Pharmacogenetics:ethical issues

What is pharmacogenetics?

The study of how genetic variation affects our response to medicines.

A pharmacogenetic test is a test to detect the presence or absence of,or change in, a particular gene or chromosome in order to predict aperson’s response to a medicine. The test could be done directly, byanalysing a person’s DNA, or indirectly, by examining the products ofthe DNA such as proteins.

Introduction

People often respond differently tothe same medicine. Few medicinesare effective for everyone; all maycause adverse side-effects andoccasionally death. These differentresponses are partly due to ourdifferent genetic make-up.Research in pharmacogeneticsinvestigates how differences in ourgenes can affect the way in whichwe respond to medicines.

Pharmacogenetics has the potentialto improve both the safety andefficacy of medicines. However,pharmacogenetic research and itsapplications raise important ethical,legal, social and regulatory issues.This important technology will

have implications for the researchand development of medicines,clinical practice and treatment, andthe use and storage of geneticinformation.

The Nuffield Council on Bioethicshas published a Report,Pharmacogenetics: ethical issues,which aims to encourage discussionof the issues and makesrecommendations for future policyand practice. This summary sets outsome of the arguments andrecommendations which arediscussed in more detail in theReport.

[Notes in square brackets throughout referto chapters and paragraphs in the Report].

a guide to the Report

21838 Pharmacogenetics 7/3/06 14:26 Page 1

2

� Variation in the way the body processes a medicine[Differentiating people]

Variation in DNA can lead to altered activity ofenzymes that are responsible for the absorption,metabolism and excretion of medicines. If amedicine is broken down too quickly it may not beeffective. Alternatively, slow metabolism can lead tothe build-up of toxic levels of a medicine. Even asingle change in the DNA can affect the way thebody processes a medicine [paras 2.11-2.14].

Example:An enzyme in the liver, CYP2D6, is involved in themetabolism of nearly one quarter of medicines,including anti-depressants and beta-blockers used totreat heart disease. Variations in the CYP2D6 genemay alter the activity of this enzyme. People withreduced levels are unable to process some medicinesproperly. In some cases the medicine will not beeffective; in other cases patients will suffer seriousside effects. For example, the painkiller codeine hasno effect on approximately 7% of Caucasiansbecause of a CYP2D6 variant. There is currently noroutine pharmacogenetic testing for CYP2D6; but adiagnostic chip that will test for a large number of2D6 variants will be introduced in 2003.

� Variation in the genetic characteristics of a disease

[Differentiating diseases]

Many diseases that are currently diagnosed as asingle clinical condition may actually have a numberof underlying causes, based on different geneticcharacteristics. Understanding more about thegenetic basis of disease may provide informationabout what medicine would be effective. Some cells,for example those in cancerous tumours, have analtered genetic make-up. This change only occurs inthe cancer cells, and not in normal tissue, and mayinfluence treatments [paras 2.15-2.16].

Example:A particularly aggressive form of breast cancer isassociated with a genetic variation which leads tooverproduction of a protein called HER2. Patientswith breast cancer can be tested to find out if theyhave high levels of HER2. If they do, the medicineHerceptin (trastuzumab) is given to treat this specifictype of cancer.

There are a number of factors that affect our responseto medicines, including age, sex, interaction with othermedicines, and diet. A genetic variant alone will rarelyprovide a precise prediction of the response of aparticular patient.

� Improving safetySome medicines have side-effects (adverse reactions)and may even occasionally cause death. If a geneticvariant is found to be associated with an adversereaction to a certain medicine, doctors could avoidprescribing the medicine to patients with this genevariant. However, there are other reasons whymedicines may be dangerous. These include errors inprescription or administration, patients’ notfollowing instructions accurately, or interactionsbetween medicines or other substances.

� Adjusting dosageGenetic information could be used to adjust thedosage of a medicine, reducing the trial-and-errorapproach which is often used today to determinethe best dose.

� Enhancing efficacyMany medicines are not effective for everyone witha particular disease. Some common treatments fordiabetes, depression and asthma are only effectivein around 60% of patients. Pharmacogenetics couldallow doctors to prescribe medicine only for thosepatients most likely to respond. Alternatively, newmedicines could be designed on the basis of geneticinformation about the cause of disease.

SCIENTIFIC BACKGROUND

How do genes affect our response to medicines?

Differences in our genes can affect the way in which we respond to medicines in two ways:

Potential benefits of pharmacogenetics

21838 Pharmacogenetics 7/3/06 14:26 Page 2

3

RESEARCH AND DEVELOPMENT

New medicines

The introduction of pharmacogenetics will have animpact on the way in which clinical trials are designedand managed. Some clinical trials already include thecollection of genetic information and this practice islikely to become more widespread [Chapter 3].

How will pharmacogenetics be used in clinicaltrials?

Pharmacogenetic analysis could influence the designof clinical trials. Smaller groups of genetically similarparticipants could be selected, potentially leading tomore reliable research results. People who are likely tosuffer adverse reactions from a new medicine could beidentified and excluded from trials, protectingresearch participants. However, there is currently notenough information available to reliably assess theimpact of pharmacogenetics on the design of clinicaltrials, and therefore also on the cost of developingmedicines.

Will pharmacogenetic testing becomemandatory in trials?

It is difficult to predict at this stage just howwidespread the use of pharmacogenetics in researchwill become. Regulatory requirements or concernabout litigation could put pressure on companies toinclude pharmacogenetic analysis in trials. However,this approach may not always be feasible orappropriate. One possibility is that pharmaceuticalcompanies will collect genetic information at the timeof a trial, but not analyse the data unless it becomesnecessary.

We recommend that the appropriate use ofpharmacogenetics in clinical trials should bepromoted. Collection of samples for possible futurepharmacogenetic analysis should be encouraged[paras 3.5-3.12].

Will pharmacogenetics be used to improveexisting medicines?

Pharmacogenetics could also be applied to existingmedicines to improve the safety and efficacy ofprescribing. The importance of such research dependson a number of factors:

� How widely used is the medicine?� What is the nature and severity of any adverse

reactions?� How accurate would a pharmacogenetic test be? � Do alternative treatments exist?

It is unlikely however that the pharmaceutical industrywould have a sufficient economic incentive to carryout this research, especially if the medicine is notcovered by patent protection. The Department ofHealth has recently announced that new funding of£4 million will be directed towards pharmacogeneticresearch over the next three years.

We recommend that pharmacogenetic research onexisting medicines should be encouraged. Fundingand support should be made available within thepublic sector and public-private partnerships shouldbe encouraged [paras 3.20-3.26].

21838 Pharmacogenetics 7/3/06 14:26 Page 3

4

It is important to ensure that patients are protected ifinformation about them is used in clinical trials. TheReport discusses a number of recommendationsrelating to consent, privacy and confidentiality[Chapter 3].

What level of anonymisation is appropriate?

The implications for patients will depend on howeasily samples can be traced back to them. The Reportconsiders the various degrees of anonymisation thatcould be used, ranging from samples that are directlyidentifiable to those that are fully anonymous. Onelikely approach would be to use anonymised data.Genetic and clinical data would be collected during atrial, but a code linking patients with their sampleswould be destroyed after the trial so they could notbe matched subsequently. However, this approachwould not allow longer term follow-up of anindividual. New samples would need to be taken frompatients if adverse reactions occurred.

We suggest that the greatest degree of anonymitycompatible with the research should be used toprotect the privacy of participants. Researchersshould explain to prospective participants how thesamples will be stored and the implications for them.[paras 3.31-3.43].

Should participants be given individual feedback?

Participants need to know whether the research islikely to give rise to information that is directlyrelevant to their health. In some cases, researchersprovide individual feedback to patients. In others,researchers offer individual test results only if patientsask for the information. Provided their data has notbeen anonymised, research participants in the UK areable to request access to information aboutthemselves. However, it is not clear to what degreeresults from pharmacogenetic trials will be reliable orclinically useful.

We recommend that the feedback of the overallresults of research should be promoted. We aresympathetic to the view that patients should begiven individual feedback if the information isclinically useful and validated. But this will not oftenbe the case with results from early-stagepharmacogenetic research. Decisions regarding theprovision of individual feedback should be explainedto the relevant research ethics committee[paras 3.44-3.49].

RESEARCH AND DEVELOPMENT

Some people believe that genetic information isfundamentally different from other forms ofmedical data. This view is called ‘geneticexceptionalism’. There are a number of reasons forthis belief, including the idea that geneticinformation is uniquely identifying and highlypredictive, for example revealing susceptibility to adisease. The information may also be of interest tothird parties including insurers and employers.However, in the context of pharmacogenetics, anundue emphasis on the fact that information isgenetic can be unhelpful. Not all genetic testsconvey highly predictive or diagnostic information

about a patient or their relatives. Equally, bloodtests, cholesterol tests and HIV tests can also givereliable predictive information, without analysis ofDNA. And non-genetic tests may also provideindirect genetic information.

Genetic information does not necessarily raisedifferent ethical issues from other types ofmedical information. We believe that the mostimportant aspect to consider is the informationthat a test reveals and its implications for thepatient, not whether the information is directlygenetic [paras 1.8-1.14].

Is genetic information special?

The use of pharmacogenetic information in clinical trials

21838 Pharmacogenetics 7/3/06 14:26 Page 4

5

How should pharmacogenetic tests beregulated?

The Medicines and Healthcare products RegulatoryAgency (MHRA) is responsible for the licensing of newmedicines and genetic tests, based on an assessmentof quality, efficacy and safety. The approval ofpharmacogenetic tests and medicines will also beunder its remit. It is important that these tests shouldbe reliable and of high quality. In some cases, amedicine will only be licensed if it is used inconjunction with a test. Herceptin, for example, islicensed for use only for patients who have beenshown to have high levels of HER2 [paras 4.3-4.6].

How will pharmacogenetic information affectdecisions about the provision of healthcare?

Healthcare providers operate on limited budgets andneed to assess whether a medicine is cost-effective,offering good value for money in light of theexpected health benefits.

Pharmacogenetics will provide information that isrelevant to these assessments by allowing moreaccurate prediction of the cost of a treatment. Ifgroups of patients are identified who are less likely torespond to a medicine, or more likely to have anadverse reaction, will it still be cost-effective to treatthem? Decisions about cost-effectiveness of treatingdifferent groups of people with the same medicinemay be affected by pharmacogenetic information.

However, an exclusive focus on cost-effectiveness,could lead to the possibility that small groups ofpeople with rare diseases or genetic variations mightnot be given treatment. Justice and equity also needto be considered. Sometimes it will be right toallocate resources to treatments or conditions thatmight otherwise not be considered cost-effective, inorder to ensure a fairer distribution of health care.

Pharmacogenetic information may affect decisionsabout which treatments to fund, by revealinginformation about the effectiveness of treatments.It is important that decisions should take intoaccount considerations of fairness. The NationalInstitute for Clinical Excellence (NICE) is responsiblefor assessing cost-effectiveness in England andWales. We endorse NICE’s approach of reviewingcases on an individual basis, considering equity aswell as cost-effectiveness [paras 4.11-4.32].

Is there a risk that some patients will beneglected?

The categorisation of patients into sub-groupsaccording to genetic features (either of their disease,or their inherited DNA), could also have an impact onthe development of new medicines. This could bebeneficial, if medicines that would otherwise havefailed could be targeted to a smaller group of patientsfor whom the medicine is likely to be safe and / oreffective. However, some of these groups might be sosmall that developing specific medicines to treat themcould be prohibitively expensive.

Incentives might be necessary to encouragepharmaceutical companies to develop medicinesthat would provide benefit to only a small numberof patients. One possibility would be to applyexisting legislation which encourages thedevelopment of medicines for ‘orphan diseases’, byoffering tax credits, incentives for research andextended patent protection.

Orphan medicine legislation may need to be reviewedif pharmacogenetic information leads to areclassification of diseases. A subgroup of a singledisease would have fewer patients, but it is not clearwhether the condition would qualify as an ‘orphandisease’ [paras 4.33-4.40].

Racial groups

There may also be stratification of patient populationsbased on racial or ethnic groupings with respect totreatment response. Some genetic variants are morecommon in certain racial or ethnic groups than in others.For example, one variant of CYP2D6 which leads to slowprocessing of many medicines, is present inapproximately 7% of Caucasians, but only 1% ofChinese. The fact that some genetic variants are more orless likely to be found within particular groups hasimplications for the design of clinical trials, and forpublic health decisions. At the same time, since there isconsiderable genetic variation within ethnic groups aswell as between them, pharmacogenetics should providea much more reliable way of predicting response to amedicine than relying on racial or ethnic classification.

We recommend that pharmacogenetic tests shouldbe validated in the populations in which they are tobe used. Those involved in pharmacogeneticsresearch should be sensitive to the potential formisunderstanding and prejudice arising from racialstereotyping [paras 4.41-4.47].

PUBLIC POLICY ISSUES

21838 Pharmacogenetics 7/3/06 14:26 Page 5

6

The introduction of pharmacogenetics will mean thatmany more patients are exposed to genetic testingthan before. Currently, patients tend only toencounter genetics in the context of serious diseasessuch as cystic fibrosis and Huntington’s disease.Pharmacogenetics could make genetic testing muchmore routine. We make a number ofrecommendations to help ensure that the delivery ofpharmacogenetics in clinical practice will be asstraightforward as possible [Chapter 5].

� Reliable and easily accessible information fromindependent sources will be important, both fordoctors and patients.

� Health professionals need to be given training tocommunicate information aboutpharmacogenetics.

� Additional resources will be needed to implementpharmacogenetic testing. Doctors will need moretime with patients, to discuss tests and to takesamples.

� New facilities for testing will be needed, allowingresults to be obtained quickly and efficiently. Testscould be carried out either in GPs’ surgeries, at ahospital, or at specialised testing facilities.

Should tests be made available directly over thecounter or on the internet?

Most people accept that not all medicines should befreely available over the counter or on the internet topatients. Should the same apply to pharmacogenetictests? The tests themselves do not pose the same directrisks as medicines, so it could be argued that patientchoice should be important. However, the nature ofthe information revealed by the test also needs to beconsidered.

We recommend that the direct provision of tests topatients needs to be assessed on a case-by-case basis.If an approved test provides clear, readilyinterpretable information about medicines that canbe purchased over the counter, the test could beprovided direct to consumers. However if, as is morelikely, a test generates complex information, withless certain predictions, it would not be appropriateto provide the test direct to the consumers. It islikely that professional advice will be needed beforeand after taking the test [paras 5.19-5.22].

Will consent forms be necessary forpharmacogenetic tests in clinical practice?

Some medical procedures, for example giving ananaesthetic, require written consent from the patient.Currently consent must be obtained for HIV tests andsome genetic tests, although not for other tests,including most routine blood tests. There has beendebate whether written consent forms and geneticcounselling will be necessary when patients havepharmacogenetic tests. Possible factors to considerinclude:

� What information about a patient’s condition mightbe revealed?

� Could the results have a profound psychologicaleffect, for example if a test reveals that a patient iseffectively untreatable?

� Will the test also reveal additional information,either now or in the future, such as likely responseto other medicines or susceptibility to an unrelateddisease? (Genetic variants can influence more thanone trait).

We recommend that each test be assessed on a case-by-case basis, according to the nature of theinformation revealed. It is important to point outthat non-genetic tests may reveal similarinformation. In most cases, written consent formswill not be required, but in some circumstances theymay be appropriate. Written information should beprovided, particularly if tests will reveal complexinformation [paras 5.9-5.17].

TREATMENT AND CLINICAL PRACTICE

21838 Pharmacogenetics 7/3/06 14:26 Page 6

7

Should patients be prescribed a medicine even ifthey do not wish to take an associated test?

Who should decide whether a patient takes apharmacogenetic test? Should patients be entitled tobe prescribed a medicine, even if they do not wish totake an associated test? This question will becomeparticularly pressing if advances in pharmacogeneticsmean that some medicines may be licensed onlybecause they can be used with a pharmacogenetictest, to ensure a patient is not at risk of a seriousadverse reaction.

Where a pharmacogenetic test is part of the licenceconditions of a medicine, it is unlikely that a doctorwould wish to prescribe the medicine without thetest, particularly if this would mean putting thepatient at risk. If the test is not part of the licencingconditions, doctors should be guided by regulatoryauthorities and professional bodies when making adecision [paras 5.23-5.29].

Who will have access to pharmacogeneticinformation?

The increasing storage of genetic information by GPsor pharmacists will raise questions about the privacyof the information, as it does in relation to thestorage of all medical data. The privacy andconfidentiality of patients must be protected[paras 5.31-5.35].

� Test results are likely to be stored in medicalrecords. We do not think that specialarrangements for storing genetic data will befeasible.

� In a few cases, information revealed bypharmacogenetic tests will be relevant to familymembers, for example if a pharmacogenetic testalso indicates susceptibility to another disease.This can be dealt with by existing practiceregarding sharing medical information.

Should pharmacogenetic information be used byinsurance companies?

Pharmacogenetic information could be relevant bothto health and life insurers, either in assessing claims orsetting premiums [paras 5.36-5.41].

� Assessing claims: Pharmacogenetic information could be useful toinform decisions about which treatments shouldbe funded for particular groups of patients. Thiswould be similar to the use of information bypublic health providers to make decisions aboutthe allocation of resources.

� Setting premiums: The Association of British Insurers has commentedthat pharmacogenetic information would not beuseful in setting premiums. This is partly becausepharmacogenetic tests will have low predictivevalue compared to tests for single-gene disorders.Other information would also continue to bemore useful in assessing risks, for exampleinformation about medical history. But there is arisk that patients will not take pharmacogenetictests because of fears about obtaining insurance.Patients could then lose the benefit of taking thetest. There is currently a moratorium in the UKuntil 2006 on the use of genetic information bylife insurers in setting premiums (except forHuntington’s disease in life insurance policies ofover £500,000).

Pharmacogenetic information falls under thismoratorium and we recommend that thismoratorium should continue.

21838 Pharmacogenetics 7/3/06 14:26 Page 7

Published by Nuffield Council on Bioethics28 Bedford SquareLondon WC1B 3JS

Telephone: 020 7681 9619Fax: 020 7637 1712Internet: www.nuffieldbioethics.org

Copies of the Report are available to download from the

Council’s website: www.nuffieldbioethics.org

For a printed copy, please e-mail bioethics@nuffieldbioethics.org

SummaryIt is difficult to predict the extent to which

‘personalised medicines’ will become a reality.

Claims of designer drugs, or ‘the right medicine, for

the right patient, at the right dose’ are misleading,

but it is important to discuss ethical, legal,

regulatory and social issues that may be raised by

improvements in predicting response to medicines.

To obtain maximum benefits from

pharmacogenetics we need to address legitimate

concerns and safeguard against inappropriate use.

There must be the right combination of constraints

and incentives to protect and promote the interests

of patients and society as pharmacogenetic testing

is more widely introduced.

21838 Pharmacogenetics 7/3/06 14:27 Page 8