Drug development: more science, more education

In 1997 the late Lew Sheiner published a paper on what he called learning and confirmation in drug development, by which he meant the different processes involved in the early and later stages of development [1]. His paper was long and difficult – Lew was not one to compromise intellectual rigour for the sake of giving you an easy ride. But it repays a careful reading, and the bottom line was that 'increased attention to science' in drug development would drive drug approval more efficiently. He meant that the types of techniques that are commonly applied during the early process of learning about a drug (dose-finding studies, carefully designed pharmacokinetic and pharmacodynamic studies, and the inclusion of heterogeneous groups of subjects, women as well as men, old as well as young, black as well as white [2]) might prolong the early phases of development but would provide information that would shorten the later process of confirmation. Such careful studies would also improve decision-making at the go/no-go stage and reduce the number of later casualties. Not much heed has been paid to this important lesson. For example, a drug entering Phase 1 trials in 2000 has only an 8% chance of reaching the market, compared with the historical success rate of about 14% [3] and the process has been estimated to be about 5% slower. Pharmacokinetic-pharmacodynamic studies, at all stages, both early and late, are thin on the ground. And we have little information about the major factors that determine the benefit to harm balance of drug therapy: dose-responsiveness, the time-course of effects, and susceptibility to beneficial and harmful effects [4]. Specifically, for all drugs we need to know more about: the between-subject variability in dose-responsiveness for both benefits and harms; the qualitative and quantitative differences between dose–response curves for benefits and harms (i.e. their relative potencies and efficacies, and the slopes of the dose-response curves for different effects); the time-courses of beneficial and adverse effects and how the benefit to harm balance varies with time; the factors, genetic and environmental, that determine individual susceptibilities to both beneficial and adverse effects. The number of new agents going through the regulatory process has fallen dramatically in recent years; for example, the number of biologics licence application submissions to the FDA, highlighted in its recent report on innovation, fell from 70 in 1993 to 21 in 2003 [3]. Sheiner proposed that more clinical pharmacologists should be involved in the early stages of drug development, and the FDA report has re-emphasized that message [3]. But, paradoxically, the number of senior clinical pharmacologists in UK universities is falling. This is in large part due to the increasing emphasis on research that will be rated highly in the UK's Research Assessment Exercise, particularly reductionist rather than functional science, and partly due to the increasing emphasis on research over teaching, to the detriment of the latter. Two papers in this issue of the Journal are germane to these problems. Fewer new drugs there may be. But are the ones we have of poorer quality? Motola et al.[5] have tried to answer that question by analysing the 151 therapeutic agents that were given regulatory approval by the European Medicines Evaluation Agency (EMEA) between 1995 and 2003. They classified them according to whether they were highly, moderately, or modestly innovative. Innovation was defined by taking into account the availability of other therapies and the effectiveness of the new agent. The algorithm was complicated, and involved value judgements about the nature of the problem and the quality of the benefit achieved. Of the 151 therapeutic agents, 121 (80%) were for what the authors considered to be serious diseases; 47 (39%) were classed as highly innovative and 19 (16%) moderately innovative; but 51 (42%) were classed as being 'merely' pharmacologically or technologically innovative. Without more detailed information it is hard to know how to interpret these last two categories. For example, insulin glargine and insulin lispro were classified as 'mere' pharmacological and technological innovations; but these new forms of insulin have made it easier for many people with diabetes to manage their condition and must have benefited them greatly. On the other hand, the category of pharmacologically innovative agents also included some me-too drugs, such as tadalafil and vardenafil, while sildenafil was counted as being highly innovative, a surely incontrovertible judgement; although the value of me-too drugs should not be disregarded [6]. Only 11 new agents were developed for treating risk factors for serious diseases, and most were pharmacological innovations, while of the 19 new agents for diseases that were not serious, most were pharmacological or technological innovations. So, overall 71 (47%) of the 151 agents were classed as being highly or moderately innovative, either because there were no alternatives or because they were highly beneficial, or both. Harms were not taken into account in this analysis. No matter how innovative a medicine is it will not be welcome if the benefit to harm balance is unfavourable; practolol, mibefradil, and rofecoxib come to mind. But harms are more difficult to detect than benefits, and comparisons of the two types of outcome are difficult. Assessing the benefit to harm balance for these new agents is a challenge that Motola et al. should consider tackling. And it would be interesting to know how innovativeness has changed over time. The most eye-catching data in this paper are the numbers of agents in the different categories that were derived from biotechnology. Of the 151 agents, 55 (36%) were so classified. But of the 71 that were thought to be highly or moderately innovative, only 19 (27%) were derived from biotechnology. This is a disappointingly low proportion, given the large amounts of resources that have been poured into this type of science since the discovery of the structure of DNA over 50 years ago, and particularly in the last 20 years, since George Koehler and César Milstein won the Nobel prize for 'pioneering contributions to the theory and techniques of immunology', primarily for their work on monoclonal antibodies. Should we be devoting so much time and energy to this line of development? Perhaps only a quarter of the budget, if these figures are to be taken at face value. Change will be slow, so great is the investment in the newer technologies and the hope and hype that surround them [7]. However, education can help in spreading the word, and that is what Stanley et al. reflect upon elsewhere in this issue, where they describe a programme for teaching medical students about drug discovery and development [8]. My own less formal experience is that teaching how drugs are discovered and developed not only informs about the processes but also gives a useful peg on which to hang important basic principles that are related to practical drug therapy. Sheiner said it, and we should keep on saying it: we need more and better science in the field of drug development and more education about how it is done and how to do it.