**5. Clinical Implementation**

#### *5.1. Pharmacogenomics*

Interindividual variability is largely heritable [1,2,7,69,111], thus stimulating interest in utilizing personalization tools such as pharmacogenomics (i.e., the impact of genome variations on individual response to therapeutics) in patient care [2,7,112–116].

#### 5.1.1. Master Regulators

Studies identifying the polymorphisms of "master regulator" genes (affecting multiple CYP enzymes) have revealed key insights, as follows [16]. The 3- -untranslated region (UTR) of the aldo-keto reductase 1D1 (AKR1D1) gene is significantly associated with mRNA expression and enzyme activity of CYP2B6, CYP2C19, CYP2C8 and CYP3A4; the AKR1D1 SNP (rs1872930) yields higher constitutive mRNA expression [16]. Identifying such master regulator genes and categorizing notable SNPs may help clinicians and pharmacists predict levels and activity of CYP450 enzymes and corresponding therapeutic consequences. As CYP450 enzymes metabolize a significant majority of medications, there is a potential role for pharmacogenomics to optimize each individual's therapeutic response and prevent adverse effects.

#### 5.1.2. Warfarin

To illustrate, a notable example in precision cardiovascular medicine is CYP2C9 genotype-guided dosing of the commonly used anticoagulant warfarin [112,117]. The Clinical Pharmacogenetic Implementation Consortium (CPIC) is an initiative focused on integrating pharmacogenomics into routine clinical care. The consortium has provided guidelines recommending the use of pharmacogenetic dosing algorithms to assist with warfarin dose based on CYP2C9 and VKORC1 genotypes, with consideration of clinical factors. Specifically, CYP2C9 alleles \*2 and \*3 in addition to \*5, \*6, \*8 and \*11 are associated with lower warfarin dosing requirements, due to decreased clearance of the S-enantiomer of warfarin [117,118]. While more is known about alleles \*2 and \*3, alleles \*5, \*6, \*8 and \*11 occur more frequently in the African American populations [117,119]. In spite of these guidelines, at present, patient dosing across much of clinical practice is guided by a dose—response —adjust cycle.

Typically, an empiric dose of warfarin is administered and patient response is tracked by measuring the international normalized ratio (INR). The dose is subsequently adjusted to fit a target threshold depending on the indication for anticoagulation. This may be, at least in part, due to the seemingly inconsistent results of various studies examining the potential impact of genetic testing to guide warfarin dosing. For example, the EU-PACT, GIFT and COAG studies are studies all investigating the utility, efficacy, cost and benefit of genetic testing to guide warfarin dosing decisions [112,117,118,120,121]. The two former studies were conducted largely in homogenous populations (>90% European ancestry), whereas the COAG study was conducted in a more diverse population in the US (28% of trial participants were African American) and did not show a great benefit to warfarin pharmacogenomics [117,120]. However, while EU-PACT administered a loading dose (according to American College of Chest Physicians guidelines) upon initiation of warfarin, the COAG study did not [121–123]. Warfarin-related variants have been less studied among African and Hispanic populations, which is notable as there is higher dose variability in these ethnic populations [112,117]. To date, no study has accounted for CYP2C9 variants more common in African Americans (\*5, \*6, \*8, \*11). It is unclear whether genetic samples currently in use for warfarin pharmacogenomics accurately represent a diverse US population. The generalizability of these trials may therefore be limited. While guidelines of governing clinical bodies at best recommend weak support of pharmacogenomic testing of warfarin dosing (Class IIB recommendation for primary stroke prevention by the American Heart Association) [124], a small study has shown promise (efficient achievement of therapeutic anticoagulation, fewer supratherapeutic INR

values and a shorter duration of low molecular weight heparin) using genotype-guided dosing [117,125]. Further investigation is required.

#### 5.1.3. Clopidogrel

To further illustrate this the potential impact of pharmacogenomics in precision medicine, some individuals with specific CYP2C19 gene mutations are poor metabolizers of the clopidogrel prodrug (the most commonly used antiplatelet agent), which can limit safety of coronary stents after percutaneous coronary intervention [7]. Some studies have demonstrated the possible cost-effectiveness of genotype-guided antiplatelet strategy [126–129], underscoring the potential utility of precision medicine to tailor medication regimens and achieve optimal patient care. In the genotype-guided antiplatelet strategy, antiplatelet utilization following acute coronary syndrome (ACS) and percutaneous coronary intervention (PCI) would be determined based on CYP2C19 genotype obtained prior to the procedure.

The desire to incorporate genotyping in routine clinical care arose from reports of clopidogrel resistance due to polymorphisms in CYP2C19 [7,117]. The CYP2C19 allele primarily responsible for decreased clopidogrel metabolism is CYP2C19\*2. Additionally, multiple alleles encoding deficient enzyme activity can manifest as a true loss-of-function and an inability to metabolize the clopidogrel prodrug. Recently, results of the POPular Genetics trial—a multicenter randomized, open label study of almost 2500 patients based in the Netherlands—demonstrated noninferiority between performing genetic testing for clopidogrel resistance prior to clinical use and using another P2Y12 inhibitor (e.g., ticagrelor or prasugrel) among patients undergoing PCI [130]. As clopidogrel (available as a generic medication) is the least expensive P2Y12 inhibitor, with the lowest bleeding risk [131,132], these results represent a promising option to reduce rates of clopidogrel treatment failure, a boon particularly to those with limited financial resources. TAILOR-PCI is a currently ongoing trial, investigating the effect of the knowledge of genotype to help guide choice of P2Y12 inhibitor In one arm of the trial prospective genotyping is pursued, with individuals possessing loss-of-function (LOF) mutations receiving an alternative P2Y12 inhibitor (e.g., ticagrelor). In the control or conventional care arm, all patients received clopidogrel after PCI and their genotype is only obtained 12 months later after completing therapy. The results of this trial could shed more light on the effect of precision medicine on patient outcomes.

On a population level, CYP450 enzyme polymorphisms can account for part of the variations in drug response common among different ethnic groups. For example, 20% of Asians are poor metabolizers of CYP2C19 (responsible for metabolizing clopidogrel, as well as phenytoin/phenobarbital, omeprazole and so on), while 7% of whites are poor metabolizers of CYP2D6 (responsible for metabolizing beta-blockers, antidepressants, opioids and so on) [1,2,69]. Discussion of CYP450 variations between ethnic groups also underscores the disparities in current literature. For example, while CYP2C9 variants are more frequent in populations of African ancestry, this population is infrequently included in clinical trials [112].

It is important to note that genetic polymorphisms in CYP2C19 do not cause clopidogrel resistance in a vacuum. Comprehensive interactions among pharmacogenomics, patient characteristics and other factors affecting the activity of P450 enzymes largely determine the level of platelet response to inhibition [7]. In addition to the presence of two or more CYP2C19 LOF alleles, type 2 diabetes and increased body mass index (BMI) are likely the most important risk factors, as they independently predict and synergistically contribute to clopidogrel resistance [7,133]. Other important modifiers include chronic kidney disease, hyperlipidemia and age > 65. Lifestyle factors can either potentiate (e.g., diet, caffeine and smoking) or interfere with (e.g., grapefruit) platelet response to clopidogrel, due to interactions with CYP450 enzyme activity [7,133]. Interestingly, a large study in patients with advanced solid tumors showed a 14-fold variation in CYP3A activity, not entirely explained by genetic polymorphisms alone [134]. Potential etiologies include increased chronic inflammatory response ((IL)-1β, TNF-α and IL-6 and cytokine activity), as well as decreased liver function [3,135,136]. This

differential CYP450 enzyme activity contributes to interindividual variation, medication bioavailability and therapeutic, null or toxic effects of medications.

#### *5.2. Modified P\*3 Pathway*

An illustration of the optimal interactions of these myriad patient characteristics with the P450 enzyme system is the P\*3 pathway [78] (Figure 3). The individual elements of the P\*3 pathway—P1, Pre-empt; P2, Predict; P3, Prevent—represent a systems-based approach to patient care. In the pathway, a patient first receives precision counseling, which is akin to genetic counseling but involves digestible information about a suite of tests in precision medicine that can coalesce to form a comprehensive risk assessment. For example, after receiving precision counseling, a patient with breast cancer may in the future receive test results that include a high-risk genetic profile—mutations in TOP2A/B (mediate response to chemotherapy), RAC2 (associates with acute cardiotoxicity phenotype) and NCF4 (associates with chronic cardiotoxicity phenotype) [137]. Systems-based techniques could integrate these separate pieces of information via mathematical modeling into a single risk factor profile accessible by clinicians to help guide the patient towards optimal therapy while minimizing cardiovascular risk [78]. This information would be made available to the patient's oncologist, cardiologist and primary care provider, all of whom ideally use the same electronic medical record (EMR). With shared decision-making, prior to chemotherapy, her cardiologist may recommend cardioprotective measures with potential clinical utility, such as prophylactic angiotensin converting enzyme inhibitor, statin, beta-blocker or dexrazoxane administered prior to each course of doxorubicin. Her oncologist can also take additional precautions, including using liposomal doxorubicin preparations, avoiding concurrent trastuzumab and anthracycline use or considering alternative therapies [78]. Patient data can then be fed back into the predictive computational models to ensure continuous learning to identify actionable items for implementation in precision Cardio-Oncology. Further, patients at high versus low risk for toxicity from various anti-neoplastic agents (e.g., tyrosine kinase inhibitors, anthracyclines, monoclonal antibodies, immunotherapies) and responders versus non-responders to cardioprotective therapy, could be stratified and identified [137].

In the case of differential metabolism, genomic and other variation leading to altered P450 enzyme activity potentially resulting in drug toxicity may serve as a single hit along the path to cardiotoxicity. The presence of patient characteristics and baseline cardiovascular risk factors or existing cardiovascular disease can increase the risk for cardiovascular toxicity. When a patient's genotype, drug exposure and other factors accumulate, such multiple hits can further increase the risk of drug toxicity and consequent cardiovascular toxicity [137,138]. Using precision medicine to, for example, uncover known causes of cardiomyopathy (e.g., Titin-truncation mutations) or other cardiovascular diseases in phenotypically normal individuals, may alert patients' cardiologists and oncologists to take measures to avoid incurring additional hits. In addition to consideration of prophylactic use of ACE inhibitors, statins, beta-blockers or dexrazoxane prior to each course of doxorubicin, other cardioprotective measures may include liposomal doxorubicin preparations, avoiding multiple cardiotoxic regimens or discussion of alternative therapies, thus potentially preventing cardiotoxicity [78,139].

Data from the entire genome can potentially eventually be combined with information about a patient's transcriptome, proteome, methylome, microbiome, metabolome, environmentome, mutanome, interactome and so on in the P\*3 pathway to potentially facilitate delivery of the right therapy to the right patient or group of patients at the right time [112]. Advances in the fields of Cardio-Oncology, precision medicine and Information Technology are therefore coalescing to create new possibilities for prevention and management of cardiac dysfunction from cancer therapy-related adverse effects. Cardioprotection in the oncologic patient involves initiation of cardiac medications in order to minimize or treat cardiotoxicity from cancer therapies, while maximizing administration of indicated cancer treatment. These medications—beta-blockers, angiotensin converting enzyme inhibitors or angiotensin receptor blockers and statins—comprise the cornerstone of cardiovascular disease risk management and treatment in the general population as well as in Cardio-Oncology [140,141]. Perhaps using

systems-based approaches in precision medicine to guide the use of these cardioprotective therapies is the panacea of prevention in Cardio-Oncology—in Preventive Cardio-Oncology.

Precision Cardio-Oncology is thus a burgeoning field that seeks to further personalize the cardiovascular care of patients in oncology for decisions related to both management and prevention of cardiovascular toxicities [142]. The goal is to achieve a maximal amount of indicated chemotherapy administered with minimal interruption, while avoiding toxicity. Other notable goals include delivering more effective and efficient care, reducing patient harm and limiting healthcare costs from inappropriate treatment [43]. Indeed, precision medicine has great potential in the care of patients taking medications metabolized by CYP450 enzymes, not only in Cardio-Oncology but in all fields of medicine.

### **6. Conclusions**

Metabolism by CYP450 enzymes can determine the bioavailability and thereby efficacy of several drugs in Cardiology and Oncology and in the emergent field of Cardio-Oncology. These enzymes can also affect drug-drug interactions between Oncology and Cardiology drugs. This can compound the use of cardiology drugs for protection from or treatment of cardiovascular toxicity. Differential metabolism of each drug can determine to a certain degree unpredictable bioavailability of the drugs in a specific individual. This can be further impacted by variations in the genome, in the context of the broader epigenome, transcriptome, proteome, microRNA regulome, microbiome, metabolome, environmentome, populome and other components of the individual as a whole organism or system, with multiple parts that can be perturbed by various cardiology or oncology drugs. Increasing knowledge and implementation of the multidimensional impact of endogenous regulatory systems on CYP450-mediated drug metabolism may help preempt drug-drug interactions, predict variations in CYP450 enzymes and prevent complications from subtherapeutic or supratherapeutic drug levels. Such a systems-based view should be considered as we move towards clinical and research practice of Precision Cardiovascular oncology, with particular attention to the role of CYP450 enzymes.

**Author Contributions:** Conceptualization, S.-A.B.; Writing—Original Draft Preparation, O.A.F. and S.-A.B.; Writing—Review & Editing, O.A.F. and S.-A.B. All authors have read and agreed to the published version of the manuscript.

**Acknowledgments:** The authors are grateful to Giselle A. Suero-Abreu at Rutgers New Jersey Medical School in Newark, N.J. for helpful discussions, suggestions and assistance with the manuscript. We are also thankful to Diana Mechelay, PharmD at CHRISTUS Good Shepherd Medical Center in Longview, T.X. for helpful discussions, suggestions and reading the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
