Oral Anticoagulants and Precision Medicine: Something Old, Something New

Over the last decade, the old “one size fits all” paradigm for drug therapy has been increasingly rejected by drug developers and patients alike, and replaced by a new paradigm, “precision medicine.” Precision medicine promises patients individualized drug therapies based on their genetic and other types of molecular profiles. Indeed, precision medicine has rapidly become the standard for drug therapy across many diseases, especially those that have been subcategorized based on their molecular characteristics. However, the battle between the old and new paradigms of drug therapy continues to rage for different pharmacological treatments, and perhaps is most evident in the area of oral anticoagulants. From 1954–2009, warfarin was the only approved oral anticoagulant on the market with a long track record of serving as the standard of care for preventing blood clots and stroke in patients with atrial fibrillation (AFib).1, 2 However, because of the narrow therapeutic index of warfarin, its wide interindividual variability, and the potential to interact with food and several medications, a need for a new generation of oral anticoagulants that overcome warfarin’s limitations became evident. This led to the development of a new class of oral anticoagulants called direct oral anticoagulants (DOACs)—including dabigatran, rivaroxaban, apixaban, and edoxaban—with a more predictable pharmacokinetic and pharmacodynamic profile, rapid onset of action, and fewer food and drug interactions.3 Evidence from randomized clinical trials has demonstrated the noninferiority or superiority of DOACs compared with warfarin with regard to stroke prevention in patients with AFib.4-7 Thus, DOACs are now the preferred alternative over warfarin for patients with AFib, except those with valvular heart disease (e.g., moderate-to-severe mitral stenosis or a mechanical heart valve).8 Although DOACs were noninferior or superior to warfarin in patients with AFib in randomized clinical trials, some can argue that these results may change in a routine practice setting—outside the controlled environment of a clinical trial (Figure 1). To address this point, in this issue of Clinical Pharmacology & Therapeutics, Huybrechts et al.9 evaluated the safety and efficacy of DOACs in comparison to warfarin in > 160,000 patients with nonvalvular AFib within a routine practice setting. Study participants treated with either warfarin or DOACs (e.g., dabigatran, apixaban, or rivaroxaban) were included in a 5-year population-based prospective monitoring program from 2010 until 2015. After conducting propensity score matching between participants treated with warfarin and participants receiving each one of the DOAC therapies, the comparison between DOACs and warfarin with respect to the risk of stroke revealed the superiority of DOACs over warfarin in patients with nonvalvular AFib. The study’s pooled hazard ratio (HR) for stroke was 0.75 (95% confidence interval (CI) 0.58–0.98) for dabigatran, 0.69 (95% CI 0.50–0.96) for apixaban, and 0.77 (95% CI 0.61–0.98) for rivaroxaban. These results are consistent with the findings from previously conducted randomized clinical trials (RE-LY,4 ARISTOTLE,5 and ROCKET AF6) that led to the approval of DOACs in patients with nonvalvular AFib for stroke prevention.8 Regarding major bleeding, in line with previous clinical trials,4-6 both dabigatran (HR 0.72; 95% CI 0.65–0.80) and apixaban (HR 0.56; 95% CI 0.49–0.64) were superior to warfarin, whereas rivaroxaban (HR 1.02; 95% CI 0.94–1.12) showed a similar risk of major bleeding compared with warfarin. The study by Huybrechts et al. provides evidence for the superiority of DOACs compared with warfarin in reducing the risk of stroke and hemorrhage in a routine clinical setting. With these results and the rapid adoption of DOACs as an alternative to warfarin, some would argue that this is the end of the warfarin era. However, given the long track record of using warfarin in clinical practice, its affordable cost, and DOACs’ limited clinical utility in special populations (e.g., children, renal impairment, and mechanical heart valves), warfarin will remain a viable and effective option for many patients worldwide—especially those in developing countries given its low cost and generic availability. Although randomized clinical trials and observational studies have shown clinical equivalence in the safety and efficacy of the generic vs. brand form of warfarin,10, 11 limited studies have focused on this comparison in elderly patients who may be sensitive to warfarin dose compared with other age groups. In this issue of Clinical Pharmacology & Therapeutics, Desai et al.12 conducted an observational study with > 33,000 patients, 65 years or older, who were treated with warfarin brand (coumadin) vs. generic and monitored to evaluate the safety and efficacy outcomes between warfarin brand vs. generic form. Consistent with previous evidence, no differences between warfarin brand vs. generic were observed with respect to effectiveness (HR 0.97; 95% CI 0.65–1.46), safety (HR 0.94; 95% CI 0.65–1.35), and all-cause mortality (HR 0.84; 95% CI 0.62–1.13). These results increase patient and healthcare provider confidence in using the generic version of this narrow therapeutic index medication. Although warfarin will remain a viable option for many patients worldwide, optimizing its dose for individual patients represents a major challenge because of its wide interindividual variability and narrow therapeutic index. This challenge is driven in part by the way this narrow therapeutic index therapy is prescribed, namely, as a “one-size-fits-all” medication. Over the past 2 decades, studies have identified intrinsic and extrinsic factors affecting warfarin dose variability. Factors like age, body surface area, smoking, race, target international normalized ratio, thrombosis indication, and amiodarone use have been shown to explain ~ 20% of warfarin dose variability.13 In this issue of Clinical Pharmacology & Therapeutics, Miano et al.14 have also highlighted the importance of baseline renal function in informing warfarin dose adjustments. Using a large cohort of hospitalized patients with AFib, Miano et al. have shown that baseline renal function plays a crucial role in identifying the magnitude of the interaction between amiodarone and warfarin. In patients with normal renal function (creatinine clearance 115 mL/min), the effect of amiodarone on warfarin varied threefold across the range of renal function, leading to an increase in warfarin sensitivity index, defined as dose-normalized international normalized ratio, by 36% compared with only 11.8% in patients with severe renal dysfunction (creatinine clearance 15 mL/min). In addition to renal function and other aforementioned factors, studying the effect of genetic polymorphisms within warfarin pharmacokinetic and dynamic genes has also highlighted the substantial role of genetics in explaining a significant percent (~ 30%) of warfarin’s dose variability.13 Perhaps, moving forward, the wealth of knowledge gained from warfarin pharmacogenomics and clinical research in the last 20 years may be applied to a precision medicine approach to warfarin dosing and help in overcoming current challenges in optimizing warfarin dose. In the last 2 decades, warfarin pharmacogenetics has made tremendous strides in identifying genetic and nongenetic factors that explained > 50% of warfarin dose variability in European-Americans.15 This wealth of knowledge paved the way for the development of several warfarin pharmacogenetic algorithms, like the Gage algorithm,13 and the International Warfarin Pharmacogenetics Consortium algorithm,16 which included both clinical and genetic variables along with race. The knowledge gained from those algorithms helped in designing three different randomized clinical trials (COAG,17 EU-PACT,18 and GIFT19) to evaluate the performance of a warfarin genotype-guided dosing approach compared with a clinical algorithm or standard dosing. The results from these three trials reported conflicting findings, in which EU-PACT and the GIFT favored warfarin genotype-guided dosing, whereas the COAG trial reported no benefit from using this pharmacogenetic-guided approach. These incongruous findings fueled discussion about the clinical utility of warfarin pharmacogenetic-guided dosing with many suggesting that the greater ethnic heterogeneity in the COAG trial (27% participants of African ancestry) compared with EU-PACT and GIFT (the majority of their participants of European descent) was one of the critical reasons behind the trials conflicting findings. A study by Drozda et al.20 further confirmed this hypothesis by showing that the lack of African-specific genetic polymorphisms (i.e., CYP2C9*5, *6, *8, and *11, and rs12777823) in genotype-guided warfarin dosing for Africans could lead to significant dosing errors in this population. This highlights the importance of identifying race-specific genetic polymorphisms and the need to better predict the right dose in different race groups. To address this point, in this issue of Clinical Pharmacology & Therapeutics, Asiimwe et al.21 conducted a meta-analysis to quantitatively estimate the effect of genetic variants, within genes involved in the pharmacokinetics and pharmacodynamics of warfarin, on warfarin maintenance dose in black-African patients. A systematic literature review was first initiated, which led to the selection of 48 studies, representing 2,336 warfarin-treated patients, to be included in this meta-analysis. The results from this meta-analysis revealed that black-Africans who carry one variant allele of CYP2C9*2 (rs1799853), CYP2C9*3 (rs1057910), CYP2C9*5 (rs28371686), CYP2C9*6 (rs9332131), CYP2C9*11 (rs28371685), or CYP2C9 gene cluster polymorphism (rs12777823) require 6.8, 12.5, 13.4, 8.1, 5.3, 4.4 mg/week less warfarin dose, respectively, compared to noncarriers. This highlights the importance of the CYP2C9 genetic polymorphisms on warfarin dose variability in black-Africans and the need to consider these polymorphisms when using genetics for optimizing warfarin dose in this population—especially the CYP2C9*5, *6, *11, and rs12777823, which are exclusively observed in patients of African ancestry and are not well represented in routine warfarin pharmacogenetic testing. For polymorphisms within warfarin’s pharmacodynamic gene, VKORC1, results revealed that carriers of one variant allele of rs9923231 (-1639G>A), rs9934438 (1173C>T), rs2359612 (2255C>T), rs8050894 (1542G>C), and rs2884737 (497T>G) require, 10.3, 11.1, 6.4, 3.78, and 8.2 mg/week, respectively, less warfarin dose compared with noncarriers. Although rs7294 (3730 G>A) variant allele carriers require 4.8 mg/week higher dose of warfarin compared with wildtype carriers. Collectively, the study by Asiimwe et al. sheds light on the importance of incorporating race-specific genetic polymorphisms in warfarin dosing algorithms. This observation aligns with the recommendation stated in the updated warfarin-pharmacogenetic dosing guidelines by the Clinical Pharmacogenetics Implementation Consortium,22 which provided guidance for warfarin pharmacogenetic testing based on race. Most actionable pharmacogenetic variants represent coding variants that result in either an amino acid substitution (missense variant) or frameshift. Indeed, the alleles of CYP2C9 that have been associated with warfarin response (CYP2C9*2, CYP2C9*3, CYP2C9*5, CYP2C9*6, and CYP2C9*11) are coding variants affecting the protein sequence of CYP2C9 either through amino acid substitutions (*2, *3, *5, and *6) or a frameshift (*11). However, one single-nucleotide polymorphism found exclusively in African Americans that associates with warfarin response, rs12777823, is found in a genomic region of the CYP2C gene cluster, underscoring the importance of noncoding alleles in interindividual variation in drug response. In this issue of Clinical Pharmacology & Therapeutics, Etheridge et al. present the largest characterization of expression quantitative trait locus (eQTL) mapping in liver samples from individuals of European ancestry. As noted by the authors, an eQTL is a genetic variant, most frequently found in a noncoding region, which alters gene expression levels through effects on gene transcription rate or transcript stability. The eQTLs may be tissue specific. That is, a single-nucleotide polymorphism may be an eQTL in one tissue and not another. Therefore, to ascertain eQTLs, tissue samples in which gene expression can be measured are needed. Primarily because of the low availability of most tissue samples, there is little information on eQTLs in various tissues. Analyzing data from > 1,000 liver samples, Etheridge et al. identify many new eQTLs in the liver, with a focus on cis-eQTLs, that is, eQTLs that are found in the locus of the gene with altered expression. As the liver is the primary organ for elimination of most prescription drugs, their study, which identifies 187,829 cis-eQTLs in over 1,000 pharmacogenes, or genes associated with drug response, represents a new resource for pharmacogenetic discovery and mechanism. As we look toward the future, it is highly expected that the cost of genotyping will keep going down, and more patients will have their genetic information available. Pharmacogenetic discovery and exploration of mechanism will continue as new data, such as eQTL data, in different tissues or in tissue samples from individuals from under-represented ancestral groups become available. In addition, perhaps with the rapid digitization of patient care, these data will become available in patients’ electronic health records. Undoubtfully, at this point, race-specific warfarin algorithms and indeed algorithms to inform dosing of other drugs will likely be a significant contributor in optimizing therapy, especially for those for which the one-size-fits-all model is broken.