Renin???angiotensin???aldosterone system polymorphisms and essential hypertension

Although a large number of studies have investigated the possible effect of polymorphisms of the renin–angiotensin–aldosterone system (RAAS) in arterial hypertension, most of the published results do not provide a clear picture of the effect of each individual gene taken alone or in combination. In the October issue of the journal, the GENIPER Project's investigators [1] tested this effect in a large number of carefully selected hypertensive subjects and controls (n = 2461) recruited in 13 Italian centres. They found small, but statistically significant, associations with similar trends in the several Italian regions. More interestingly, they provide arguments for an additive effect of the individual variants of the genes studied, although no synergistic interaction was observed. As in other studies addressing the possible relationships between genetic polymorphisms and complex traits, several questions need to be asked to evaluate the strengths and limitations of the presented results. Is the study reliable in its setting and its results? In case of replication, how do these results compare to those already published in the literature? What is their contribution to the available scientific knowledge? One of the main issues related to case–control studies is statistical power. This power is not evaluated in most reports despite the fact that its computation is not difficult [2]. Evaluation of statistical power is important because studies testing the replication of a given association in independent samples are not often satisfying the level of power required to detect the initially described association, as demonstrated by Lalouel and Rohrwasser for the angiotensinogen (AGT) gene [3]. Statistical power depends not only on the number of cases and controls, but also on each allele frequency and on the a priori genetic hypothesis on the mode of transmission of the effect. It would have been interesting to have such information for each of the genes tested in the study reported by Castellano et al. [1], although reasonable numbers of well-selected cases and controls are undoubtedly achieved in that particular report. The second issue that frequently arises from multicentre studies is the possible relative heterogeneity of the population studied, both on its genetic background and on the environmental pressure to which it can be submitted. In that respect, it was surprising to observe a significant North–South difference in the angiotensin I-converting enzyme (ACE) I/D polymorphism allele frequency but not in the other polymorphisms tested. This difference would not have been surprising for the M235T (identified as T/C 704 in this report) polymorphism of the AGT gene, which is particularly sensitive to the ethnic background [4]. Differences in the ACE D allele frequency have also been previously observed in Caucasians compared to African and Asian populations, but usually not in a single country. Whether the difference observed in the GENIPER project between the North (0.61) and the South (0.67), including Sardinia, reflects a particular genetic ancestry remains to be elucidated. Nevertheless, this mild geographical difference in allele frequency did not account for the lack of association observed with that particular allele, the results being homogeneous across the regions. Finally, it is also important to note that the genotyping was performed with appropriate techniques and repeatability. In that respect, the web access to the genotyping procedures, the efforts made for intra-repeatability of the assays within the same laboratory and across laboratories, the double-blind scoring of each genotype, and the additional systematic amplification of the DD genotypes with a technique that recognizes the insertion-specific sequences, are witnesses of the quality of the genotypic measurements, which are not often given in case–control studies [5]. The use of a checklist, such as that proposed by Little et al. [6] for appraising studies of gene-diseases associations, should increase the quality of the studies submitted for publication and facilitate the reviewers evaluation. Are the results comparable to previously published similar reports? The fact that the ACE I/D polymorphism gave negative results is not surprising because the majority of studies have found no association between that gene or locus and essential hypertension [7]. Positive associations have been observed in men in two large studies [8,9] but have not been documented further. However, for this particular genem, as for the others, one of the limits of the GENIPER results is the analysis of a single polymorphism at each given gene. Analysis of one polymorphism gives information on that particular genetic variation but does not allow a more general conclusion to be drawn for the gene itself. More than 100 varying sites have now been detected at the ACE locus that resolve in approximately 15 distinct haplotypes [10]. Even though the ACE I/D polymorphism explains a large part of the variance of plasma ACE concentration [11], it is only a neutral marker in strong linkage disequilibrium with the functional variant that appears to be located in the 3′ part of the gene. In addition, the overall genetic variance of serum ACE is now explained by two major intragenic sites which may act through additive effects [12]. Thus, only one part of the D allele will bear one or two functional variant(s) that impact on ACE tissular expression and possibly on essential hypertension. Genotyping a reasonable number of non-redundant polymorphisms (approximately five) at a given gene gives a more reasonable chance of testing the implication of the entire gene in a particular trait. The impact of the AGT gene on arterial hypertension, which was confirmed by the GENIPER Project's investigators, is probably (with the alpha-adducin) one of the most solid among the multitude of candidate genes and polymorphisms that have been tested to date [4]. As expected, the odds ratio of patients bearing the 704T (or 235T in protein sequence) allele for being hypertensive is relatively low (1.35), and is consistent with studies compiled in several meta-analyses [13]. Whether this mild effect is (or not) dependent on age, sex and pregnancy, and whether it is also related in part to the impact of angiotensinogen on adipocytes and body mass index (which was greater in hypertensive subjects than controls in the GENIPER report), has not yet received a convincing answer. As for the ACE gene, a large number of polymorphisms have been detected at the AGT gene. The C/T 704 (M235T) polymorphism at the AGT gene is in strong linkage disequilibrium with other nucleotide substitutions in exon 2 (T174M), intron 1 and in the 5′ part of the gene [14]. Several of them might affect, alone or in combination, the transcription of the gene and thus AGT expression [15]. The association detected with the M235T allele could then reflect an association with one or the other polymorphism(s). In that respect, a large-scale population-based study in Denmark emphasized the association of the T174M polymorphism on plasma AGT gene concentration and arterial hypertension [16]. Dividing the effect of one genotype into more complex haplotypes provides a chance to observe what is going on in real life (i.e. the possible effect of several polymorphisms acting in combination at a given locus) [17]. The results observed by the GENIPER Project's investigators with the other polymorphisms are more provocative. They found a mild association with well-known polymorphisms of the angiotensin I-receptor (AT1R) and CYP11B2 genes and hypertension. However, the so-called ‘at-risk’ alleles (A1166 and C-344, respectively) detected in their study have been defined as low-risk alleles in several other studies. This is particularly true for the A1166C polymorphism of the AT1R gene, for which the C allele has been associated with hypertension in the initial report by Bonnardeaux et al. [18] as well as with acute blood pressure response to angiotensin II [19] and other cardiovascular phenotypes [20]. However, some of these associations have been disputed in other studies, among them a previous study from the North of Italy [21]. The inverse relationship reported here confirms that this polymorphism does not explain a reliable genetic part of essential hypertension and the possibility of spurious results in case–control studies. By its role in the final step of the aldosterone synthesis, the CYP11B2 gene encoding the aldosterone synthase is also an attractive candidate for human hypertension. It is located on chromosome 8q, in a region possibly linked to systolic blood pressure [22]. A polymorphism was identified in the promoter region of the gene (C-344T), located at a steroidogenic transcription factor SF-1 binding site that could play an albeit non-proven role in the expression of the gene [23]. Together with this putative function, the high allele frequency of this polymorphism (0.45–0.55) has stimulated numerous case–control studies in essential hypertension and primary aldosteronism. Some initial positive reports with these clinical phenotypes, or with plasma aldosterone concentration, suggested that the –344T allele could favour sodium retention and high blood pressure [24]. However, contradictory results have also rapidly been obtained according to the populations and the biochemical and clinical parameters studied, and associations have also been found with the –344C allele [25]. The GENIPER results are in favour of a mild association with hypertension, with an odds ratio of 1.3 for being hypertensive for subjects bearing the –344T allele. Whether this association represents a true meaningful effect in the general population, such as an alteration of the adrenal biosynthesis in essential hypertension, or reflects a random balance of negative and positive results, will require further studies focusing on intermediate biological phenotypes [26]. Probably, one of the most interesting parts of the GENIPER report relies on the search for interactions between these genetic polymorphisms. In a first analysis, the combination of at-risk genotypes led to an odds ratio for hypertension similar to those observed for each single polymorphism, thus not showing any additive effect. However, another statistical approach based on the use of each genotype combination, the number of at-risk alleles and the corresponding odds ratio, showed an interesting significant trend towards a higher risk of hypertension for genotype combinations with more than three at-risk alleles. A gene–gene interaction is logical if several genes are considered as belonging to the same enzymatic cascade, and this has been suggested by some reports [27,28]. Although these studies face the problem of stratification, and thus the risk of an increased probability of false positive results, the use of sophisticated statistical methods [29] and replication of the results in other well-performed and powerful studies should help to better estimate the role of the RAAS polymorphisms in essential hypertension. In conclusion, the GENIPER study suggests a possible mild and interacting impact of the RAAS genetic polymorphisms on blood pressure regulation in the entire population. The weakness of their combined effect might be explained by the fact that the phenotype (hypertension as a qualitative trait, basal blood pressure) corresponds to a non-accurate phenotype/measurement that has not been determined in rigourous standardized environmental conditions. If it is assumed, as is probably the case, that these polymorphisms exert their effect only in certain conditions, a ‘stress-the-genotype’ approach might help us to unravel their true pathophysiological influence. This approach is well illustrated by the studies performed by Montgomery et al. [30] to evaluate the effect of the ACE I/D polymorphism on cardiovascular response to exercise. Some studies also suggested that the AGT genotype might influence the blood pressure response to non-pharmacological therapy, especially as a marker of salt-sensitivity [31,32]. Undoubtedly, the use of this ‘stress-the-genotype’ strategy to explore gene–environment interactions might also be the key to evaluating the complex effects of RAAS genetic polymorphisms.