can contribute to the development of better and more sophisticated approaches to discovery safety assessment. Thus, the resources and momentum behind a problem solving effort, particularly around a late-stage clinical candidate present a unique opportunity to develop a more stable and valid science base upon which to build a more rational approach to safety assessment in early drug discovery. The capture, enhancement and exploitation of this science base is presently the focus of large centrally funded ‘predictive toxicology’ initiatives in both Europe (http://ec.europa.eu/research/health/imi/index_en.html) and the US (http://www.fda.gov/oc/initiatives/criticalpath/whitepaper.html).
What does an increase in the sophistication of a safety science base look like and to whom should it be directed? When faced with such an enormous problem it is always good to address the basics first. The choice of therapeutic target is one critical area and many advances have been made in recent years in the characterisation of a protein’s role in various tissues. This ‘target safety’ aspect is not a one-time exercise to be carried out at the beginning of a project, but, as the recent experience with Vioxx has shown us, a constant vigil to relate all aspects of the complicated life of a therapeutic target to the adverse event signals coming from our preclinincal and clinical studies.
However, it is the area of chemical design that perhaps the most value can be gained from translating safety data to real, tangible decisions. This is a long, tedious process but several successful examples have been identified in which a valid set of decision-making tools can be used to warn for chemical designs that contain inherent safety liabilities. In some cases this can lead to decisions before synthesis is even undertaken. I other cases the level of confidence is such that the tool sparks the decision to do further experiments in more sophisticated model systems to investigate the probability of a real safety problem. The keys to either of these scenarios is first, a clear strategy behind the decision-making tools such that results from a simple test (QSAR, in vitro) can be followed up and confirmed in a relevant in vivo test and second, adequate throughput to facilitate iterative design. Without this confirmation mechanism very little about the real risk of a compound or series can be concluded and without adequate throughput the chemistry will be mired in indecision.
Will it help?
With the goal of integrating as much of this rapidly expanding sophistication into chemical design, the potential for reducing safety-related failures exists, but is far from certain. One certain effect will be that promising compounds will fail less often for the reasons of today. Rather, the problems of the future will be qualitatively different, but if we have done a conscientious job in addressing the problems of today using knowledge from detailed mechanistic studies as well as screening, we will free up more time and resources within our safety groups for problem-solving around more of tomorrow’s problems. That, despite the real potential for continued failure, is progress and thus a success in itself.
Scott Boyer, Astra Zeneca