**4. Cost-effectiveness analysis of pharmacogenetic testing**

Many would argue that clinical practice guidelines should just focus on whether pharmacogenetic testing improves effectiveness and ignore cost considerations. However, decision making about the widespread use of genotyping also depends on its costeffectiveness. This means that even if authorities were to recommend genotyping patients prior to cardiovascular therapy based on proof of effectiveness, the recommendation might not easily be implemented without the support of other stakeholders. One important stakeholder is the payer, such as a health insurance company and its attitude can be an instrumental factor in the successful implementation of pharmacogenetic testing. Health insurance companies may require proof of cost-effectiveness - and some estimates of budget impact - before considering reimbursement.

A cost-effectiveness analysis (CEA) compares the total costs and effectiveness of two or more different treatment strategies. All sorts of costs must be considered here, including not just the cost of genotyping, but also the cost of monitoring and the cost of cardiovascular events that occur later in time. While costs are all expressed in the same way (money!), effectiveness can be defined in different ways. The definition of effectiveness determines how cost-effectiveness is expressed. For example, effectiveness can focus on the risk of an adverse event and the difference in effectiveness between two treatments can be expressed as the absolute reduction in risk of an event. The cost-effectiveness of one treatment versus another will then be expressed as the extra cost to avoid one adverse event (calculated by dividing the difference in costs by the reduction in risk). However, since this expression of

Future of Pharmacogenetics in Cardiovascular Diseases 217

Until now, only the most obvious gene-drug interactions have been detected since these are least complicated to detect when researchers are looking for causal SNPs. However, rare SNPs with large effects might as well be of importance, but it is a challenge to find large numbers of cases that are required to obtain enough power in pharmacogenetic studies when looking at smaller effects or lower allele frequencies (Daly, 2010). A trend is observed that larger studies are being performed and meta-analyses are carried out to investigate these less frequent genetic profiles. Several techniques are further developed and might lead to new insights in

This type of study investigates the association between drug response and previously identified candidate genes. These candidate genes might play a relevant role in the pharmacokinetics or pharmacodynamics of the drug and might therefore be, for example, the metabolizing enzyme or the target protein. An example is the use of candidate gene approaches for the understanding of the overall drug response to coumarins. (Daly, 2010). In 1992, Rettie *et al*. indicated *CYP2C9* as main metabolizing enzyme of warfarin (Rettie et al., 1992). A few years later, Furuya *et al.* first reported that SNPs in this gene affect the stable coumarin maintenance dose (Furuya et al., 1995). A decade later, VKORC1 was identified as the target enzyme of the coumarins (Rost et al., 2004; Li et al., 2004) and studies confirming the association between *VKORC1* genotypes and stable coumarin maintenance dose followed. Another example is the role of the *CYP2C19* genotype on the clopidogrel (Hulot, 2006) therapy response and how the treatment with tamoxifen is influenced by the *CYP2D6*

Since 2007, genonome wide association (GWA) studies have become more frequently applied in the pharmacogenetics field. This resulted in novel identified associations between drug response and variations in DNA (Daly, 2010). In CVD, GWA studies resulted in confirmation of the already available knowledge, rather than in newly identified interactions. For clopidogrel, the influence of *CYP2C19* was confirmed (Schuldiner et al., 2009) and for statin induced muscle symptoms an association with *SLCO1B1* was found (SEARCH Collaborative Group, 2008) in a GWA study. In a GWA study on acenocoumarol maintenance dose, an additional effect was found for polymorphisms in *CYP4F2* and *CYP2C18* (Teichert et al., 2009b). These GWA studies led to more knowledge about several drug-gene interactions, but the causality of the relationship is not always clear in these studies. Another difficulty with this type of analyses is the need of large patient numbers

DNA sequencing is the determination of the nucleotide bases in DNA. In contrast to GWA studies, where tag SNPs are used to cover as much of the variation within the gene as possible, this technique will determine the exact order of nucleotides in DNA. Instead of tag SNPs that are usually markers for the causal SNP - and thereby introduce noise because they

the pharmacogenetic research field. We will discuss them in this paragraph.

**5. Pharmacogenetic developments** 

**5.1 Candidate-gene studies** 

genotype (Hoskins, 2009).

**5.3 Sequencing** 

**5.2 Genome-wide association studies** 

because of the correction for multiple testing.

cost-effectiveness is very disease-specific, it is difficult, if not impossible, to compare the cost-effectiveness of different treatments for different diseases with each other and this comparability is valuable when making budget allocation decisions. For this reason, some authorities or health insurance companies require a cost-utility analysis. In a cost-utility analysis (CUA), the health gains acquired by a new treatment are expressed in Quality Adjusted Life Years (QALYs), which can be compared more easily with other treatments, also in other diseases, than the cost per adverse event avoided.

Several economic evaluations (such as CEAs and CUAs) have been performed for coumarin derivatives. The problem with these analyses is that no robust data on the effectiveness of genotyping are available yet; the large RCTs that can provide this data are still ongoing (van Schie et al., 2009; French et al., 2010). This current lack of evidence results in a wide variability in cost-effectiveness ratios among the studies that have been done, ranging from dominance (where use of genotyping reduces costs and increases health) to a very high incremental cost of \$347,000 per QALY gained (Verhoef et al., 2010). The costs of genotyping are also not clear yet. In literature, the estimated cost of genotyping for *CYP2C9* ranges from \$67 to \$350 and the estimated cost of genotyping both *CYP2C9* and *VKORC1* ranges from \$175 to \$575. Recently, a Point-Of-Care Test (POCT) for genotyping *CYP2C9* and *VKORC1*  has been developed. With this test, the patient's genotype can be determined in the physician's office within 2 hours and this is estimated to cost less than \$50 per patient for both *CYP2C9* and *VKORC1* (Howard et al., 2011). The costs of genotyping are expected to decrease even further, with increased usage. This will also influence the chance that pharmacogenetic testing is cost-effective.

Decisions about whether or not to implement pharmacogenetic testing in clinical practice will differ among different countries. This difference can be caused by several factors. Firstly, the amount of money society is willing to pay varies among different countries. For example, this 'willingness to pay' is approximately \$50,000 per QALY gained in the US or £20,000–30,000 (approximately \$33,000-50,000) per QALY gained in the UK (National Institute for Health and Clinical Excellence [NICE], 2008). Secondly, the costs, not only of genotyping but also of the consequences like bleeding events, are not identical in all countries. Next to this, the effectiveness of genotyping can also be higher in one country than in another. This is for example possibly the case with coumarin derivatives. In some countries the standard care is already of very high quality, with specialized anticoagulation clinics to monitor the effect of the drug, while in other countries this is not the case and there is still room for further improvement.

As mentioned before, the use of pharmacogenetics in treatment with a certain drug can only be recommended if information on effectiveness and costs of genotyping is available, although it is not clear what level of evidence is needed for a valid decision. Obviously, it is impossible to obtain perfect evidence. Therefore, value of information (VOI) analyses could be performed to establish the cost–effectiveness of further research on the efficiency of the strategy. If the costs of performing this research are greater than the benefits of the additional information, then it would not be worthwhile to conduct this research (Sculpher & Claxton, 2005). The parameters that have the greatest influence on the uncertainty regarding the cost–effectiveness of genotyping should be the main focus of future studies in this area. The costs of conducting these studies should also be considered. However, this will also depend upon the regulatory environment, and VOI forms only a part of the picture.
