Kininogen gene (KNG) variation has a consistent effect on aldosterone response to antihypertensive drug therapy: the GERA study

Maja Barbalic, Gary L. Schwartz, Arlene B. Chapman, Stephen T. Turner, Eric Boerwinkle

Abstract

Recent experimental and clinical studies suggested that apart from playing an essential role in blood pressure homeostasis, aldosterone is involved in the pathophysiology of cardiovascular and renal diseases by inducing structural changes in the heart, kidney, and vessel wall. The interindividual variation of aldosterone response to antihypertensive treatment is considerable, and is at least partially explained by genetic variation. In this study, we investigated aldosterone response to two antihypertensive drugs—a thiazide diuretic and an angiotensin receptor blocker (ARB). Genetic variations in 50 candidate genes were tested for association with aldosterone response in four independent samples: African American (AA) responders to a diuretic (n = 289), AA responders to an ARB (n = 252), European American (EA) responders to a diuretic (n = 295) and EA responders to an ARB (n = 300). Linear regression was used to test the association with inclusion of age, sex, and body mass index as covariates. The results indicated the existence of one or more variants in the kininogen gene (KNG) that influence interindividual variation in aldosterone response. The significant association was replicated in three of four studied groups. The single nucleotide polymorphism rs4686799 was associated in AA and EA responders to the diuretic (P = 0.04 and P = 0.07, respectively), and rs5030062 and rs698078 were significantly associated in EA responders to the diuretic (P = 0.05 and P = 0.01) and EA responders to the ARB (P = 0.04 and P = 0.02). Although the clinical implication of KNG gene variation to antihypertensive drug response is yet to be determined, this novel candidate locus provides important new insights into drug response physiology.

  • antihypertensive treatment
  • candidate genes

aldosterone is the final endocrine signal in the renin-angiotensin-aldosterone system and has an important role in blood pressure homeostasis. Primary aldosteronism, a condition prevalent in up to ∼15% of hypertensive patients, has been shown to enhance the risk of cardiovascular and kidney diseases (15, 21, 22). Independent of its effects on blood pressure (BP), recent studies suggest that aldosterone might induce structural changes in the heart, kidney, and vessel wall causing myocardial fibrosis, nephrosclerosis, and vascular inflammation (9, 23). All of this points to aldosterone's central role in multiple cardiovascular disease processes and pathologies.

There is substantial interindividual variation in BP response to antihypertensive medications, and at least a proportion of this variation is due to genetic differences among individuals (13). In addition, antihypertensive medication, either directly or indirectly, influences the renin-angiotensin-aldosterone system and alters circulating aldosterone levels in blood (24). For example, thiazide diuretics are known to raise blood aldosterone levels, and angiotensin receptor blockers (ARBs) are known to lower blood aldosterone levels (5, 7).

Here we report the results of a large-scale candidate gene study investigating the genetic determinants of aldosterone response to two antihypertensive treatments. We gave higher priority to a large-scale candidate gene study compared with a genomewide association study because these important genes have not been fully investigated in previous pharmacogenetic studies of aldosterone. Future studies will consider a genomewide association approach to identify novel genes underlying aldosterone response. We sought to identify associations that were replicated among four independent samples of hypertensive subjects: African American (AA) responders to a diuretic, European American (EA) responders to a diuretic, AA responders to an ARB, and EA responders to an ARB. We further investigated the contribution of the identified “aldosterone single nucleotide polymorphisms (SNPs)” to systolic (SBP) and diastolic (DBP) blood pressure response to antihypertensive treatment.

METHODS

Study population.

Phenotype data and biological samples were collected as part of the Genetic Epidemiology of Responses to Antihypertensives (GERA) study (5, 7). The study was designed to determine whether measured variation in genes predicts interindividual differences in BP response to diuretic therapy (protocol 1, Ref. 7) and to an angiotensin II receptor blocker (protocol 2, Ref. 5) in community-based samples of hypertensive AA and EA. Briefly, men and women with essential hypertension between 30 and 59 yr of age were recruited at Emory University in Atlanta, Georgia (AA) and the Mayo Clinic in Rochester, Minnesota (EA). Hypertension was defined as a BP ≥ 140/90 mmHg or previous diagnosis of essential hypertension and current treatment with antihypertensive medications. The participants were instructed to maintain a standard sodium intake of 2 mmol·kg−1·day−1 throughout the study period. Antihypertensive medications were discontinued, and subjects were evaluated every other week during a 4- to 6-wk washout phase. Once stable elevation of BP (>90 mmHg diastolic but <180/110 mmHg) was achieved, antihypertensive therapy was administered for 4–6 wk. In protocol 1, 289 AA and 295 EA were treated with 25 mg of hydrochlorothiazide, taken orally once daily for 4 wk. In protocol 2, an independent sample of 252 AA and 300 EA were treated with 16 mg of an ARB (candesartan) taken orally once daily for 2 wk, with an increase to 32 mg daily for 4 more weeks. Dietary compliance was monitored by measurement of 24-h urine sodium excretion and dietary recall diaries. The study was approved by the Institutional Review Boards of the Mayo Clinic and Emory University.

BP was measured and blood was obtained for biochemical measurements at the end of the washout period (baseline) and at the end of 4–6 wk of antihypertensive treatment. Morning fasting blood samples were drawn after 30 min in a seated position.

Serum aldosterone concentrations were determined by radioimmunoassay using reagent kits purchased from Diagnostic Products (Los Angeles, CA). Plasma renin was measured by radioimmunoassay of angiotensin I according to the method of Sealey and Laragh (23a, 23b) with GammaCoat PRA Radioimmunoassay Kits purchased from DiaSorin (Stillwater, MN).

Study sample and genotyping.

In total, 530 AA (289 treated with hydrochlorothiazide and independent sample of 252 treated with ARB) and 550 EA (295 treated with hydrochlorothiazide and independent sample of 300 treated with ARB) were genotyped with a custom Illumina chip designed to study 1,536 SNPs within 56 candidate genes for hypertension pharmacogenetics (Table 1). The SNPs were selected based on the R2 measure of linkage disequilibrium so that they “tag” other common SNPs throughout each gene. The software Tagger (www.broad.mit.edu/mpg/tagger/) and HapMap Phase 2 (www.hapmap.org) and Perlegen African-Americans (www.ncbi.nlm.nih.gov/snp/) were used for SNP selection. After removal of SNPs with missing genotype calls >5%, minor allele frequency <2%, and deviation from Hardy-Weinberg equilibrium at P < 0.001, the number of remaining SNPs was 1,320 in AA and 1,262 in EA. Individuals missing >10% of the SNP calls were removed from the analysis, leaving 289 AA and 295 EA treated with a diuretic and 252 AA and 301 EA treated with an ARB.

View this table:
Table 1.

Genes, chromosome numbers, and number of SNPs and Meff (effective number of independent SNPs) in each gene

Statistical analysis.

The association with candidate gene SNPs was tested for baseline aldosterone level and for aldosterone response to antihypertensive treatment. To analyze the response of aldosterone level independent of renin activity, we used the residuals of the renin on aldosterone linear regression. Aldosterone response was defined as the difference between the baseline residuals and the after-treatment residuals.

Linear regression was used to test the association with age, body mass index (BMI), and sex as covariates. A one degree of freedom additive model was applied because it has been shown to perform well under a variety of genetic models, which are not known a priori (17). The analysis was performed in Plink 1.01. To avoid the colinearity problem that may arise from the correlation between adjacent SNPs within a gene or gene region, we used SNPSpD software (14, 20) to determine the effective number of independent SNPs (Meff) within the examined genes (Table 1). If the genes were <500 kb apart they were combined to estimate the Meff. Subsequently, to account for multiple statistical testing within each gene, the adjusted P values were determined by applying the Sidak correction using the estimated Meff in each gene. The false discovery rate q values (2) were also calculated with Meff and the equation from Ref. 14.

SNPs found to be associated with aldosterone response were subsequently tested for their association with SBP and DBP response to antihypertensive treatment. The association analysis was performed in Plink using an additive model with age, sex, BMI, and baseline SBP or DBP as covariates.

Statistical significance was defined by P < 0.05 or q < 0.05 after taking into account multiple comparisons. Replication has become the sine qua non of genetic association studies (6). The study design used here permitted us to assess replication among the four independent samples—between two drugs within an ethnic group and between ethnic groups for each drug separately.

RESULTS

Sample characteristics at the baseline examination and after treatment in AA and EA for the two drugs are presented in Table 2. Average plasma aldosterone levels increased significantly after treatment with a diuretic in AA and in EA and decreased significantly after treatment with an ARB in both groups (P < 0.0001 in each case). However, plasma renin levels increased significantly after both treatment drugs in all four groups (P < 0.0001 in each case) (Table 2).

View this table:
Table 2.

Sample characteristics measured at end of washout period (baseline) and after antihypertensive treatment

The statistically significant results for the genotype-phenotype association analyses for baseline aldosterone levels and aldosterone response are shown in Tables 3 and 4, respectively. For baseline aldosterone level, there was no SNP that had an adjusted P value <0.05 that replicated in another group. In contrast, the analysis of aldosterone response to antihypertensive treatment revealed significant associations that were replicated across three of four examined samples. Replication was found for SNPs within the kininogen gene (KNG) between AA and EA responders to a diuretic and between the two EA samples in response to the two different drugs. The SNP rs4686799 was significantly associated in AA and EA responders to a diuretic (P = 0.0437 and P = 0.0760, respectively). Two SNPs, rs5030062 and rs698078, were significantly associated in EA responders to a diuretic and in EA responders to an ARB (Table 4). Figure 1 shows the graphical relationship between aldosterone response and genotype for the three SNPs, rs4686799, rs5030062, and rs698078. In all cases, the general relationship between aldosterone response and genotype is the same between the two drugs. For rs4686799 and rs698078, the relationship is also the same between the two ethnic groups. For rs5030062, however, the relationship is different between EA and AA. In EA, the direction of the aldosterone response effect is AA > Aa > aa. In AA, the direction of the aldosterone response effect is aa > Aa > AA. Examination of the linkage disequilibrium structure of these three SNPs provides further insight into these observations (Table 5). In EA, rs4686799, rs5030062, and rs698078 are in high linkage disequilibrium. In AA, rs698078 and rs5030062 are in high disequilibrium, but they are not in disequilibrium with rs4686799. In AA, only rs4686799 had a significant effect on aldosterone response. Therefore, these data likely map the aldosterone response effect to those regions of KNG that are in high linkage disequilibrium with rs4686799.

Fig. 1.

Mean difference in aldosterone response [standardized residuals (st. res.)] among genotypes for 3 replicated single nucleotide polymorphisms. Blue line, European Americans treated with diuretic; green line, African Americans treated with diuretic; red line, European Americans treated with angiotensin receptor blocker; violet dashed line, African Americans treated with angiotensin receptor blocker.

View this table:
Table 3.

SNP associations (P or q <0.05) with aldosterone baseline levels in each sample

View this table:
Table 4.

SNP associations (P or q <0.05) with aldosterone response in each sample

View this table:
Table 5.

Linkage disequilibrium structure of the three associated SNPs in each sample

We next analyzed the involvement of these kininogen SNPs in the response of SBP and DBP to diuretic and ARB therapy. None of these SNPs was significantly associated with either SBP or DBP response. Haplotype analyses did not change this conclusion (data not shown).

DISCUSSION

This is the first study to examine the genetic determinants of aldosterone response to antihypertensive treatment. We report a significant and consistent effect of KNG gene variation on aldosterone response to two different commonly prescribed antihypertensive medications. In particular, the SNP rs468799 had a significant and consistent effect on aldosterone response to a diuretic in both AA and EA, and in EA the SNP rs698078 had a significant and consistent effect on aldosterone response to both a diuretic and an ARB. The association of rs698078 with aldosterone response was consistent in AA for both drugs but did not reach statistical significance after accounting for multiple comparisons. The importance of this finding is strengthened by the replication of the association between groups that were given two different drugs and between AA and EA responders to the same drug. The differences in linkage disequilibrium pattern in AA and EA within KNG allowed us to narrow the region to one in high disequilibrium with rs4686799. It is interesting that in the groups taking two different drugs, the SNPs' effects act in the same direction, suggesting that the putative KNG variant contributes to interindividual aldosterone response to antihypertensive treatment in the same manner regardless of the treatment used.

KNG encodes for kininogen, a high-molecular-weight precursor of bradykinin. Bradykinin is a well-known vasodilator with an important role in BP homeostasis (8, 16). Bradykinin's release from kininogen is mediated by kallikrein, a serine protease that liberates kinins from kininogens (3). By activating bradykinin subtype B2 receptors on endothelial cells (1, 11), bradykinin stimulates release of vasodilating agents: nitric oxide (NO), prostacyclin, and endothelium-derived hyperpolarizing factor (EDHF) (4). NO is also known to be one of the inhibitors of aldosterone synthesis (12), which may be a possible link relating bradykinin and aldosterone production.

It has been reported that AA have markedly lower kallikrein levels than EA, with AA hypertensive subjects generally showing the lowest measured levels (19). Lack of a significant association of KNG on aldosterone response in the sample of AA responders to an ARB could be related to a diminished role of the kallikrein-kinin system in AA.

The KNG variants observed to be significantly associated with aldosterone response were not significantly associated with SBP or DBP response to antihypertensive treatment. This is disappointing because the kallikrein-kinin system is known to be involved in BP regulation and response to antihypertensive drug therapy. As a vasodilator, bradykinin counterbalances the action of angiotensin II on BP control and renal blood flow and protects against the effects of vasoconstrictor agents. Knockout of the B2 receptor in mouse disrupts vasodepressor response to bradykinin and leads to exaggerated response to angiotensin-converting enzyme (ACE) inhibitors due to unbalanced vasoconstriction effects of the renin-angiotensin-aldosterone system (10, 18). It could be hypothesized that the effect found to be prominent in aldosterone response variability or possible effects of these variants on unmeasured response variability of the kallikrein-kinin system are muted by the complexity of BP response regulation. Indeed, aldosterone response was not a predictor of BP response in the GERA studies (5, 7).

Alternatively, it could be hypothesized that the genes' effect on aldosterone response is separate from the genes' effect on BP response to each of the drugs. Regardless, the results presented here provide additional insight into the physiological response to these commonly prescribed BP-lowering medications.

In summary, we report evidence that one or more variants within the kininogen gene influences the variability of aldosterone response to antihypertensive treatment among hypertensive patients. The finding was confirmed by the replicated association across three samples and indicates interesting insights in antihypertensive drug response physiology.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grants HL-74735 and HL-53335 and the Mayo Foundation.

Acknowledgments

We are thankful for the technical assistance of Zhiying Wang, Megan Grove, Jodie Van De Rostyne, Jeremy Palbicki, Robert Tarrell, and Prabin Thapa.

Footnotes

  • Address for reprint requests and other correspondence: E. Boerwinkle, Human Genetics Ctr., 1200 Herman Pressler, Houston, TX 77030 (e-mail: eric.boerwinkle{at}uth.tmc.edu).

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 23a.
  25. 23b.
  26. 24.
View Abstract