Pharmacogenetics—expectations and realityBMJ 2004; 329 doi: https://doi.org/10.1136/bmj.329.7456.4 (Published 01 July 2004) Cite this as: BMJ 2004;329:4
- Geoff Tucker, professor of clinical pharmacology ()
- Academic Unit of Clinical Pharmacology, Pharmacokinetics and Pharmacogenetics Group, University of Sheffield, Royal Hallamshire Hospital, Sheffield S10 2JF
Thirty years have passed since Mike Rawlins, the current chairman of the National Institute for Clinical Excellence (NICE), coauthored a small but perfectly formed book entitled Variability in Human Drug Response.1 In the interim we have witnessed the waxing and waning of clinical pharmacology and the inexorable rise of genomic medicine. Genomic medicine has generated many expectations with regard to the advent of “personalised medicine” and “individualised prescriptions,” fuelled by the pace of technological advances in genotyping; enthusiasts extrapolating beyond small proof of principle and retrospective studies; and a few apparent success stories, such as the treatment of breast cancer with trastuzumab (Herceptin) and of HIV with abacavir (Ziagen).
But the promise of pharmacogenetics has largely remained unfulfilled. In general, drug response and toxicity are likely to be a complex function of the influence of many genes interacting with environmental and behavioural factors. Trastuzumab is effective in only the 15-20% of breast cancer patients who respond positively to a test for mutations in the tumour that over-expresses human epidermal growth factor receptor (HER)-2.2 And about half of white, male, HIV positive patients with specific variations in the HLA-B gene are likely to develop severe reactions to abacavir.3 These examples remain controversial with regard to the specificity, exclusivity, cost, and reliability of the associated genetic testing; they also represent cases in which gene penetrance and frequency are relatively high.
Outside clinical pharmacology, poor prescribing skills, interactions between drugs and between drugs and herbs, and lack of adherence to treatment are insufficiently acknowledged as causes of therapeutic failure or adverse drug reactions. Although genetic testing is often held up as a way of improving compliance, this phenomenon has a notable behavioural component that is independent of drug and disease.4
Given that genetic factors need to be put into perspective, the challenge now is to assemble large, prospective, multidisciplinary, multicentre projects to assess the real clinical and economic value of predictive genetic testing in drug therapy. This coincides with the new dawn of clinical investigation ushered in by the perceived failure of new drug development and a recent flurry of position papers.5 6 Whereas the pharmaceutical industry seems largely to be taking a “wait and see” attitude with regard to targeted treatment solutions as opposed to the traditional “one size fits all” approach, the UK Department of Health has taken the initiative in calling for research proposals for the development of genetic tests for existing drug therapies that have a greater than 50% chance of reaching the bedside in five years' time. This may be a tall order, but it is one that focuses attention on the key requisites for evaluating the potential cost effectiveness of pharmacogenetic strategies to help healthcare providers.
Several primary characteristics that will enhance the cost effectiveness of pharmacogenetic testing have been proposed.7 These include severe clinical or economic consequences that can be avoided through the use of a test, difficulty in monitoring of drug response using current methods, lack of an alternative drug with equivalent therapeutic profile and price, the existence of a well established association between genotype and clinical phenotype, the availability of a rapid and inexpensive genetic test, and a relatively high frequency of the variant gene. The table lists some examples of relatively low hanging fruit with regard to prospective evaluation, albeit with varying degrees of concordance with these criteria. A recent example of the development of a predictive dosage algorithm incorporating genetic testing showed that 39% of the variance in the maintenance dose of warfarin could be explained by a combination of genetic, clinical, and demographic factors.8 The use of the algorithm also more than halved the risk of adverse drug reactions. Further refinement of the model could lead to a cost effective improvement in the use of warfarin.
In the treatment of complex diseases, such as cancer and hypertension, and in the prediction of adverse drug reactions in general, the science has not advanced much beyond the fishing expedition or stone turning approach to identify important combinations of genetic determinants of drug response. Accordingly, in these areas—with the exception of the use of genetic tumour markers—the promise of relatively straightforward predictive tests is likely to be much further down the line.
Several criteria can be proposed for the conduct of prospective studies to develop predictive genetic tests of drug response. The major candidate genes should be well documented as being functionally relevant and should cover all aspects of the pharmacology and toxicology (receptors, enzymes, transporters, immunomodulators). The possibility of significant global and regional variation in the relevant genes should be addressed. Demographic and environmental factors (for example, drug interactions) should also be considered as covariates. Drug exposure and compliance should be assessed rigorously to reduce “noise” in the data. The study should have adequate power—assuming that the statisticians can agree on methods for assessing multiple gene effects. Once a predictive model with data from an initial set of patients is developed, this should be tested prospectively in a further set. A professional cost benefit analysis should be included.
Potential constraints on implementing the promise of precise prescriptions include general ethical and racial issues,9 the disincentives to clinical investigation imposed by recent inflation in the associated paperwork, costs involved in gaining ethical and regulatory approval, and considerations of patients' and doctors' acceptance of genetic testing.10 Additional cost implications may occur in countries where healthcare is not free at the point of delivery. Historically, methods of predicting drug responses, such as renin profiling and acetylator phenotyping, have had a very low uptake in the United Kingdom. Genotyping may be different? Time, as ever, will tell.
Competing interests None declared.