Personalized drug therapy; the genome, the chip and the physician
Article first published online: 20 JUN 2005
British Journal of Clinical Pharmacology
Volume 60, Issue 1, pages 1–4, July 2005
How to Cite
Lewis, L. D. (2005), Personalized drug therapy; the genome, the chip and the physician. British Journal of Clinical Pharmacology, 60: 1–4. doi: 10.1111/j.1365-2125.2005.02457.x
- Issue published online: 20 JUN 2005
- Article first published online: 20 JUN 2005
Currently the concept of personalized medicine is receiving much attention [1, 2], and is also a focus topic of drug regulatory authorities such as The Food and Drug Administration [3, 4]. This concept is viewed by some as the epitome of the future of 21st century medicine. But how modern is the concept? Some medical historians would contest that the often cited axiom ‘Primum non nocere’- first do no harm (often attributed to Hippocrates or Galen but possibly originally stated by the English physician Thomas Sydenham)  is a long espoused principle of medical practice that clearly relates to personalized drug therapy. Thus the underpinning principle of ‘personalized drug therapy’ though often promoted as modern, may have been a component of the alert physician's therapeutic decision-making algorithms for centuries.
So what precisely is meant by personalized medicine or personalized drug therapy? Perhaps the question, if asked from a clinical pharmacologist's perspective should be stated as what is individualized drug therapy? One definition, is that it is medical practice with the knowledge and expertise to optimize drug therapy in the face of the many factors that combine to determine individual patient variability in drug response:
- • patient factors; for example genetics, age, genetics, gender, concurrent disease, concurrent drug therapy and environmental agents such as smoking and ethanol
- • drug factors; including pharmacokinetics, pharmacodynamics, adverse effects and drug interactions.
Promotion of the concept and the practice of individualization of drug therapy has been a prime mission of the discipline of clinical pharmacology, and its widespread acceptance would bring about the abandonment of the ‘one drug or one dose fits all’ paradigm.
However, in the current scientific environment the term personalized medicine (personalized drug therapy) seems inexorably linked with the rapidly evolving science of pharmacogenetics. Therefore a logical paradigm is that personalized drug therapy based on pharmacogenetics is really a component of the practice of individualized drug therapy as defined previously. Pharmacogenetics could provide the pivotal force to propel the practice of individualized drug therapy to be the standard of care in the not to distant future. The scientific knowledge base of pharmacogenetics has expanded rapidly over last few years. Evidence that supports this is that between 1961 and May 2005 there were 2993 listings in PubMed that contained the term pharmacogenetics . Between 1961–64 there were only 20, but from January to the end of May 2005, there were 339 such publications listed.
This issue of the Journal contains two papers that describe studies of the human CYP2C drug metabolizing isoenzyme family and its pharmacogenetics [7, 8]. CYP2C isoenzymes share considerable sequence homology but have differences in site and level of expression as well as drug substrate specificity. The main family members, CYP2C8, 2C9, 2C18 and 2C19, are collectively responsible for the Phase-I metabolism of approximately 20% of prescribed drugs [1, 2, 9]. Little is currently known about the tissue profile of protein expression or substrates of CYP2C18. The description of the crystal structure of CYP2C8 has allowed new probe substrates and inhibitors to be developed  and the first case study of a CYP2C8 polymorphism associated with cerivastatin-induced rhabdomyolysis has been described . Currently, however, the clinical relevance of CYP2C8 polymorphisms [12–14] is not fully defined. The literature indicates that the most important members of the CYP2C family from a human drug metabolism perspective are CYP2C9 and CYP2C19 [15–17]. CYP2C9 is important in the metabolism of drugs such as warfarin (S-warfarin), sulphonylureas (tolbutamide, glipizide, etc.), irbesartan, losartan, several NSAIDs and celecoxib [15, 17–19]. CYP2C19 plays a significant role in the metabolism of drugs such as the proton pump inhibitors and the anti-epileptic drugs diazepam and phenytoin [15, 17, 18]. Two genotypes of CYP2C9, 2C9*2 and 2C9*3 have reduced activity and occur in 0.2–1% of Caucasians [15, 17]. In the case of CYP2C19, although there are several inactive variants, the two variants, 2C19*2 and 2C19*3 account for 95% of individuals with low enzymatic activity . CYP2C19 poor metabolizer genotypes are found in 2–3% of Caucasians, 4% in Africans/African Americans and 10–25% in South East Asians .
Holstein et al., in this issue of the Journal, report a study in which they investigated the hypothesis that the CYP2C9 genotypes *2/*3 and *3/*3 (slow metabolizer genotypes that are reported to have only 10–20% of the activity of the wild type CYP2C9*1 genotype ) were risk factors for the severe hypoglycemia associated with sulphonylurea therapy in Caucasian patients as a result of compromised metabolic clearance. The study was a retrospective case-control cohort study of patients with severe hypoglycemia. The authors observed a significantly higher incidence of the CYP2C9 slow metabolizer genotype in the severely hypoglycemic patients compared with that in the control groups. Caution is warranted in interpreting/extrapolating these preliminary findings into the realm of ‘personalized medicine’ as the study was retrospective and has the potential for cohort selection bias.
Genetic variation in CYP2C9 activity and concordance with resultant clinical drug effect data is probably best understood for warfarin. Warfarin metabolism mediated by CYP2C9 has a significant influence on the prothrombin time (therapeutic and toxic), because the S-enantiomer of warfarin, the more potent of its two enantiomers in producing its anticoagulant effect, is predominantly metabolized by CYP2C9. Retrospective studies in patients on chronic maintenance warfarin therapy suggested that the dose of warfarin was strongly related to the CYP2C9 genotype [22–24]. The maintenance dose of warfarin in CYP2C9*2/*3 heterozygotes was lower than in patients who are homozygous normal (CYP2C9 *1/*1) and the lowest maintenance doses were required by the few patients who were CYP2C9*3 homozygotes. Patients with low CYP2C9 activity were also at increased risk of bleeding complications from warfarin. A small (n = 48), non randomized, prospective study of CYP2C9 genotype-based warfarin dosing in orthopaedic patients suggested encouragingly effective anticoagulation in most patients, but a small number of the patients showed evidence of high prothrombin times (INR > 4) . Furthermore, studies now suggest that polymorphic variants of the vitamin K epoxide reductase complex (the target of warfarin), subunit 1-VKORC1-contribute along with the CYP2C9 polymorphisms to interpatient variability in warfarin dosage [26, 27]. These findings emphasize that a focus on single-gene effects accurately predicting clinical drug effects may be misleading and that the paradigm should be revised to involve genotyping of multiple genes, each partly contributing to the genetic variability in drug response.
The advancing science of pharmacogenetics should promote the optimal design of randomized studies of appropriate genotype vs non genotype based dosing (for example for warfarin or sulfonylureas) and study therapeutic and adverse drug induced outcomes. This will best determine the clinical utility of such testing and its contribution to personalized drug therapy. For these drugs measurement of prothrombin time (INR) and blood glucose/HbA1c is simple, rapid and currently represents the best overall measure of pharmacokinetic and pharmacodynamic variability in any individual.
In another report in this issue of the Journal, Miura et al. studied the effect of fluvoxamine (an inhibitor of CYP1A2 and CYP2C19) on the pharmacokinetics of single oral doses of lansoprazole in 18 normal healthy Japanese subjects of known CYP2C19. The subjects were of known CYP2C19 genotype (six homozygous extensive metabolizers, 2C19*1/*1; six heterozygous extensive metabolizers [n = 3 *1/*2 and n = 3 *1/*3] and six heterozygous poor metabolizers [n = 5 *2/*2 and n = 1 *2/*3]). This was a randomized, placebo controlled crossover study of the effect of fluvoxamine (25 mg twice daily for 6 days) or placebo on the pharmacokinetics of a single 60 mg oral dose of racemic lansoprazole, in which the plasma concentration versus time profile of (R-)- and (S-)-lansoprazole and the metabolite lansoprazole sulfone were studied. The researchers concluded that fluvoxamine increased the plasma concentrations of (S-)-lansoprazole more than that of (R-)-lansoprazole and, in extensive CYP2C19 metabolizer subjects, CYP2C19 played a greater role in the absorption and elimination of (S-)- lansoprazole than of (R-)- lansoprazole. These authors have previously reported similar associations between CYP2C19 genotype and the effect of fluvoxamine on omeprazole pharmacokinetics . The clinical relevance of these findings is not straightforward to interpret because the subjects were healthy volunteers, and the percentage of subjects who were CYP2C19 poor metabolizers was intentionally selected to be much higher than in many patient groups.
In the case of CYP2C19 the data concerning the relationship between genotype and clinical drug effect is best understood for proton pump inhibitor therapy when used to treat peptic ulcers. There are marked differences in plasma concentrations of proton pump inhibitors in people with different CYP2C19 genotypes and these are mirrored by drug-related modulation of gastric pH . It would therefore not be unexpected that the healing of peptic ulcers has been associated with a gene-dose effect for CYP2C19 . In addition, the Helicobacter pylorii cure rate when treated with a proton pump inhibitor plus amoxicillin was related to CYP2C19 genotype [29, 31]. These genotype-related differences were less pronounced when standard triple antihelicobacter drug therapy was employed. But patients not cured by the triple drug regimen were often homozygous for CYP2C19 (*1/*1 – extensive metabolizers)  and if they were retreated with higher doses of proton pump inhibitors, a cure was achieved. Randomized prospective data showing benefit of CYP2C19 genotyping in improving clinical outcomes with proton pump inhibitor therapy (as a component of standard triple therapy) for treating Helicobacter pylori infection, or as monotherapy for gastro-oesophageal reflux disease are currently not available.
The reader is referred to other sources for details on the extensive literature on the genetic polymorphisms in certain drug receptors (e.g. beta receptors, epidermal growth factor receptor), drug transporter proteins (e.g. P-glycoprotein, [ABC-B1]) or other drug metabolizing enzymes such as CYP2D6 or thiopurine methyl transferase (TPMT) [1, 15, 33, 34]. The current data for genotyping TPMT in patients with leukaemia who are to receive 6-mercaptopurine or azathioprine perhaps best illustrates the clinical and pharmacoeconomic benefit from using such technology [1, 33, 35]. It is noteworthy that at the end of 2004, the US Food and Drug Administration approved a microarray chip designed to determine the presence or not of 31 well characterized polymorphisms in CYP2D6 and CYP2C19 in any individual patient . This microarray chip and the associated hardware needed to define a patient's genotype for these drug metabolizing enzymes is very expensive. The clinical utility of such prospective genotyping is still being evaluated.
In conclusion, currently, the basic and clinical scientific literature and pharmacogenotyping technology does not generally, nor specifically (as in the case of CYP2C isoenzymes) justify genotyping as a routine part of a paradigm to achieve individualized drug therapy. Perhaps, one exception to this is in the case of thiopurine therapy and TPMT genotyping, where there are clear scientific and clinical data to support such a paradigm. However, when scientific advances occur rapidly, such an overall viewpoint may have to be revised in the foreseeable future . What clinical pharmacologists must ensure is that physicians have adequate training in pharmacology and therapeutics, to develop a sound knowledge of the multiple patient (including pharmacogenetic) and drug factors that contribute to individual pharmacokinetic and pharmacodynamic variability. This should be considered a pivotal component in facilitating any physician's ability to individualize drug therapy.
- 3http://www.fda.gov/cder/genomics/scienceForum2005.pdf (accessed May 31st 2005).
- 6http://www.ncbi.nlm.nih.gov/entrez/query.fcgi (accessed May 31st 2005).
- 18http://medicine.iupui.edu/flockhart/table.htm (accessed May 31st 2005).
- 36Food and Drug Administration. Medical devices: clinical chemistry and clinical toxicology devices; drug metabolism enzyme genotyping system. Fed Regist 2005; 70: 11865–87.