Experimental Methods and Materials

As with the ACE gene, a number of approaches have been adopted to evaluate the hypothesis that the AGT gene plays a role in BP control and each will be considered in-turn. With respect to these approaches, linkage analysis, cross-sectional prevalence analysis and meta-analyses have all been performed in large studies to evaluate the role of the AGT gene. However, as yet, large prospective studies with incidence data have not been reported on. As with the ACE gene, although a number of genome-wide association or linkage studies have been performed which incorporate markers for the AGT gene as part of the “gene chips”, these will not be discussed for the reasons previously indicated.

That is, to achieve statistical power when employing Bonferroni or modified Bonferroni corrections to adjust for what is often thousands of comparisons in these studies, would require probability values which are so large that few loci contributing to a polygenic effect would achieve significance. As with the ACE gene, again with respect to cross-sectional data, although a number of small studies have been published, I will only review the outcomes of large population-based studies (>1000 participants), as these are less likely to produce false positive or false negative outcomes.

Strong linkage and family-based association data both verify and refute a role for the AGT gene in contributing toward BP. In this regard, there are four studies that support linkage of the AGT gene to hypertension (Jeunemaitre et al 1992, Caulfield et al 1994, Caulfield et al 1995, Baker et al 2007) and five studies that do not support linkage or family-based associations (Brand et al 1998, Niu et al 1999, Niu et al 1998, Niu et al 1999^b, Wang et al 1999). In linkage studies, a wide range of sample sizes have been

employed. In this regard, one of the studies that failed to show linkage (Brand et al 1998) had an impressive sample size (n=350 families and 630 affected sibling pairs) and most likely included data from a number of the groups that had previously reported on linkage or a lack thereof. A limitation of the large study that failed to show linkage (Brand et al 1998) was that the method of analysis employed is sensitive to allele frequency estimations when the parental genotypes are not known. Nevertheless, in that study (Brand et al 1998) the same analytical methods were employed as in most of the studies demonstrating linkage between the AGT gene and hypertension (Jeunemaitre et al 1992, Caulfield et al 1994, Caulfield et al 1995). Moreover, in support of this large study that failed to show linkage of the AGT gene to hypertension (Brand et al 1998), using analytical techniques which do not rely on precise specifications of the genetic models, in 335 hypertensives from 315 nuclear families, a second group failed to show excess transmission of AGT gene variants to hypertensive Chinese (Niu et al 1999, Niu et al 1999^b). What should nevertheless be underscored is that unlike the ACE gene where linkage to diastolic, but not systolic BP was identified, presently there are no linkage studies that have attempted to assess the contribution of the AGT gene to either systolic or diastolic BP considered as separate traits and which pathophysiologically are to some extent driven by separate mechanisms.

Similar to the linkage data, large (>1000 participants) cross-sectional and association studies provide evidence both for and against a role for the AGT gene in contributing toward BP. Variants of the AGT gene have been shown to be a risk factor for hypertension in 9100 randomly selected participants of the Copenhagen Heart Study (this association was in women only) (Sethi et al 2001) and in 2461 participants sampled

from the general population in Italy (Castellano et al 2003). A role for the AGT gene in BP control is also supported by an association between AGT genotype and ambulatory pulse pressure in 1425 individuals from 248 families each with one proband (Baker et al 2007), and home BP measurements in 1245 individuals randomly selected from the general Japanese population (Matsubara et al 2003). A confounding characteristic of these studies is nevertheless that the authors used different AGT gene markers to show these effects. These markers included the T→M at amino acid position 174 [C→T substitution at nucleotide position +521 (Sethi et al 2001), the M235T variant (Castellano et al 2003), T+31C (Matsubara et al 2003), and the C→T substitution at nucleotide position -532 (Baker et al 2007) within the AGT gene. Thus, the exact causal variant of the AGT that contributes toward BP control still remains uncertain. Last, recent analysis of the Framingham data in 2155 participants shows an association between AGT genotype and BP using the general estimations equation (Levy et al 2007). However, this analysis was part of a much larger study intended to generate rather than test hypotheses and hence the outcomes were not corrected for multiple genotyping.

As with the major studies showing positive associations, there are similarly key studies which have failed to show a relationship between the AGT gene and BP. In 4322 participants of the NHLB Family Blood Pressure Program, (which nevertheless pooled data from different study designs including sib-pair, sib-ship and extended family designs)(Province et al 2000), and in 904 randomly selected African-Americans (Larson et al 2000), no relationship between G→A ( -6) polymorphism and hypertension was noted. Furthermore, despite showing strong heritability estimates for the BP values obtained, in 1006 participants of a genetically isolated Dutch population sample, no

relationship between the M235T variant and BP was noted (van Rijn et al 2007). Last, in a large case-control study in 1358 participants, participants homozygous for the T allele of the AGT M235T variant even had a decreased risk for hypertension (Mondry et al 2005).

Three meta-analyses conducted on a substantial number of small and large case-control studies, with overall large total sample sizes (n=5493-18704), support a role for the AGT gene in contributing toward hypertension (Sethi et al 2003, Staessen et al 1999, Kunz et al 1997, Kato et al 1999). These meta-analyses nevertheless indicate that a publication bias in favour of positive outcomes may have occurred.

2.2.2.5 Potential explanations for discrepancies in data assessing the