Pharmacogenomics

How much to prescribe to whom

Over the last 10 years we have developed methods to understand more precisely the individual’s metabolism of certain types of drugs through the use of genetic testing for the genes that determine enzyme activity but this has largely been at the experimental level and confined to hospital patients.

This now means we are at the stage of technically being able to help front line clinicians prescribe more accurately and safely. To do this we need to be able to easily obtain the material needed for genetic testing, do the test, and deliver that result back to the clinician in a timely and meaningful manner. This article addresses the background to this and the ongoing work to demonstrate feasibility for this

Drugs including Warfarin can cause significant harm

The overall incidence of serious adverse drug effects to drugs is 6.7% of hospitalized patients making these reactions between the fourth and sixth leading cause of death ^[2]. Four percent of hospital admissions are related to potentially preventable drug side effects. Drugs most commonly implicated in causing these admissions included low dose aspirin, diuretics, Warfarin, and non-steroidal anti-inflammatory drugs other than aspirin, the most common reaction being gastrointestinal bleeding ^[3]. The risk of dying associated with a major bleed associated with Warfarin is 13% ^[4] so this is one of the most important drugs responsible for adverse effects and potentially avoidable deaths.

Warfarin is one of the most difficult drugs to control.

Warfarin is a notoriously difficult drug to start and control. There is a 10 fold variation between individuals in the dose required to achieve optimum therapeutic anticoagulation but to date we have not been able to identify individuals who may need more or less WarfarinThe starting dose is generally 5 milligrams and while there is some adjustment for age and body weight using algorithms this has not led to significant decreases in hospitalizations or complications associated with Warfarin. In a similar vein the alteration of dose is usually one or two milligrams based on the INR and the previous dose or doses. Whether changing dose or starting treatment no allowance is currently made for the individual’s likely rate of metabolism of Warfarin ^[5].

Point of Care to the Lab

How might genetics help?

Clinically available Warfarin is a mixture of (R) and (S) Warfarin and the anticoagulant effect of S Warfarin is three to five times that of R Warfarin. S Warfarin is metabolized almost exclusively by CYP2C9 to its major metabolite (S)-7-hydroxywarfarin. Although as many as 30 genes are involved in the metabolism of warfarin ^[6] CYP2C9 is one of the two that have been investigated to a greater extent. There are 11 variant alleles of CYP2C9 however there are two common variants that lead to reduced Warfarin metabolism. Approximately 20% of Caucasians Warfarin-treated patients carry a variant allele of CYP2C9.

Taking into account this genetic variation in combination with patient demographics might improve safety through more accurate dosing ^[7]. The presence of the variant genotype is associated with 2.2 times the risk of bleeding and above range INR levels. ^{[8, 9]}. Having an INR level outside the optimal range is associated with a 29% increase in mortality, and a 10% increase in stroke compared with patients who are optimally controlled ^[10]. In addition it has been shown that the presence of variant alleles is associated with the patient taking a longer time to achieve control for the Warfarin. It may take as much as twice the average time (4 days) to achieve control with all the additional expense and inconvenience that entails ^[11].

Pharmacogenomics in primary care

There have been a handful of studies that have explored testing for variants of CYP2C9 in hospital settings but this is not the setting where the majority of prescribing takes place. For that we have to start undertaking work in primary care. The involvement of family physicians in routine genetic testing of patients prior to starting or changing medication may seem like a dream and is certainly a challenge. The barriers are not only the technical but how the information should be used within the practice. It is not even known whether such an approach is feasible.

Scalability

Our research team is currently addressing this very problem. Can we actually set up a system where, in primary care, we can routinely take blood, perform the genetic test, and have that result help the clinicians make decisions about drug dosing? The principle behind this study is to determine whether we can test for the gene that predicts drug metabolism but in a way that is not “research” based. We want to know if it is as easy to do as, for example, measuring cholesterol. The reason for this is the large number of drugs that potentially may benefit from this approach. So for that reason we need to make sure the methods used can be scaled up to perhaps as many as 30% of patients.

Pharmacogenomics in practice

The feasibility study will test the following steps

1. Can we easily perform the test in family practice?

Our study will start with the patient already on Warfarin. They will be approached to join the study and after consenting will have blood taken. During this part of the study we will look at how much information patients want and assess their attitude to this approach. The time taken for sharing this information will be one of the outcomes of this study.

OSCAR is an Electronic Health Record (EHR) system running at an academic family practice based at the Queen Elizabeth Health Complex (QEHC). It is part of the scalability of the process that the request can be generated from an electronic medical record. For this project we will be using an open source medical record. The reason for this is that we can share openly with all electronic medical record suppliers the process and algorithms that will be used in the process and hence enable effective knowledge translation.

For the blood test we shall be using both a normal tube of blood and also testing a blotting paper test. The latter might be an easier way of doing this in a large scale manner. The sample will be transmitted to the laboratory using normal transport methods.

2. Can we routinely and automatically determine the gene?

Beacon is the pharmacogenomics (PGx) guidance engine for pharmacogenomics test delivery and is embedded in the PHIMS framework (Pharmacogenomics Health Information Management System). The PGx Centre receives the sample and after the (Warfarin) test has been run, data will be transferred to the PHIMS system.  The Beacon system will analyze the sample to determine patients genotype for single nucleotide polymorphisms (SNPs) that affect Warfarin metabolism or sensitivity (cytochrome P450 complex (CYP2C9)). They will then run an established Warfarin dosing algorithm that includes genetic phenotype such as that produced by Lenzini and colleagues. ^[12]

This will generate a suggested dosing and send this result back to the PGx OSCAR system as a part of the patients electronic medical record within 24 hours.

3. Can this information be transmitted to the clinician in a meaningful way?

At this stage the algorithm within OSCAR will display to the clinician the advice regarding dosage taking into account current and previous levels of INR, previous doses of Warfarin, demographic information as well as the genetic marker. From this advice the clinician can make a decision about appropriate therapeutic dosing.

Implications

The overall process for this single drug is really quite straightforward from the clinician’s point of view. What has been overlooked in this article is the many years of work that went into the determination of the genetic markers that predict Warfarin metabolism and secondly the subsequent work that went into making that test readily available in an automated system. Both these stages have taken many years. Based on how this was done it is expected that future tests will be faster in the development stage. What we will be faced with is the prospect of multiple genetic tests to determine individual’s likely metabolic process for each drug the doctor and they may be considering. The implications are daunting but we are already receiving automated reports that incorporate cholesterol into a decision support module predicting cardiac risk. This process will be no different from that but what will be necessary is an underlying technological platform in the shape of clinical decision support modules embedded into electronic medical records.

Feasibility

So is this pie in the sky? The economic implications of not doing this are staggering but drug side effects account today for a sizeable proportion of the health budget. As therapeutics becomes more complex, and more people are taking medication pharmacogenomics becomes a cost effective approach. Our team believes that day has come and family physicians need take on this additional responsibility.

references

1. Gasche, Y., et al., Codeine Intoxication Associated with Ultrarapid CYP2D6 Metabolism. N Engl J Med, 2004. 351(27): p. 2827-2831.
2. Lazarou, J., B.H. Pomeranz, and P.N. Corey, Incidence of Adverse Drug Reactions in Hospitalized Patients: A Meta-analysis of Prospective Studies. JAMA, 1998. 279(15): p. 1200-1205.
3. Pirmohamed, M., et al., Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ, 2004. 329(7456): p. 15-19.
4. Linkins, L.A., P.T. Choi, and J.D. Douketis, Clinical impact of bleeding in patients taking oral anticoagulant therapy for venous thromboembolism: a meta-analysis. Ann Intern Med, 2003. 139(11): p. 893-900.
5. Shine, D., et al., A randomized trial of initial warfarin dosing based on simple clinical criteria. Thromb Haemost, 2003. 89(2): p. 297-304.
6. Wadelius, M. and M. Pirmohamed, Pharmacogenetics of warfarin: current status and future challenges. Pharmacogenomics J, 2007. 7(2): p. 99-111.
7. Gage, B., et al., Use of Pharmacogenetic and Clinical Factors to Predict the Therapeutic Dose of Warfarin. Clin Pharmacol Ther, 2008.
8. Higashi, M.K., et al., Association between CYP2C9 genetic variants and anticoagulation-related outcomes during warfarin therapy. JAMA, 2002. 287(13): p. 1690-8.
9. Sanderson, S., J. Emery, and J. Higgins, CYP2C9 gene variants, drug dose, and bleeding risk in warfarin-treated patients: a HuGEnet systematic review and meta-analysis. Genet Med, 2005. 7(2): p. 97-104.
10. Jones, M., et al., Evaluation of the pattern of treatment, level of anticoagulation control, and outcome of treatment with warfarin in patients with non-valvar atrial fibrillation: a record linkage study in a large British population. Heart, 2005. 91(4): p. 472-7.
11. Takahashi, H. and H. Echizen, Pharmacogenetics of CYP2C9 and interindividual variability in anticoagulant response to warfarin. Pharmacogenomics J, 2003. 3(4): p. 202-14.
12. Lenzini, P.A., et al., Laboratory and clinical outcomes of pharmacogenetic vs. clinical protocols for warfarin initiation in orthopedic patients. J Thromb Haemost, 2008.

There are outcome measures relating to patients in terms of bleeding and INR control as well as frequency of blood tests and tablet dose changes. In terms of the health professionals there may also be outcome questions about time that it takes to share information, lack of confidence about that knowledge as it is new, and the use of the electronic decision support. If we are to see the introduction of this sort of “personalized prescribing support” what are the most important questions that we need answered. Clearly we might need some sort of Delphi process to explore this and, while we will be setting up an expert steering group with patients, I would like to gain some feel from you about where we should focus our energy in terms of outcomes.