Genetics of blood pressure: time to curate the collection

Abstract

The genetics of blood pressure (BP) is all about discovery and understanding, but it is certainly not for the faint hearted. Despite heroic effort, the question we posed nearly 15 years ago [1] regarding the whereabouts of BP genes remains largely unanswered. It has not been an easy journey. The early candidate gene studies in rats [2] and humans [3] fell out of favor because of inconsistency between studies and the fact they were small in size or scope. The exciting leap from the gene to the genome and the advent of genome-wide association studies (GWAS) was hoped to solve the previous difficulties. Their unbiased interrogation of the genome using millions of markers in thousands of individuals was expected to have the oomph to unearth the location of the DNA variants that would explain the 40% of BP variability commonly attributed to genes (i.e. heritability). Unfortunately, the early GWAS essentially failed to locate anything of substance for BP [1,4]. The response was to build bigger and bigger conglomerates to study hundreds of thousands of individuals and use denser genome maps [5]. As a result, three lessons were learned. First, as with candidate gene analyses, giant genomic studies are not immune from inconsistency and disagreement. Second, the polygenic nature of BP was affirmed by the detection of multiple (about 80) common variants (each present in >5% and typically around 30% of the population), but each estimated to have unmeasurable (≤1 mmHg) effects on individual BP. Third and most surprising, instead of the expected 40%, all known GWAS variants together accounted for only about 3% of BP variance. It was little compensation that BP was not the only phenotype for which GWAS had failed to discover the variants to explain heritabilities and the concept of ‘missing heritability’ was born. There are many possible explanations for the mysterious missing heritability, but one fashionable explanation is that rare variants might be the missing pieces of the jigsaw [6]. This is conceivable, as rare variants slip through the net of a GWAS scan because it is designed to detect common variants. Therefore, the search for low frequency (1–5%) and rare (<1%) variants began. The first targets were the exons or protein-coding regions of genes. Early studies in and adjacent to the exons of three candidate BP genes discovered 30 different rare variants (frequency <0.05%) that were estimated to affect BP by a measurable decrement – lowering BP by about 6 mmHg compared with controls [7]. If three genes could yield 30 mutations from 3000 or so individuals [7], imagine what screening all 20 000 genes in tens and hundreds of thousands of study participants could reveal. So, moving from the exon to the exome – the 1–2% of the genome comprising all the exons – was a natural step made possible by rapidly advancing molecular technology. The study appearing in this issue by Nandakumar et al.[8] and two recent others [9,10] are among the first to put the exomes to the test for BP. Before passing judgement, it needs to be appreciated that these analyses are not the gold standard for exomic interrogation. Ideally, this would involve deep sequencing of the exomes, an approach underway for a variety of complex diseases [6]. However, costs are still relatively high and instead it is more economical for very large studies to use custom genotyping arrays known as exome chips. But the trade-off is a more limited survey of the exome. Also, the DNA variants on the array were derived from mostly European sequences and their utility in non-Europeans is less certain. Finally, the exome array comprises mostly low and rare variants, but 11% of the 250 000 chip variants are common, including almost 6000 from the GWAS catalogue. The two largest exomic analyses were conducted in populations composed mostly of European ancestry and together they studied nearly 700 000 individuals [9,10]. The sum of discovery from these efforts was 61 new BP loci – some for systolic, some diastolic, some pulse pressure, some increasing BP, and some lowering it. Yet, despite the majority of the array variants being either low frequency or rare, all but six of the 61 new loci were common variants with an average population frequency of 34% and a mean estimated BP effect of 0.25 mmHg. The remaining six loci were either low frequency or rare variants, with an average estimated BP effect of 2.2 mmHg. The overall picture looked more like another GWAS and could not have been further removed in either nature or number of variants expected from the prior screening of exons in candidate genes [7]. And frustratingly, still the vast majority of heritability remained missing. The Nandakumar et al.[8] study provides a slightly more encouraging perspective. Conducted exclusively in study participants of African ancestry (but relying on the essentially European array) the discoveries were exclusively rare variants for 10 genes, many of which are plausible candidates in a fun