Pavel HAMET,MD, PhD
CRCHUM, Centre de Recherche du Centre Hospitalier de l’Université de Montréal
Montréal (Québec) – CANADA
Hypertension is generally recognized as a complex disease, with a significant genetic component interacting with known and unknown environmental factors. The genetics of Mendelian types of hypertension has been largely resolved, but that of its polygenic counterpart requires further study. Recent skepticism about the usefulness of genetic analysis stems from the relatively small contribution of individual genetic markers to overall blood pressure variance. Here, we discuss several explanations behind this finding. Included are disease heterogeneity, the need for separate analysis by sex and age, ethnic complexity, and the need for analysis of genetic versus environmental contributions. The extent to which genetic determinants modulate the efficacy of a therapeutic agent is, again, better understood for monogenic components. The technology is currently available to dissect the individual and combined pharmacogenomic components of blood pressure and of medication response; however, it must be emphasized that clinical trial testing of these new approaches will be required. At present, the most significant progress has been achieved in accompanying development of novel medications with genetic determinants of their eventual safety and responsiveness in the so-called companion diagnostic paradigm. Along those lines, we propose to pursue network-based identification by comparative network analysis in responders and nonresponders, eventually leading to druggable candidate development.
Genetic determinants of cardiovascular disease (CVD), and hypertension (HT) in particular, were demonstrated initially by classic studies using correlations between (i) twins, parents and their natural and adopted children, and reconstituted families, and (ii) heritability of blood pressure (BP) within families.1 These studies illustrated that about 30% to 50% of HT has a genetic determinant, and that the remaining percentage is environmentally influenced. Furthermore, the genetic component is larger in subjects under 50 years of age: the ratio of intrafamilial prevalence of HT to that observed in the same general population, called λs, is 3.7 in young subjects, and is only 2.0 in older subjects from the same families, as determined in our cohort from a relatively isolated population from Saguenay-Lac St-Jean in Quebec, Canada.2 Initial studies of segregation of genetic polymorphisms with- in genes of the renin-angiotensin system were introduced in experimental animals, particularly rats,3 and soon after in humans.4 The greatest success in elucidating the attributes of genetic HT came from the brilliant studies by Lifton’s group,5 which investigated monogenic forms of this disease in familial HT syndromes associated with sodium transport mechanisms. Resolution of these Mendelian forms of HT has been aided by pedigree studies, the monogenic involvement, and the absence of mutation in the general population. Major technological developments have facilitated a recent burst of activity in regards to the more complex polygenic form of so-called essential HT. Whereas only a few years ago merely 500 polymorphic markers were known for all chromosomes, there were 50 000 known single nucleotide polymorphisms (SNPs) in the year 2000, and we are currently using over 1 million SNPs and copy number variants (CNV) to pinpoint associations down to a gene, its regulatory elements, or the intergenic regions, which frequently contain relevant structures such as microRNA. We also expect much needed progress in transcriptomic and epigenomic studies to integrate geneenvironment interactions into our understanding of biological systems. The term “essential” HT should then be modified to distinguish between monogenic and polygenic disorders, and eventually to further differentiate specific etiological entities based upon their predominant genetic architecture. For example, we have demonstrated that HT in families with obesity and HT in those without obesity is determined by distinct genetic loci.6
Major achievements in genome-wide association studies (GWAS) of HT have been reported. Their success has been limited by the requirement of a very large sample size and a need for resolution of a small portion of the BP variance component (less than 0.5 mm Hg BP variance is usually associated with a single SNP, even when highly significantly associated with BP). It should also be noted that a highly significant contribution of SNPs in very few subjects does not exclude unknown causes of monogenic contribution. What are the potential reasons for these characteristics? In our opinion, they include:
_ Heterogeneity of the disease. We have mentioned that HT with obesity and HT without obesity have distinct genomic architectures. In our studies, we have demonstrated linkage for HT in the proximal part of chromosome 1 in families with high prevalence of obesity; no such linkage for HT was observed in families with low prevalence of obesity in the same population.6
_ Sex- and age-dependent genotypes. In most epidemiological studies, including genomic studies, adjustments are made for sex and age. However, it has been clearly demonstrated in both human and animal studies that such adjustments may obscure contributions that are present in one or the other sex only and/or at a certain age. In our systematic approach, we have demonstrated that an SNP on chromosome 12 (rs575121) was significantly associated with elevated systolic BP in men only, and with a contrasting genotypic effect in women.7
_ Geo-ethnicity (intraethnic admixture). It is generally accepted that the difference in susceptibility to disease between various ethnic groups is mainly due to allelic frequency. Thus, it is important to observe whether or not the allelic frequency of a certain SNP is significantly related to the penetrance of the disease, as is the case in broad ethnic groups, such as Africans, Asians, and whites.8 The FTO gene is one example of genes shown to have different impacts on risk prevalence in different ethnic groups. For instance, one FTO allele (rs6499640) has been shown to be significantly associated with body mass index, and though it has a high frequency in European countries, it is rare in Asians. This has become even more relevant, as we have demonstrated that this gene is related not only to obesity, but also to HT, as confirmed in two different cohorts,9 in populations in which HT is frequently associated with obesity.
_ Genes versus environment (allelic penetrance). In HT, this exploration is still in its early stages, but at least one example should be mentioned about the impact of BP treatment. The fact that most of the studies are performed only in treated subjects is certainly diluting the power of resolution of the studies. In our French Canadian cohort, we performed fine phenotyping that included withdrawal of medication for lipids and HT, which frequently is not feasible in large cohort studies (23rd Scientific Meeting of the International Society of Hypertension, Abstract 215, Vancouver 2010). This allowed us to identify several SNPs within the FTO gene that are highly significantly associated with diastolic BP, but that appeared only after withdrawal of medication. No linkage or association was detected in the same subjects before medication withdrawal.
Several reports have shown that most treatments used in chronic diseases have an average individual response rate of 50%,10 ranging from 25%for antineoplastic agents to 80% for analgesics. In between, we find a 30% response rate for Alzheimer’s, 48% for osteoporosis, 57% for diabetic agents, 60% for asthma, and 62% for depression. An additive response of combined classes of medication constitutes, from a pharmacogenomic point of view, evidence for distinct genomic targets. Diuretics, such as indapamide, angiotensinconverting enzyme (ACE) inhibitors, and/or calcium channel blockers (CCBs), each acting on their sets of gene targets, elicit an additive response rate when they are combined in the same individual. Pioneering work by Giuseppe Bianchi demonstrated the clinical impact of adducin gene polymorphism as a partial determinant of diuretic responsiveness, and underscored the importance of the particular context of ethnicity and gene-gene interaction.11 It is important to realize that therapeutic agents, when they are ineffective, may even have an opposite negative impact in some subjects. One such example is the response to corticosteroids in subjects with specific sequence variants in the corticotropin-releasing hormone receptor 1 gene, CRHR1, which may result in a positive or negative impact on pulmonary function.12 Clear evidence concerning treatment of HT is incomplete, and there is a need for major clinical trials in which polymorphisms in relevant target genes and metabolizing genes in the subjects are ascertained, in order to initiate development of more personalized therapies for individual subjects or specific groups of subjects.
Pharmacogenomics as a science was pioneered by Motulsky and colleagues13 in their investigation of the need to avoid specific anesthetic agents, and is based on detection of susceptibility to certain adverse drug reactions. More recently, David MacLennan’s group has demonstrated such susceptibility in malignant hyperthermia populations. They describe mutations in the ryanodin receptor type 1 gene, RYR1, a sarcoplasmic reticulum calcium release channel, and the CACNA1S gene that codes for the α1 subunit of the voltagegated calcium channel receptor, known as dihydropyridine receptor.14 Subjects with mutations within these genes are obliged to avoid volatile anesthetics, which could cause a lethal hypermetabolic crisis that is associated with this autosomal dominant disorder. The polymorphism within these genes was recently reported to be related to sensitivity of hypertensive subjects to CCBs. Specific and sometimes unexpected therapeutic responses are described in monogenic forms of severe HT, including glucocorticoid-mediated attenuation of HT in Laidlaw’s syndrome of glucocorticoid-remediable hyperaldosteronism targeting CYP11B1 and CYP11B2. Triamterene, usually a weak antihypertensive, is lifesaving in patients with Liddle’s syndrome, targeting the genetic abnormality of the βsubunit of the epithelial sodium channel, SCNN1B and SCNN1G. In general, regarding the polygenic and complex form of essential HT, several levels of genetic determinants will have to be considered when aiming at genetically designed BP-lowering therapy.
_ Phenotypically prescribed drugs have variable efficacy: the rate of absent or incomplete response varies from 10% to 30% for ACE inhibitors, 15% to 25% for β-blockers, reaching 30% to 70% for statins, and 40% to 70% for β2-agonists.
_ The BP change observed during treatment in trials of most currently used antihypertensive agents, including converting enzyme inhibitors, is usually between 5 to 10 mm Hg, but has been demonstrated to vary from an extreme of 25 mm Hg BP reduction to even an increase in BP in up to 15% of subjects. This is a reality not frequently reported in the literature. The specific genomic architecture of nonresponders, and even opposite responders, is unknown at present.
_ Certain polymorphisms have been determined for some therapeutic targets, including a polymorphism in the β1-adrenergic receptor motif, which alters cardiac function and β-blocker response in human heart failure: subjects homozygous for arginine 389 have been shown to benefit from a 38% reduction in mortality when treated with bucindolol, which contrasts with the effect in the 30% of subjects with a glycine 389 polymorphism, who had no clinical response compared with placebo-treated subjects.14,15 The above mentioned sensitivity to CCBs in subjects with polymorphisms within the CACNA1 gene is another example.
_ Naturally, polymorphisms in drug-metabolizing enzymes is an important factor that is well known, described, and included in the Medical Compendium designed for physicians; yet, these polymorphisms are not widely taken into consideration at present due to the lack of availability of pharmacokinetic diagnostics in the majority of clinical settings.
Most recently, associations of hypertension drug target genes with BP and hypertension have been evaluated and validated in 86 000 individuals.16 The strongest signals were in the ADRB1 and AGT genes. Additional signals were described in the genes ACE and CACNA1A.16 Hundreds of SNPs will have to be included as a base unit of genetic scores rather than as isolated genes. This new approach, called variable set enrichment analysis, should be applied for its practical utility; however, it needs to be validated in prospective, genetically stratified therapeutic clinical trials in the future.17
In 1898, Sir Archibald Garrod coined the term “chemical individuality” to describe inherited predispositions to metabolizing sulfonyl drugs. It took 100 years, until 1998, for Herceptin, a drug effective in 25% of the subpopulation of breast cancer patients, to be approved by the Food and Drug Ad ministration (FDA) for use in tandem with genetic diagnostic testing on this eve of personalized health care. Personalized medicine, sometimes called “precision medicine,” is defined by the international PersonalizedMedicine Coalition for its clinical applications, aiming to shift the emphasis inmedicine from reaction to prevention. Other goals include selection of optimal therapy, reduction of the use of the trial-and-error model in prescribing, and safer use of drugs through avoiding adverse reactions. Yet, major developments in patient-centered medicine at present are based on “companion diagnostics.” This term has been coined to characterize the development of a diagnostic tool alongside that of medication, such that medication will be given only to subjects that satisfy the predefined criteria, ie, those predicted to respond to the drug and for whom the drug is predicted to be safe. The first model, the Herceptin model, was actually based on a failed clinical trial. Further research, however, demonstrated that the drug was effective in a subset of subjects with breast cancer, and its genetic detection is now widely recognized as a prerequisite for prescription of this costly medication, which has significant side effects, but is effective in a significant subset defined by the presence of human epidermal growth factor receptor 2 (HER2) on breast cancer cells, providing a clear benefit to those patients. This area has been recently summarized by Hamburg and Collins,18 respectively, the Commissioner of the FDA and Director of the National Institutes of Health (USA). Currently, there are 3 FDA-approved drugs used with companion diagnostics in clinical practice: i) Herceptin (traztuzumab), which targets HER2 to treat metastatic breast cancer, and the approved companion diagnostics—an immunohistochemistry test and a gene amplification test; ii) Erbitux (cetuximab), which targets epidermal growth factor receptor (EGFR) to treat metastatic colorectal cancer, and its companion diagnostic— an immunohistochemistry test; iii) Glivec (imatinib), which targets the cell surface tyrosine kinase receptors in gastrointestinal tumors, and its companion diagnostic—an immunohistochemistry test.
As mentioned above, most clinical applications of genomics are in the area of monogenic diseases, and include detection and prevention, such as in monogenic HT. Only future prospective evaluation of the efficacy of genetic prediction in clinical trials, randomized using tools of genomic stratification, will pave the way to application of individually targeted medicine in hypertension.
Athorough analysis of the theme of genomics in drug discovery and development has been published as a monograph by Semizarof et al,19 demonstrating the importance of genomics in new target identification. This new classification of disease should take into account expression profiles and gene copy number in addition to genomic sequence. As such, in June 2011, the FDA approved determination of the copy number of the HER2 gene as a diagnostic tool for better Herceptin use in targeted individuals. To our knowledge, this is the first time that a drug therapy is to be applied subsequent to prediagnosis based on copy number variance. Furthermore, animal models of disease should be submitted to genetic profiling in concordance with human genetic architecture. The next step is target validation, the targeted subtraction or addition of genes in vitro and in vivo, followed by validation of gene function using microarray and pathway analysis. Lead compound identification and optimization and in vitro biomarker discovery includes the identification of polymorphisms, copy number, gene expression, methylation, other epigenetic markers, and microarray profiling as associated with drug sensitivity.
Another area where genomics can be of help is the dissection of druggable targets, summarized as the druggable genome. An update by Russ and Lampel20 proposes the evaluation of the most frequent targets, including rhodopsin-like G protein– coupled receptors (GPCR), protein kinases, ion channels, proteases, and enzymes, as they have classically been used by the pharmaceutical industry before the use of novel genomic technologies. It is clear that the “one gene, one target” paradigm, which in the past was considered the goal for the industry, is progressively being abandoned, as molecular targets and their therapeutic utility are determined by the balance of their actions and not by their absolute specificity. The pleotropic impact of drugs is of highest interest, as we are learning more and more about the balance of metabolic networks, which will be the future blueprint for drug discovery. As proposed by Barabási and his colleagues21: “Advances in genome analysis, network biology, and computational chemistry have the potential to revolutionize drug discovery by combining system-level identification of drug targets with the atomistic modeling of small molecules capable of modulating their activity.” The future direction in therapeutic areas such as HT and diabetes will be based on past successes, yet be guided by the evidence that residual risk is still present and can be targeted genetically. Therefore, we propose the first phase should be organization of network-based identification, followed by unmet needs identification, validation, and optimization. As a next step, outcome-dependent comparative network analysis should be performed in responders and nonresponders, based on their respective genomic signatures. We believe that this will accelerate the progress toward identification of new targets in nonresponders as well as new targets in responders. Virtual screening of druggable candidates should be performed in parallel to this step, leading to further validation, optimization, and the addition of novel technologies, such as deep sequencing.
1. Hamet P, Pausova Z, Tremblay J, Deng A. Hypertension as a genetic disease. In: Mancia G, Chalmers J, Julius S, Saruta T, Weber MA, Ferrari AU, eds. Manual of Hypertension. 1st ed. Churchill Livingstone; 2002.
2. Hamet P. The renin-angiotensin system: where do we stand, and what is the future? Am J Hypertens. 2005;18(9, pt 2):125S-126S.
3. Jacob HJ, Lindpaintner K, Lincoln SE, et al. Genetic mapping of a gene causing hypertension in the stroke-prone spontaneously hypertensive rat. Cell. 1991; 67(1):213-224.
4. Jeunemaitre X, Soubrier F, Kotelevtsev YV, et al. Molecular basis of human hypertension: role of angiotensinogen. Cell. 1992;71(1):169-180.
5. Lifton RP. Individual genomes on the horizon. N Engl J Med. 2010;362(13): 1235-1236.
6. Pausova Z, Gaudet D, Gossard F, et al. Genome-wide scan for linkage to obesity- associated hypertension in French Canadians. Hypertension. 2005;46(6): 1280-1285.
7. Seda O, Tremblay J, Gaudet D, et al. Systematic, genome-wide, sex-specific linkage of cardiovascular traits in French Canadians. Hypertension. 2008;51(4): 1156-1162.
8. Ding K, Kullo IJ. Geographic differences in allele frequencies of susceptibility SNPs for cardiovascular disease. BMC Med Genet. 2011;12:55.
9. Pausova Z, Syme C, Abrahamowicz M, et al. A common variant of the FTO gene is associated with not only increased adiposity but also elevated blood pressure in French Canadians. Circ Cardiovasc Genet. 2009;2(3):260-269.
10. Spear BB, Heath-Chiozzi M, Huff J. Clinical application of pharmacogenetics. Trends Mol Med. 2001;7(5):201-204.
11. Citterio L, Lanzani C, Manunta P, Bianchi G. Genetics of primary hypertension: the clinical impact of adducin polymorphisms. Biochim Biophys Acta. 2010; 1802(12):1285-1298.
12. Tantisira KG, Lake S, Silverman ES, et al. Corticosteroid pharmacogenetics: association of sequence variants in CRHR1 with improved lung function in asthmatics treated with inhaled corticosteroids. Hum Mol Genet. 2004;13(13): 1353-1359.
13. Ozdemir V, Motulsky AG, Kolker E, Godard B. Genome-environment interactions and prospective technology assessment: evolution from pharmacogenomics to nutrigenomics and ecogenomics. OMICS. 2009;13(1):1-6.
14. Kraeva N, Riazi S, Loke J, et al. Ryanodine receptor type 1 gene mutations found in the Canadian malignant hyperthermia population. Can J Anaesth. 2011; 58(6):504-513.
15. Liggett SB, Mialet-Perez J, Thaneemit-Chen S, et al. A polymorphism within a conserved β1-adrenergic receptor motif alters cardiac function and β-blocker response in human heart failure. Proc Natl Acad Sci U S A. 2006;103(30): 11288-11293.
16. Johnson AD, Newton-Cheh C, Chasman DI, et al. Association of hypertension drug target genes with blood pressure and hypertension in 86 588 individuals. Hypertension. 2011;57(5):903-910.
17. Yang W, de Las Fuentes L, Davila-Roman VG, Charles GC. Variable set enrichment analysis in genome-wide association studies. Eur J Hum Genet. 2011;19 (8):893-900.
18. Hamburg MA, Collins FS. The path to personalized medicine. N Engl J Med. 2010;363(4):301-304.
19. Semizarov D, Blomme E, Bhathena A. Genomics in drug discovery and development. Hoboken, NJ: John Wiley & Sons, Inc; 2009.
20. Russ AP, Lampel S. The druggable genome: an update. Drug Discov Today. 2005;10(23-24):1607-1610.
21. Shen Y, Liu J, Estiu G, et al. Blueprint for antimicrobial hit discovery targeting metabolic networks. Proc Natl Acad Sci U S A. 2010;107(3):1082-1087.
Keywords: companion diagnostics; copy number variant; hypertension; personalized medicine; pharmacogenomics; single nucleotide polymorphism