Genetic susceptibility for human familial essential hypertension in a region of homology with blood pressure linkage on rat chromosome 10
Genetic susceptibility for human familial essential hypertension in a region of homology with blood pressure linkage on rat chromosome 10Cécile Julier1,*, Marc Delépine1, Bernard Keavney1, Joseph Terwilliger1,2, S. Davis3, Daniel E. Weeks1,3, Thuan Bui1, Xavier Jeunemaître4, Gilberto Velho5, Philippe Froguel6, Peter Ratcliffe1, Pierre Corvol4, Florent Soubrier5 and G. Mark Lathrop1
1The Wellcome Trust Centre for Human Genetics, University of Oxford, Windmill Road, Oxford OX3 7LD, UK, 2Columbia University Department of Psychiatry and Columbia Genome Centre, 722 West 168th Street, New York, NY 10032, USA, 3Department of Human Genetics, University of Pittsburgh, Pittsburgh, PA 15261, USA, 4Laboratory of Experimental Medicine, INSERM U36, Collège de France, 3 rue d'Ulm, 75005 Paris, France, 5INSERM Unité 358, Hôpital St Louis, 16 rue de la Grange aux Belles, 75010 Paris, France and 6CNRS EP 10, Institut Pasteur of Lille, 1 rue Calmette, 59019 Lille, France
Received June 13, 1997;Revised and Accepted August 26, 1997
Hypertension is a significant risk factor for heart attack and stroke and represents a major public health burden because of its high prevalence (e.g. 15-20% of the European and American populations). Although blood pressure is known to have a strong genetic determination, the genes responsible for susceptibility to essential hypertension are mostly unknown. Loci involved in blood pressure regulation have been found by linkage in experimental hereditary hypertensive rat strains, but their relationship to human hypertension has not been extensively investigated. One of the principal blood pressure loci has been mapped to rat chromosome 10 and we have undertaken an investigation of the homologous region on human chromosome 17 in familial essential hypertension. Affected sib-pair analysis and parametric analysis with ascertainment correction gave significant evidence of linkage (P <0.0001 in some analyses) near two closely linked microsatellite markers, D17S183 and D17S934, that reside 18 cM proximal to the ACE locus in the homology region. Our results indicate that chromosome 17q could contain a susceptibility locus for human hypertension and show that comparative mapping may be a useful approach for identification of such loci in humans.
The multifactorial basis of hypertension is well established and 30-50% of the variation in blood pressure between individuals is attributed to genetic factors (1 ). However, the genes responsible for susceptibility to hypertension and blood pressure variation are mostly unknown. Some progress towards their identification has been made, using two principal approaches: linkage analyses in families segregating for rare forms of familial hypertension and exploration of candidate genes. The first of these has led to the detection of linkage and the identification of genetic alterations at four genes: the 11[beta]-hydroxylase in glucocorticoid-suppressible hyperaldosteronism (2 ); the [beta] and [gamma] subunits of the epithelial sodium channel in Liddle's syndrome (pseudoaldosteronism) (3 ,4 ); the 11[beta]-hydroxysteroid dehydrogenase gene in the syndrome of apparent mineralocorticoid excess (5 ). However, these syndromes account for a small fraction of hypertension, with quasi-Mendelian inheritance and particular phenotypes, and it is therefore expected that other genes are also involved in the disorder. Using a candidate gene approach, the angiotensinogen gene (AGT) has been reported to contribute to susceptibility in some affected sib-pair studies (6 ,7 ) but not others (E.Brand, N.Chatelain, G.Bianchi, T.de Bruijn, M.Caufield, J.Connell, B.Keavney, S.Schmidt, H.Schunkert, H.Schuster, A.Sharma, X.Jeunemaitre and F.Soubrier, in preparation). It is unlikely that AGT or any other candidate genes that have been tested to date account for a majority of the genetic susceptibility to hypertension.
Genetic studies of hereditary hypertension in the rat, and particularly genome-wide scans, have led to the detection of several loci involved in blood pressure regulation (8 -12 ). Regions of homology in human represent candidate regions for containing susceptibility genes for hypertension, whose exploration is an alternative to investigation of rare syndromes or candidate genes. One of the principal loci identified in crosses involving spontaneously hypertensive stroke-prone (SHR/SP) and Dahl salt-sensitive hypertensive rats has been localised near the angiotensin I converting enzyme gene (ACE) on rat chromosome 10 (8 ,9 ,13 ,14 ). This locus was estimated to account for >20% of the total variance of the systolic and diastolic blood pressures after NaCl loading in the F2(SHRSP * WKY) cross (8 ).
Based on comparative mapping data, the human homologues of the corresponding gene (or genes) should reside within a contiguous segment of human chromosome 17q, which contains the ACE gene. Although ACE has been examined genetically for linkage to human hypertension in one study with negative results (15 ), a systematic investigation of the complete homology region has not been made. Based on such an investigation in a large number of human hypertension families, we present evidence of linkage to a putative susceptibility locus for hypertension within the homology region on chromosome 17q but distinct from ACE.
The region of homology to the rat hypertension susceptibility loci BP-SP1 was defined based on comparative maps of rat chromosome 10, mouse chromosome 11 and human chromosome 17, in part from the compilations provided by the Mouse Genome Database that are available on the World Wide Web (16 ). As shown in Figure 1 , the chromosomal region that spans this location is extensively conserved between these species, extending from MYHSE/Myhse/MYH3 on human 17p to the distal end of human 17q. Because of the wide confidence interval for the placement of the blood pressure locus in the rat, we decided to explore the whole homology region for a putative susceptibility locus for hypertension and selected 21 microsatellite markers to screen (Fig. 1 ). The markers cover 110 cM and extend over the region of homology between rat chromosome 10 and human chromosome 17p and ~20 cM distally to GH on 17q. The mean heterozygosity of the markers is 0.75 (0.49-0.94) and the mean recombination distance between adjacent loci is 5.9% (0.1-15.0%). A higher density of markers was selected in the region proximal to or nearby GH/ACE, with a mean recombination frequency of 2.9% between the 13 markers in the D17S946-D17S789 interval (~30 cM).
We studied a panel of 357 French and UK families containing 384 sibships with one or more hypertensive offspring and a total of 518 affected sib-pairs. Strict diagnostic criteria were applied so that the disease phenotype would correspond to the upper 1-5% of the blood pressure distribution (see Materials and Methods). A small number of families were ascertained through probands with non-insulin-dependent diabetes mellitus (29 families) or insulin-dependent diabetes mellitus (27 families); these families contained at least one hypertensive patient in addition to the diabetic proband. Further details on the families and clinical phenotypes are provided in Table 1 .
The relative power of standard non-parametric test statistics and their behaviour under the absence of linkage have been found to vary when compared in different data sets (S.Davis and D.E.Weeks, in preparation). Because of this, we decided to apply four statistical methods to the chromosome 17q data so that the results could be compared. These were: (i) affected sib-pair analysis with the SIBPAL program (17 ); (ii) affected sib-pair analysis with the SIBPAIR program (18 ,19 ); (iii) parametric MOD score analysis in which maximum likelihood analysis is performed conditional on the phenotype status of family members as an ascertainment correction (20 ); (iv) affected sib-pair analysis with the GENEHUNTER program (21 ).
Simulations were used to examine the behaviour of the four statistical methods in our family panel under various genetic models, with or without linkage to a marker locus. Figures 2 and 3 illustrate the results of these studies for a marker with allele frequencies identical to those estimated for D17S934 (one of the markers exhibiting the strongest evidence of linkage in the real data) and for a hypothetical, fully informative marker locus. The first issue to be examined was the adequacy of large sample or other theoretical approximations for the calculation of P values. In the absence of linkage to the marker, Figure 2 shows that the empirical distributions were either closely approximated by the theoretical distributions (SIBPAL and SIBPAIR) or that the latter gave conservative P values (parametric analysis and GENEHUNTER). These results justify the use of the theoretical approximations as a conservative approach for assigning significance to linkage data from these families.
Figure 3 illustrates the power to detect linkage with P <= 0.001 or P <= 0.0001 as a function of the affected sib-pair identity-by-descent (IBD) probabilities under various genetic models. As explained in Materials and Methods, the simulations assumed a disease prevalence of 5% and a dominant susceptibility locus with two alleles, one of which conferred increased risk for hypertension. The penetrance associated with the susceptibility genotypes was varied to obtain IBD probabilities in the interval 51-59% (compared with 50% in the absence of linkage). With any of the first three statistical methods, the power to detect a susceptibility locus tightly linked to D17S934 at a significance level ofP <= 0.001 was >= 70% when the average affected sib-pair IBD in the sample was >= 57-58%. For a fully informative marker, greater power was found even with the more stringent condition ofP <= 0.0001 for similar IBD. Since the simulations assumed no recombination between the marker and disease loci, these are estimates of the maximum power that could be achieved under these genetic models with a dense map covering the homology region.
Table 1 (a). Number of families and affected sib-pairs divided into subsets according to ascertainment criteria
Family panel
Number of affected individuals in sibship
Number of nuclear families
Number of pedigreesa
Number of affected sib-pairs
1
2
3
4
5
6
HTE
23
101
39
7
3
1
174
154
305
HTO
89
56
15
4
1
0
166
147
135
HTD
26
9
0
0
0
1
36
29
24
HTN
5
12
6
4
0
0
27
27
57
Totals
143
178
60
15
4
2
402
357
518
No. with affected parent(s)
37
30
12
2
0
0
81
NA
78
HTE, panel of essential hypertension from France; HTO, panel of essential hypertension from Oxford; HTD, panel with hypertension and insulin-dependent diabetes; HTN, hypertension and non-insulin-dependent diabetes. Further details on the different family panels is provided in Materials and Methods.
aSome multi-generational pedigrees contain two or more nuclear families
Table 1(b). Phenotypes in the four groups of families as a function of hypertension status for individuals with a positive or negative diagnosis of hypertension
Disease status
SBP untreated (mmHg)
SBP treated (mmHg)
DBP untreated (mmHg)
DBP treated (mmHg)
Age at diagnosis
Age at time of study
BMI (kg/m2)
HTE (French)
HT+ (n = 567)
161.1 ± 21.3a (n = 111)
154.9 ± 22.3 (n = 456)
103.2 ± 11.0a (n = 111)
94.7 ± 12.5 (n = 456)
40.2 ± 10.8
53.0 ± 11.8a
25.7 ± 3.9a
HT- (n = 141)
125.4 ± 10.2
80.0 ± 6.8
NA
44.8 ± 12.0
23.5 ± 3.5
HTO (Oxford)
HT+ (n = 207)
162.8 ± 15.8a (n = 95)
153.8 ± 24.0 (n = 112)
102.0 ± 7.7a (n = 95)
89.7 ± 13.0 (n = 111)
44.4 ± 9.9
52.0 ± 10.4a
26.9 ± 3.9a
HT- (n = 256)
122.6 ± 10.1
75.2 ± 8.1
NA
43.8 ± 10.7
25.3 ± 3.8
HTD (French)
HT+ (n = 57)
188.3 ± 19.9a (n = 45)
161.8 ± 18.5 (n = 12)
100.2 ± 13.5a (n = 43)
90.0 ± 10.4 (n = 12)
44.3 ± 10.3
53.6 ± 11.6c
25.4 ± 4.3c
HT- (n = 112)
126.2 ± 9.9
73.6 ± 7.5
NA
47.4 ± 12.3
23 ± 73.2
HTN (French)
HT+ (n = 67)
186.9 ± 17.2a (n = 59)
163.8 ± 25.6 (n = 8)
102.5 ± 10.8a (n = 59)
92.5 ± 13.9 (n = 8)
50.6 ± .7.2
60.9 ± 8.7
29.3 ± 4.7c
HT- (n = 21)
131.9 ± 17.2
75.0 ± 7.6
NA
57.1 ± 12.8
26.2 ± 2.9
Significance of the difference between HT+ and HT- groups: aP < 0.0001; bP < 0.001; cP < 0.01.
n, number of individuals.
HT+, hypertensives; HT-, normotensives; SBP: systolic blood pressure; DBP: diastolic blood pressure.
Table 1(c). Phenotypes in HTD and HTN families as a function of diabetes and hypertension status for individuals with a positive or negative diagnosis of hypertension
Disease status
SBP untreated (mmHg)
SBP treated (mmHg)
DBP untreate (mmHg)
DBP treated (mmHg)
Age at onset (years)
Age (years)
BMI (kg/m2)
IDDM patients
HT+ (n = 20)
191.2 ± 16.9a (n = 17)
150.7 ± 11.0 (n = 3)
100.6 ± 11.4a (n = 17)
86.7 ± 5.8 (n = 3)
38.8 ± 9.2
47.7 ± 9.3
23.9 ± 3.3
HT- (n = 30)
127.0 ± 10.6
74.1 ± 7.9
NA
45.1 ± 9.1
23.7 ± 3.6
NIDDM patients
HT+ (n = 71)
188.1 ± 19.4a (n = 60)
157.3 ± 17.9 (n = 11)
101.8 ± 11.2a (n = 59)
89.1 ± 13.8 (n = 11)
50.7 ± 6.9
60.5 ± 8.4
29.4 ± 4.8c
HT- (n = 27)
131.5 ± 15.6
75.2 ± 7.1
NA
55.7 ± 13.1
26.0 ± 3.5
Non-diabetics
HT+ (n = 33)
183.9 ± 18.1a (n = 27)
178.3 ± 23.2 (n = 6)
101.8 ± 14.3a (n = 26)
96.7 ± 8.2 (n = 6)
46.6 ± 10.2
57.2 ± 12.4b
25.6 ± 4.2c
HT- (n = 76)
125.5 ± 9.7
73.7 ± 8.0
NA
47.9 ± 13.3
23.6 ± 2.9
Significance of the difference between HT+ and HT- groups: aP <0.0001; bP <0.001; cP <0.01. n, number of individuals. HT+, hypertensives; HT-, normotensives; SBP: systolic blood pressure; DBP: diastolic blood pressure.
The 24 microsatellite markers shown in Figure 1 were initially characterised in French families from the HTE, HTN and HTD panels. Several markers between D17S946 and D17S789 provided suggestive evidence of linkage (P <0.01) and the subset of markers included in this interval were characterised in the total family panel (Table 2 a). The strongest support for linkage with the SIBPAL and SIBPAIR affected sib-pair methods was found at D17S183(P = 0.0001 with the SIBPAIR program). Although D17S183 has only 47% heterozygosity, it is closely linked to D17S934 (no obligate recombinants in the family panel), a marker with 83% heterozygosity. D17S934 produced similar, although slightly less significant results, with the affected sib-pair methods. The parametric analyses provided broadly similar evidence for linkage, but here the most significant result was at D17S934:P = 0.0002 (or P = 0.00004) assuming a 5% (or 1%) disease prevalence. As observed in the simulations, the GENEHUNTER NPL statistic was often much less significant than those obtained with the other methods, possibly because of its conservative behaviour. However, the strongest evidence of linkage from GENEHUNTER was also obtained at D17S934 (P = 0.008).
The simulation results in Figure 3 suggest that it might be possible to obtain additional support for linkage by combining D17S934 or D17S183 with other markers to increase the information content. To examine this possibility, we undertook further linkage tests with pairs of markers using the parametric and GENEHUNTER approaches for multilocus analysis. Table 2 b shows the results of multilocus analyses with all marker pairs that included D17S934.
The strongest evidence for linkage was obtained in parametric analysis when D17S934 was combined with GH, the marker with the highest heterozygosity. These tests achieved a significance of P = 0.00002 at 5% prevalence and P = 0.000004 at 1% prevalence. The maximum likelihood placement of the susceptibility locus was in the interval between D17S934 and GH at 13 cM from the latter (Fig. 4 ). Assuming 5% prevalence, the parameter estimates at this location were 0.02 for the frequency of the susceptibility allele and 0.21 for the penetrance of the susceptibility genotypes. All four family panels contributed to the support for linkage within the interval D17S934-GH (P = 0.05, HTE; P = 0.04, HTO; P = 0.01, HTN; P = 0.02, HTD, at 5% prevalence). Thus, there appears to be no evidence of heterogeneity between families from France and Oxford or between families with and without diabetic probands. Other analysis methods also led to the same conclusion.
Our linkage results support the hypothesis of a susceptibility locus for human hypertension on chromosome 17q. We selected chromosome 17q for study because of its homology to a region of rat chromosome 10 that contains one of the principal loci implicated in the regulation of blood pressure in experimental models of hereditary hypertension. The blood pressure locus on rat chromosome 10 was first detected in an F2 cohort derived from a SHRSP * WKY cross (8 ,9 ), where it was estimated to contribute >20% of the total variance of the systolic and diastolic blood pressures after NaCl loading (8 ). Subsequently, the same region has been found to be linked to blood pressure variation in several other crosses involving SHR/SP (13 ) or Dahl hypertensive rats (14 ).
Rather than test a single marker or candidate gene as in past studies of the ACE region in human hypertension, e.g. Jeunemaitre et al. (15 ), we took account of the imprecise localisation of the blood pressure locus and examined markers throughout the region of homology of rat chromosome 10 and human chromosome 17. Families containing a total of 518 affected sib-pairs were ascertained for the study and strict diagnostic criteria were applied to obtain a phenotype that corresponded to 1-5% population prevalence of the disease. By increasing the stringency of the phenotype definition, we increased the power to detect linkage. Moreover, a substantial number of families contained three or more affected sib-pairs or one or more affected parent. In simulations, this ascertainment led to a greater average affected sib-pair IBD probability in the family panel compared with the expectation in randomly selected affected sib-pairs, also increasing the power of the study (results not shown). For a dominant susceptibility allele with 0.05 frequency and a highly informative marker tightly linked to the susceptibility locus, simulations suggested that it would be possible to detect linkage with at least 70% power when the expected affected sib-pair IBD probability in the sample was in the range 0.57-0.58 or greater.
Nominal evidence of linkage (P <0.01) to hypertension was found at several markers spanning the region of homology. The strongest evidence came from two closely linked markers, D17S183 and D17S934, that map 18 cM proximal to the ACE locus, where we obtained P <= 0.0001 in some affected sib-pair and parametric analyses. The combination of D17S183 or D17S934 with GH, the most informative of the markers studied, gave test statistics with a significance of P <0.00001 in multilocus parametric linkage analysis. Many other pairwise marker analyses were also significant with P <0.0001, some of which are shown in Table 2 b.
On the other hand, non-parametric linkage analysis with the GENEHUNTER method did not reach the same level of significance. Variation in the performance of different statistical methods in the presence of linkage has been described previously (S.Davis and D.E.Weeks, in preparation). In particular, the GENEHUNTER statistic has been found to be less powerful, often giving non-significant test statistics when other methods are significant, similar to the pattern observed here in simulation studies. However, because of our observation that the GENEHUNTER statistic appears to be conservative based on these simulation studies, it might be possible to modify the method in order to obtain empirical P values that may be similar to those obtained in the other methods. It should also be noted that the pairwise marker data analysed with the GENEHUNTER and the parametric methods were not identical, as it was necessary to reduce the size of some pedigrees in the GENEHUNTER calculations, as discussed in Material and Methods. When parametric analysis was performed on the reduced data for a subset of loci pairs, the tests were found to be less significant, although still more so than GENEHUNTER (results not shown). These comparisons point to the need for careful evaluation of statistical methods as part of linkage studies of multifactorial diseases.
ACE maps to the region of blood pressure linkage on rat chromosome 10 and it is also in the homology region that we have targeted in human hypertension. Because of the important role of ACE in the renin-angiotensin system, it has been studied previously as a candidate gene in hypertension, but the results have been largely negative (22 ). In particular, Jeunemaitre et al. (15 ) found no evidence for linkage to hypertension with the GH microsatellite, near to the ACE locus, in an independent panel of families. In our family panel, which contained a larger number of sib-pairs and which was selected under a different protocol, we obtained P = 0.003 by affected sib-pair analysis with GH. However, the strongest evidence of linkage was found with D17S183 and D17S934, which are 18 cM proximal to ACE. In the parametric analyses for the marker pairs D17S183-GH and D17S934-GH the maximum likelihood placement of the putative susceptibility locus was 13 cM from GH/ACE. The latter were outside the 95% support interval, suggesting that another gene (or genes) from the homology region could be implicated in hypertension.
Congenic strains bred from Dahl salt-sensitive hypertensive rats gave evidence for a QTL with a major effect on blood pressure localised within a 31 cM segment that includes ACE (23 ). The congenic segment maps to the homology region in which we have found linkage in human hypertension families and lends support to the hypothesis that the same genes could be implicated in hypertension in humans. Recent data from crosses involving inbred hypertensive strains of rat suggest that a blood pressure locus on rat chromosome 10 may reside at a distance from the ACE gene or that two linked loci may be involved (24 ,25 ). Other candidate genes, e.g. phenylethanolamine N-methyltransferase (26 ), and transcripts have been mapped to the interval, but at present no evidence is available to implicate any of these in hypertension in either the rat or human. The identification of the gene or genes responsible for hypertension susceptibility will most likely require a systematic investigation of the region in both species. If homologous genes are involved, a powerful approach will be to combine fine mapping of congenic lines in the rat and human families.
Traditional approaches to gene identification have been successful for rare Mendelian syndromes associated with severe hypertension (27 ), but to date these genes have not been implicated in essential hypertension. Previous genetic investigations of essential hypertension have mostly focused on key candidate genes from the renin-angiotensin system (28 ) and significant evidence of linkage in more than one family panel has been reported for angiotensinogen (6 ,7 ). Our study illustrates an alternative approach which, as a first stage, targets homology regions based on blood pressure linkage in the rat and does not rely on a priori testing of candidate genes. The next step will be to seek confirmation of these linkage results in other large panels of families containing multiple cases of hypertension with a similar disease definition to those used here.
Patients were diagnosed with hypertension if: (i) multiple measurements of diastolic blood pressure exceeded 95 mm Hg at the time of examination; (ii) the patient had a clinical record with multiple diastolic blood pressure readings over 95 mm Hg prior to treatment; (iii) the patient was on long-term anti-hypertensive treatment with one or more drugs. All patients were 60 or younger at time of onset of hypertension. Subjects with diastolic blood pressure <90 and systolic blood pressure <140 at >30 years were considered unaffected; otherwise the phenotype was assumed unknown. Comparison with unpublished survey data from the UK suggests that the disease prevalence under these ascertainment criteria will be between 1 and 5% after adjustment for age and BMI.
The families were divided into four panels based on additional ascertainment criteria: (i) HTE (essential hypertension from France), families ascertained through patients from the Hypertension Clinic at Broussais Hospital and from other centres in France (6 ); (ii) HTO (essential hypertension in families collected in the region of Oxford, UK), families ascertained from the hypertension clinic at the John Radcliffe Hospital, Oxford, or through general practices in the Oxford area and containing at least one individual with a diagnosis of hypertension; (iii) HTD (hypertension and insulin dependent diabetes), French families initially collected for studies of insulin-dependent diabetes mellitus (IDDM) with at least one offspring having been diagnosed with IDDM and at least one offspring (not necessarily diabetic) diagnosed with hypertension; (iv) HTN (hypertension and non-insulin-dependent diabetes mellitus), French families initially collected for studies of non-insulin-dependent diabetes mellitus (NIDDM) with at least two offspring having been diagnosed with NIDDM and at least one offspring (not necessarily diabetic) diagnosed with hypertension. In the HTE panel, families with a parental history of hypertension were preferentially included. In the HTO families, the third diagnostic criterion was replaced by the requirement of treatment with two drugs. In the HTD and HTN panels, criteria (i) or (ii) were required for a positive diagnosis of hypertension. Additional selection criteria applied to some probands were BMI <27 kg/m2 (HTE) or BMI <30 kg/m2 (HTO) and reported alcohol consumption <= 14 U/week for females or <= 21 U/week for males (HTE, HTO). Use of oral contraception or secondary causes of hypertension were exclusion criteria for the diagnosis of hypertension.
Reported variability in linkage results obtained with different non-parametric statistical methods and computer implementations (S.Davis and D.E.Weeks, in preparation) led us to apply four different methods of analysis to the data for the purpose of comparison:SIBPAL. The SIBPAL program (17 ) is one of the most commonly applied of the non-parametric methods. SIBPAL provides affected sib-pair and IBD regression statistics. We used only the former here. Information on half-sibs is incorporated into the affected sib-pair statistic, but the results cannot be reliably interpreted if the number of half-sibs is small, as it is in our sample. To remove the effect of half-sibs, we divided extended pedigrees into nuclear families prior to applying SIBPAL. The program estimates IBD sharing at a single marker locus from parental and offspring data for each affected sib-pair and treats pairs from within the same family as independent.SIBPAIR. The SIBPAIR program implements the algorithm described in Satsangi et al. (20 ) and Kuokkanen et al. (19 ). It is essentially equivalent to the calculation of lod scores under the assumption of a simple recessive disease model which takes into account non-independence of affected sib-pairs from the same family and missing data on parents. This method has been shown to conform closely to its asymptotic null distribution in small samples and was amongst the most powerful of the non-parametric test statistics examined by Davis and Weeks (in preparation).Parametric analysis. Parametric linkage analysis was undertaken with a modified version of the VITESSE (30 ) program. Parameter estimates and test statistics were calculated conditional on the trait phenotype status of all the family members (mod scores) as a correction for ascertainment (20 ). We assumed a two-allele, dominant model for the trait locus which allowed for incomplete, non-zero penetrance of the three genotypes while constraining the overall disease prevalence to a predetermined value (here either 5 or 1%, representing the extremes of the estimated prevalence for hypertension defined with our diagnosis criteria). For a single marker locus, this is equivalent to maximising the lod score over all genetic models satisfying these constraints. In the tests of linkage with a single marker locus, we imposed the additional constraint of no recombination between the putative susceptibility locus and the marker locus, as in Davies et al. (31 ); this constraint is used since the recombination rate may be confounded with other parameters. Under the hypothesis of dominance, we assumed that the susceptibility allele d had frequency p with penetrance fd- for the two genotypes Dd and dd. The parameters p and fd- were estimated and the penetrance fDD associated with the DD genotype was calculated as a function of these estimates and the fixed disease prevalence (r) as fDD = {r - fd-[1 - (1 - p)2]}/(1 - p)2. A likelihood ratio test statistic was calculated against the null hypothesis in which the susceptibility allele frequency was 0 or, equivalently, q = f = r. In the parametric multilocus analysis the test statistic was calculated at different placement of the trait locus with respect to the markers. The likelihood ratio is calculated against the null hypothesis in which the susceptibility allele frequency is 0. The test statistic has been compared with a [chi]2 distribution with 2 degrees of freedom to obtain significance levels. Simulations, some of which are described below, showed this to be conservative.GENEHUNTER. The NPL score statistic was calculated with the GENEHUNTER program (21 ) with the `all pairs' option. Because of the large size of some of the pedigrees in our study, the statistical calculations could not be made with the complete data. We divided the pedigrees into nuclear families and removed a number of individuals from the larger sibships to overcome this. Significance levels are evaluated with the perfect data approximation as explained in Kruglyak et al. (21 ).
The parametric and GENEHUNTER programs were applied to analyse multilocus marker data. The SIBPAL and SIBPAIR programs are restricted in application to a single marker locus.
Allele frequency estimates were obtained prior to linkage analysis based on the allele counts using one randomly selected individual for which genotypes were available from each pedigree. The analyses were performed in the combined panel of families, as well as in the four sub-panels. The linkage results on the sub-panels reported in the text were calculated with the allele frequency estimates from the combined family panel. Recalculation with estimates from the sub-panels did not modify these significantly, as the allele frequencies did not differ significantly between sub-panels of families.
The performance of the test statistic was evaluated by simulations with SLINK (32 ) under a dominant single locus model with two alleles, d and D, having allele frequencies of 0.05 and 0.95 respectively. The simulations were made conditional on the observed phenotype status in the family panel. For the power calculations, we simulated 500 replicates under models in which the penetrance associated with the dd and dD genotypes was 0.10, 0.13, 0.15, 0.17 or 0.20. The penetrance of the DD genotype was adjusted so as to obtain an overall prevalence of 5% for the disease in the general population as discussed above. Simulations were made under the null hypothesis with the penetrance for all three genotypes equal to the population prevalence; 1500 replicates were made to examine the null distributions of the test statistics.
In the power calculations shown here, the marker and trait loci are assumed to be completely linked. We considered two situations: one in which the marker was completely informative and the other in which the marker allele frequencies were equal to those at D17S934. The test statistics were calculated with the allele frequency estimates fixed at the values used in the simulations. In the results presented in Figures 2 and 3 , the parametric analyses were undertaken assuming 5% prevalence; however, neither the power nor the performance of the statistic under the null hypothesis were substantially different when the simulated data were analysed with 1% prevalence.
We thank M.Marmot and H.Hemingway for providing the blood pressure survey data from the Whitehall II study. Financial support has been provided by the Wellcome Trust, INSERM, the French Ministry of Research and the EC Concerted Action `Molecular Genetics of Hypertension'. B.K. is supported by a MRC Training Fellowship. G.M.L. holds a Wellcome Trust Principal Fellowship. S.D. and D.E.W. were supported in part by NIH grant HG00719. Portions of the results of this paper were obtained using the program package SAGE.
1 Ward,R. (1995) In Laragh,J.H. and Brenner,B.M. (eds), Hypertension: Pathology, Diagnosis and Management. Raven Press, New York, NY, Vol. 1, pp. 67-88.
2 Lifton,R.P., Dluhy,R.G., Powers,M., Rich,G.M., Gutkin,M., Fallo,F., Gill,J.R.,Jr, Feld,L., Ganguly,A., Laidlaw,J.C., Murnaghan,D.J., Kauggman,C., Stockigt,J.R., Ulick,S. and Lalouel,J.M. (1992) Hereditary hypertension caused by chimaeric gene duplications and ectopic expression of aldosterone synthase. Nature Genet., 2,66-74.MEDLINE Abstract
3 Shimkets,R.A., Warnock,D.G., Bositis,C.M., Nelson Williams,C., Hansson,J.H., Schambelan,M., Gill,J.R.,Jr, Ulick,S., Milora,R.V., Findling,J.W., Canessa,C.M., Rossier,B.C. and Lifton,R.P. (1994) Liddle's syndrome: heritable human hypertension caused by mutations in the beta subunit of the epithelial sodium channel. Cell, 79,407-414.MEDLINE Abstract
4 Hansson,J.H., Nelson Williams,C., Suzuki,H., Schild,L., Shimkets,R., Lu,Y., Canessa,C., Iwasaki,T., Rossier,B. and Lifton,R.P. (1995) Hypertension caused by a truncated epithelial sodium channel gamma subunit: genetic heterogeneity of Liddle syndrome. Nature Genet., 11,76-82.MEDLINE Abstract
5 Mune,T., Rogerson,F.M., Nikkila,H., Agarwal,A.K. and White,P.C. (1995) Human hypertension caused by mutations in the kidney isozyme of 11[beta]-hydroxysteroid dehydrogenase. Nature Genet., 10,394-349.MEDLINE Abstract
6 Jeunemaitre,X. et al. (1992) Molecular basis of human hypertension: role of angiotensinogen. Cell,71,169-180.MEDLINE Abstract
7 Caulfield,M., Lavender,P., Farrall,M., Munroe,P., Lawson,M., Turner,P. and Clark,A.J. (1994) Linkage of the angiotensinogen gene to essential hypertension. New Engl. J. Med., 330,1629-1633.MEDLINE Abstract
8 Hilbert,P., Lindpaintner,K., Beckmann,J.S., Serikawa,T., Soubrier,F., Dubay,C., Cartwright,P., De Gouyon,B., Julier,C., Takahasi,S., Vincent,M., Ganten,M., Georges,M. and Lathrop,M. (1991) Chromosomal mapping of two genetic loci associated with blood-pressure regulation in hereditary hypertensive rats. Nature, 353,521-529.MEDLINE Abstract
9 Jacob,H.J., Lindpaintner,K., Lincoln,S.E., Kusumi,K., Bunker,R.K., Mao,Y.P., Ganten,D., Dzau,V.J. and Lander,E.S. (1991) Genetic mapping of a gene causing hypertension in the stroke-prone spontaneously hypertensive rat. Cell, 67,213-224.MEDLINE Abstract
10 Dubay,C., Vincent,M., Samani,N.J., Hilbert,P., Kaiser,M.A., Beressi,J.P., Kotelevtsev,Y., Beckmann,J.S., Soubrier,F., Sassard,J. and Lathrop,G.M. (1993) Genetic determinants of diastolic and pulse pressure map to different loci in Lyon hypertensive rats. Nature Genet., 3,354-357.MEDLINE Abstract
11 Pravenec,M., Gauguier,D., Schott,J.J., Buard,J., Kren,V., Bila,V., Szpirer,C., Szpirer,J., Wang,J.M., Huang,H., St. Lezin,E., Spence,M.A., Flodman,P., Printz,M., Lathrop,G.M., Vergnaud,G. and Kurtz,T. (1995) Mapping of quantitative trait loci for blood pressure and cardiac mass in the rat by genome scanning of recombinant inbred strains. J. Clin. Invest., 96,1973-1978.MEDLINE Abstract
12 Schork,N.J., Krieger,J.E., Trolliet,M.R., Franchini,K.G., Koike,G., Krieger,E.M., Lander,E.S., Dzau,V.J. and Jacob,H.J. (1995) A biometrical genome search in rats reveal the multigenic basis of blood pressure variation. Genome Res., 5,164-172.MEDLINE Abstract
13 Nara,Y., Nabika,T., Ikeda,K., Sawamura,M., Endo,J. and Yamori,Y. (1991) Blood pressure cosegregates with a microsatellite of angiotensin I converting enzyme (ACE) in F2 generation from a cross between original normotensive Wistar-Kyoto rat (WKY) and stroke-prone spontaneously hypertensive rat (SHRSP). Biochem. Biophys. Res. Commun., 181,941-946.MEDLINE Abstract
14 Deng,Y. and Rapp,J.P. (1992) Cosegregation of blood pressure with angiotensin converting enzyme and atrial natriuretic peptide receptor genes using Dahl salt-sensitive rats. Nature Genet., 1,267-272.MEDLINE Abstract
15 Jeunemaitre,X., Lifton,R.P., Hunt,S.C., Williams,R.R. and Lalouel,J.M. (1992) Absence of linkage between the angiotensin converting enzyme locus and human essential hypertension. Nature Genet., 1,72-75.MEDLINE Abstract
16 Nadeau,J.H., Grant,P.L., Mankala,S., Reiner,A.H., Richardson,J.E. and Eppig,J.T. (1995) A Rosetta stone of mammalian genetics.Nature, 373,363-365.MEDLINE Abstract
17 Department of Biometry and Genetics, LSU Medical Centre (1994) SAGE: Statistical Analysis for Genetic Epidemiology, release 2.2. Department of Biometry and Genetics, LSU Medical Centre, New Orleans, LA.
18 Satsangi,J., Parkes,M., Louis,E.L.H., Kato,N.K.W., Terwilliger,J.D., Lathrop,G.M., Bell,J.I. and Jewell,D.P. (1996) Two-stage genome-wide search in inflammatory bowel disease: evidence for susceptibility loci on chromosomes 3, 7, and 12. Nature Genet., 14,199-202.MEDLINE Abstract
19 Kuokkanen,S., Sundvall,M., Terwilliger,J.D., Tienari,P.J., Wikstrom,J., Holmdahl,R., Pettersson,U. and Peltonen,L. (1996) A putative vulnerability locus to multiple sclerosis maps to 5p14-p12 in a region syntenic to the murine locus Eae2. Nature Genet., 13,477-480.MEDLINE Abstract
20 Hodge,S.E. and Elston,R.C. (1994) Lods, wrods, and mods: the interpretation of lod scores calculated under different models. Genet. Epidemiol., 11,329-342.MEDLINE Abstract
21 Kruglyak,L., Daly,M.J., Reeve-Daly,M.R. and Lander,E.S. (1996) Parametric and nonparametric linkage analysis: a unified multipoint approach. Am. J. Hum. Genet., 58,1347-1363.MEDLINE Abstract
22 Villard,E. and Soubrier,F. (1966) Molecular biology and genetics of the angiotensin-I-converting enzyme: potential implications in cardiovascular diseases. Cardiovascular Res., 32,999-1007.
23 Dukhanina,O.I., Dene,H., Deng,A.Y., Choi,C.R., Hoebee,B. and Rapp,J.P. (1997) Linkage map and congenic strains to localise blood pressure QTL on rat chromosome 10. Mamm. Genome,8,229-235.MEDLINE Abstract
24 Deng,A.Y. and Rapp,J.P. (1995) Locus for the inducible, but not a constitutive, nitric oxide synthase cosegregates with blood pressure in the Dahl salt-sensitive rat. J. Clin. Invest., 95,2170-2177.MEDLINE Abstract
25 Kreutz,R., Hubner,N., James,M.R., Bihoreau,M.T., Gauguier,D., Lathrop,G.M., Ganten,D. and Lindpaintner,K. (1995) Dissection of a quantitative trait locus for genetic hypertension on rat chromosome 10. Proc. Natl. Acad. Sci. USA, 92,8778-8782.MEDLINE Abstract
26 Hoehe,M.R., Plaetke,R., Otterud,B., Stauffer,D., Holik,J., Byerley,W.F., Baetge,E.E., Gershon,E.S., Lalouel,J.M. and Leppert,M. (1992) Genetic linkage of the human gene for phenylethanolamine N-methyltransferase (PNMT), the adrenaline-synthesizing enzyme, to DNA markers on chromosome 17q21-q22. Hum. Mol. Genet., 1,175-178.MEDLINE Abstract
27 Lifton,R.P. (1996) Molecular genetics of human blood pressure variation. Science, 272,676-680.MEDLINE Abstract
28 Soubrier,F. and Lathrop,M. (1995) The genetic basis of hypertension: an update on recent studies. Curr. Opin. Nephrol. Hypertension, 177-181.
29 Vignal,A., Gyapay,G., Hazan,J., Nguyen,S., Dupraz,C., Cheron,N., Becuwe,N., Tranchant,M. and Weissenbach,J. (1993) In Adolphs,K.W. (ed.), Methods in Molecular Genetics: Gene and Chromosome Analysis. Academic Press, San Diego, CA, Vol. 1, pp. 211-221.
30 O'Connell,J.R. and Weeks,D.E. (1995) The VITESSE algorithm for rapid exact multilocus linkage analysis via genotype set-recoding and fuzzy inheritance. Nature Genet., 11,402-408.MEDLINE Abstract
31 Davies,J.L., Kawaguchi,Y., Bennett,S.T., Copeman,J.B., Cordell,H.J., Pritchard,L.E., Reed,P.W., Gough,S.C., Jenkins,S.C., Palmer,S.M., Balfour,K.M., Rowe,B.R., Farrall,M., Barnett,A.H., Bain,S.C. and Todd,J.A. (1994) A genome-wide search for human type 1 diabetes susceptibility genes. Nature, 371,130-136.MEDLINE Abstract
32 Weeks,D.E., Ott,J. and Lathrop,G.M. (1990) SLINK: a general simulation program for linkage analysis. Am. J. Hum. Genet., 47,A204.
*To whom correspondence should be addressed. Tel: +44 1865 740 017; Fax: +44 1865 742 196; Email: cecile@well.ox.ac.uk +The SIBPAIR program is part of the ANALYZE programme package which is available by anonymous ftp from flemming.well.ox.ac.uk. MOD score analysis has been implemented as part of the VITESSE program package which is available from watson.hgen.pitt.edu.
This page is maintained by OUP admin. Last updated Sat Oct 18 13:40:20 BST 1997
. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1997
