How genomic scans can help predict altered drug response or disease
Genomic scans are now available to identify genetic variants associated with disease and drug response. Alain Li Wan Po, Peter Fardon, Candy Cooley, Sarah Warburton and Colin Barker illustrate the potential usefulness of genomic scans in informing prescribing, using a hypothetical scenario in a community pharmacy as an example
Whole genome sequencing is now possible and several personal genomic profiles, including those of scientists James Watson (one of the co-discovers of DNA) and Craig Venter, have been posted on the internet for research purposes.
More limited scans to identify genetic variants associated with disease and drug response can be purchased for less than $1,000, and this has generated considerable public interest. Some pharmacogenetic testing are already in routine use in hospital practice (see Panel below), notably in oncology and serious infections clinics.
Pharmacogenetic tests in routine use
Nucleic acid amplification test. A positive test result allows pharmacists to supply azithromycin over-the-counter for both the patient tested and his or her sexual partner(s).
Epidermal growth factor receptor
The test identifies over-expression of epidermal growth factor receptors (EGFRs) and gene amplication. This is a data-sheet requirement for the prescribing of cetuximab, which is used in the treatment of colorectal cancer.
The prescribing of trastuzumab (used to treat breast cancer) requires identification of human epidermal growth factor receptor-2 tyrosine kinase receptor over-expression.
The test for HLA-B*5701 identifies individuals who are hypersensitive to abacavir, used in the treatment of HIV and AIDS.
KRAS mutation test
The prescribing of panitumumab for advanced colorectal cancer requires (in the EU) identification of functional KRAS and EGFR over-expression.
Presence of Bcr-Abl translocation (Philadelphia chromosome)
The presence of the Bcr-Abl translocation (or Philadelphia chromosome) is a diagnostic feature of chronic myelogenous leukaemia.
The tuberculosis test is used to identify multidrug-resistant TB rapidly in developing countries.
Analytically reliable genomic scans can now be obtained and various health correspondents of the press have already commented on their own experience of having their genome scanned fully or scanned for a panel of genetic variants associated with disease and altered drug response.
Although the cost, at the equivalent of less than a year’s season ticket for a premiership football team, is still regarded as expensive, there is little doubt that prices will keep dropping as the technology develops. Some claim that soon a 10-minute genome scan will be commercially available.1
Closer to practice is the availability of scans to identify specific polymorphisms, reported to be associated with risk of disease or drug response. Specifically, any doctor can now order a cytochrome p450 (CYP) profile, including CYP2D6, for a patient, and a recent survey indicates that over 80 health-related genetic tests are available direct to the consumer in the US.2 Twelve of these were for pharmacogenomic testing, the most common category.
The US company 23andMe (named after the 23 pairs of human chromosomes) provides genetic profiles with single nucleotide polymorphisms (SNPs), which is reported to be associated with, albeit often poorly quantified, disease risk.
Sergey Brin, co-founder of Google, has had his scan done by 23andMe, which was founded by his wife Anne Wojcick. It revealed that he had the G2019S mutation, which is associated with an increased risk of Parkinson’s disease, as well as a family history of Parkinson’s disease on his mother’s side.3
23andMe is embarking on a study of genetic associations with parkinsonism and participants are asked to pay $25 instead of the usual price $399 if they are willing to participate in the study.4
With a target of 10,000 subjects, the power of the study will be far beyond that of studies published so far and estimates of risk are likely to be more reliable than smaller observational studies. Google and 23andMe may well produce results that surprise sceptics.
When deCODE genetics, the Icelandic company that also markets a personal genome service started in business, few predicted that it would make the fundamental genomic discoveries that are now widely acknowledged. Despite the enlightening disease-gene associations, the reliability of risk predictions for most diseases that are polygenic and show close interplay with environmental risk factors remains controversial.
Similarly, the translation of observed pharmacogenetic associations into clinically useful tests before prescribing has been successful in only a few cases.5,6
Yet in some cases pharmacogenetic testing is already in clinical use for informing treatment decisions and, in other cases, is a requirement specified in the drug datasheet or label before prescribing. In most such cases, the test is a DNA test rather than testing for the presence of a genetic sequence variation. The challenge for clinicians is therefore to appreciate when a genetic association is potentially important to the care of an individual patient.
The Watson genomic code
James Watson’s genetic code is in the public domain. Software tools are also available to allow searching for relevant genetic variants, most notably SNPs, which are known to be associated with altered drug response. Part analysis of Professor Watson’s genetic code is also available on the internet, which highlights some genetic variants associated with drug response. In this article, we will concentrate on the specifics of the scenario (Panel below), but highlight other relevant issues as well.
A scenario in a pharmacy
James Watson, a 74 year old, calls in during one of his trips to London. He is generally healthy but occasionally develops bad headaches, which he ascribes to pressure from the media. He says that the latest episode was brought on by some comments he made at a lecture which, in his view, were misrepresented.
He asks for your advice on what tablets he could take. His paracetamol tablets do not seem to have worked and he has tried other over-the-counter analgesics as well.
Being a pioneer in DNA research and genomics research, he presents you with his personal genomic code (jimwatsonsequence.cshl.edu/cgi-perl/gbrowse/jwsequence) and asks you to give your advice. He volunteers that he is lactose intolerant. What do you do?
1. Run for the door and call the police
2. Ignore his genomic code and proceed normally with a medicines review taking account of relevant clinical information and information held on your drug-drug interaction databases
3. Have a stab at identifying major issues identifiable from his genetic code, including genetic variants which flag drug-drug interactions
Let us assume you decide to be brave and attempt to read his code to give him some informed advice. What are you likely to find in his personal genomic code?
Metabolic enzyme polymorphism
Watson’s genetic code shows the presence of an SNP of the CYP2D6 gene, which is associated with poor metabolism of a range of CYP2D6 drug substrates, including codeine.
Codeine needs to be metabolised to morphine for analgesic efficacy. Therefore, given the impaired metabolic pathway, less morphine would be formed. Moreover, given the small amounts of codeine present in OTC analgesic tablets that contain codeine, such products are unlikely to be any more effective for him than single-ingredient tablets, such as paracetamol.
Dihydrocodeine is another opiate analgesic available OTC in combination with other analgesics (eg, paracetamol). The only structural difference from morphine is the presence of an additional double-bond in the structure.
Biotransformation of dihydrocodeine to dihydromorphine is also mediated by CYP2D6.7 Although in extensive metabolisers the pharmacokinetics of both dihydrocodeine and dihydromorphine following ingestion of dihydrocodeine are linear8 (ie, a doubling of dose leads to a doubling of drug exposure), this may not be the case in poor metabolisers.
Therefore, in extensive metabolisers, a higher dose may be required to lead to proportionately larger effects (both positive and adverse). In poor metabolisers, this may not be the case and overdosing may lead to non-linear effects.7
The implication of the recent work on the pharmacogenetics of dihydrocodeine is that one would probably expect less pharmacogenetic variability in response to dihydrocodeine than codeine. However, this has not yet been adequately validated by pharmacodynamic and clinical studies. Any greater benefit from dihydrocodeine than codeine because of the former derivative’s analgesic activity in experimental pain models has yet to be proven.
Therefore, if a dihydrocodeine analgesic combination is recommended, it may well be more for the placebo additive effect than for anything else. This raises many ethical issues, although many clinicians are happy about prescribing poorly validated medicines believing them to be potentially beneficial.9
Since Professor Watson has informed you that he is lactose intolerant, there is no need to pursue this further in the genetic code because a patient may still develop an undesirable phenotype in the absence of known causative polymorphisms. Therefore, for Professor Watson, formulations containing lactose as tablet excipient should be avoided.
In addition to polymorphisms affecting metabolic pathways, drug receptors or targets involved in the pharmacological effect (also called pharmacodynamic effect) may also show functional mutations. The vitamin K epoxide reductase complex subunit 1 (VKORC1), a target for the anticoagulant warfarin, is an example.
Professor Watson’s genetic code also shows an SNP in the VKORC1 gene with reference sequence rs8050894. This polymorphism is in strong linkage disequilibrium (tends to be inherited together) with an SNP in the promoter region, which is strongly associated with warfarin sensitivity, leading to a lower dose requirement.10
So, if Professor Watson is on warfarin, then there is a need to consider carefully any use of analgesics that contain aspirin, which may independently increase the risk of bleeding and possibly complicate therapy. This emphasises the importance of a thorough drug history.
Although we have adopted a light-hearted approach to this discussion, the implications for practice are real and potentially serious. Although personal whole genomic profiling is still costly and hence not routinely done, there is little doubt that technically, reliable scans can, and will, be done at increasingly affordable prices.
The quantification of risk of disease from genotyping is now widely accepted not only for monogenic diseases, but also for some polygenic diseases, such as breast cancer. However, genotyping for susceptibility genes with low penetrance is still debated.
The clinical use of genotyping for drug-response is established for only a few genetic variants. Testing for HLA-B*5701 to predict potentially lethal abacavir (a nucleoside analog reverse transcriptase inhibitor used to treat HIV and AIDS) hypersensitivity is probably the best example.
For most genetic variants associated with drug response, the clinical use is less established, and genotyping for CYP2D6 is an example. It could be argued that knowing that a patient has a genotype which predicts a poor metaboliser phenotype should alert the clinician in the same way that a potential drug interaction leading to altered drug pharmacokinetics would.
A clinician would not prescribe an interacting drug without weighing the harm-benefit balance. An SNP may highlight the same risk as a drug interactant and should be treated accordingly.
One of the major challenges to the application of pharmacogenetic testing (including CYP2D6 testing) in clinical practice is that, in most cases, the relationship between response and genetic variant is imperfect. Rarely does a genetic variant always predict a particular response. The extent to which a test is useful depends on the accuracy of predictions in routine patient care.
Moreover, even for a genetic variant strongly associated with a drug response, the latter is often altered by other factors, such as age, organ function, concomitant drug ingestion and other variants not only of the same gene but also of other genes involved in the metabolic or response pathways of the drug concerned.
For example, after codeine ingestion, any morphine formed is conjugated to the 3- and 6-glucuronides through the intermediary of the enzyme uridyl glucuronosyltransferase 2B7 (UGT2B7). Similar metabolic conjugations occur for any dihydromorphine formed after taking dihydrocodeine. Although morphine-3-glucuronide is inactive, the 6-glucuronide is active.
A recent case-control study suggests that breastfed infants of mothers who were treated with codeine for obstetric pain and who were ultra-rapid CYP2D6 metabolisers and were UGT2B7*2 homozygotes (individuals with both alleles of this type) had a greater risk of developing potentially life-threatening central nervous system depression.11
The scenario of a patient coming in with his or her genetic metabolic profile is no longer fictional. Neither is the usefulness of genotyping for predicting likely drug response in secondary care. The clinical professions in primary care need to be ready for this pharmacogenetics challenge. If you knew that a patient was taking a drug which interferes with one that you intend to prescribe, would you not take that into account?
Knowing that a patient has a genetic variant which nullifies or amplifies the action of a drug should surely be given the same careful consideration during the drug prescribing process.
The science of medicine requires that the risks and benefit be traded off. The art of medicine is the integration of results of this deliberation with the patient’s preferences to arrive at the best course of action for the patient.
Alain Li Wan Po is professional lead in pharmacogenetics, Peter Farndon is director, Candy Cooley is centre manager, Sarah Warburton is NHS national trainer for laboratory geneticists and Colin Barker is an education officer, all at the National Genetics Education and Development Centre, Birmingham
Cytochrome p450 (CYP)
Cytochrome p450 is the superfamily of enzymes most extensively involved in the metabolism of drugs. Among the families of particular interest are CYP2C9, CYP2C19, CYP2D6 and CYP3A4. Drugs are often metabolised by more than one of the CYP enzymes.
Genetic variants of the CYP genes may lead to loss of enzyme activity and, in some cases, increased activity. Genetic variants that occur with a frequency of 1 per cent or more are called polymorphisms.
The genotype of a person is defined by the variant carried in relation to a specific gene, usually at a specific position (locus).
For example, the CYP2D6 genotype of Professor Watson, in relation to the reference sequence, is defined by the nucleotide change 100C>T (ie, cytosine to thymine at position 100 of the genetic code).
For this reason, Professor Watson’s genotype would be referred to as being of the T form. This corresponds to an amino acid change P34S (ie, proline to serine at amino acid 34) in the normal CYP2D6 enzyme.
Penetrance is defined as the proportion of individuals with a mutation that causes a particular disorder who exhibit clinical symptoms of that disorder. A condition is said to show complete penetrance if clinical symptoms are present in all individuals who have the disease-causing mutation. Conversely, reduced or incomplete penetrance is shown when clinical symptoms are not always present in individuals who have the disease-causing mutation.
An example of an autosomal dominant condition showing incomplete penetrance is breast cancer due to mutations in the BRCA1 gene. Females with a mutation in this gene have an 80 per cent lifetime risk of developing breast cancer.
The phenotype is the observable characteristic of a person in relation to a specific genetic variant. In the present context, the CYP2D6 phenotype of Professor Watson is “slow metaboliser” of drugs that are substrates of this enzyme.
The sequences of alleles with single nucleotide polymorphisms are defined and these sequences are given specific codes or reference sequence (rs) codes (eg, rs1065852 is one of the CYP2D6 variant alleles present in James Watson’s genome).
Single nucleotide polymorphism (SNP)
A polymorphism describes a variant of a gene which occurs with a prevalence of at least 1 per cent in a given population. An SNP is a polymorphism involving a single nucleotide change.
More recently, the 1 per cent frequency qualifier has been loosened and rarer variants are also referred to as polymorphisms. The nucleotide change may be silent in that the same amino acid is coded. Coding SNPs where the amino acid is changed (also called non-synonymous SNPs) are more likely to have functional consequences such as loss or impaired of enzyme activity.
However, this is not necessarily so, if the change results in an amino acid similar in characteristic to the one it replaces, the resulting protein is not structurally or functionally changed significantly.
1. Hayden EC. Accessible genomes move closer. Nature 2008;455:1014.
2. Goddard KA, Robitaille J, Dowling NF, Parrado AR, Fishman J, Bradley LA et al. Health-related direct-to-consumer genetic tests: a public health assessment and analysis of practices related to internet-based tests for risk of thrombosis. Public Health Genomics 2009;12:92–104.
3. Simpson A. Google co-founder Sergey Brin reveals Parkinson’s Disease fear. Available at www.telegraph.co.uk (accessed 23 October 2009).
4. Pollack A. Google co-founder backs vast Parkinson’s study. Available at www.nytimes.com (accessed 23 October 2009).
5. Katsanis SH, Javitt G, Hudson K. Public health. A case study of personalized medicine. Science 2008;320:53–4.
6. The Royal Society. Personalised medicines: hopes and realities. London: The Royal Society; 2005. p56.
7. Fromm MF, Hofmann U, Griese EU, Mikus G. Dihydrocodeine: a new opioid substrate for the polymorphic CYP2D6 in humans. Clinical Pharmacology and Therapeutics 1995;58:374–82.
8. Ammon S, Hofmann U, Griese EU, Gugeler N, Mikus G. Pharmacokinetics of dihydrocodeine and its active metabolite after single and multiple oral dosing. British Journal of Clinical Pharmacology 1999;48:317–22.
9. Tilburt JC, Emanuel EJ, Kaptchuk TJ, Curlin FA, Miller FG. Prescribing “placebo treatments”: results of national survey of US internists and rheumatologists. BMJ 2008;337:a1938.
10. Scott SA, Edelmann L, Kornreich R, Desnick RJ. Warfarin pharmacogenetics: CYP2C9 and VKORC1 genotypes predict different sensitivity and resistance frequencies in the Ashkenazi and Sephardi Jewish populations. American Journal of Human Genetics 2008;82:495–500.
11. Madadi P, Ross CJ, Hayden MR, Carleton BC, Gaedigk A, Leeder JS et al. Pharmacogenetics of neonatal opioid toxicity following maternal use of codeine during breastfeeding: a case-control study. Clinical Pharmacology and Therapeutics 2009;85:31–5.
Citation: The Pharmaceutical Journal URI: 10985487
Recommended from Pharmaceutical Press