Introduction
What is precision medicine? The concepts of “P4” systems medicine (predictive, preventive, personalized, and participatory medicine) and individualized medicine have been proposed in recent years, and have a meaning similar to precision medicine. P4 medicine, proposed by Leroy Hood, denotes a revolution in medicine [1–3]. P4 medicine promises more than a shift from a population-based “one size fits all” medicine to a particular individual-based “personalized medicine”; it also represents a movement from a reactive to a prospective and proactive approach to disease prevention, and emphasizes the importance of physician and patient participation in health care. P4 medicine is based on holistic and systems approaches to disease, emerging technologies, and analytical tools. In 2014, Topol [4] suggested the term “individualized medicine” to avoid confusion with personalized medicine, and he explained that individualized medicine relates not only to medicine that is particularized to individual patients but also to the future impact of digital technology on individual health care. Using multiple biological omic tools, we can digitize a human and create a geographic information system for the individual. In 2011, a committee representing the National Academy of Sciences in the United States proposed the term “precision medicine,” which refers to “tailoring of medical treatment to the individual characteristics of each patient through the ability to classify individuals into subpopulations with development of the Knowledge Network of Disease” [5].
Enthusiasm for precision medicine has grown significantly in the last few years, especially after President Obama announced the precision medicine initiative (PMI) in his State of the Union Address on January 20, 2015. President Obama asked Congress for $215 million in his 2016 budget to fund the PMI and lead to a new era of medicine. Shortly afterward, the Chinese Ministry of Science and Technology held the first National Expert Conference of Precision Medicine in March 2015, and planned to invest ¥20 billion by 2030 to support the PMI. In 2016, the Chinese Ministry of Science and Technology started the National Key Research and Development Program on Precision Medicine. The US PMI used $130 million to fund a research cohort and $70 million to intensify efforts to identify molecular drivers of cancer and to apply that knowledge to drug development. In contrast, the Chinese precision medicine program includes five subjects: development of a new generation of omic technologies for clinical application; a large-scale population-based cohort study; an integration, storage, usage, and sharing platform to obtain big precision medical data; precision strategies for disease prevention, diagnosis, and therapy; and the construction of precision medicine application and demonstration systems.
Cardiovascular disease (CVD) is the leading cause of death worldwide. Patients with the same disease tend to receive similar treatment: so-called one size fits all medicine. However, treatment effects differ across individuals. Thus individual treatment based on a person’s unique characteristics could enable the provision of more precise care, which may eventually maximize treatment benefits and minimize harm.
Two components are critical in precision medicine. The first is a large-scale cohort, which provides the basis of genetic and risk factor studies. The second is the study of genomics and pharmacogenomics, which identifies genetic characteristics to help predict and treat disease. Big data can be produced by the collection of medical records from cohort and omic studies. However, the analysis and integration of these data to find associations between phenotypes and genotypes is a difficult task.
Large-Scale Cohorts
Both in the United States and in China, a major aim of the PMI is to develop large-scale research cohorts. Several famous cohorts for cardiovascular research were developed before the concept of precision medicine was proposed. In 1948, the US Public Health Service embarked upon the first large-scale cardiovascular study: the Framingham Heart Study [6, 7]. Over the past 67 years, three generations of volunteers have participated in this study and more than 4000 research papers have been published based on studies using this cohort. For decades, the study has provided substantial insights into CVD epidemiology and risk factors. In 2006, the SNP Health Association Resource Program began, which supports genome-wide genotyping across all the Framingham cohorts. Hundreds of common genetic variants that affect the risk of CVD have been identified among 550,000 SNPs [7]. The Chinese Multi-provincial Cohort Study (CMCS) is a prospective cardiovascular cohort with more than 30,000 participants recruited from 11 provinces of China since 1992 [8]. Investigators have evaluated the validity of using the Framingham prediction model for the Chinese population, by comparing the results of the CMCS and Framingham studies. They found that the original Framingham model overestimated the coronary heart disease risk for CMCS participants, and recalibrated the Framingham functions using the CMCS cohort to improve the performance of Framingham functions in the Chinese population [8]. The China Kadoorie Biobank was set up to investigate the main genetic and environmental causes of common chronic diseases in the Chinese population; during 2004–2008, over 0.5 million participants were recruited [9].
One project of the Chinese precision medicine Program is to develop a CVD cohort of more than 50,000 participants and a follow-up greater than 5 years. Another project fund is allocated for precise subtyping of CVD based on multiple omic data obtained from cohort studies.
One study using a Chinese hypertension cohort showed that 75% of hypertension patients in China have hyperhomocysteinemia, a much higher level than found in other countries [10]. Moreover, the China Stroke Primary Prevention Trial confirmed the efficacy of folic acid (which can reduce the level of homocysteine) therapy in primary stroke prevention among Chinese hypertension patients [11]. Using a Chinese hypertrophic cardiomyopathy (HCM) cohort, Zou et al. [12] found a greater frequency of multiple mutations in Chinese HCM patients than in the white population, and described the correlation between genotypes and phenotypes.
These cohorts will provide big data, which are so large and complex that they cannot be analyzed with traditional data processing methods. Big data are commonly characterized by the “3Vs”: volume, velocity, and variety [13–15]. Volume refers to the amount or size of the data, which determines whether they can be considered as big data. Velocity depicts the speed of data processing. Scientists need to analyze increasing volumes of data with greater rapidity, which requires new software and hardware support. Variety describes the number of different types of data. Several types of data can be used to analyze biological samples, such as genomic, transcriptomic, proteomic, and metabolomic data. Greater data variety provides more opportunities to understand diseases and biological processes. How are big data applied in cardiovascular precision medicine? The sources of big data include clinical registries, electronic health records, patient-reported information, biometric data, medical imaging data, multiomic data, and Internet data. These data are further combined into an analytical platform and used in the prediction of risk and resource use, population management, drug and medical device surveillance, disease and treatment heterogeneity, clinical decision support, quality of care and performance measurement, public health, and research, all of which aim to improve cardiovascular quality and outcomes of care [15]. Although big data provide opportunities to implement cardiovascular precision medicine, physicians and scientists also face challenges, such as how to integrate data from various sources. We need to develop new analytical tools and apply these tools in clinical practice.
Multiomics
Genomics
On what basis should patients with the same disease receive tailored therapy? Genetic features might be a suitable foundation. The Human Genome Project, which was completed in 2003, provided the first opportunity to understand all human genes [16]. Genetic variations play an important role in the pathogenesis of multiple CVDs. HCM is the leading cause of sudden death in young people, and more than 50% of HCM patients have a family history of the disease [17]. Studies have identified disease-causing mutations in more than 11 genes that encode cardiac sarcomere proteins; these account for 80% of familial HCM cases [18–20]. To identify the function of these variations detected in HCM patients, mice containing these mutation sites have been produced. One recent study designed a small molecule, MYK-461, that reduces cardiomyocyte contractility by decreasing the adenosine triphosphatase activity of cardiac myosin heavy chain, and eventually suppresses the development of HCM [21] in mice with mutations in the β-cardiac myosin heavy chain 7 gene (MYH7). The study suggested a paradigm for treating diseases on the basis of their genetic features.
Aortic aneurysm often results from cystic medial degeneration, and can lead to dissection or rupture. As many as 40% of people with aortic dissections die instantly [22]. At least 19% of patients have a family history of thoracic aortic aneurysm (TAA), and the presentation shows classic Mendelian inheritance, indicating the contribution of a single gene to this disease [23]. Some familial TAAs have syndromic presentations, such as Marfan syndrome and Loeys-Dietz syndrome, which result from mutations in the fibrillin 1 gene (FBN1) and genes encoding transforming growth factor β receptors respectively. In some cases, the TAA treatment strategy depends on the patient’s genotype. According to European and American guidelines on treatment of TAA, regardless of the cause, patients who have a maximal aortic diameter of 55 mm or greater should undergo surgery [24, 25]. However, earlier interventions have been proposed for aortic diameters greater than 42 mm in patients with Loeys-Dietz syndrome [24, 25] because TAA growth in patients with Loeys-Dietz syndrome is much faster (more than 10 cm/year), resulting in a mean age of death of 26 years [26]. In the past two decades researchers have identified increasing numbers of genes responsible for nonsyndromic TAA. However, most TAA cases are sporadic, and the genetic predisposition of this disease is still under investigation. Using genome-wide association studies, LeMaire et al. [27] found that common SNPs in a region encompassing FBN1 are associated with sporadic TAA. Using next-generation sequencing technology, we identified FBN1 variants in 15.75% of patients with sporadic TAA and dissections, and expanded the mutation spectrum of FBN1, which will be helpful for genetic counseling of Chinese patients [28].
Atrial fibrillation (AF) is the most frequent cardiac arrhythmia. There are more than 10 million AF patients in China, and the disease is responsible for one-third of strokes in elderly people. In 2003, Chinese clinicians identified the first genetic mutation underlying AF in a four-generation family [29]. More than 20 genes associated with AF have been identified, which facilitates the individual treatment of AF.
Pharmacogenomics
Many drugs do not work in the same way for every patient. Many medications produce benefits in some patients but no response or adverse reactions in other patients. Pharmacogenomics offers the opportunity to predict drug response on the basis of a patient’s genotype, and to provide clinical decision support to increase the safety and effectiveness of drug therapy. This makes it an important area of precision medicine [30]. In the early years of the first decade of this century, the US Food and Drug Administration (FDA) began to require the pharmaceutical industry to submit pharmacogenomic information [31]. In 2009 the Clinical Pharmacogenetics Implementation Consortium was formed, and it has published multiple dosing guidelines for drugs. To date, more than 200 drug labels contain pharmacogenomic information approved by the US FDA, the European Medicines Agency, the Pharmaceuticals and Medical Devices Agency in Japan, and Health Canada. In the field of CVDs, pharmacogenomic tests have been increasingly used to provide clinically valuable information about individualized prescriptions of drugs such as warfarin, clopidogrel, aspirin, statins, and antihypertensive drugs.
Warfarin, the most commonly used oral anticoagulant drug for CVDs, is the leading cause of hospitalization for adverse drug events in the United States [32]. Variants of the cytochrome P450 (CYP) family 2 subfamily C member 9 gene (CYP2C9) and the vitamin K epoxide reductase complex subunit 1 gene (VKORC1) are associated with sensitivity to warfarin, and up to 60% of the variability in the dose requirement of warfarin can be explained by clinical factors plus common variants of CYP2C9 and VKORC1 [33–35]. Thus, in 2007, the US FDA approved pharmacogenomic data for warfarin labeling, which was further revised in 2010 to produce a dosing table based on CYP2C9 and VKORC1 genotypes, indicating the desirability of genetic testing before warfarin use.
Clopidogrel is an oral thienopyridine prodrug that is widely used in antiplatelet therapy for patients undergoing percutaneous coronary intervention. Conversion of clopidogrel to its active metabolite requires the participation of multiple CYPs in the liver, particularly CYP2C19. The common loss-of-function allele CYP2C19*2 (681G>A, rs4244285) is associated with impaired clopidogrel antiplatelet activity [36–38]. In addition to CYP2C19, variants of the genes ABCA1, CES1, P2RY12, and PON1 contribute to the interindividual variability of clopidogrel responsiveness [39].
Statins are the most commonly used cholesterol-lowering drugs, and can inhibit the activity of 3-hydroxy-3-methylglutaryl-CoA reductase, a key enzyme for cholesterol synthesis. Variants of the genes encoding 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) and low-density lipoprotein receptor (LDLR) are associated with variability in the effect of statins in lowering low-density lipoprotein cholesterol levels [40–42]. Other variants of genes such as APOE, CYP3A4, CYP2C9, and ABCB1 are also associated with efficacy of statin response by affecting low-density lipoprotein transport or statin pharmacokinetics [42]. Another critical application of pharmacogenetics in statin therapy is to predict individual adverse drug reactions. SLCO1B1 521C polymorphism decreases simvastatin transport into hepatocytes and thus increases systemic simvastatin concentrations, which leads to increased risk of myopathy. On the basis of these findings, SLCO1B1 genetic testing is recommended before simvastatin is used [43, 44].
There is evidence that two genes have a genetic association with the antihypertensive effect of drugs. The NEDD4L polymorphism rs4149601G>A predicts greater sensitivity to thiazide diuretics, and polymorphisms of the β1-adrenergic receptor gene (ADRB1) are associated with the efficiency of the blood pressure lowering response to β-blockers [45]. Homozygous ADRB1 389R carriers show greater antihypertensive effects and greater increases in left ventricular ejection fraction when treated with β-blockers [46, 47].
The use of large-scale cohorts in pharmacogenomic studies will make it easier to establish links between genetic variants and drug response. This will greatly help physicians to select drugs. At the same time, new technologies for sequencing have been developing rapidly, which has enabled many hospitals to begin genetic testing.
Other Omics
Precision medicine is based on increased knowledge of the relation between genotype and phenotype, which characterizes the expression of disease in individuals. Unlike cancer, most CVDs are complex and involve substantial gene-environment interactions; thus, genomic studies are insufficient to solve all the challenges in this area. Other omic technologies, such as transcriptomics, proteomics, metabolomics, epigenomics, exomics, glycomics, and phenomics, should also be used in cardiovascular precision medicine.
Proteomics and metabolomics are mass spectrometry–based omics. Proteomics is widely used for identification of biomarkers, diagnosis, and therapy selection [48]. With use of a targeted proteomics approach, osteoprotegerin, growth/differentiation factor 15, and matrix metalloprotease (MMP)-12 have been identified as new risk factors for carotid artery plaque in a large human sample [49]. Other biomarkers, such as trimethylamine N-oxide, MMP-2, MMP-9, MMP10, neutrophil gelatinase-associated lipocalin, and galectin 3 are also associated with CVD [36, 37, 50, 51]. Using a holistic proteomic approach, Hage et al. [39] identified novel biomarkers of inflammation that predict severity and prognosis in heart failure patients with preserved ejection fraction. Proteomics has also been used to explore the effects of drugs, such as antiplatelet agents [52]. Studies have analyzed the platelet protein map [40, 52, 53], and compared the platelet proteome profiles of aspirin-resistant and aspirin-sensitive patients with acute coronary syndrome [38]. These two groups show differences in the proteins involved in energy metabolism, cytoskeleton, oxidative stress, and cell survival pathways [38]. Individual proteomics analysis could help to predict which antiplatelet drug is suitable for a patient at risk of thrombosis.
Precision medicine has been recognized as a new era in health care. With multiomic analysis, data collected from analysis of genetic variants, biomarker detection, and phenotype assessment can be integrated and used to improve strategies for preventing, diagnosing, and treating diseases in individual patients. Precision medicine both offers promise and presents challenges to the future care of individuals with CVDs.