The molecular revolution in biology and medicine has already had an enormous effect upon the drug discovery process and seems destined to have a similar impact upon development.
Between the mid-1960s and mid-80s the world pharmaceutical industry’s capacity to innovate was beginning to be limited by the number of clearly identified biological targets. In consequence many companies worked on the same targets and, not surprisingly, they often produced very similar drugs. Almost all the pre-clinical pharmacodynamic and pharmacokinetic research was carried out in isolated tissues or intact animals and there was a substantial degree of uncertainty about the validity of extrapolations from animals to man. In a remarkably short time genomics has changed the situation from a limited range of targets in animals to a profusion, most of which are human proteins or protein targets in pathogenic organisms. Every week, as sequencing proceeds apace, several new human genes are being discovered. Bioinformatics specialists use this information to look for analogies with known proteins or functional homologies in simpler organisms like C. elegans. Recombinant DNA technology has made it relatively easy to express human proteins in cultured cells and automated high throughput screening has made it possible to screen thousands of chemical compounds a day for activity on the target protein. Legacy chemical archives are being supplemented, and may in time be supplanted, by use of combinatorial chemistry which can achieve the levels of productivity required to feed the high throughput screens. Lead compounds derived from pharmacological screens can themselves be screened for properties such as solubility, permeability and metabolic stability to (human) cytochrome P450s. The front end of pharmaceutical discovery research has increased its efficiency by orders of magnitude but, in doing so, has created a range of new problems. One of the most important of these is target validation.
When a molecular biologist clones a new gene it is often possible to guess at its functional class from a deduced amino acid sequence and homology with other proteins of known function e.g. a 7-transmembrane spanning receptor or a protease. Identification of the tissues in which the gene is expressed may give a clue to function but it is still a long step to know whether a gene that looks as though it codes for a receptor or an enzyme and is expressed in interesting areas of a tissue such as brain is a worthwhile pharmaceutical target. Several methods can be used. If the proposed target is a receptor its affinity for its natural ligand can be used to find the ligand, a method known as ligand fishing. There have been some spectacular successes from this approach e.g. nociceptin/nocistatin and orexin. Once the ligand is identified it is relatively straightforward to delineate its function using the techniques of experimental physiology. Another approach is to use gene knock out, or if the knock out is lethal to the embryo, knock down using the cre/lox method. In this method the cre-recombinase is used to achieve excision of a gene flanked by two lox sequences. A third approach is to obtain diseased human tissue and use array technology to investigate the expression of a large number of potentially disease-related genes, including the one of current interest. Changes in the amount and timing of gene expression may allow an inference to be drawn about its potential role in the disease process. Animal models of disease have been invaluable in conditions such as hypertension but are of very limited value in psychiatric disease. An area which is growing in importance is to identify allelic polymorphisms in man that predispose to, or protect from disease. In reality none of these methods provides a robust validation of the target but they may establish a strong enough hypothesis to justify a drug discovery effort.
The details of the processes for developing a sustainable chemical lead through high throughput screening, combinatorial chemistry, molecular modelling and structure activity relationships are beyond the scope of this review but molecular biology is having an impact here too. Most obviously, lead compounds can be screened for their metabolic stability, the P450 isoforms which they use for metabolic degradation and their ability to inhibit important isoforms such as CYP 3A4. With use of these methods many drug interactions should be avoidable, or, if that is not possible, predicted well in advance. Using rates and routes of metabolism derived from these data and studies with human hepatocyte prediction of in vivo kinetics in man is improving. Molecular methods can be extended to toxicological screening by investigating the expression of genes concerned with tissue repair using array technology as early, and more sensitive, indicators of tissue injury than traditional morphological methods.
The next big challenge is the move into man. At least in theory, studies in normal volunteers should be more predictable. The new drug will, almost certainly, be active on the human protein target and many of the pharmacokinetic parameters should have been predicted with reasonable accuracy. Of course, the response in an integrated conscious human may still be rather different and retains a substantial element of the unexpected. Human volunteers develop unexpected headaches, nausea and vomiting, dizziness etc despite the best efforts to predict them. Demonstration of a known pharmacodynamic action in man although frequently difficult is often possible. Clinical pharmacologists have proved to be imaginative in using both direct measurements, challenge techniques and new techniques like position emission tomography with 11C labelled ligands to define the pharmacological action and dose-response curve of a new drug in man. The real challenge is to obtain an early indication of therapeutic activity, in the knowledge that the majority of the new approaches will fail.
The link between a specific pharmacodynamic action and a therapeutic effect ranges between direct and immediate, e.g. the use of (−)-thyroxine in myxoedema, to (apparently) indirect and poorly understood e.g. dopamine D2-receptor antagonists in schizophrenia. The standard method of demonstrating proof of therapeutic concept is to carry out a randomised, controlled, phase II trial with adequate statistical power to permit a robust conclusion. Often such a trial will involve 300–400 patients, last many months and cost several million pounds. With more speculative new approaches this is a high cost to pay because it limits the number of options that can be tested. With a profusion of potential targets there is much to be gained in earlier identification of promising therapeutic approaches, and for industry corresponding commercial advantage, if it is possible to gain insight into potential therapeutic activity by use of smaller scale and more intensive studies in patients. This had led to a renaissance in interest in the discipline of ‘Experimental Medicine’ or ‘Experimental Therapeutics’ which brings together advanced techniques of clinical measurement with the best available chemical, cellular and molecular assay methods. The ghost of the last 1960s style clinical investigator is escaping from the polymerase chain reaction (PCR) oven and materialising again at the bedside (but this time with a PhD in his or her knapsack). It is not an exaggeration to say that if we are to realise, reasonably quickly, the enormous potential therapeutic benefits of the molecular revolution in drug discovery we must move quickly to establish ‘Experimental Medicine’ both in clinical academic centres and in industry.
Once the therapeutic hypothesis has been established an even larger question remains about the impact of genomics on the way drugs are used. There are widely conflicting views. One extreme holds that the physician of the future will routinely use a high density DNA chip to test for allelic polymorphisms before initiating treatment and the choice of drug and dose may be largely based on the readout from the chip. The sceptics argue that the, sometimes very wide, inter individual differences in drugs response caused by polymorphisms of CYP2D6 or CYP2C19 have had remarkably little effect upon prescribing and doubt that the, individually much smaller, contributions from an array of single nucleotide polymorphisms will have any greater effect upon prescribing practice. Probably the truth is somewhere in between.
The enthusiasts undoubtedly underestimate variability in drug response due to environmental factors and variable compliance. In the MRC hypertension trial the winter/summer variation in systolic pressure was almost as great as the difference between active treatment and placebo. One participant in a recent pharmacogenomics meeting commented, wryly, that the most important pharmacogenetic discovery would be the gene which determined non-compliance with prescribed treatment. But the sceptics also have their problems. No-one can pretend that the current ‘one tablet once a day’ approach to therapeutics is optimal. If physicians will not take the trouble to titrate treatment perhaps they will pay more attention to a genetic test that prompts them to do so. And once there are tests on the market all the usual marketing tools will come into play—and they will have an effect upon prescribing behaviour. Obviously all this will have to be tested in large scale trials to define dose-response curves and the feasibility of predicting optimal drug, optimal dose and adverse reaction profile. Even if the genetic promise is not fully delivered, a re-kindled interest in the individualisation of treatment will be a worthwhile gain.
The founding fathers of British clinical pharmacology were always clear that the discipline was ‘Clinical Pharmacology and Therapeutics’ with a strong base in the clinic. Time has proven that was a far sighted position. When we founded the Journal 25 years ago we did not carry ‘Therapeutics’ into the title, preferring the snappier BJCP, but the intention was always there. Paradoxically, the molecular era has reinforced the need for experimental medicine/therapeutics as a vital element of clinical pharmacology.