Genome-wide mapping of human loci for essential hypertension

MECHANISMS OF DISEASE

Mechanisms of disease

Genome-wide mapping of human loci for essential hypertension Mark Caulfield, Patricia Munroe, Janine Pembroke, Nilesh Samani, Anna Dominiczak, Morris Brown, Nigel Benjamin, John Webster, Peter Ratcliffe, Suzanne O’Shea, Jeanette Papp, Elizabeth Taylor, Richard Dobson, Joanne Knight, Stephen Newhouse, Joel Hooper, Wai Lee, Nick Brain, David Clayton, G Mark Lathrop, Martin Farrall, John Connell, for The MRC British Genetics of Hypertension Study

Summary Background Blood pressure may contribute to 50% of the global cardiovascular disease epidemic. By understanding the genes predisposing to common disorders such as human essential hypertension we may gain insights into novel pathophysiological mechanisms and potential therapeutic targets. In the Medical Research Council BRItish Genetics of HyperTension (BRIGHT) study, we aim to identify these genetic factors by scanning the human genome for susceptibility genes for essential hypertension. We describe the results of a genome scan for hypertension in a large white European population. Methods We phenotyped 2010 affected sibling pairs drawn from 1599 severely hypertensive families, and completed a 10 centimorgan genome-wide scan. After rigorous quality control, we analysed the genotypic data by non-parametric linkage, which tests whether genes are shared in excess among the affected sibling pairs. Lod scores, calculated at regular points along each chromosome, were used to assess the support for linkage. Findings Linkage analysis identified a principle locus on chromosome 6q, with a lod score of 3·21 that attained genome-wide significance (p=0·042). The inclusion of three further loci with lod scores higher than 1·57 (2q, 5q, and 9q) also show genome-wide significance (p=0·017) when assessed under a locus-counting analysis.

Clinical Pharmacology and Barts and The London Genome Centre, William Harvey Research Institute, Barts and The London, Queen Mary’s School of Medicine, Charterhouse Square, London EC1M 6BQ, UK (Prof M Caulfield FRCP, P Munroe PhD, J Pembroke RGN, R Dobson BSc, J Knight PhD, S Newhouse MSc, J Hooper BSc); Cardiology, University of Leicester, Glenfield Hospital, Leicester, UK (Prof N Samani FRCP); Division of Cardiovascular and Medical Sciences, University of Glasgow, Western Infirmary, Glasgow, UK (Prof A Dominiczak FRCP, W Lee PhD, N Brain MSc, Prof J Connell FRCP); Clinical Pharmacology and the Cambridge Institute of Medical Research, University of Cambridge, Addenbrooke’s Hospital, Cambridge, UK (Prof M Brown FRCP, D Clayton MA); Medicine and Therapeutics, Aberdeen Royal Infirmary, Aberdeen, UK (J Webster FRCP); Nuffield Department of Clinical Medicine and Department of Cardiovascular Medicine, University of Oxford, Wellcome Trust Centre for Human Genetics, Oxford, UK (Prof P Ratcliffe FRCP, S O’Shea DPhil, E Taylor PhD, M Farrall FRCPath); UCLA School of Medicine, Los Angeles, CA, USA (J Papp PhD); and Centre National de Genotypage, Evry, France (Prof G M Lathrop PhD) Correspondence to: Prof Mark Caulfield (e-mail: [email protected])

2118

Interpretation These findings imply that human essential hypertension has an oligogenic element (a few genes may be involved in determination of the trait) possibly superimposed on more minor genetic effects, and that several genes may be tractable to a positional cloning strategy. Lancet 2003; 361: 2118–23 See Rapid review page 2149

Introduction Hypertension is a common, modifiable risk factor for coronary artery and cerebrovascular diseases. These diseases are major causes of morbidity and mortality, accounting for more than 12 million deaths annually worldwide and incurring a huge concomitant burden on health-care resources (http://www.who.int/en/index. html). Segregation and twin analyses show that a third of the between-individual variation in blood pressure is heritable in human beings.1 Some insights into the likely genetic architecture of human hypertension can be gained from studies of rat QUANTITATIVE TRAIT LOCI, in which a series of blood-pressure genes has been mapped.2 In human beings, the molecular basis of several, rare Mendelian traits in which hypertension is a major feature have been identified.3,4 However, identification of individual genes contributing to common essential hypertension has proved more difficult. The Medical Research Council BRItish Genetics of HyperTension (BRIGHT) study aims to identify genes that confer susceptibility to essential hypertension by doing a genome-wide linkage screen. We describe the results of our assessment of extreme-concordant sibling pairs to map loci for human blood pressure with a GENOME-WIDE SCAN among white European families.

Methods Family ascertainment We recruited severely hypertensive families from the Medical Research Council General Practice Framework and other family-physician practices in the UK, based on affected sibling pairs of white British ancestry up to the level of grandparents. We obtained ethics-committee approval from local research committees of the partner institutes, and participants’ fully informed written consent. Each family contained at least two affected siblings, in whom onset of hypertension was diagnosed before age 60 years and who had sitting blood-pressure values of 150/100 mm Hg or higher, based on one reading, or 145/95 mm Hg or higher as a mean of three readings. These criteria correspond to the threshold for the top 5% of blood-pressure distribution in a contemporaneous health screening survey of 5000 UK men and women in 1995 (N Wald and M Law, personal communication). We excluded hypertensive individuals THE LANCET • Vol 361 • June 21, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.

MECHANISMS OF DISEASE

GLOSSARY GENOME-WIDE SCAN

Use of evenly spaced markers throughout the genome to identify disease regions. LINKAGE DISEQUILIBRIUM

Also referred to as allelic association, this is the occurrence of alleles at two loci more frequently than expected given the known allele frequencies and recombination fraction between the two loci. LOCUS

Chromosomal location of a gene or particular DNA sequence. LOD SCORE

Lod scores are used to assess the statistical support for linkage and are defined as the common logarithmic likelihood ratio of the data under linkage (alternative hypothesis) and independent segregation (null hypothesis) models. MICROSATELLITE

A DNA sequence where short sequences are repeated in tandem arrays. QUANTITATIVE TRAIT LOCI

A chromosomal region which contains a gene or genes responsible for the quantitative trait under investigation. SINGLE-NUCLEOTIDE POLYMORPHISM

A DNA sequence variation that occurs when a single nucleotide (A, T, C, or G) in the genome sequence is changed.

who: self-reportedly consumed more than 21 units of alcohol per week; had diabetes; had intrinsic renal disease; had a self-reported history of secondary hypertension that was corroborated by the family physician; or had coexisting illness. We aimed to recruit hypertensive individuals with body-mass index less than 30 kg/m2. Phenotyping and genotyping Data on personal medical history, socioeconomics, genealogy, family history of cardiovascular disease, current medication, and prediagnosis blood pressures were collected from all participants by nursing teams in six centres. Blood pressures at phenotyping were recorded with the Omron HEM-705CP portable, semiautomated, oscillometric device (Omron Healthcare, Mannheim, Germany), and 24 h ambulatory blood pressures were measured with Spacelabs, model 90207 (Spacelabs, Wokingham, UK). The anthropometric measurements taken were waist-to-hip ratio, skin-fold thickness, and body-mass index. The biochemical variables measured were 24 h creatinine clearance, urinary electrolytes, and albumin-to-creatinine ratio. Each affected sibling underwent electrocardiography at phenotyping. All measurements were recorded by standard operating procedures (http://www.brightstudy.ac.uk). For genotyping, DNA was extracted from whole blood by a modified salting-out method, and quantified with a picogreen assay. The Linkage Marker Set MD 10 (Applied Biosystems, Foster City, CA, USA) formed the core marker set for the genome-wide screen. These markers are distributed at an average marker density of 10 centimorgan ([cM] roughly every 10 million bases in the genome). This set was supplemented with MICROSATELLITES D1S2660, D5S1953, D14S1044, D18S468, D19S886, and D19S894 from the Applied Biosystems HD 5 Marker Set to replace failed MD 10 markers. Microsatellite markers D5S1969, D5S427, D5S398, D5S1359, D5S2089, D5S2019, D5S1988, D5S1977, D5S646, and D5S672 on chromosome 5 were

genotyped for preliminary grid tightening in regions highlighted from interim analysis, thereby extracting more information. Before statistical analysis, we did rigorous genotype quality assurance to ensure accurate binning of alleles. All phenotypic data were stored in a purpose-written relational database, in which all anonymised phenotypic data are maintained according to family structure. Genotype data were stored in a separate relational database, with query tools to run data quality checks on genotype consistency and accuracy of control genotypes within the BRIGHT dataset. Statistical analysis We estimated marker allele frequencies by maximum likelihood, with use of the SPLINK (version 1.9) computer program,5 and checked for convergence by simple counts. A kinship analysis,6 done with the RELPAIR program (version 0.90), identified several halfsibling pairs that had been collected and recorded as fullsibling pairs. The family structures were revised accordingly before Mendelian consistency checks were made with the Pedcheck7 program; inconsistent genotypes were subsequently marked as missing. The genetic map for the MD 10 markers and the supplementary markers in regions of interest were derived from the sex-averaged maps published by the Co-operative Human Linkage Centre (http://gai.nci.nih.gov/CHLC) and the Centre for Medical Genetics (http://research.marshfieldclinic.org/ genetics/). We did non-parametric linkage analysis with the computer programme Merlin (version 0.9.0),8 in combination with MLSix to compute multipoint maximum LOD SCORE statistics9,10 for affected full-sibling and half-sibling pairs at regular intervals along each chromosome. In regions of linkage, the peak lod score is associated with a measure (recurrence risk ratio in full siblings ␭sib) of the genetic effect size for this susceptibility gene. We calculated the 95% CI for the ␭sib estimate with a bootstrapping technique in which the identity by descent vectors for sibling relationships were randomly resampled with replacement; lod scores were maximised for each resample and the ␭sib estimates were recorded and sorted and the 2·5 percentile point and the 97·5 percentile point of the series were reported. We used a computer simulation exercise to calculate critical maximum lod-score thresholds for genome-wide significance at the 5% level, thereby implementing a LOCUS-counting strategy.11 From the BRIGHT dataset, 1000 replicates were generated by gene dropping with the Merlin programme.8 This procedure models the precise family structures, patterns of absent genotypes, marker locations, and allele frequencies noted in the experimental dataset. These replicates were generated under the null hypothesis of no linkage; maximum-lod-score statistics were computed with Merlin and MLSix, and the results from each genome-wide scan were processed to count the number of independent linked regions greater than the desired maximum-lod-score threshold. Independent linked regions were defined as regions of linkage that were separated by at least 40 cM. The 5% significance threshold for n-independent linked regions is defined as the maximum-lod-score statistic for which 50 of the 1000 replicates included n or more independent linked regions. A significant excess of independent linked regions in the experimental data is indicated if the maximum-lod-score statistics exceed the appropriate threshold. We used several interlinking programs to collate feature data on our four main regions of interest. Our

THE LANCET • Vol 361 • June 21, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.

2119

MECHANISMS OF DISEASE

Demographics Phenotypic feature Mean (SD) age (years) At diagnosis At phenotyping Blood pressure Mean (SD) prediagnosis systolic (mm Hg) Mean (SD) prediagnosis diastolic (mm Hg) Mean (SD) systolic at phenotyping (mm Hg) Mean (SD) diastolic at phenotyping (mm Hg) Proportion with systolic in top 5% for age (%) Proportion with diastolic in top 5% for age (%) Median (IQR) body-mass index (kg/m2) Proportion with body-mass index >30 kg/m2 (%) Median (IQR) waist-to-hip ratio Men Women

48 (11) 64 (9·3) 172 (18) 104 (8·8) 156 (21) 94 (11) 97 93 27 (25–30) 21 0·94 (0·90–0·98) 0·82 (0·78–0·87)

Table 1: Demographics of 1599 families phenotyped and genotyped

software retrieved RefGene, mRNA, and expressed sequence tag accessions, SINGLE-NUCLEOTIDE POLYMORPHISMS (Single-Nucleotide Polymorphisms Consortium and National Institutes of Health), and microsatellite locations and positional information relative to latest freeze of the UCSC genome annotation (Golden Path, UCSC Genome, Bioinformatics, http://genome.ucsc. edu) and captured ENSEMBL annotations to facilitate cross-referencing. Role of the funding source The sponsor of the study had no role in the design of the study, data collection, data analysis, data interpretation, the writing of the report, or the decision to study.

Results We phenotyped 3599 individuals from 2010 affected relative pairs, identified in 1599 families. We successfully identified people who had severe hypertension at diagnosis (table 1). Many of the participants had remained hypertensive on treatment and not reached British Hypertension Society targets (<140/85 mm Hg) at phenotyping. 21% of participants had body-mass index values higher than 30 kg/m2, most of whom had been identified as lean hypertensives at diagnosis, but by the time of recruitment their body-mass index had increased. In the genome scan of 1599 families we identified one chromosomal region with a lod score of more than 3·14 (6q, maximum lod score 3·21, table 2). Computer simulations suggest that this result is significant at the genome-wide level after allowance for multiple testing (p=0·042, table 3). In addition, three regions had lod scores higher than 1·57 (2q, 5q, and 9q, table 2). Computer simulations for a locus-counting analysis suggest that these linkages collectively attain genomewide significance (p=0·017, table 3),11 which makes falsepositive signals unlikely. We reanalysed data excluding sibling pairs in which one or two siblings had body-mass index of 30 kg/m2 or greater. The maximum lod scores for chromosomes 6, 5, and 9 were similar to those Maximum lod score Chromosomal location 2q 1·76 5q 1·85 6q 3·21 9q 2·24

Nearest marker

Interval size (cM) ␭sib (95% CI)

D2S142 D5S2019 D6S281 D9S290

25 27 16 25

1·05 (1·01–1·10) 1·05 (1·02–1·13) 1·09 (1·04–1·16) 1·15 (1·04–1·26)

Table 2: Summary of genome-wide screen results on 1599 families

2120

Estimated maximum-lod-score Observed maximum threshold lod score Number of independent linked regions 1 3·14 2 2·20 3 1·80 4 1·57

3·21 2·24 1·85 1·76

p

0·042 0·035 0·043 0·017

Table 3: Locus counting analysis

calculated for the complete dataset (6q 3·13, 5q 2·27, and 9q 2·12); the lod score on chromosome 2q dropped to 0·98 and the fourth largest maximum lod score mapped to chromosome 12 (1·78). Therefore, the locuscounting analysis of the reduced dataset was significant genome-wide. The interim analyses enabled us to direct the placement of additional markers around maximumlod-score peaks. This has enabled us to do preliminary grid tightening to a density of 2 cM under the region of interest on chromosome 5q; the lod score remained similar (1·85). We have computed a genome-wide exclusion map to monitor our progress in mapping genes for hypertension (figure). By use of a strict criterion of maximum lod score –2, we have excluded linkage from 76% of the autosomal genome for a susceptibility gene that accounts (under a multiplicative model of gene interaction) for a third (␭sib =1·21) of the total familial clustering (␭sib=1·78) calculated under a multifactorial threshold model for families ascertained by the BRIGHT criteria. A complementary exclusion map can be computed with bootstrap estimates of the 95% CI for ␭sib associated with the maximum lod score for each location. With this criterion, 99% of the genome is excluded for a gene that accounts for a third of the predicted familial clustering and 74% of the genome for a gene that accounts for a fifth of the predicted clustering (␭sib =1·12).

Discussion The achievement of genome-wide significance has proved elusive in many linkage studies of complex traits.12,13 Our results are, therefore, encouraging, given the prevalence of hypertension and the low familial recurrence risk. Our genome-wide scan for hypertension differs from others because we focused on large-scale recruitment of severely affected sibling pairs who had early-onset hypertension; this feature might identify families with strong genetic influences for hypertension. By choosing a stringent threshold of the top 5% of bloodpressure distribution, we enriched our population for linkage information to blood-pressure loci. The impact on power by relaxing this stringency to the top 20% is striking: 6570 affected sibling pairs or 18 380 unselected affected sibling pairs drawn randomly from the general population would be required to map a locus with 6% heritability. Most of our families are based on affected sibling pairs with only 8% of families, including parents and some with additional unaffected siblings. We calculate that we have extracted 64% of the maximum linkage information that such family units can yield, which suggests that gridtightening within our regions of interest should form a profitable part of a future fine mapping strategy. We did no subgroup analysis for the primary linkages. We intend to take a unified approach that avoids some of the difficulties of multiple subgroup analysis, especially in terms of weakening in power.13 We will introduce covariates for identity by descent-sharing within a logistic regression framework. The likelihood for such models may be calculated from posterior identity by descent-

THE LANCET • Vol 361 • June 21, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.

MECHANISMS OF DISEASE

Chromosome 1

Chromosome 9

Chromosome 17

Chromosome 2

Chromosome 10

Chromosome 18

Chromosome 3

Chromosome 11

Chromosome 19

Chromosome 4

Chromosome 12

Chromosome 20

Chromosome 5

Chromosome 13

Chromosome 21

Chromosome 6

Chromosome 14

Chromosome 22

Chromosome 7

Chromosome 15

Chromosome X

Chromosome 8

Chromosome 16

0

0

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

MLS MLSX

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

Lod score/␭sib

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

0 3·5 3·0 2·5 2·0 1·5 1·0 0·5 0

50 100 150 200 250 300 350 Location (Kosambi cM)

50 100 150 200 250 300 350 Location (Kosambi cM)

50 100 150 200 250 300 350 Location (Kosambi cM)

Maximum-lod-score plots for all chromosomes MLS=maximum lod score. MLSX=Exclusion limit presented as genetic effect size, scaled as a recurrence risk ratio for siblings (␭sib) excluded at lod-score threshold of ⫺2·0. Each panel shows the chromosome number with the genetic length of the chromosome in cM on the abscissa.

THE LANCET • Vol 361 • June 21, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.

2121

MECHANISMS OF DISEASE

sharing probabilities under the hypothesis of no linkage.14,15 By use of extensions of this approach we propose to do further analyses at points in the genome where analysis suggests linkage.15 The advantage of this approach is to investigate which shared phenotypic characteristics that contribute to linkage signals are possessed by families. The genome-wide screens published to date for hypertension genes are diverse in phenotype, ethnic origin, selection criteria, numbers and structures of families, and range from analysis of single large pedigrees to sibling-pair resources of up to 599 families.18–21 The US Family Blood Pressure Program investigators have published genome scan data from four partner networks that used distinct phenotyping strategies and reflect the ethnic demography of the US population.18–21 In these interim publications no individual network-based genome-scan or the meta-analysis attained genome-wide significance. In several other studies linkage to several chromosomal regions has been suggested. There is a region of overlap with our findings on chromosome 2 (140–170 cM). This region was linked to hypertension in Chinese sibling pairs and Finnish twins, and linkage is suggested in a discordant sibling-pair screen.22–24 In addition, linkage signals are located proximally to the BRIGHT locus on chromosome 2p, documented by two genome scans in families of African ancestry genome scan.17,25 Furthermore, this region has also been linked to the hypertension-associated phenotype pre-eclampsia in some genome-wide screens.26,27 Given the size of the BRIGHT population, it is unsurprising that findings differ between our and previously published studies. Preliminary analysis of each region with a lod score higher than 1·57 has identified a small number of good candidates that map to the chromosome 2 and 9 regions of interest. These candidates include the genes encoding serine-threonine kinases (STK39, STK17B, chromosome 2q), a protein kinase (PKNBETA on 9q), G-protein coupled receptors (GPR21 on 9q33, GPR107 9q, and GPR1 on 2q33) and a potassium channel (KCNJ3 on 2q24.1). Several blood-pressure quantitative trait loci from hypertensive rat models exhibit homology with the BRIGHT regions of interest. For instance, rat 3q11-q23 is homologous to the human 9q region, rat 2q14-q16 with human chromosome 5p region, and rat chromosome 9q22-q34 is homologous to the human 2q locus. We are assembling large-scale case-control and parentoffspring trio resources so that after grid tightening we will be able to finely map regions of interest with direct and indirect LINKAGE DISEQUILIBRIUM mapping together with comparative genetics of experimental hypertension. These studies will be complemented by further investigation of high-quality candidates in transgenic models and in-vitro tissue-expression profiling. We acknowledge that the chromosomal regions we identified in this study are unlikely to reflect all genetic effects on hypertension. There are probably other genes implicated in human hypertension at heritabilities below the limit of detection of an extreme concordant siblingpair approach. These might be detectable in association studies of candidate genes. None of our loci encompass previously postulated candidate genes from physiological understanding of normal blood-pressure control, or from smaller-scale genetic studies. The potential of our results might be the discovery of previously unrecognised biochemical pathways, or processes as the root cause of this major risk factor for cardiovascular disease.

2122

Contributors M Caulfield is the study coordinator and contributed to study design, experimental work, analysis, and writing of the paper. P Munroe contributed to study design, experimental work, analysis, and writing of the paper. J Pembroke is the nursing coordinator and contributed to study design and experimental work. N Samani, A Dominiczak, M Brown, N Benjamin, J Webster, and P Ratcliffe contributed to study design and coordinated sample collection in the partner institutes. S O’Shea, J Papp, S Newhouse, J Hooper, W Lee, and N Brain did genotyping and quality control. E Taylor did the statistical analysis. R Dobson, and J Knight undertook quality control. D Clayton contributed to study design, statistical analysis, and interpretation. M Lathrop contributed to study design. M Farrall contributed to study analysis and interpretation, and writing of the paper. J Connell chaired the investigators’ steering group, contributed to study design, and coordinated sample collection. All researchers contributed to critical appraisal of the paper.

Conflict of Interest statement None declared.

Acknowledgments We thank the BRIGHT nurses: Maggie Bruce, Joan Jamieson, Anne Nixon, Katie Witte, Aberdeen; Sandy Colville-Stewart, Joanne Kent, Jane Pheby, London; Sue Hood, Jill Hunt, Debbie Lloyd, Diane Picton, Cambridge; Lynne Gatherer, Clare Gemmell, Kirsteen Gilmour, Katherine MacFarlane, Fiona Porteus, Glasgow; Karen Edwards, Sarah Hampson, Jay Hussein, Sarah Mellor, Leicester; and Polly Whitworth, and Anna Zawadzka, Oxford. We thank the technical and informatics team: Jennifer Anderton, Kay Davies, Tom Longmore, Steve McGinn, Fiona Quarmby, Trina Smith, Gerovie Soutter, Nadia Thornton, Rob Lawrence, Oxford; Suzi Leibel, Charles Mein, Dan Swan, Barts and The London Genome Centre; and Delphine Bacq, Christine Betard, Arnaud Lemanque, Delphine Torchard, Centre National de Genotypage. We thank Joe Terwilliger, Bernard Keavney, Nicholas Wald, and Malcolm Law for advice; Madge Vickers, Tom Meade, and the Medical Research Council General Practice Framework and other participating family practices for recruitment; and the families who made the study possible. This study was supported by the UK Medical Research Council (grants G9521010 and G000666), who hold observer status on BRIGHT investigators steering group, the British Heart Foundation (grant PG/02/128), the Wellcome Trust, the British Hypertension Society, and the Barts and the London Charitable Foundation. We thank the Scottish Higher Education Council and Centre National de Genotypage for genotyping and infrastructure support.

References 1 2 3 4 5

6 7

8

9

10 11

12

13 14

Luft FC. Twins in cardiovascular genetic research. Hypertension 2001; 37: 350–56. Jacob HJ, Kwitek AE. Rat genetics: attaching physiology and pharmacology to the genome. Nat Rev Genet 2002; 3: 33–42. Lifton RP, Charavi AG, Geller DS. Molecular mechanisms of human hypertension. Cell 2001; 104: 545–56. Wilson H, Disse-Nicodeme S, Choate K, et al. Human hypertension caused by mutations in WNK kinases. Science 2001; 293: 1107–11. Holmans P, Clayton D. Efficiency of typing unaffected relatives in an affected sib-pair linkage study with single locus and multiple tightly linked markers. Am J Hum Genet 1995; 57: 1221–32. Boehnke M, Cox NsJ. Accurate inference of relationships in sib-pair linkage studies. Am J Hum Genet 1997; 61: 423–29. O’Connell JR, Weeks DE. PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 1998; 63: 259–66. Abecasis GR, Cherny SS, Cookson WO, Cardon LR. Merlin-rapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet 2002; 30: 97–101. Risch N. Linkage strategies for genetically complex traits, II: the power of affected relative pairs analysis. Am J Hum Genet 1990; 46: 229–41. Holmans P. Asymptotic properties of affected-sib-pair linkage analysis Am J Hum Genet 1993; 52: 362–74. Wiltshire S, Cardon LR, McCarthy MI. Evaluating the results of genomewide linkage scans of complex traits by locus counting. Am J Hum Genet 2002; 71: 1175–82. Altmuller J, Palmer LJ, Fischer G, Scherb H, Wjst M. Genomewide scans of complex human diseases: true linkage is hard to find. Am J Hum Genet 2001; 69: 936–50. Leal SM, Ott J. Effects of stratification in the analysis of affected sib pair data: benefits and costs. Am J Hum Genet 2000; 66: 567–75. Rice JP, Rochberg N, Neuman RJ, et al. Covariates in linkage analysis. Genet Epidemiol 1999; 17 (suppl 1): S691–95.

THE LANCET • Vol 361 • June 21, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.

MECHANISMS OF DISEASE

15 Holmans P. Detecting gene-gene interactions using affected sib pair analysis with covariates. Hum Hered 2002; 53: 92–102. 16 Samani NJ. Genome scans for hypertension and blood pressure regulation. Am J Hypertens 2003; 16: 167–71. 17 Cooper RS, Luke A, Xiaofeng Z, et al. Genome scan among Nigerians linking blood pressure to chromosomes 2, 3 and 19. Hypertension 2002; 40: 629–33. 18 Rao DC, Province MA, Leppert MF, et al. A genome-wide affected sibpair linkage analysis of hypertension: the HyperGen network. Am J Hypertens 2003; 16: 148–50. 19 Kardia SLR, Rozek LS, Krushkal J, et al. Genome-wide linkage analyses for hypertension genes in two ethnically and geographically diverse populations. Am J Hypertens 2003; 16: 154–57. 20 Ranade K, Hinds D, Hsiung A, et al. A genome-scan for hypertension susceptibility loci in populations of Chinese and Japanese origins. Am J Hypertens 2003; 16: 158–62. 21 Province MA, Kardia SLR, Ranade K, et al. A meta-analysis of genome-wide linkage scans for hypertension: the National Heart Lung and Blood Institute Family Blood Pressure Program. Am J Hypertens 2003; 16: 144–47.

22 Krushkal J, Ferrell R, Mockrin SC, Turner ST, Sing CF, Boerwinkle E. Genome-wide linkage analyses of systolic blood pressure using highly discordant siblings. Circulation 1999; 23: 1407–10. 23 Perola M, Kainulainen K, Pajukanta P, et al. Genome-wide scan of predisposing loci for increased diastolic blood pressure in Finnish siblings. J Hypertens 2000; 11: 1579–85. 24 Zhu DL, Wang HY, Xiong MM, et al. Linkage of hypertension to chromosome 2q14-q23 in Chinese families. J Hypertens 2001; 1: 55–61. 25 Hunt SC, Ellison RC, Atwood LD, Pankow JS, Province MA, Leppert MF. Genome scans for blood pressure and hypertension: the National Heart, Lung, and Blood Institute Family Heart Study. Hypertension 2002; 40: 1–6. 26 Arngrimsson R, Sigurard ttir S, Frigge ML, et al. A genome-wide scan reveals a maternal susceptibility locus for pre-eclampsia on chromosome 2p13. Hum Mol Genet 1999; 8: 1799–805. 27 Moses EK, Lade JA, Guo G, et al. A genome scan in families from Australia and New Zealand confirms the presence of a maternal susceptibility locus for pre-eclampsia, on chromosome 2. Am J Hum Genet 2000; 67: 1581–85.

Uses of error Waiting lists: irritation or death sentence? Julian Gunn Percutaneous coronary intervention (PCI) is becoming commonplace and routine. My patients expect to have an average of two vessels stented via a 6 French (2 mm) catheter in the femoral artery and, at the end of the procedure, have a collagen sealing device fitted, enabling mobilisation at 2 h. 40% of them go home the same day. With such a routine procedure, the loss of a patient is particularly shocking. A 52-year-old man presented with angina. His treadmill test showed reversible ischaemia and he underwent cardiac catheterisation in May, 2001. This showed good left ventricular function and severe diffuse disease affecting the bifurcation of the left anterior descending artery and its second diagonal branch. The circumflex and dominant right coronary arteries were not significantly stenosed. He was referred to me for PCI. As is common in the UK, he spent 5 months on a waiting list. The procedure was technically difficult. Even passing a coronary guidewire down the two arteries was challenging. I inflated an intra-arterial balloon inflation and deployed a stent. A filling defect appeared in the lumen; often indicative of a thrombus or a mural dissection. I gave him abciximab; a glycoprotein IIbIIIa inhibitor. Flow became sluggish in the whole left coronary artery. I inserted an intra-aortic balloon pump

to support the circulation but the blood pressure continued to fall and, ultimately, all blood flow ceased in the coronary artery. My patient drifted into unconsciousness and, despite aggressive efforts at resuscitation, died in front of me. Breaking the news to his wife and 14-year-old daughter, when they returned from a shopping trip, was an experience which left an indelible impression upon me. Interventional cardiologists will be familiar with the problem of acute vessel closure during PCI, but watching all blood flow slow down and stop was highly unusual. I could not account for it. I suspected extensive intracoronary thrombosis precipitated by the exposure of a large amount of plaque consequent upon balloon injury. I had to persuade the coroner’s officer to open an inquest. Necropsy showed that the whole coronary tree was heavily diseased with atheroma, but also, that the right coronary artery was completely and chronically occluded. This must have occurred (without the patient presenting to hospital) while he had been on the waiting list. If I had re-checked the patency of the right coronary artery at the beginning of the procedure, I would not have undertaken such a complex intervention on his only other coronary artery, and a schoolgirl would still have a father.

Cardiovascular Research Group, University of Sheffield, Northern General Hospital, Sheffield S5 7AU, UK (J Gunn MRCP)

THE LANCET • Vol 361 • June 21, 2003 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet.

2123