Common polymorphisms at multiple blood pressure-related loci from distinct biological pathways are likely to contribute to the genetic component of essential hypertension (EH) in humans [
Approximately 4 million Africans were brought to Brazil as slaves over a period of four centuries. Before the abolition of slavery in Brazil (in 1888), many communities, named
In this paper, we present the results of a carefully conducted family-based association study (652 participants from 97 informative families) in semi-isolated Brazilian populations of African ancestry, which includes individuals from remnants of
Participants were ascertained from African-derived Brazilian populations named
Briefly, the sample consisted of individuals aged over 17 years with clinical and anthropometric data. Participants were sampled from 12
Clinical evaluation occurred at multiple visits performed between 2003 and 2010. Due to the fact that
Blood pressure was measured after 15 min of rest in a sitting position. Two measurements were obtained by a physician from each participant at 5 min intervals. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) values were means of the two physician-obtained measurements.
The primary phenotype was the risk of essential hypertension classified according to the World Health Organization (WHO) criteria (
Normotensive control participants were defined as those individuals with a SBP < 140 and DBP < 90 and no history of use of antihypertensive drugs.
We used the same rational described previously to choose the set of seven variants from six genes [
We hypothesized that seven variants from six major candidate genes directly involved in the (1) renin-angiotensin-aldosterone system: angiotensin-I-converting-enzyme (
DNA was extracted from whole blood using standard procedures. Genotypes for the
Data were expressed as means ± standard deviation (SD), median (interquartile range), or absolute number (percentage) when appropriate. The
We performed two approaches of analysis: (i) a family-based design and (ii) an unrelated case-control design (Figure S1, online only material).
Family-based association test (FBAT) analyses were performed with the FBAT program. The FBAT framework uses a variety of generalized linear models to perform tests similar to the transmission-disequilibrium test, allowing for the analysis of complex pedigrees [
Under a family-based design, we investigate all possible two-way multilocus effects using the recently proposed flexible family-based multifactor dimensionality reduction (FAM-MDR) approach [
In sensitivity analyses, we also performed analysis considering 384 unrelated participants. For single-locus analysis, we used the MAX3 statistic which selects the largest test statistic from the dominant, recessive, and additive models [
From a total of 1521 potentially eligible inhabitants across 12
Descriptive statistics of 652 individuals from 12
|
|
---|---|
Age (years) | 43.5 (17–91) |
Male (%) | 45.6% |
SBP (mmHg) | 131.8 ± 25.6 |
DBP (mmHg) | 82 ± 13.7 |
Hypertension* | 41.6% |
BMI (Kg/m2) | 24.7 ± 4.48 |
BMI > 25 Kg/m2 | 41.9% |
*Hypertension was defined as a SBP ≥ 140 mmHg and/or a DBP ≥ 90 mmHg or use of antihypertensive medications.
The prevalence of essential hypertension (EH) was 40.5% in women and 44.6% in men, a higher frequency when compared to admixed Brazilian populations [
In the single-locus FBAT analysis, no variant was significantly associated with any of the three examined blood pressure-related phenotypes, namely systolic and diastolic blood pressure (SBP and DBP, resp.) levels and EH (Table
FBAT analysis for blood pressure-related traits in
Gene | Variant | rs |
|
Phenotypes | |||
---|---|---|---|---|---|---|---|
EH | SBP | DBP | |||||
|
I/D | rs1799752 | 0.50 | NIF | 97 | 93 | 95 |
|
0.774 | 0.155 | 0.123 | ||||
|
Glu298Asp | rs1799983 | 0.16 | NIF | 63 | 63 | 61 |
|
0.101 | 0.050 | 0.079 | ||||
|
C825T | rs5443 | 0.58 | NIF | 86 | 84 | 82 |
|
0.793 | 0.417 | 0.342 | ||||
|
G-350A | rs5441 | 0.68 | NIF | 79 | 76 | 76 |
|
0.063 | 0.672 | 0.565 | ||||
|
M235T | rs699 | 0.27 | NIF | 96 | 92 | 94 |
|
0.608 | 0.293 | 0.579 | ||||
|
c.-154+20128C>A | rs3755351 | 0.34 | NIF | 97 | 93 | 95 |
|
0.196 | 0.209 | 0.137 | ||||
|
A486V | rs1801058 | 0.24 | NIF | 70 | 69 | 68 |
|
0.096 | 0.229 | 0.230 |
NIF: number of informative families.
The family-based multifactor dimensionality reduction (FAM-MDR) method was applied to blood pressure as a continuous phenotype (i.e., both SBP and DBP in mmHg). As shown in Table
Best two-locus models from FAM-MDR analyses.
Phenotype | Model |
|
|
---|---|---|---|
SBP |
|
8.203 | 0.533 |
|
6.318 | 0.819 | |
|
6.043 | 0.837 | |
|
4.172 | 0.981 | |
|
3.864 | 0.984 | |
|
3.146 | 0.997 | |
| |||
DBP |
|
14.064 | 0.040 |
|
8.500 | 0.460 | |
|
8.014 | 0.512 | |
|
7.970 | 0.512 | |
|
6.763 | 0.681 | |
|
6.364 | 0.732 |
SBP: systolic blood pressure. DBP: diastolic blood pressure.
We next performed a case-control study with unrelated subjects only, which yielded a total of 384 unrelated participants, whose clinical characteristics are shown in Table
Clinical characteristics of the unrelated
Normotensives ( |
Hypertensives ( |
| |
---|---|---|---|
Gender | |||
Men | 98 (47.57) | 79 (44.38) | 0.540 |
Women | 108 (52.43) | 99 (55.62) | |
Age (years) | 32 (24.5–44.7) | 55.7 (42.4–67.4) | <0.001 |
Adjusted SBP (mmHg)* |
|
|
<0.001 |
Adjusted DBP (mmHg)* |
|
|
<0.001 |
BMI (Kg/m2) |
|
|
<0.001 |
*Adjusted for the antihypertensive use according the proposition by Tobin et al
Table
Genotype frequencies for the seven studied polymorphisms in normotensive and hypertensive
Gene/Status | Genotype, |
|
|
| ||
---|---|---|---|---|---|---|
|
I/I | I/D | D/D | |||
Hypertensive | 39 (21.91) | 92 (51.69) | 47 (26.40) | 0.474 | 0.518 | 0.227 |
Normotensive | 54 (26.21) | 92 (44.66) | 60 (29.13) | |||
|
Glu/Glu | Glu/Asp | Asp/Asp | |||
Hypertensive | 127 (71.35) | 47 (26.40) | 4 (2.25) | 0.842 | 0.150 | 0.922 |
Normotensive | 151 (73.30) | 50 (24.27) | 5 (2.43) | |||
|
C/C | C/T | T/T | |||
Hypertensive | 42 (23.60) | 74 (41.57) | 62 (34.83) | 0.999 | 0.592 | 0.054 |
Normotensive | 22 (10.68) | 111 (53.88) | 73 (35.44) | |||
|
G/G | A/G | A/A | |||
Hypertensive | 93 (52.25) | 60 (33.71) | 25 (14.04) | 0.028 | 0.297 | 0.595 |
Normotensive | 106 (51.46) | 82 (39.81) | 18 (8.74) | |||
|
Thr/Thr | Thr/Met | Met/Met | |||
Hypertensive | 91 (51.12) | 73 (41.01) | 14 (7.87) | 0.425 | 0.259 | 0.482 |
Normotensive | 123 (59.71) | 68 (33.01) | 15 (7.28) | |||
|
A/A | A/C | C/C | |||
Hypertensive | 27 (15.17) | 94 (45.63) | 83 (46.63) | 0.256 | 0.661 | 0.565 |
Normotensive | 22 (10.68) | 68 (38.20) | 90 (43.69) | |||
|
Ala/Ala | Ala/Val | Val/Val | |||
Hypertensive | 98 (55.06) | 72 (40.45) | 8 (4.49) | 0.893 | 0.253 | 0.586 |
Normotensive | 117 (56.80) | 72 (34.95) | 17 (8.25) |
HWE: Hardy-Weinberg equilibrium;
Haplotypic analyses of both G-350A and C825T
Analysis of haplotypic effects (G-350A/C825T) on blood pressure-related phenotypes.
Trait |
|
Haplotypes | Haplotypic effects | LD measures | LR test | ||||
---|---|---|---|---|---|---|---|---|---|
G(-350A) | C825T | OR (95% CI) |
|
|
|
|
| ||
Hypertension | 384 | G | C | reference | — | 0.40 | 0.10 | 5.29 (3) | 0.152 |
G | T | 0.586 (0.358–0.958) | 0.033 | ||||||
A | C | 0.591 (0.336–1.040) | 0.068 | ||||||
A | T | 0.865 (0.451– 1.659) | 0.661 | ||||||
| |||||||||
|
|
||||||||
| |||||||||
SBP (mmHg) | 384 | G | C | reference | — | 0.40 | 0.10 | 7.04 (3) | 0.071 |
G | T | −2.208 (−6.734–2.318) | 0.339 | ||||||
A | C | −0.482 (−5.95–4.986) | 0.862 | ||||||
A | T | −1.361 (−7.39–4.67) | 0.658 | ||||||
| |||||||||
|
|
||||||||
| |||||||||
DBP (mmHg) | 384 | G | C | reference | — | 0.40 | 0.10 | 2.82 (3) | 0.420 |
G | T | −1.247 (−3.878–1.383) | 0.352 | ||||||
A | C | −2.328 (−5.447–0.792) | 0.143 | ||||||
A | T | −1.996 (−5.596–1.605) | 0.277 |
GMDR-based models were constructed to exhaustively identify all possible two- to four-locus models that potentially have an influence on the risk of hypertension and/or modulate SBP and DBP levels in
Best predictive models from GMDR analyses.
Model | Test accuracy | CVC |
|
---|---|---|---|
Hypertension | |||
|
0.4884 | 4 | 0.5324 |
|
0.5569 | 4 | 0.0792 |
|
0.5850 | 3 | 0.0222 |
| |||
Systolic blood pressure | |||
|
0.5220 | 5 | 0.2644 |
|
0.5752 | 5 | 0.0444 |
|
0.5282 | 4 | 0.2440 |
| |||
Diastolic blood pressure | |||
|
0.6031 | 9 | 0.0044 |
|
0.5677 | 3 | 0.0724 |
|
0.6347 | 6 | 0.0014 |
CVC: cross-validation consistency.
It has been shown that analyses focusing on biologically plausible candidate genes might be a strategy to circumvent the problem of limited coverage of current genome-wide association (GWA) platforms. This strategy holds the promise of increasing the density of markers in regions coding for known blood pressure-related effectors that are likely to be missed by GWA studies in hypertension [
Here, we present a comprehensive investigation about the potential role of seven variants (from six candidate genes directly involved in the blood pressure regulation) in the susceptibility of essential hypertension in a semi-isolated, African-derived population (
Using different methodological approaches (single-locus, haplotypic, and multilocus effects) and study designs (family-based and unrelated case-control designs), our investigation did not suggest an important contribution of the studied markers in the risk of essential hypertension in
Meta-analyses addressing the role of the C825T polymorphism at the
Recently, gene-centric approaches have revealed
Major limitations of our study include: first, the small number of participants investigated, that is, our samples are underpowered to detect typical main effects of genetic variants associated with common traits and second, lack of information on serum lipids as well as accurate information on smoking habits. In fact, lipids were not investigated due to technical restrictions. Furthermore, we observed that there was inaccurate reporting of smoking status by participants mainly due to heterogeneity in the definition of smoking. For example, a high proportion of participants reported to smoke homemade cigarettes, but failed to consider this as a “smoking habit”. As a result, both lipids and smoking status were not included in adjusted analyses.
Furthermore, our results should be carefully interpreted because correction for multiple testing was not performed in our analyses. As this is a hypothesis testing study, we considered that correction for multiple testing would be overly conservative and might lead to a substantial loss of statistical power.
Finally, it is worth of mentioning that the present study does not invalidate the hypothesis of gene-gene interactions/multilocus effects between or among the studied variants nor single-loci effects on blood pressure. Our null associations might plausibly be associated with low statistical power and the reduced panel of markers investigated.
In conclusion, our results do not support the hypothesis of significant association between the seven investigated variants and the risk of essential hypertension in
The authors thank all individuals from