This is a mini-review of applied pharmacogenomics that is composed for clinicians, especially those providing primary care. In the first half, office scenarios of 7 best known and most clinically relevant drug-gene pairs are presented in brief format, with main references attached to each. In the second half, the background knowledge and current implementations are introduced as a primer leading towards further exploration by interested individuals.
Keywords: Pharmacogenomics; Implementation; Primary care
“CPIC1 guidelines are designed to help clinicians understand HOW available genetic test results should be used to optimize drug therapy, rather than WHETHER tests should be ordered. A key assumption underlying the CPIC guidelines is that clinical high-throughput and pre-emptive genotyping will become more widespread, and that clinicians will be faced with having patients’ genotypes available even if they have not explicitly ordered a test with a specific drug in mind. “ —— Pharmgkb.org
Have you been wondering why codeine appears so often in adverse drug reaction (ADR) profile? Have you ever felt frustrated by a patient with a long list of active medications and a longer list of ADRs who needs another something? Have you always wished to know how likely the alerted event is actually going to happen? I try to shed some light on Pharmacogenomics (PGx), a relatively obscure subject to clinicians, yet a soon-to-be vital tool of day-to-day practice of medicine. A series of clinical scenarios common in primary care are composed to demonstrate the drug-gene pairs best known to today’s applied PGx. This is followed by an outline of applied PGx: its foundation-laying technologies, cumulating knowledge, blooming industries and forthcoming clinical implementation.
Part 1. Best Known Drug-Gene Pairs In Primary Care
I have no reason to doubt that this will be my daily routines within 5 years. Actually it is already the look of some Dutch physician’s life.
Ms. A is a 23 year-old college student who complains of bad bronchitis cough depriving her sleep right prior to her final exams. She requires “something more potent than the overthe- counters”. As I attempt to prescribe guaifenesin-codeine, an alert pops out, “Caution! CYP2D62 ultrarapid metabolizer (UM) *1A/*2×N3. See recommendations (Table 1).” I counsel her to avoid codeine and inform future prescribers of her CYP2D6 genotype.
Core literatures: FDA released its strongest warning in 2013 that alerts the danger of CYP2D6 ultrarapid metabolizers taking codeine, shortly after the deaths of 3 children in 2012 (pharmgkb.org > codeine > drug labels). Historically, case reports were featured by post-tonsillectomy children and neonates breast-fed by post-cesarean mothers (PMID4: 22492761, 19692698, 15625333, 16920476, 23614474, 23709324). All 3 PGx associations published dosing guidelines on codeine-CYP2D6 ( pharmgkb.org > codeine > dosing guidelines; PMID: 24214521, 24458010, 21412232). Mechanism of this classic interaction can be simply put as a genetic amplification of CYP2D6 metabolized codeine-to-morphine transformation; CPIC guideline has an excellent illustrated interpretation of it (PMID: 24458010). Similar mechanism is shared by tramadol, oxycodone and hydrocodone, though to a lesser extent (PMID: 18713907, 12920424).
Mrs. B is a 75 year-old retired nurse who is admitted for an ischemic stroke that is attributed to extensive atherosclerotic vascular disease. She has been on simvastatin 20mg daily without side effects for many years. Alert appears while simvastatin is being re-dosed: “SLCO1B1 diplotype *1A/*5 translates to an intermediate risk for myopathy due to one of two copies being decreased function allele (*5).” She is counseled on the better chance for tolerance if lower dose of a more potent statin less affected by low SLCO1B1 activity is given instead of increasing simvastatin to 80mg; rosuvastatin 20mg is chosen. Enclosed in the alert is an interpretation by PGx consultants: “SLCO1B1 transports statins into hepatocytes and its dysfunction slows down statin clearance leading to increased myopathy. While *1A allele has normal function, *5 contains “C” at rs41490565, encoding a decreased activity transporter.“
Core literatures In 2008, the single nucleotide variant SLCO1B1* 5 was detected with solid significance by genome-wide sequencing, and then verified in large number of myopathy cases to be associated to simvastatin-induced myopathy (PMID: 18650507). CPIC dosing guideline focuses on simvastatin for it is the most affected, though other statins also are influenced to a lesser degree (pharmgkb.org > simvastatin > dosing guidelines; PMID: 24918167). Pravastatin was shown in a randomized study to be a reasonable first choice statin for carriers of the SLCO1B1*5 allele; and women may benefit from increased surveillance for symptoms (PMID: 19833260).
I receive a letter from the cardiologist of my patient Mr. C, a 64 year-old chef. It is a summary of the patient’s hospitalization for acute MI, coronary artery stenting 2 days ago and plan of post-MI care. A literature review is attached to the letter and starts as this: “Clopidogrel is a pro-drug that needs activation by CYP2C19 for platelet suppression effect. Poor metabolizers (PM) of this enzyme should receive alternative anti-platelet agents e.g. prasugrel or ticagrelor, due to the risk of increased re-thrombosis.” A consent is also attached and signed by patient to disclose CYP2C19 genotype.
Core literatures Cases of antithrombosis failure due to lack of clopidogrel activation by CYP2C19 have explained all (PMID: 21270785, 21215696). FDA released boxed warning in 2009 that recommends CYP2C19 genotyping prior to prescribing clopidogrel (pharmgkb.org > clopidogrel > drug labels). Both CPIC and DPWG provide dosing guidelines (pharmgkb.org > clopidogrel > dosing guidelines; PMID: 23698643, 21412232).
Mrs. D, a 42 year-old high school teacher with breast cancer is here to receive PGx testing results. She has a history of estrogen- positive cancer, has been taking tamoxifen for 3 years since lumpectomy and chemo-radiation therapy and is doing well. She did develop chemo-induced menopause despite young age. As I open her report, several recommendations are displayed with one highlighted at the top: “Studies with good quality suggest increased risk of breast cancer relapse when tamoxifen is used on CYP2D6 poor or intermediate metabolizers (IM), avoid concomitant use of CYP2D6 inhibitors and consider aromatase inhibitors (AI: anastrozole, letrozole, or exemestane) in postmenopausal women.” I explained above message to her. A return visit with her oncologist to discuss alternative hormone suppression agents is requested.
Core literatures Tamoxifen is another pro-drug activated by CYP2D6 (to endoxifen). Evidences are not as sufficient and strong, maybe due to long follow-up needed to reach significant difference in outcomes, and also need for larger sample to reveal findings in subset analysis. CPIC has not recommended any actions based on CYP2D6 genotype, but DPWG has its dosing guideline (pharmgkb.org > tamoxifen > dosing guidelines; PMID: 21412232). Studies thus far important are copied here (PMID: 19809024,18024866, 16361630, 23213055, 23570465, 23764426) including a meta-analysis (PMID: 24329190).
Coumadin is initiated on Mr. E, an 80-year-old caucasian gentleman for newly discovered atrial fibrillation. CYP2C9 and VKORC1 genotypes, *1/*3 and “A/A” at rs99232316, along with other indices auto-populate into the electronic form used to initiate coumadin clinic enrollment. As I complete and sign the order, the recommended warfarin daily dose is provided: “Highly sensitive to warfarin. Maintenance dose of 1.8mg with optional first day loading dose of 2.7mg is recommended”. For easier operation, I prescribe 2.5mg on day 1 and 2.0mg on day 2 and day 3. On day 4 I receive his first INR from home monitor along with the recommended dosing for the next 5 days; it is signed and sent back to Mr. E.
Core literatures Genotype-guided coumadin dosing is probably the single one example of PGx implementation that has gone the farthest. Several large-scale randomized controlled trials comparing genotype-guided to traditional dosing have been completed (PMID: 24251360, 24251361, 24251363). Currently the most widely used algorithm was developed by Washington University and is free for public use (www.warfarindosing. org; PMID: 18305455). CPIC dosing guideline (pharmgkb.org > warfarin > dosing guidelines; PMID: 21900891) and main references (PMID: 19031075; 19300499, 19228618, 18574025) are collected here.
Mrs. F, a 47-year-old Chinese descent is here for physical exam. A new drug, carbamazepine, is reflected as interval change of her medication profile. It was started 1 month ago by a neurologist for trigeminal neuralgia, and she had not ever used carbamazepine before that. As I update the active medication list in EMRG, a message jumps out: “HLA-B*15:02 and HLA-A*31:01 both significantly increase the risk of carbamazepine-induced hypersensitivity reactions including Steven-Johnson Syndrome / Toxic Epidermal Necrolysis (SJS/TEN). Individuals carrying at least one copy of either allele are not recommended to start carbamazepine if naïve to the medication.” Mrs. F reports that a genetic test was done by her neurologist prior to starting arbamazepine and she brings me a copy of the report. A panel of clinically relevant HLA alleles are tested: “Mrs. F carries one copy of HLA-B*58:01 (heterozygote for the allele) which renders her susceptible to allopurinol-induced severe cutaneous adverse reactions (SCAR). All other alleles tested are negative including HLA-B*15:02 and HLA-A*31:01. The report is sent to our PGx consultants to be EMR-charted for future reference.
Core literatures Association of HLA-B*15:02 and carbamazepine- induced SJS/TEN (but not other reactions) was first reported in Han-Chinese (PMID: 15057820), then replicated quickly by multiple asian ethnic groups (PMID: 23132554). HLA-A*31.01 was discovered to be related to carbamazepine- induced hypersensitivity reactions (not limited to SJS/ TEN) via GWAS in 2011 in two different ethnic groups (European and Japanese) by two different teams (PMID: 21428769, 21149285). In 2007, FDA released boxed warning that requires HLA-B*15:02 screening prior to carbamazepine prescription; information on HLA-A*31:01 was added in 2013; drug package inserts in Japan also contain the warning on HLA-B*15:02 ( pharmgkb.org > carbamazepine > drug labels; PMID: 23895776). CPIC and CPNDS guidelines were published in 2013 and 2014 successively (pharmgkb.org > carbamazepine > dosing guidelines; PMID: 23695185, 24597466). There are only few drugs by far that have a known HLA allele that confers clinically significant outcomes. Among them, allopurinol- HLA-B*58:01 pair is the most pertinent to primary care, therefore main articles are listed for reference (PMID: 15743917, 18192896, 19933789, 21912425). Anti-HIV drug abacavir causes systemic hypersensitivity syndrome (SHS) with extremely high mortality in carriers of HLA-B*57:01; preemptive testing is obligated by FDA regardless of ethnicity (pharmgkb.org > abacavir > drug labels)
As I preview the chart of my next patient Miss. G, a 35 year-old graduate student of sociology with a history of bipolar type II with major depressive episodes, anxiety and fibromyalgia, as a message catches attention: “More than 3 ADRs detected share a common CYP450 enzyme. Consider genotyping for minor metabolizer status.” There are 9 medications listed under her ADR profile (Table 2A).
She tells me during the interview that she stops taking medications all together, and struggles to manage her mood and pain by counseling and exercise. Genotyping of CYP450 panel comes back in 3 days; diplotypes of crucial CYPs are translated into metabolism phenotypes with the help of PGx consultants (Table 2B). A new list of drugs addressing her problems with no or minimal dependence on deficient CYPs are constructed (Table 2C). Gabapentin, desvenlafaxine, lamotrigine and clonazepam are selected and planned to be initiated in an orderly fashion under close observation.
Table 2. CYP dependence analysis provided by PGx consult
prove to benefit an individual drug’s therapeutic outcome, a finely designed trial comparing genotype-guided versus traditional prescription of all medications used for major depression patients reflects overall benefit (PMID: 24018772). This reversed the failed initial attempt to guide psychotropic drug dosing, as it turned out to be an issue of medication selection rather than dose adjustment. DPWG7 guidelines feature most of the widely used SSRIs, SNRIs, TCAs and antipsychotics whose metabolism largely rely on CYP2 family enzymes (Pharmgkb. org > PGx drug dosing guidelines > DPWG).
Part 2. Brief Digest Of Applied PGx
Pharmacogenetics (PGt) studies how drug response is affected by interindividual variations, and PGx is simply PGt on genomic scale. Box 1 takes an overview at a half century’s history of PGt-PGx.
Box 1. Milestones of pharmacogenetics-pharmacogenomics
1957 Pharmacogenetics described and named
1960 Differential elimination of isoniazid related to slow or rapid acetylation which is under genetic control
1975 CYP2D6 polymorphism described after wide interindividual range of plasma concentrations of debrisoquine and sparteine cause orthostatic hypotension and fetal death respectively
1993 CYP2D6 ultrarapid metabolizer reported
1994 CYP2C9 and 2C19 reduced activity alleles discovered
2000 PharmGKB founded
2001 CYP2D6, 2C19 and 2C9 genotypes-based dosing guidelines of psychotropic drugs published
2005 First-generation haplotype map of human genome completed by International HapMap project
2006 AmpliChip® became first commercial PGx array approved by FDA
2006 DPWG incorporated G-Standard into nationwide EMR in Netherland
2007 FDA released warning on clopidogrel and CYP2C19.PM
2011 CPIC published first set of dosing guidelines
2011 GWAS8 related HLA-A*13:01 to carbamazepine-induced hypersensitivity reactions
2013 FDA released warning on codeine and CYP2D6.UM
PGx research starts with looking for an association between a drug response trait and a polymorphism (majority SNP9). Traditionally a candidate SNP was tested in a case-control study; if the SNP’s frequency was significantly different between cases and controls, an association was identified. Upon advent of the era of next-generation sequencing, thousands of randomly selected SNPs scattered throughout human genome can be tested within a short time and an affordable budget, revealing SNP biomarkers that have never been hypothesized to play a role in the process. This is called genome-wide association study, or GWAS, as illustrated in Box 2.
Box 2. Genome-wide association study
Clinical applications of PGx are not new to today’s medicine: much so popular have been targeted therapies for cancers and autoimmune disorders in the past decade. There are however a lot more to it, on the road of translation into clinics and hospitals. Box 3 demonstrates the spectrum of drug-gene interactions, by an artificial classification.
PD interactions are less well established due to complexity of the exact mechanism of action for most drugs, but are getting enormous boluses of freshly cooked data from GWAS to be analyzed.
Our knowledge is more advanced in PK interactions. Among all PK genes, CYPs have received highest attention due to the vital role P450s play in biotransformation of about 80% of all drugs in clinical use, and also because of their extensive polymorphism.
Box 4. Cytochrome P450 superfamily
Human CYP superfamily has 57 genes and 59 pseudogenes divided among 18 families and 43 subfamilies. Only CYP1, 2 and 3 families metabolize drugs, or exogenous toxic compounds, principally in the liver; other CYPs participate in endogenous compounds metabolism and are present in most organs of the body. CYPs catalyze majority of phase I reactions, which introduce a polar group into the substrates rendering them hydrophilic for subsequent reactions. Phase I usually deactivates a drug but occasionally it activates a pro-drug. Phase II and III conjugates and excretes the metabolites respectively.
Though accounts for only 2~4% of all hepatic CYPs and largely non-inducible, CYP2D6 is involved in the metabolism of 25% of drugs (PMID: 19817501). CYP2D6 is the most highly polymorphic CYP gene. This is hypothesized to represent a diversity in plant detoxification that evolved under survival pressure during periods of food constraint. The presence of numerous pseudogenes and high activity at CYP2D locus are thought to have produced the large numbers of CYP2D6 alleles (PMID: 15492763).
CYP3A4/5, the 2 major isoforms in CYP3A subfamily expressed in hepatocytes, are the CYPs present in largest quantity in human adults; it is highly inducible and less polymorphic than CYP2 family genes; it contributes to the metabolism of 50~60% of drugs (Pharmgkb.org > CYP3A4 > VIP, PMID: 24926778).
It has been 9 years since DPWG guidelines were incorporated into G-standard, the Dutch national drug database embedded in the nationwide EMR. In the US, large quantities of evidences have been accumulated and appraised by PharmGKB since 2000, and almost 30 guidelines have been developed by CPIC since 2009 (PMID: 22992668). PGx implementation efforts are unprecedented today. Box 5 introduces the various resources and platforms in a speedily maturing environment for applied PGx.
PGx trait, like other individual parameters considered in pharmacotherapy including age, renal function, liver function, drug interactions and more, is part of the profile and has its limitations. Nevertheless, like other non-genetic factors, it needs not be perfect to provide useful information to the prescriber. Individualization has always been an ideal of medicine. PGx is necessary but not sufficient to individualized pharmacotherapy.
Box 6. Numbers in pharmacogenomics
383 severe adverse events could have been prevented if PGx information was considered in a PCMH with 52942 patients over a 5-year period (PMID: 22739144)
4000–6000 USD higher costs per year are used to treat an UM or PM compared to an IM or EM patient (PMID: 24018772)
69 drugs, 6% of all prescription drugs, taken by 25% of American patients, contained human PGx biomarkers in their FDA labels in 2008; 62% of these biomarkers were CYP genes (PMID: 18657016)
30~40% of interindividual coumadin dosing variations can be attributed to VKORC1 and CYP2C9 polymorphisms; 13.3 individuals need to be genotyped to prevent one case to get supraor subtherapeutic warfarin
In the field of PGx, we have come a long way, and it is high time to face the ground-breaking step forward.