Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

The association between apolipoprotein E gene polymorphisms and essential hypertension: a meta-analysis of 45 studies including 13 940 cases and 16 364 controls

Abstract

The apolipoprotein E single-nucleotide polymorphisms are among the potential candidate genes that may serve as modulators in susceptibility to essential hypertension. In an effort to clarify earlier inconclusive results, we performed a meta-analysis of population-based case–control genetic association studies. Random-effects methods were applied on summary data in order to combine the results of the individual studies. We identified in total 45 studies, including 13 940 hypertensive cases and 16 364 controls. The contrast of E4 carriers versus non-carriers yielded an overall odds ratio (OR) of 1.16 (95% confidence interval (CI): 1.02, 1.31), whereas the contrast of E4 allele versus the others in a subtotal of 6617 cases and 7330 controls, yielded an OR of 1.39 (95% CI: 1.12, 1.72). There was moderate evidence of publication bias in both contrasts, which was eliminated after excluding studies not in Hardy–Weinberg equilibrium. Subgroup analyses revealed that significant estimates arose from studies on Asian populations, as opposed to the Caucasian ones. Furthermore, no evidence of publication bias was demonstrated in the comparisons within this subgroup. Our results are consistent with recent meta-analyses but show that the association is weaker than that has been previously demonstrated. Further studies are needed in order to fully address questions about the etiological mechanism of the particular association, as well as to study the effect in populations of African descent.

Introduction

A family history of hypertension is present in as many as 70–80% of all patients with essential hypertension and in 50% of hypertensive children. It has been estimated that genetic factors account for 30 of the variation in blood pressure (BP) in various populations and as much as 60–70 of hypertension in families.1

Although a number of individual genes and genetic factors have been linked to the development of essential hypertension, multiple genes most likely contribute to the development of the disease in a particular individual. Apolipoprotein E (ApoE) was first identified in 1973 in human very low density lipoprotein2 and has been implicated in the development of hypertension. ApoE is the primary ligand for two receptors, the low density lipoprotein (LDL) receptor found on the liver and other tissues, and an apoE-specific receptor found on the liver. Allelic variation in apoE is consistently associated with plasma concentrations of total cholesterol, LDL cholesterol and apolipoprotein B (the major protein of LDL, very low density lipoproteinL and chylomicrons). There are three different apoE alleles in humans: E2, which has cysteine residues at positions 112 and 158; E3, which occurs in 60–80 of Caucasians and has cysteine at position 112 and arginine at position 158; and E4, which has arginine residues at positions 112 and 158. The various apoE isoforms interact differently with specific lipoprotein receptors, ultimately altering circulating levels of cholesterol. While apoE4 and E3 bind with approximately equal affinity to lipoprotein receptors, apoE2 binds with <2 of this strength.3 The difference in uptake of postprandial lipoprotein particles results in differences in regulating hepatic LDL receptors, which in turn contributes to genotypic differences in total and LDL cholesterol levels via the lipid-lowering effect of apoE2 and the lipid-increasing effect of apoE4.

Dyslipidemia and hypertension occur together more often than can be explained by chance. Adverse lipid levels appear to predate the onset of hypertension by years. Atherogenic dyslipidemias could lead to hypertension by several mechanisms including atherosclerosis leading to reduced elasticity, endothelial dysfunction resulting in reduced nitric oxide production and lipid-mediated damage to the renal microvasculature.4 Finally, dyslipidemia and hypertension represent two of several components of the metabolic syndrome that may share common mechanistic pathways.

Smoking-related oxidative stress causes tertiary conformational alterations and disrupts the functional integrity of apoE, which likely interferes with its role in regulating plasma cholesterol homeostasis. Studies which explored the gene–environment interactions between apoE gene polymorphisms and cigarette smoking in coronary heart disease (CHD) patients showed that smoking interacted with apoE on CHD risk where the E2 allele counteracted CHD risk from smoking, while the E4 allele was seen to potentiate this risk.5 However, it is not known whether the synergistic effect between smoking and apoE influences hypertension.

Several studies have evaluated the relationship between apoE genotypes (particularly the E4 allele) and the incidence of hypertension, often yielding conflicting results. However, these reports may have been both underpowered to detect the true relationship and also subject to publication bias. To overcome these issues, we conducted a carefully designed and complete meta-analysis of all available population-based case–control (either a retrospective or a cross-sectional one) genetic association studies relating apoE polymorphisms and hypertension.

Materials and methods

Retrieval of published studies

We used PubMed (http://www.ncbi.nlm.nih.gov/sites/entrez) and Scopus (http://www.scopus.com) to identify case–control studies containing data of apoE genotypes on hypertension among unrelated subjects. We additionally required that there were sufficient published data on the genotypes or allele frequencies for determining an estimate of relative risk (that is, odds ratio (OR)) and its corresponding confidence interval (CI). We extended our search in studies correlating apoE with other morbidities if data associating apoE and hypertension were present. Studies evaluating secondary hypertension or other types of monogenic hypertension were excluded. The queries used for the search were ‘Apolipoprotein E or ApoE’ combined with ‘gene or polymorphism or variant or allele or mutant or mutation’ and ‘hypertension or BP or systolic BP (SBP)’, all of which were MeSH terms (Medical Subject Headings in the US National Library of Medicine). We also searched publications mentioned in the ‘related articles’ option in MEDLINE (Bethesda, MD, USA), as well as reference lists of all retrieved studies to identify for other relevant publications. If there were multiple publications from the same study group, the most complete and recent results were used. We did not restrict on the basis of the language and we translated studies in languages other than English in order to avoid local literature bias.6 To avoid selection bias, no study was rejected because of poor quality data.7 Furthermore, no quality scoring was attempted, as it involves subjective assignment of points, and modern approaches advocate against it.7 Instead, we considered as a better choice to try investigating the sources of heterogeneity directly, incorporating various study-level covariates in a meta-regression, as well as performing subgroup and sensitivity analyses (see Statistical analysis section). We conducted a systematic computerized literature search for papers published as of 20 June 2010. The full text of the retrieved articles was read to decide whether information on the topic of interest was included.

Data extraction

Two reviewers (SS, PGB) independently examined the retrieved articles in order to extract the information needed using a data collection form. From each study the following data were abstracted: publication data, first author’s last name, year of publication, ethnicity and country of the population studied, study design, the number of individuals in both the case and control groups with each different genotype tested and average case and control characteristics (systolic BP, diastolic BP, anti-hypertensive medications, possible co-morbidities, smoking habits, alcohol consumption, latitude, details about matching of cases and controls) within each study. Information on Hardy–Weinberg equilibrium (HWE) among the controls was collected or calculated if the genotype data were available. Following data extraction, discordances were adjudicated by discussion between the authors and a consensus was reached.

Statistical analysis

The OR was used to compare contrasts of the alleles and genotypes between cases and controls. In case of zero cell counts a continuity correction was applied by adding 0.5 to all the cells of the contingency table. Initially, for avoiding multiple comparisons comparing the effect of the genotypes against a reference genotype, we used a multivariate random-effects method of meta-analysis that takes into account the pairwise correlations of the ORs.8 Genotype E3/3 was chosen as reference category (baseline) for this analysis as it is the most common among the healthy and diseased subjects with a frequency of about 67%, as well as because the literature suggests that this is the ‘wild-type’ genotype. Afterwards, we proceeded by grouping the genotypes and allele carriers in order to derive a summary OR for the most likely genetic model of inheritance. We calculated combined ORs along with their 95% CIs for each genotype or allele contrast (that is, E2 carriers versus the others, E2 alleles versus the others and so on) using a standard random-effects method.9 Subgroup analyses were conducted appropriately in order to investigate the effect of dichotomous variables (racial descent of the populations, studies involving subjects with co-morbidities, deviations from HWE, and so on), whereas meta-regression analysis was carried out for continuous variables. The inconsistency index I2 (ranging from 0–100%) was also calculated for the percentage of observed between-study variability that was due to heterogeneity rather than chance.10 χ2 tests for heterogeneity were also performed to assess between-study and between-group heterogeneity.

The combined ORs along with their corresponding 95% CIs, were estimated using the random-effects method,9 which in the presence of heterogeneity is more appropriate as it is prudent to take into account an estimate of the between-study variance (τ2). If heterogeneity is absent, τ2 equals zero and thus the random-effects and the fixed-effects models provide essentially the same results. Using the random-effects models, we could also extend the analysis by incorporating various study-level covariates as linear predictors, in order to estimate the extent to which these covariates explain the observed heterogeneity. This approach thus yields a random-effects meta-regression,11 where the component of between-studies variance is estimated by the restricted maximum likelihood method. Random-effects meta-regressions were performed using as covariates various average study characteristics, such as the mean SBP, mean diastolic BP, anti-hypertensive medications, co-morbidities, the latitude of the city in which the study was conducted and so on.

Publication bias or other small-study-related bias was evaluated using the rank correlation method of Begg and Mazumbar.12 For the same purpose, we also conducted a random-effects weighted regression11 of logORi against its estimated standard error (sei). This method constitutes the random-effects analog of the Egger’s regression method13 and it was performed as the fixed-effects regression method of Egger might provide false-positive evidence for publication bias in the presence of heterogeneity. In addition, clustering the available studies according to their sample size and performing subgroup analyses should be considered as an equivalent (and may be more crude) approach.

In an attempt to identify potential influential studies, separate analyses were undertaken. This was achieved by performing a sensitivity analysis, in which we calculated the random-effects estimates by removing an individual study each time and then checked whether any of these estimates can bias the results if the overall significance of the estimate is altered. Furthermore, as another kind of sensitivity analysis, we undertook separate analyses for ‘normal’ hypertensive subjects excluding studies performed on various populations in which the cases or controls suffered from other diseases (for instance familial hypercholesterolemia, CHD or cerebrovascular disease) or in populations in which the genotype distribution in controls deviated from the HWE.14 For testing deviations from the HWE in controls and given that we have to deal with multiple alleles, we used a specialized method along with the accompanied software (http://www.biology.ualberta.ca/jbrzusto/hwenj.html).15

Likewise, cumulative meta-analysis16 was performed in order to identify a possible trend of the combined estimate over years, a situation that often introduces a special form of bias (the so called ‘Proteus phenomenon’ or the ‘winner’s curse’) in genetic association studies.17 For the detection of the time-trend, we used the standard cumulative meta-analysis approach,16 which consists of visually inspecting the plot and a recently proposed regression-based method.18

For all analyses performed, the statistical package Stata 10 (Stata Corporation, College Station, Texas, USA) was used. In all analyses statistically significant results were declared as those with a P-value<0.05.

Results

The initial literature search yielded 421 published articles. After that and the subsequent screening, we came up with 45 research studies reporting data for a healthy (normotensive) group and a hypertensive patients group. Some of these papers contained information about populations of non-distinct racial descent and thus they were considered as mixed populations.19, 20, 21, 22 Two initially identified studies were excluded, as they contained overlapping subsets of individuals with other studies already included.22, 23 Several studies were discarded because they did not report data that could be used in the analysis (that is, the genotypes, the E4 carriers or the E4 allele frequencies for the contrasts between cases and controls) or they did not report data for the control group. Usually, these were studies in which hypertension was not among the primary end points.

The identified studies contained in total information for 16 364 healthy (normotensive) subjects and 13 940 hypertensive patients. Two of the identified studies used a family-based design20, 24 and one used a subset of individual twin participants.25 Two studies were written in languages other than English (one in French and one in Portuguese)26, 27 and these were retrieved, translated and included in the analysis. We found 27 studies on Caucasian or European-descent populations, 14 studies on Asian-descent populations, whereas 4 studies reported data incorporating populations of mixed origin (White/African-American/Hispanic/Asian/other or recruitment from different countries or no data for the racial descent). Only 21 of the included studies contained complete information about the patients’ and controls’ genotypes. For these studies we found that, in six studies the control groups were indeed in HWE.23, 28, 29, 30, 31, 32 For the remaining 24 studies that did not report full genotype data, it is apparent that deviations from HWE could not be detected. The detailed characteristics of each study (country conducted, racial descent of subjects involved, characteristics of cases and controls, sample size, and so on.) are summarized in Table 1 and the results according to the populations studied in Table 2. The pooled frequency of the E4 allele in hypertensive cases was 11.5% and in normotensive controls 10.4%. However, in the control subjects of Caucasian populations it was found to be 12.9%, whereas in Asians 9.4%.

Table 1 Detailed characteristics of the studies included in the meta-analysis
Table 2 Results of subgroup analyses and heterogeneity diagnostics

In two of the contrasts examined, significant overall associations were found between the apoE polymorphism and the risk for hypertension (Figures 1 and 2). The multivariate random-effects method8 yielded a non-statistically significant OR for the contrasts of each of the five different genotypes compared with the wild-type E3/3 but it should be mainly attributed to the small sample size used (17 studies). We then, proceeded by collapsing the genotypes and performing traditional univariate meta-analyses using standard random-effects methods.9 The contrast of the E4 carriers versus non-carriers yielded a significant estimate (OR 1.16, 95% CI: 1.02–1.31, P-value=0.021) while populations of Asian descent constantly presented a more prominent risk (OR 1.41, 95% CI: 1.05–1.90, P-value=0.022) in a total of 13 940 cases and 16 364 controls. In this analysis, which was based on a total of 45 studies, there was clear evidence of heterogeneity between racial groups (P-value=0.039) indicating that there are significant differences of the estimates according to racial descent. The allele-based contrast, based on data from 21 studies, revealed also a statistically significant OR for the contrast of E4 allele versus the others (OR 1.39, 95% CI: 1.12–1.72, P-value=0.003) for the overall population and an OR equal to 1.49 (95% CI: 1.05–2.13, P-value=0.026) for the Asians in a subtotal of 6617 cases (13 234 alleles) and 7330 controls (14 660 alleles). However, in this analysis differences between races did not reach statistical significance (P-value=0.45). Nevertheless, the smaller subset of studies in this analysis should be held responsible for this result. In both contrasts considered (E4 carriers versus non-carriers and E4 allele versus the others) there was a statistically significant between-study heterogeneity as indicated by the P-value of the corresponding test (P<0.001 in both cases) and the large value of the I2 index (68.9 and 79.2%, respectively).

Figure 1
figure1

Forest plot for the results of the meta-analysis of the E4 versus non-E4 carriers contrast. The random-effects method of DerSimonian and Laird was used with inverse-variance weights. The size of the estimate of each study is inversely proportional to its variance. We also list the results of subgroup analyses and heterogeneity diagnostics.

Figure 2
figure2

Forest plot for the results of the meta-analysis of the E4 allele versus the other alleles contrast. The random-effects method of DerSimonian and Laird was used with inverse-variance weights. The size of the estimate of each study is inversely proportional to its variance. We also list the results of subgroup analyses and heterogeneity diagnostics.

In the influential analysis of E4 carriers versus non-carriers, there was evidence for two studies that had a greater effect to the overall significance of the estimates. The first study was that of Niu et al.,32 which if removed produced an overall estimate of 1.09 (95% CI: 0.98, 1.21) and the second one was that of Dembinska-Kiec et al.,33 which, if removed produced an overall estimate of 1.12 (95% CI: 0.99, 1.25). The same studies influenced the analysis of E4 allele versus the others and produced an overall estimate of 1.24 (95% CI: 1.04, 1.49) and 1.30 (95% CI: 1.06, 1.59), respectively, when removed. After removing each of the remaining studies and recalculating the overall estimate and the 95% CI for the remaining studies, the significance of the OR remained nearly unchanged.

In the cumulative meta-analysis of both contrasts, however, there was strong evidence suggesting that the first published study (Eto et al.34) that reported a significant association for the E4 carriers, or the second one published (Isbir et al.28) for the E4 alleles triggered the subsequent publication of other studies that tried to replicate the initial results. Even though, for both contrasts, the overall estimate obtained by the meta-analysis is significantly different than that of the first or the second study, respectively (0.67 versus 1.16 for the E4 carriers and 1.78 versus 1.39 for the E4 allele contrast). The inspection of the cumulative meta-analysis plots for the E4 carriers showed no evidence for trend of the effect estimates over time and the same conclusions were drawn from the formal regression-based statistical tests18 (Figure 3).

Figure 3
figure3

Cumulative meta-analysis plot for the results of the meta-analysis of the E4 versus non-E4 carriers contrast. The studies are sorted by the year of publication. A slope significantly different from zero indicates time-trend related bias. The regression-based test indicated no such bias. Vertical lines represent the 95% CI. The two regression lines, excluding the first study and including all studies, nearly coincide.

Performing subgroup analyses for the two contrasts we found that there were significant differences between subjects of Caucasian and Asian racial descent, as the sub-analyses of Asians indicated a significant association in both contrasts (P-values 0.022 and 0.026, respectively), whereas the studies involving Caucasians yielded roughly significant estimates in E4 allele contrasts (P-value 0.057) and non-significant in E4 carriers contrasts (P-value 0.293) (Figures 1 and 2). Furthermore, in both contrasts, studies in HWE produced a significant association whereas studies not in HWE did not. Excluding studies performed on hypertensive populations with co-morbidities (patients suffering from familial hypercholesterolemia, CHD, cerebrovascular disease, and so on) did not influence both contrasts significantly (Figures 1 and 2). Performing random-effects meta-regressions on various study-level covariates (SBP, diastolic BP, anti-hypertensive medications, latitude) in the subset of studies that reported these figures also failed to provide any significant findings. We also, intended to perform analyses for environmental factors that might influence the outcome such as smoking and alcohol consumption but the available data precluded these analyses. For instance, only three studies provided data for smoking habits for both apoE and cases/controls. Similarly, only one study provided data for smoking as a potential confounder of the apoE-hypertension association and another one included data for alcohol consumption between apoE genotypes.

There was moderate evidence for publication or other small-study-related bias in both contrasts examined. When we restricted the analysis to studies on populations that were in HWE plus those in which deviations from HWE could not be detected, any evidence for publication bias was removed using Begg’s test (P-value 0.087 for the E4 carriers contrasts and P-value 0.452 for the E4 allele contrasts) or the random-effects regression method (P-values equal to 0.193 and 0.215, respectively) (Figures 4 and 5). Subgroup analyses revealed no publication bias in populations of Asian descent.

Figure 4
figure4

Funnel plot for the results of the meta-analysis of the E4 versus non-E4 carriers contrast including all studies. Asymmetry of the plot indicates publication or other small studies related bias. The results of the three formal tests for detecting such bias are listed.

Figure 5
figure5

Funnel plot for the results of the meta-analysis of the E4 versus non-E4 carriers contrast excluding studies deviating from HWE. Asymmetry of the plot indicates publication or other small studies related bias. The results of the three formal tests for detecting such bias are listed.

Performing subgroup analyses after grouping the studies according to the number of included cases yielded a positive trend for the E4 carriers versus non-carriers contrast, as the OR for studies with fewer than 200 cases was equal to 1.35 (95% CI: 1.07, 1.69), the OR for studies with cases ranging between 200 and 500 was 1.14 (95% CI: 0.85, 1.54) and the OR for studies with more than 500 cases was 0.96 (95% CI: 0.89, 1.04). This is another way of looking at the results for a possible publication bias, as this analysis suggests that smaller studies (with large variance) are more likely to favor the investigated association (larger OR). In the contrast of E4 allele versus the others, the smaller studies once again yielded a larger OR (1.77 (95% CI: 1.36, 2.31)), the intermediate a smaller OR (1.45 (95% CI: 0.58, 3.65)) and the larger studies an even smaller one (0.94 (95% CI: 0.82, 1.06)).

Discussion

To our knowledge, this is the third published meta-analysis investigating the association between apoE gene polymorphisms and hypertension. In two recently published meta-analyses performed by the same research team, an attempt to explore this association has been performed, but the population studied was significantly smaller, generated wide CIs indicative of insufficient study power and overestimated the correlation giving a twofold increased risk of developing hypertension in Chinese individuals carrying the E4 allele.35, 36 The meta-analysis performed here includes in total 45 studies with 13 940 cases and 16 364 controls and incorporated various parameters, the influence of which is often underestimated. These include the search for ‘gray’ literature,37 the inclusion of non-English research papers6 and the sensitivity analysis of studies not in HWE.14 A problem in pooling data we have to address here is that of unpublished data. This problem relates to studies being executed, but not reported, usually because a significant association has not been found. Meta-analyses that exclude unpublished data likely overrepresent studies with statistically significant findings and inflate effect size estimates.37 We tried to retrieve data from conference abstracts/presentations, supplements, proceedings of meetings or symposiums but our search yielded a limited number of data, most of which were incomplete or non-accessible. Additionally, we could not identify any registries with gene–gene or gene–environment interactions and hypertension.

We have shown that the E4 allele is associated with an increased risk for hypertension. We identified two influential studies (Niu et al.32 and Dembinska-Kiec et al.33) both of which were performed in ‘normal’ populations with essential hypertension (Chinese and Polish, respectively) and, in the cumulative meta-analysis of both contrasts they were clearly responsible for more significant evidence favoring the association. The common feature of these studies is the significant discordance in the E4 allele frequency between controls and cases (5.72% in controls versus 20.07% in cases and 15.38% versus 40.83%, respectively). We should emphasize that when including only the studies in HWE and the studies performed on populations in which deviations from HWE could not be detected, the evidence of publication bias is significantly reduced, the results are more homogeneous and the magnitude or the significance of the association is not altered. In conclusion, the indications of publication bias, although disturbing and worth noticing, should be attributed mainly to the heterogeneity of the results.

Subgroup analysis including studies not in HWE failed to reproduce a relationship between apoE and hypertension resulting in a non-significant estimate for both contrasts (Figures 1 and 2). This difference may be attributed to the fact that hypertension was defined as SBP>140mmHg in the majority of studies in HWE (5 out of 6) but only in 8 out of 15 studies not in HWE, while the rest of them used higher levels of SBP. Moreover, studies in HWE concern mainly ‘normal’ individuals with hypertension whereas studies not in HWE include hypertensive subjects with co-existing diseases. Thus, we may conclude that the confounding effect of HWE found in this meta-analysis could be explained by the different characteristics of the individuals included. Control subjects in studies not in HWE demonstrate higher BP levels and a greater morbidity compared with those in HWE.

In both contrasts examined, a constant finding was the more prominent association of the polymorphism with hypertension in Asian populations, as well as the homogenously statistically significant results, which in all cases yielded a P-value<0.03, though χ2 test for heterogeneity between the racial groups did not reach statistical significance in E4 allele contrast. No publication biases were demonstrated in the comparisons within this subgroup. Neither the Begg’s test nor the more sensitive random-effects regression method showed evidence of publication bias in E4 carriers contrasts (P-values equal to 0.827 and 0.814, respectively) and in E4 allele contrasts (P-values equal to 0.602 and 0.285, respectively). Based on the results obtained from the larger subset of 45 studies (that is, the E4 carriers contrast) we found that the risk for hypertension is higher mainly in populations of Asian origin. This is an important finding, suggesting that populations of Asian origin are at elevated risk for hypertension due to the presence of the E4 allele, in contrast to Caucasians. The reasons for this discrepancy may be attributed to other, not yet determined, genetic factors influencing the outcome, to some other unmeasured environmental variable or to lifestyle preferences. As a matter of fact, a large multi-center study has shown that there is an inter-correlation between SBP, the frequency of some identifiable alleles and the geographic latitude, and that these latter factors account for a substantial proportion of the worldwide variation of BP, suggesting that the current epidemic of hypertension is due to exposures of the modern period interacting with ancestral susceptibility.38 The epidemiological causability relationship deciphered by this work concludes that an increase in latitude is associated with lower frequency of these alleles, as well as with increased SBP. The so called ‘negative confounding’ effect does not occur with apoE isoforms, as at high latitudes E4 allele frequency tends to be high and SBP is lower.

In our meta-analysis, data for smoking were available in 22 out of the 45 studies. The distribution of smoking across apoE genotypes or alleles was provided in 18 studies but only 10 of them included data for smoking between cases and controls. In the majority of these studies, the influential effect of smoking to the association between apoE and disease could not be estimated as data for smoking were given separately for apoE and cases/controls groups, respectively. Additionally, in all apart from one study,39 hypertension was not of the primary end points. Conclusions related to smoking as a potent confounder of the apoE-hypertension association could not be elicited directly from this meta-analysis.

Our study has a number of limitations. We did not manage to include unpublished data but the large number of cases and controls compared with previous meta-analyses gives sound evidence for the validity of the associations examined. Additionally, the nature of our study does not allow us to make inferences about causation, but merely the description of associations. Furthermore, we were unable to retrieve data on various potential confounders (total cholesterol (TC), LDL, high density lipoprotein (HDL), triglycerides (TG), and so on.) from the original publications, and only nine of them reported all the appropriate figures for the pre-specified covariates. The heterogeneous criteria used for the definition of hypertension (average BP>140/90mmHg, or 150/90, or 160/90, or >160/95, or the use of anti-hypertensive medications) had a variable effect in the dichotomous classification of subjects into cases and controls. We attempted to correct for major confounders but we cannot rule out the effect of residual confounding by additional factors such socio-economic status, exercise or various lifestyle and dietary habits.

In conclusion, our data are in agreement with the view that carriers of the apoE4 allele demonstrate an increased risk of hypertension and this correlation is more prominent in populations of Asian descent, although this relationship is less marked than the previous meta-analyses have demonstrated. A causability relationship associates apoE gene polymorphisms with dyslipidemia and subsequently with hypertension; the biological mechanism of the proposed association remains to be elucidated. Additional and more carefully designed cross-sectional and prospective studies are needed in order to establish a more consistent view of these interrelations. For instance, performing large genetic association studies in hypertensive patients and normotensive controls, stratified by their lipid profile (TC, LDL, HDL and TG) will minimize the potential confounding by other factors predisposing to components of the metabolic syndrome. The underlying molecular causal pathways that confer susceptibility to hypertension are warranted to be established to provide biological or clinical validations of our findings.

References

  1. 1

    Ward R . Familial aggregation and genetic epidemiology of blood pressure. In: Laragh J, Brenner B, (eds). Hypertension-Pathophysiology, Diagnosis and Management. Raven Press: New York, 1990 pp 81.

    Google Scholar 

  2. 2

    Shore VG, Shore B . Heterogeneity of human plasma very low density lipoproteins. Separation of species differing in protein components. Biochemistry 1973; 12 (3): 502–507.

    CAS  Article  Google Scholar 

  3. 3

    Siest G, Pillot T, Regis-Bailly A, Leininger-Muller B, Steinmetz J, Galteau MM et al. Apolipoprotein E: an important gene and protein to follow in laboratory medicine. Clin Chem 1995; 41 (8 Pt 1): 1068–1086.

    CAS  PubMed  Google Scholar 

  4. 4

    Schaeffner ES, Kurth T, Curhan GC, Glynn RJ, Rexrode KM, Baigent C et al. Cholesterol and the risk of renal dysfunction in apparently healthy men. J Am Soc Nephrol 2003; 14 (8): 2084–2091.

    CAS  PubMed  PubMed Central  Google Scholar 

  5. 5

    Zak I, Niemiec P, Balcerzyk A, Krauze J . Combined ‘pro-atherosclerotic’ variants of the ACE and APOE genes increase the risk of the coronary artery disease associated with the presence of cigarette smoking. Acta Cardiol 2008; 63 (6): 741–747.

    Article  Google Scholar 

  6. 6

    Pan Z, Trikalinos TA, Kavvoura FK, Lau J, Ioannidis JP . Local literature bias in genetic epidemiology: an empirical evaluation of the Chinese literature. PLoS Med 2005; 2 (12): e334.

    Article  Google Scholar 

  7. 7

    Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283 (15): 2008–2012.

    CAS  Article  Google Scholar 

  8. 8

    Bagos PG . A unification of multivariate methods for meta-analysis of genetic association studies. Stat Appl Genet Mol Biol 2008; 7 (1): Article31.

  9. 9

    DerSimonian R, Laird N . Meta-analysis in clinical trials. Control Clin Trials 1986; 7 (3): 177–188.

    CAS  Article  Google Scholar 

  10. 10

    Higgins JP, Thompson SG, Deeks JJ, Altman DG . Measuring inconsistency in meta-analyses. BMJ 2003; 327 (7414): 557–560.

    Article  Google Scholar 

  11. 11

    Thompson SG, Sharp SJ . Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med 1999; 18 (20): 2693–2708.

    CAS  Article  Google Scholar 

  12. 12

    Begg CB, Mazumdar M . Operating characteristics of a rank correlation test for publication bias. Biometrics 1994; 50 (4): 1088–1101.

    CAS  Article  Google Scholar 

  13. 13

    Egger M, Davey Smith G, Schneider M, Minder C . Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315 (7109): 629–634.

    CAS  Article  Google Scholar 

  14. 14

    Trikalinos TA, Salanti G, Khoury MJ, Ioannidis JP . Impact of violations and deviations in Hardy-Weinberg equilibrium on postulated gene-disease associations. Am J Epidemiol 2006; 163 (4): 300–309.

    Article  Google Scholar 

  15. 15

    Guo SW, Thompson EA . Performing the exact test of Hardy-Weinberg proportion for multiple alleles. Biometrics 1992; 48 (2): 361–372.

    CAS  Article  Google Scholar 

  16. 16

    Lau J, Schmid CH, Chalmers TC . Cumulative meta-analysis of clinical trials builds evidence for exemplary medical care. J Clin Epidemiol 1995; 48 (1): 45–57 discussion 59-60.

    CAS  Article  Google Scholar 

  17. 17

    Ioannidis JP, Trikalinos TA . Early extreme contradictory estimates may appear in published research: the Proteus phenomenon in molecular genetics research and randomized trials. J Clin Epidemiol 2005; 58 (6): 543–549.

    Article  Google Scholar 

  18. 18

    Bagos PG, Nikolopoulos GK . Generalized least squares for assessing trends in cumulative meta-analysis with applications in genetic epidemiology. J Clin Epidemiol 2009; 62 (10): 1037–1044.

    Article  Google Scholar 

  19. 19

    Terry JG, Howard G, Mercuri M, Bond MG, Crouse JR . Apolipoprotein E polymorphism is associated with segment-specific extracranial carotid artery intima-media thickening. Stroke 1996; 27 (10): 1755–1759.

    CAS  Article  Google Scholar 

  20. 20

    Djousse L, Myers RH, Province MA, Hunt SC, Eckfeldt JH, Evans G et al. Influence of apolipoprotein E, smoking, and alcohol intake on carotid atherosclerosis: National Heart, Lung, and Blood Institute Family Heart Study. Stroke 2002; 33 (5): 1357–1361.

    Article  Google Scholar 

  21. 21

    Sachdev PS, Parslow R, Wen W, Anstey KJ, Easteal S . Sex differences in the causes and consequences of white matter hyperintensities. Neurobiol Aging 2009; 30 (6): 946–956.

    CAS  Article  Google Scholar 

  22. 22

    Fuzikawa AK, Peixoto SV, Taufer M, Moriguchi EH, Lima-Costa MF . Association of ApoE polymorphisms with prevalent hypertension in 1406 older adults: the Bambui Health Aging Study (BHAS). Braz J Med Biol Res 2008; 41 (2): 89–94.

    CAS  Article  Google Scholar 

  23. 23

    Resnick HE, Rodriguez B, Havlik R, Ferrucci L, Foley D, Curb JD et al. Apo E genotype, diabetes, and peripheral arterial disease in older men: the Honolulu Asia-aging study. Genet Epidemiol 2000; 19 (1): 52–63.

    CAS  Article  Google Scholar 

  24. 24

    Profenno LA, Faraone SV . Diabetes and overweight associate with non-APOE4 genotype in an Alzheimer’s disease population. Am J Med Genet B Neuropsychiatr Genet 2008; 147B (6): 822–829.

    Article  Google Scholar 

  25. 25

    Carmelli D, Swan GE, Reed T, Miller B, Wolf PA, Jarvik GP et al. Midlife cardiovascular risk factors, ApoE, and cognitive decline in elderly male twins. Neurology 1998; 50 (6): 1580–1585.

    CAS  Article  Google Scholar 

  26. 26

    Bennouar N, Allami A, Laraqui A, Azeddoug H, El-Kadiri N, Benkouka F et al. [Apolipoprotein E and angiotensin-converting enzyme gene polymorphisms as risk factors of coronary disease]. Ann Biol Clin 2004; 62 (3): 295–304.

    CAS  Google Scholar 

  27. 27

    Carmo MM, Lima FL, Rodrigues MO, Albergaria I, Fonseca A . Influence of the APOE genotypes in some atherosclerotic risk factors. Acta Med Port 2008; 21 (5): 433–440.

    Google Scholar 

  28. 28

    Isbir T, Yilmaz H, Bihorac A, Akoglu E . Mild-to-moderate hypertension and apolipoprotein E gene polymorphism. Am J Hypertens 1997; 10 (7 Part 1): 827–828.

    CAS  PubMed  Google Scholar 

  29. 29

    Corbo RM, Vilardo T, Ruggeri M, Gemma AT, Scacchi R . Apolipoprotein E genotype and plasma levels in coronary artery disease. A case-control study in the Italian population. Clin Biochem 1999; 32 (3): 217–222.

    CAS  Article  Google Scholar 

  30. 30

    Yilmaz H, Isbir T, Agachan B, Aydin M . Is epsilon4 allele of apolipoprotein E associated with more severe end-organ damage in essential hypertension? Cell Biochem Funct 2001; 19 (3): 191–195.

    CAS  Article  Google Scholar 

  31. 31

    Li X, Du Y, Huang X . Association of apolipoprotein E gene polymorphism with essential hypertension and its complications. Clin Exp Med 2003; 2 (4): 175–179.

    CAS  Article  Google Scholar 

  32. 32

    Niu W, Guo X, Su Y, Qiu C . Apolipoprotein E and low-density lipoprotein receptor gene polymorphisms in dyslipidemias-associated essential hypertension. J Hum Hypertens 2007; 21 (4): 337–339.

    CAS  Article  Google Scholar 

  33. 33

    Dembinska-Kiec A, Kawecka-Jaszcz K, Kwasniak M, Guevara I, Pankiewicz J, Malczewska-Malec M et al. Apo E isoforms, insulin output and plasma lipid levels in essential hypertension. Eur J Clin Invest 1998; 28 (2): 95–99.

    CAS  Article  Google Scholar 

  34. 34

    Eto M, Watanabe K, Chonan N, Ishii K . Familial hypercholesterolemia and apolipoprotein E4. Atherosclerosis 1988; 72 (2-3): 123–128.

    CAS  Article  Google Scholar 

  35. 35

    Niu W, Qi Y, Qian Y, Gao P, Zhu D . The relationship between apolipoprotein E epsilon2/epsilon3/epsilon4 polymorphisms and hypertension: a meta-analysis of six studies comprising 1812 cases and 1762 controls. Hypertens Res 2009; 32 (12): 1060–1066.

    CAS  Article  Google Scholar 

  36. 36

    Niu WQ, Qi Y . Meta-based evidence for apolipoprotein E varepsilon2/varepsilon3/varepsilon4 polymorphism in association with hypertension among Chinese. J Hum Hypertens 2011; 25 (12): 725–731.

    CAS  Article  Google Scholar 

  37. 37

    Conn VS, Valentine JC, Cooper HM, Rantz MJ . Grey literature in meta-analyses. Nurs Res 2003; 52 (4): 256–261.

    Article  Google Scholar 

  38. 38

    Young JH, Chang YP, Kim JD, Chretien JP, Klag MJ, Levine MA et al. Differential susceptibility to hypertension is due to selection during the out-of-Africa expansion. PLoS Genet 2005; 1 (6): e82.

    Article  Google Scholar 

  39. 39

    Karvonen J, Kauma H, Kervinen K, Ukkola O, Rantala M, Paivansalo M et al. Apolipoprotein E polymorphism affects carotid artery atherosclerosis in smoking hypertensive men. J Hypertens 2002; 20 (12): 2371–2378.

    CAS  Article  Google Scholar 

  40. 40

    Kalmijn S, Feskens EJ, Launer LJ, Kromhout D . Cerebrovascular disease, the apolipoprotein e4 allele, and cognitive decline in a community-based study of elderly men. Stroke 1996; 27 (12): 2230–2235.

    CAS  Article  Google Scholar 

  41. 41

    Schmidt R, Schmidt H, Fazekas F, Schumacher M, Niederkorn K, Kapeller P et al. Apolipoprotein E polymorphism and silent microangiopathy-related cerebral damage. Results of the Austrian Stroke Prevention Study. Stroke 1997; 28 (5): 951–956.

    CAS  Article  Google Scholar 

  42. 42

    Kessler C, Spitzer C, Stauske D, Mende S, Stadlmuller J, Walther R et al. The apolipoprotein E and beta-fibrinogen G/A-455 gene polymorphisms are associated with ischemic stroke involving large-vessel disease. Arterioscler Thromb Vasc Biol 1997; 17 (11): 2880–2884.

    CAS  Article  Google Scholar 

  43. 43

    Kimura H, Suzuki Y, Gejyo F, Karasawa R, Miyazaki R, Suzuki S et al. Apolipoprotein E4 reduces risk of diabetic nephropathy in patients with NIDDM. Am J Kidney Dis 1998; 31 (4): 666–673.

    CAS  Article  Google Scholar 

  44. 44

    Niemi M, Kervinen K, Rantala A, Kauma H, Paivansalo M, Savolainen MJ et al. The role of apolipoprotein E and glucose intolerance in gallstone disease in middle aged subjects. Gut 1999; 44 (4): 557–562.

    CAS  Article  Google Scholar 

  45. 45

    Dzimiri N, Meyer BF, Hussain SS, Basco C, Afrane B, Halees Z . Relevance of apolipoprotein E polymorphism for coronary artery disease in the Saudi population. Arch Pathol Lab Med 1999; 123 (12): 1241–1245.

    CAS  PubMed  Google Scholar 

  46. 46

    Meigs JB, Ordovas JM, Cupples LA, Singer DE, Nathan DM, Schaefer EJ et al. Apolipoprotein E isoform polymorphisms are not associated with insulin resistance: the Framingham Offspring Study. Diabetes Care 2000; 23 (5): 669–674.

    CAS  Article  Google Scholar 

  47. 47

    Gonzalez-Amieva A, Lopez-Miranda J, Fuentes F, Castro P, Marin C, Lopez-Granados A et al. Genetic variations of the apolipoprotein E gene determine the plasma triglyceride levels after heart transplantation. J Heart Lung Transplant 2000; 19 (8): 765–770.

    CAS  Article  Google Scholar 

  48. 48

    Guz G, Nurhan Ozdemir F, Sezer S, Isiklar I, Arat Z, Turan M et al. Effect of apolipoprotein E polymorphism on serum lipid, lipoproteins, and atherosclerosis in hemodialysis patients. Am J Kidney Dis 2000; 36 (4): 826–836.

    CAS  Article  Google Scholar 

  49. 49

    Imazu M, Yamamoto H, Toyofuku M, Watanabe T, Okubo M, Egusa G et al. Association of apolipoprotein E phenotype with hypertension in Japanese-Americans: data from the Hawaii-Los Angeles-Hiroshima Study. Hypertens Res 2001; 24 (5): 523–529.

    CAS  Article  Google Scholar 

  50. 50

    Bednarska-Makaruk M, Broda G, Kurjata P, Rodo M, Roszczynko M, Rywik S et al. Apolipoprotein E genotype, lipid levels and coronary heart disease in a Polish population group. Eur J Epidemiol 2001; 17 (8): 789–792.

    CAS  Article  Google Scholar 

  51. 51

    Schwanke CH, da Cruz IB, Leal NF, Scheibe R, Moriguchi Y, Moriguchi EH . Analysis of the association between apolipoprotein E polymorphism and cardiovascular risk factors in an elderly population with longevity. Arq Bras Cardiol 2002; 78 (6): 561–579.

    CAS  Article  Google Scholar 

  52. 52

    Katsuya T, Baba S, Ishikawa K, Mannami T, Fu Y, Inamoto N et al. Epsilon 4 allele of apolipoprotein E gene associates with lower blood pressure in young Japanese subjects: the Suita Study. J Hypertens 2002; 20 (10): 2017–2021.

    CAS  Article  Google Scholar 

  53. 53

    Mozas P, Castillo S, Reyes G, Tejedor D, Civeira F, Garcia-Alvarez I et al. Apolipoprotein E genotype is not associated with cardiovascular disease in heterozygous subjects with familial hypercholesterolemia. Am Heart J 2003; 145 (6): 999–1005.

    CAS  Article  Google Scholar 

  54. 54

    Qiu C, Winblad B, Fastbom J, Fratiglioni L . Combined effects of APOE genotype, blood pressure, and antihypertensive drug use on incident AD. Neurology 2003; 61 (5): 655–660.

    Article  Google Scholar 

  55. 55

    Pezzini A, Grassi M, Del Zotto E, Bazzoli E, Archetti S, Assanelli D et al. Synergistic effect of apolipoprotein E polymorphisms and cigarette smoking on risk of ischemic stroke in young adults. Stroke 2004; 35 (2): 438–442.

    CAS  Article  Google Scholar 

  56. 56

    de Leeuw FE, Richard F, de Groot JC, van Duijn CM, Hofman A, Van Gijn J et al. Interaction between hypertension, apoE, and cerebral white matter lesions. Stroke 2004; 35 (5): 1057–1060.

    CAS  Article  Google Scholar 

  57. 57

    Ranjith N, Pegoraro RJ, Rom L, Rajput MC, Naidoo DP . Lp(a) and apoE polymorphisms in young South African Indians with myocardial infarction. Cardiovasc J S Afr 2004; 15 (3): 111–117.

    CAS  PubMed  Google Scholar 

  58. 58

    Kang JH, Logroscino G, De Vivo I, Hunter D, Grodstein F . Apolipoprotein E cardiovascular disease and cognitive function in aging women. Neurobiol Aging 2005; 26 (4): 475–484.

    CAS  Article  Google Scholar 

  59. 59

    Um JY, Hwang CY, Hwang WJ, Kang SD, Do KR, Cho JJ et al. Association between iris constitution and apolipoprotein e gene polymorphism in hypertensives. J Altern Complement Med 2004; 10 (6): 1101–1105.

    Article  Google Scholar 

  60. 60

    Bhavani AB, Sastry KB, Reddy NK, Padma T . Lipid profile and apolipoprotein E polymorphism in essential hypertension. Indian Heart J 2005; 57 (2): 151–157.

    CAS  PubMed  Google Scholar 

  61. 61

    Rowan E, Morris CM, Stephens S, Ballard C, Dickinson H, Rao H et al. Impact of hypertension and apolipoprotein E4 on poststroke cognition in subjects &gt;75 years of age. Stroke 2005; 369: 1864–1868.

    Article  Google Scholar 

  62. 62

    Pitsavos C, Choumerianou DM, Skoumas J, Maumus S, Stefanadis C, Dedoussis GV et al. Apolipoprotein E polymorphism is not associated with lipid levels and coronary artery disease in Greek patients with familial hypercholesterolaemia. Clin Exp Med 2005; 5 (4): 196–201.

    CAS  Article  Google Scholar 

  63. 63

    Baum L, Lam LC, Kwok T, Lee J, Chiu HF, Mok VC et al. Apolipoprotein E epsilon4 allele is associated with vascular dementia. Dement Geriatr Cogn Disord 2006; 22 (4): 301–305.

    CAS  Article  Google Scholar 

  64. 64

    Formiga F, Alia P, Navarro MA, Pujol R . Apolipoprotein e genotypes in nonagenarians. J Am Geriatr Soc 2006; 54 (9): 1471–1473.

    Article  Google Scholar 

  65. 65

    Wohlin M, Sundstrom J, Lannfelt L, Axelsson T, Syvanen AC, Andren B et al. Apolipoprotein E epsilon4 genotype is independently associated with increased intima-media thickness in a recessive pattern. Lipids 2007; 42 (5): 451–456.

    CAS  Article  Google Scholar 

  66. 66

    Wang B, Zhao H, Zhou L, Dai X, Wang D, Cao J et al. Association of genetic variation in apolipoprotein E and low density lipoprotein receptor with ischemic stroke in Northern Han Chinese. J Neurol Sci 2009; 276 (1-2): 118–122.

    CAS  Article  Google Scholar 

  67. 67

    Ruixing Y, Jinzhen W, Weixiong L, Yuming C, Dezhai Y, Shangling P . The environmental and genetic evidence for the association of hyperlipidemia and hypertension. J Hypertens 2009; 27 (2): 251–258.

    Article  Google Scholar 

Download references

Acknowledgements

We would like to thank the two anonymous reviewers whose comments helped in improving the quality of the manuscript.

Author information

Affiliations

Authors

Corresponding author

Correspondence to P G Bagos.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Stoumpos, S., Hamodrakas, S., Anthopoulos, P. et al. The association between apolipoprotein E gene polymorphisms and essential hypertension: a meta-analysis of 45 studies including 13 940 cases and 16 364 controls. J Hum Hypertens 27, 245–255 (2013). https://doi.org/10.1038/jhh.2012.37

Download citation

Keywords

  • apolipoprotein E
  • meta-analysis
  • polymorphism
  • random effects

Further reading

Search

Quick links