Candidate genes, association studies and haplotype analysis in the search for the genetic basis of hypertension

The case for the candidate gene approach The research paradigms striving to unfold the genetic basis of complex diseases such as hypertension can be classified into two major categories, as also illustrated in Figure 1: (i) the traditional deductive candidate gene approach, which deals primarily with previously identified genes that are thought to participate in the pathophysiology of the disease, based on knowledge acquired from traditional physiology; and (ii) the more advanced inductive approach, which is based on the rapidly evolving high throughput technologies of genomics, transcriptomics and proteomics. However, it must be realized that the two approaches are closely intertwined, as the latter high tech research strategies, which may be considered perhaps more up-to-date, ultimately identify genes that have been previously identified as likely to be involved in the pathogenesis of the disease, or to lead to the identification of novel genes. Either way, a list of genes is generated, each of which then needs to be tested for its relevance to disease by the traditional candidate gene approach. Thus, the study of candidate genes is a vital intermediary common step in the overall research paradigm. In this issue of the journal, Herrmann et al. [1] report on their study focusing on one candidate gene, angiotensinogen, for which they aimed to identify an association with blood pressure phenotypes.Fig. 1: Research paradigm for complex diseases.The case for angiotensinogen Of all the known candidate genes for hypertension, why did the authors choose to study the angiotensinogen gene in particular? Angiotensinogen is an integral part of the renin–angiotensin–aldosterone system (RAAS), which has attracted much renewed interest over the past decades, in particular following the advent of angiotensin-converting enzyme inhibitors and angiotensin receptor blockers and their major influence on the course of hypertension, as demonstrated by innumerable clinical studies. Of all the RAAS components, Hermann et al. [1] chose angiotensinogen as the focus of their study. Is this justifiable? Because circulating levels of angiotensinogen are well below the Km of renin for its substrate, angiotensinogen is likely to be rate limiting in the cascade that leads downstream to the production of angiotensin II. When levels of angiotensinogen increase, angiotensin II production increases. Therefore, angiotensinogen is indeed central in the RAAS cascade and of definite interest as a candidate gene in relation to hypertension, in which the RAAS is thought to play a central role. The case for association studies Once a candidate gene has been identified, the investigator's task remains to provide unequivocal proof for its participation in the pathophysiology of disease. A functional role for a candidate gene can be shown by knockout, knockin, transgenic and congenic-related technologies. However, these technologies can only be applied in experimental models (e.g. in rodents), but not in humans. At present, association studies are the only means to confirm a functional role for a candidate gene in human populations. Even then, a ‘confirmed’ association does not necessarily imply proof of ‘cause'. Studies showing association can at best ‘suggest’ cause. Thus, the study by Herrmann et al. [1] does not provide definitive proof but is, nonetheless, definitely consistent with and suggestive of a pathophysiological role of angiotensinogen or of variations in the gene promoter in essential hypertension in humans. This general limitation of association studies should be born in mind amongst those who seek to unfold the pathophysiological basis of hypertension in humans through genetic studies. The case for haplotype analysis in association studies In their study, Herrmann et al. [1] were looking for an association between haplotypes of the angiotensinogen gene and blood pressure phenotypes. In the past, the traditional candidate gene approach has been based on attempts to detect associations between single polymorphisms within the various candidate genes and the respective phenotypes of interest. The authors quote seven such studies in which an association between AGT polymorphisms and essential hypertension was sought. In three of these studies, a positive association was reported, while no association was found in the other four. On screening PubMed, over 200 publications can be found regarding the link between genetic variants of the angiotensinogen gene and hypertension, demonstrating the widespread interest amongst researchers in this type of study and in the relationship between angiotensinogen and hypertension. One of the first studies to report a linkage between single genetic variants of the angiotensinogen gene and hypertension in humans was by Jeunemaitre et al. [2] in 1992. Many of the studies that followed either confirmed or did not find such an association, thus rendering the association between the angiotensinogen variants and hypertension inconclusive. One possible reason for the discrepant results is that many of the ‘older’ studies used single polymorphism analysis, which may be a too simplistic approach in that it does not take into account allele to allele crosstalk or interaction. Herrmann et al. [1] remind us of this through the use in their study of an alternative approach, that is based not on one single nucleotide polymorphism related to one gene, but rather on haplotype analysis entailing a combination of multiple polymorphisms within one gene. In 1997, Jeunemaitre et al. [3] were again among the first to introduce haplotype analysis of the angiotensinogen gene in essential hypertension. The complexity of haplotype analysis Many researchers may be unfamiliar with the concept of haplotype analysis. If haplotype is defined as a combination of multiple alleles on one chromosome, haplotype analysis refers to the simultaneous study of multiple alleles, as opposed to the study of single alleles at a time. The alleles composing the haplotype are considered polymorphic if a greater than 1% frequency of alleles is found for a particular gene sequence. The application of haplotype analysis for detecting associations is not novel but has not proved attractive to many researchers to date, most likely because of its complexity. To study an association between a single nucleotide polymorphism (SNP) and a phenotype is relatively easy, both statistically and conceptually. Furthermore, it does not require a large population size. To study an association with two nucleotide polymorphisms simultaneously is already more complex. In a previous study investigating two of the angiotensinogen polymorphisms in relation to blood pressure phenotypes in normotensive families, no association was found [4]. In the current study, the authors extended the search for an association using four instead of two polymorphisms, and this time in a hypertensive population. The use of three, four or more SNPs simultaneously significantly complicates the haplotype analysis. The investigators in the current study are therefore to be commended for their audacity in tackling four SNPs simultaneously in the haplotype analysis. However, there is one major problem that arises with such a complex study design. As the number of genotype combinations increases exponentially, the study population needs to be increased in parallel to very large numbers for there to be sufficient power to detect an association. This is well exemplified in the study by Herrmann et al. [1] in which only a few haplotypes could be analysed because of the small number of subjects resulting from the necessary fragmentation of the study population into sub-genotypes, highlighting one severe limitation of the study. The analysis comprising simultaneously looking at all four SNPs as part of the ‘haplotype analysis’ is thus interesting and perhaps a promising approach, but it is unquestionably problematic because of the large potential genomic combinations and the requirement of a very large study population. Another limitation of the study by Herrmann et al. [1] is that it focused on only one gene, the angiotensinogen gene, while disregarding other genes within the RAAS and overlooking potential gene-to-gene interaction. The fact that Herrmann et al. [1] only focused on angiotensinogen may not be surprising considering the complexity of single gene analysis with four SNPs. Multiple gene analysis, each with several variants within each gene, whether in the encoding or the regulatory areas, would have imposed too heavy a burden on the statistical handling of the data and would require a much larger, probably unobtainable, study population size. The case for studying the promoter region in the angiotensinogen gene In their haplotype analysis, Herrmann et al. [1] focused not on the encoding regions of the angiotensinogen gene, but rather on the promoter region. They looked simultaneously at four different promoter haplotypes: alleles C-532T and A-20C, C-18T, and G–6A. These investigators were not the first to focus on the promoter region of the angiotensinogen gene. Nonetheless, their study is important in that it reminds us that, even if associations are not found within the encoding exons of a particular gene, there is a need to look for polymorphisms of functional and regulatory significance within other parts of the gene, whether it be the introns, upstream within the promoter region or downstream. The downside of this realization is that the complexity of gene regulation and the increasing number of functional polymorphisms that continuously accrue, further increase the complexity of haplotype analysis and the difficulty in recruiting sufficiently large study populations to detect association. The case for the phenotype Herrmann et al. [1] report blood pressure data obtained from 327 untreated hypertensive patients. In 212 of these, 24-h blood pressure monitoring was peformed. The authors defined those patients who had never been treated before or who had stopped antihypertensive medications for at least 4 weeks for any reason as ‘untreated'. They ruled out secondary forms of hypertension, leaving only patients with ‘essential hypertension’ within the study group. Based on our current state of knowledge, they could not have done any better in categorizing the blood pressure phenotype. However, is ‘essential hypertension’ in the population at large a homogeneous enough clinical entity to allow association studies of the kind they performed? It is quite likely that, in the perhaps not too distant future, we will have to subdivide this group of ‘essential hypertensive’ patients into sub-entities, and only then we will have realized the futility of our current efforts. And yet, based on the data currently available, we are unable to subdivide the population of ‘essential hypertensives’ at present, and if we wish to carry out association studies with hypertension, then we need to resort to what is likely a heterogeneous population and to work under the most likely false assumption that it is homogenous, for the lack of a better option. To the credit of Herrmann et al. [1], they understood the complexity of blood pressure as a phenotype, beyond the labelling of ‘essential hypertension', and subdivided high blood pressure into its various components: systolic, diastolic, daytime, night-time and 24-h averages. It is noteworthy that, even when they began their analysis by looking at single AGT variants before going to haplotype analysis, they had already detected significant associations relating to systolic blood pressure, diastolic blood pressure and average daytime and night-time blood pressure. These findings are important because they stress the need to study blood pressure not as one amorphic value, but to study the various components comprising the blood pressure phenotype separately, just as the authors did. The quality of their phenotype was unquestionably further improved by adding the ambulatory blood pressure measurements to resting blood pressure, which helped them exclude those patients who did not answer to the criteria defining hypertension. The limitations of association studies in human populations The ultimate target of research seeking the genetic basis of hypertension is, and always has been, human populations. However, it should be realized that there are major limitations to association studies performed in humans. One limitation is that, as discussed above, an association does not prove cause. In fact there is no additional experimental tool that allows us to proceed in human populations beyond the positive association in terms of proof causation. This is the point at which there is a need to move away from human studies back to experimental animal work, which, for lack of a better option, can provide proof of causation. The second problem is that human populations tend to be heterogeneous and an association that is detected in one population may not be detected in another. Thus, the association that was found in Caucasians by Herrmann et al. [1] may not necessarily be universal. It is noteworthy that many of the positive associations previously reported with regard to the relationship between angiotensinogen and hypertension were found in Japanese and Chinese populations, and less so in Caucasians. Conclusions Where does all this leave us? In terms of the pathophysiology of hypertension, the possible association of the angiotensinogen protein with hypertension has been known for a decade or more, if not in all then at least in some populations, and thus the information conveyed in this study is not novel. The importance of the study by Herrmann et al. [1] lies primarily in that it stresses the complexity of diseases such as hypertension and that the commonly reported search for associations between single gene variants and hypertension may be too simplistic. This study is also of major importance in that it helps highlight the well-known concept that variants in the regulatory regions, and not necessarily in the encoding regions, may be just as important in modulating the phenotype (i.e. blood pressure in this case). The haplotype analysis approach, although much more complicated and requiring significantly larger study populations than the traditional single gene association studies, appears to be more logical and comprehensive, and thus stands a better chance of success in achieving the goals set forth by this type of research.