Pharmacogenetics begins to deliver on its promises

Pharmacogenetics promises the safer use of drugs through personalised prescribing that is informed by the genetic make-up of the individual, cancer tumour, or invading micro-organism. Although the first inherited molecular marker for a severe drug induced toxic effect was identified over half a century ago, until recently, few such markers were routinely used by prescribers. Even pharmacogenetics pioneers were beginning to wonder whether it was all hype. However, rapid progress is being made, and an increasing number of drugs have been licensed for use with a companion genetic test, both to improve the chances of success and to reduce the risk of severe side effects. This progress is typified in a linked article reporting that genetic testing helped to reduce patients’ risk of a rare but serious adverse reaction to allopurinol.1

Abnormal reactions to foods and chemicals have been recorded since antiquity. Pythagoras, better known for his geometry, commented on hypersensitivity to the broad bean (favism); a reaction now known to be associated with hereditary deficiency in glucose-6-phosphate dehydrogenase (G6PD). Over 2000 years later, Charles Darwin observed that hair and skin colour was “correlated . . . with a complete immunity from the action of certain vegetable poisons, and from the attacks of certain parasites,” and that this immunity seemed “to be partly inherent.”2

With the introduction of synthetic drugs, this interplay between nature, nurture, and variability in drug response became apparent. Some individuals given the antimalarial primaquine developed haemolytic anaemia while most did not. Furthermore, black patients were more susceptible to this adverse effect.3 Similar observations in the 1950s led to the birth of pharmacogenetics; the study of hereditary influences on drug response with Mendel’s laws of inheritance and Garrod’s “inborn errors of metabolism” providing the theoretical framework.

Subsequent studies showed that wide variability in dose requirement was often attributable to genetic variation (for example, warfarin). Others discovered that some cancer drug responses depended on specific genetic variations, and ethnic variability in response could sometimes be explained by differences in the prevalence of the variants. Pharmacogenetic targeting is now at the core of drug discovery research, and many highly successful, molecularly targeted agents have emerged within a widening therapeutic spectrum. Prescribed with companion tests, many have become so called “blockbuster” drugs, such as imatinib and trastuzumab.

Pharmacogenetics advanced still further with improved understanding of adverse drug reactions, once termed “idiosyncratic” and generally thought to be immunologically based. In 1980, Baruj Benacerraf, Jean Dausset, and George Snell were awarded the Nobel prize for identifying the genetic locus later shown to be the source of much of this idiosyncracy: the major histocompability complex or the human leucocyte antigen (HLA).

An illustration of the participation of HLA in drug reactions is the strong association between hypersensitivity to abacavir and the HLA-B*57:01 allele.4 5 Another example is the association of the HLA-B*15:02 allele with carbamazepine induced severe cutaneous adverse reactions (SCARs), including the potentially lethal Stevens-Johnson syndrome and toxic epidermal necrosis.6

In this week’s issue of The BMJ, Ko and colleagues report a prospective evaluation of HLA-B*58:01 genotyping for preventing allopurinol associated SCARs.1 The researchers genotyped 2910 patients eligible for allopurinol treatment but not previously exposed to the drug, and prescribed alternative or existing treatments for the 571 patients with the target allele. Using historical incidence as a comparator, the researchers expected to see seven cases of SCARs but found none; a highly significant difference.

Are these results generalisable enough to prompt routine genetic testing for patients needing allopurinol? Not yet. Participants were all Han Chinese from Taiwan, a population with a high prevalence of the HLA-B*58:01 allele (about 10%). While the association has been reported in other populations, there are several reasons for caution.7 8 9 10 The risk allele frequency is low in white populations (<1%). More importantly, even in other Asian populations, the strength of association between allele and reaction is weaker. Almost all carriers of HLA-B*58:01 (about 98%) do not develop SCARs, and in some ethnic groups (including white populations), many patients with the adverse effect do not carry the risk allele.11

Even when genetic prediction of risk is robust, there are important trade-offs when considering routine testing before treatment. Ko and colleagues chose benzbromarone as the main alternative to allopurinol, a drug with a well defined efficacy and safety profile. Although no major adverse effects were observed, benzbromarone has been withdrawn from the market in many countries because of its potential for hepatotoxicity. The other alternative, febuxostat, is a new and expensive drug with a poorly defined, long term safety profile that includes reports of potentially lethal liver failure and cardiovascular events.12 If genotyping is not available, prescribers could be tempted to substitute allopurinol for less safe or effective alternatives, to the detriment of most patients.

Pharmacogenetic studies have made important contributions to the safer use of drugs. Studies such as the present analysis by Ko and colleagues make adverse reactions—once classified as type B or idiosyncratic and bizarre—more predictable and avoidable.

Fifty years ago, a BMJ editorial commented: “To prevent attacks persons deficient in G6PD must avoid all potentially harmful drugs and foods. There is a case to be made for the use of routine screening tests, such as the spot-test, on all males of Mediterranean, Asian, or African extraction before treatment with sulphonamides, aspirin, phenacetin, and other drugs.”13 Such routine testing was never adopted because implementation of pharmacogenetic testing requires careful consideration of the benefits and harms involved, especially if the distribution of a risk allele among different ethnic populations is highly variable.