Amy I. Lynch, Eric Boerwinkle, Barry R. Davis, Charles E. Ford, John H. Eckfeldt, Catherine Leiendecker-Foster, Donna K. Arnett. Pharmacogenetic Association of the NPPA T2238C Genetic Variant With Cardiovascular Disease Outcomes in Patients With Hypertension. JAMA. 2008;299(3):296–307. doi:10.1001/jama.299.3.296
Author Affiliations: Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis (Drs Lynch and Eckfeldt and Ms Leiendecker-Foster); Human Genetics Center, University of Texas Health Science Center at Houston (Dr Boerwinkle)
and University of Texas School of Public Health (Drs Davis and Ford),
Houston; and Department of Epidemiology, University of Alabama at Birmingham (Dr Arnett).
The NPPA gene codes for the precursor of atrial natriuretic polypeptide, suggesting that NPPA may modulate the efficacy of some antihypertensive drugs.
To test whether participants with minor NPPA alleles in the T2238C or G664A variants had different rates of cardiovascular disease or blood pressure (BP) changes than common allele homozygotes when treated with a diuretic vs other antihypertensive medications.
Design, Setting, and Patients Post hoc analysis of 38 462 participants with hypertension from ALLHAT, a multicenter randomized clinical trial conducted in the United States and Canada. Genotyping was performed from February 2004 to January 2005.
Participants were randomly assigned to receive a diuretic (chlorthalidone;
n = 13 860), a calcium antagonist (amlodipine; n = 8174),
an angiotensin-converting enzyme inhibitor (lisinopril; n = 8233),
or an α-blocker (doxazosin; n = 8195).
Main Outcome Measure
The primary outcome measure was coronary heart disease (CHD),
defined as fatal CHD or nonfatal myocardial infarction (mean follow-up,
4.9 years). Secondary outcomes were stroke, all-cause mortality, combined cardiovascular disease outcomes, and 6-month systolic and diastolic BP changes. Genotype × treatment interactions were tested where genotypes were modeled additively and dominantly.
Depending on genotype, the event rates per 1000 person-years were 15.3 to 19.7 for CHO, 9.6 to 15.4 for stroke, and 27.4 to 30.7
for all-cause mortality. For the NPPA T2238C variant, lower event rates were found for the C allele carriers than for the TT homozygous individuals when comparing chlorthalidone and amlodipine (CHD: CC = 0.86; TC = 0.90; TT = 1.09; P = .03 [dominant model]; stroke: CC = 1.18;
TC = 0.82; TT = 1.26; P = .01
[additive and dominant models]; all-cause mortality: CC = 0.87;
TC = 0.98; TT = 1.12; P = .05
[dominant model]). Combined end points yielded similar results. Consistent with these clinical findings, 6-month changes in systolic BP for those with the CC genotype showed larger reductions with chlorthalidone (−6.5 mm Hg) than with amlodipine (−3.8 mm Hg), lisinopril (−2.4 mm Hg), or doxazosin (−3.8 mm Hg). Among those with the TT genotype, systolic BP differences between drugs were less (range,
−5.4 to −7.5 mm Hg; P value,
<.001 to .003 for interaction); diastolic BP showed similar results.
We found no pharmacogenetic associations with the NPPA G664A variant.
The NPPA T2238C variant was associated with modification of antihypertensive medication effects on cardiovascular disease and BP. Minor C allele carriers experienced more favorable cardiovascular disease outcomes when randomized to receive a diuretic,
whereas TT allele carriers had more favorable outcomes when randomized to receive a calcium channel blocker.
Approximately 71 million individuals in the United States have 1 or more types of cardiovascular disease (CVD), at least 65 million of whom have hypertension.1
Although control of hypertension has been improving in recent years, among those treated, only about two-thirds have their hypertension controlled.2
Seeking ways to reduce CVD morbidity and mortality by tailoring treatment to a patient's particular genotype is a laudable goal. To date, studies of gene polymorphisms in hypertension candidate genes, such as angiotensin-converting enzyme (ACE)
and the angiotensin II receptor, have been shown to predict response to treatments such as ACE inhibition and angiotensin II blockade.3 However, the use of information on genetic variability to predict response to antihypertensive therapy and, thus,
guide therapeutic choices, has yet to be realized in the clinical setting.
The NPPA (atrial natriuretic precursor A) gene is an attractive candidate for pharmacogenetic research. Found on chromosome 1p36, NPPA encodes the precursor from which atrial natriuretic polypeptide (ANP) is derived. Atrial natriuretic polypeptide controls extracellular fluid volume and electrolyte homeostasis, acting as a diuretic.4
Animal studies show that genetically reduced ANP can lead to salt-sensitive hypertension and genetically increased ANP can lead to hypotension.5,6
Research has also shown an association between ANP (and the NPPA gene)
and CVD outcomes such as stroke, heart failure, left ventricular hypertrophy,
coronary artery disease, and hypertension, as well as CVD risk factors such as insulin sensitivity and resistance, with studies generally showing that minor allele carriers have poorer outcomes in cases in which genetic effects are observed.7- 16
Our objective was to test whether participants with hypertension in the Genetics of Hypertension Associated Treatment (GenHAT) Study with minor NPPA genotypes (NPPA G664A and NPPA T2238C) randomized to the diuretic chlorthalidone had different outcomes with regard to 7 CVD measures than their counterparts who were randomized to other classes of antihypertensive medication. We sought to determine whether there was a detectable pharmacogenetic association of NPPA variants with CVD and blood pressure lowering among those randomized to 1 of 4 antihypertensive medications.
Participants were part of the Genetics of Hypertension Associated Treatment (GenHAT) study, an ancillary study to the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT).
ALLHAT was a randomized, double-blind, multicenter clinical trial (623 clinical centers) with 42 418 hypertensive participants aged 55 years or older who had 1 or more additional risk factors for CVD. ALLHAT was designed to determine if the incidence of fatal coronary heart disease (CHD) and nonfatal myocardial infarction in high-risk hypertensive patients was lower with treatment using each of 3 antihypertensive drug classes: a calcium channel blocker (amlodipine), an ACE inhibitor (lisinopril), and an α-adrenergic blocker (doxazosin) compared with treatment using a diuretic (chlorthalidone). The ratio of drug class assignment was 1:1:1:1.7, respectively, and doses were titrated to attain blood pressure control to less than 140/90 mm Hg. Secondary outcomes included all-cause mortality, stroke, and other cardiovascular events.17
GenHAT genotyped variants in several hypertension-related genes in 39 114 ALLHAT participants with available DNA, making the study design a post hoc subgroup analysis of a randomized clinical trial. The goal of GenHAT was to understand gene-treatment interactions on CVD outcomes and blood pressure lowering. Approximately half (46%)
of the participants were women, and about half (47%) were of white non-Hispanic race/ethnicity (race/ethnicity data self-reported from 5 study-defined categories: white, black, Asian/Pacific Islander,
American Indian/Alaskan native, or other; Hispanic status was also self-reported).18
Participants were excluded from these analyses if they were missing genotype data (n = 686
for NPPA T2238C and n = 665 for NPPA G664A); therefore, there were 38 428
participants included in NPPA T2238C analyses and 38 449 included in NPPA G664A analyses.
There were 38 462 participants with data available for at least 1 variant.
Complete descriptions of both ALLHAT and GenHAT have been previously published.17,18 This research was reviewed and approved by local institutional review boards, and all participants provided written informed consent. Genetic data were made anonymous since GenHAT identifiers were unique and the code that links the 2 has been destroyed for participants from the Veteran Affairs'
study sites. The key for the remaining participants is offline and stored in a locked file.
ALLHAT participants were randomized to treatment between February 1994 and January 1998. The follow-up period ended in March 2002. In keeping with a priori stopping guidelines for ALLHAT, after a January 2000 data review, the doxazosin arm was discontinued because of futility for the primary end point and a significantly higher incidence of CVD, particularly congestive heart failure, when compared with chlorthalidone treatment.19
Outcomes of interest in this analysis were fatal CHD and nonfatal myocardial infarction (hereafter referred to as CHD), stroke, heart failure (fatal, requiring hospitalization, or treated in an outpatient setting), all-cause mortality, end-stage renal disease, combined CHD (including CHD plus coronary revascularization plus hospitalized angina)
and combined CVD (including combined CHD plus stroke plus treated angina without hospitalization plus heart failure [fatal, requiring hospitalization, or treated in an outpatient setting] plus peripheral arterial disease [in-hospital or outpatient revascularization]). Outcomes were reported by clinical investigators, and documentation (death certificate, hospital discharge summary) was submitted for any outcome involving death or hospitalization. National databases were also used to identify deaths occurring among participants lost to follow-up.
Outcomes were not mutually exclusive; the primary outcome (CHD) was included in the combined CHD outcome, and the combined CHD outcome (as well as stroke and congestive heart failure) was included in the combined CVD outcome, each outcome being more inclusive than the previous one. We also analyzed blood pressure change from baseline to 6 months after randomization. The values represent the average of 2 seated blood pressure measurements. A detailed description of outcome ascertainment for ALLHAT has been previously published.17
GenHAT genotyped 2 single-nucleotide polymorphisms (SNPs) in the NPPA gene between February 2004 and January 2005: NPPA G664A (SNP database ID rs5063)
and NPPA T2238C (SNP database ID rs5065).
The multilocus assay panel including these 2 NPPA SNPs became available in the fall of 2003. DNA was isolated on FTA paper (Fitzco Inc, Maple Plain, Minnesota) from blood samples. Genotyping was performed using amplified DNA products of a multiplex polymerase chain reaction and detected using a colorimetric reaction of allele-specific oligonucleotide probes hybridized to a nylon membrane, as described previously.20
Stata software, version 9.2 (Stata Corp, College Station, Texas)
and HaploView software, version 4.0 (Daly Laboratory at the Broad Institute, Cambridge, Massachusetts) were used for statistical analyses.
Data analysis for this report began in August 2006. Hardy-Weinberg equilibrium tests21
were performed using the χ2 goodness-of-fit test. Linkage disequilibrium (R2) between the 2 NPPA markers was assessed using HaploView. Cox regression was used to test the main effects of NPPA genotypes and treatment assignment and genotype × treatment interactions on clinical outcomes, resulting in hazard ratio (HR)
and ratio of hazard ratios point estimates, respectively. The test of interest in assessing the pharmacogenetic (ie, gene-treatment interaction)
association is the significance of the ratio of hazard ratios point estimate. The assumption of proportional hazards was assessed by observing log-log plots and testing treatment × ln(time) interactions (a time-dependent covariate). If violated, logistic regression was performed, resulting in odds ratio (OR) and ratio of odds ratios point estimates. Linear regression was used to test the main gene effects and genotype × treatment interactions on blood pressure change 6 months after randomization. We modeled the genotypes both additively in 3 categories and collapsed into 2 categories, resulting in dominant models of inheritance: CC + CT vs TT for NPPA T2238C, and AA + AG vs GG for NPPA G664A. For testing main effects of genotypes on outcomes, we adjusted for age, sex, race (white, black, Asian/Pacific Islander, American Indian/Alaskan Native, or other), Hispanic status (self-reported as yes/no/don't know), baseline body mass index, diabetes status, baseline total cholesterol, smoking status, education (years),
and baseline systolic and diastolic blood pressures.
To test genotype × treatment interactions (the pharmacogenetic association), we created a genotype × treatment parameter and did 4 comparisons for the clinical outcomes: chlorthalidone vs amlodipine, chlorthalidone vs lisinopril, and chlorthalidone vs amlodipine plus lisinopril for the full follow-up time (mean, 4.9
years) and chlorthalidone vs doxazosin for the follow-up period until the discontinuation of the doxazosin arm (mean, 3.2 years). For blood pressure change, we compared chlorthalidone vs amlodipine, chlorthalidone vs lisinopril, chlorthalidone vs doxazosin, and chlorthalidone vs amlodipine + lisinopril + doxazosin.
When evidence of an interaction was noted, we tested for 3-way interactions with sex, diabetes status, and race by adding the appropriate 3-way interaction term to the model. Racial differences with regard to the effect of ACE inhibitors vs diuretics on stroke and combined CVD have been previously reported for ALLHAT, with black participants having greater treatment effect with diuretics than nonblacks.19
Since the minor allele frequency for NPPA T2238C differed significantly between black and nonblack participants (0.39 for black vs 0.15 for nonblack), we assessed whether any observed genotype × treatment interaction could be explained by an underlying ethnicity × treatment interaction by adding an ethnicity × treatment parameter to the model for stroke and combined CVD. The NPPA G664A variant did not show the same minor allele frequency differences (minor allele frequency = 0.059 for black participants and 0.052
for nonblack participants).
We performed 72 tests of pharmacogenetic associations (9 outcomes [7 clinical outcomes + 2 blood pressure measures] × 4
treatment comparisons × 2 gene variants), which translate into a statistically significant P value after Bonferroni correction of .05/72 ≈ .0007. However,
a Bonferroni correction may be unduly conservative; the 72 tests were not independent, since in all cases we were comparing the chlorthalidone treatment group with the other treatment groups, and several of the clinical outcomes were nested within each other. For example, the primary CHD outcome is included in the combined CHD outcome, which is then included in the combined CVD outcome. The published GenHAT design outlined 6 primary hypotheses, which did not include testing the pharmacogenetic associations of the NPPA gene.
Secondary questions were to be considered exploratory and, therefore,
would not be adjusted for multiple comparisons.18 In keeping with the original study design,
these results should be interpreted as exploratory.
Study power was calculated for the 6 primary genotype × treatment interaction (pharmacogenetic association) hypotheses described in the GenHAT design article. Post hoc power calculations for genotype × treatment interactions were performed for each of the NPPA variants using a method based on the work of Peterson and George.22
For the primary outcome of CHD in the dominant models, there was 80% power to detect a ratio of hazard ratios of 1.18 to 1.27 for the NPPA T2238C variant and 1.30 to 1.45 for the NPPA G664A variant,
depending on the treatment comparison. These calculations are based on 2-sided tests with individual α levels of .0007 to correct for multiple testing and, thus, to correspond to an overall α
level of .05. The range in detectable ratio of hazard ratios for secondary outcomes was 1.52 to 2.00 and 1.12 to 1.15 for end-stage renal disease (least number of events) and combined CVD (greatest number of events),
respectively, for NPPA T2238C, and of 1.85
to 2.70 and 1.20 to 1.24 for end-stage renal disease and combined CVD, respectively, for NPPA G664A. The power to detect genotype × treatment interactions on blood pressure change in the range of 2 to 3 mm Hg was estimated to be more than 80%.
Baseline characteristics for the 38 462 participants, as well as blood pressure 6 months after randomization, according to treatment assignment are shown in Table 1. There were no differences in baseline values among treatment groups, with the exception of mean total cholesterol (P = .02). There were differences among the treatment groups in blood pressure values 6 months after randomization: the chlorthalidone group had the lowest mean systolic blood pressure of the 4 treatment groups, whether or not participants were previously taking blood pressure–lowering medication (all participants,
treated and untreated at baseline: P < .001).
For diastolic blood pressure 6 months after randomization, the amlodipine group had the lowest mean diastolic pressure, except among participants previously untreated for high blood pressure, who had lower diastolic blood pressure when randomized to lisinopril, though the effect in this subgroup was not statistically significant (all participants, P < .001; treated at baseline, P < .001; untreated at baseline, P = .16).
The NPPA G664A genotype frequencies were in Hardy-Weinberg equilibrium when tested race-specifically.
The NPPA T2238C genotype frequencies were in Hardy-Weinberg equilibrium for the white, Asian/Pacific Islander,
and American Indian/Alaskan native groups but not for the black (P = .004) and other (P = .004)
groups. Missing values for genotype data did not differ across randomized treatment group (by χ2 test, P = .20
for NPPA T2238C and P = .26
for NPPA G664A). The International HapMap Project reports a linkage disequilibrium parameter for the NPPA T2238C and NPPA G664A SNPs of 0.01 for the data from Utah residents with ancestry from northern and western Europe (CEU) and 0.04 for participants from Yoruba in Ibadan, Nigeria (YRI). The GenHAT data show a similar degree of linkage disequilibrium for the black and nonblack participant subcategories: R2 estimates = 0.009 for nonblack participants and 0.039 for black participants.
Hazard ratios for main effects of the NPPA variants on clinical outcomes are provided in Table 2. While there was some evidence of association of the NPPA variants with CVD outcomes in unadjusted models,
the only association with a P<.05 in the adjusted models was the effect of NPPA G664A on combined CVD in the dominant genetic model, which is not statistically significant if corrected for multiple comparisons. We found no evidence of an epistatic effect between the 2 loci for any outcome (data not shown).
The effects of ALLHAT treatment assignments on CVD outcomes have been previously published.19,23,24
Comparable with the ALLHAT findings, we found evidence of significant effects of treatment assignment on stroke, heart failure, and combined CVD in the GenHAT subpopulation. For stroke, chlorthalidone assignment was borderline protective compared with lisinopril assignment (HR,
0.88; 95% confidence interval [CI], 0.78-1.00; P = .05).
For heart failure, those assigned to chlorthalidone had the most favorable outcome (OR, 0.70; 95% CI, 0.63-0.78; P < .001
for chlorthalidone vs amlodipine and OR, 0.83; 95% CI, 0.74-0.93; P = .001 for chlorthalidone vs lisinopril).
For combined CVD, chlorthalidone assignment was protective compared with assignment to lisinopril (HR, 0.91; 95% CI, 0.87-0.96; P = .001 for chlorthalidone vs lisinopril).
Using data with follow-up to the point in time when the doxazosin arm was discontinued for comparability, the doxazosin group had a higher risk of stroke (HR, 1.32; 95% CI, 1.14-1.53; P < .001),
heart failure (OR, 1.88; 95% CI, 1.65-2.11; P < .001),
combined CVD (HR, 1.22; 95% CI, 1.15-1.29; P < .001),
and, to a lesser extent, combined CHD (HR, 1.09; 95% CI, 1.00-1.18; P = .04) than the chlorthalidone group (data not shown).
Table 3 presents outcome frequencies and rates by genotype group, genotype-specific treatment effects, and the results of pharmacogenetic association tests for chlorthalidone vs amlodipine, chlorthalidone vs lisinopril, and chlorthalidone vs amlodipine + lisinopril using Cox regression (or logistic regression where appropriate). For pharmacogenetic tests, a significant genotype × treatment P value indicates that the effect of the minor allele differs by treatment (or conversely,
that the effect of treatment differs by genotype). For the chlorthalidone vs doxazosin comparisons (Table 4),
we used follow-up data only to the point of discontinuation of the doxazosin arm.
There was evidence of a pharmacogenetic association for the NPPA T2238C variant with CHD, stroke, all-cause mortality, combined CHD, and combined CVD when comparing chlorthalidone and amlodipine. The genotype-specific chlorthalidone vs amlodipine treatment effect indicated that the TT genotype demonstrated a greater risk of these outcomes when treated with chlorthalidone vs amlodipine compared with the other genotypes. For stroke, there was also evidence for the chlorthalidone vs amlodipine plus lisinopril comparison: similar to the previous finding, the TT genotype was associated with a greater risk when treated with chlorthalidone vs other treatments compared with the other genotypes. These results are consistent with our hypothesis that for participants assigned to chlorthalidone, those with at least 1 copy of the minor C allele had lower risk of disease and/or death compared with those assigned to amlodipine (and amlodipine plus lisinopril for stroke), while those in the chlorthalidone group with the TT genotype had higher risk of disease and/or death than those assigned to amlodipine.
The one exception is for stroke among the CC homozygous participants,
who experienced higher rates when randomized to chlorthalidone vs amlodipine. These results do not reach statistical significance when corrected for multiple comparisons, though the consistency of the association and the correlated structure of the outcomes lend credibility to the results. We found no evidence of pharmacogenetic associations with clinical outcomes for chlorthalidone vs lisinopril, for chlorthalidone vs doxazosin comparisons with the NPPA T2238C variant, or for any NPPA G664A tests.
Table 5 presents the main effects of the NPPA variants and genotype ×
treatment effects on blood pressure change (6-month blood pressure − baseline blood pressure). We did not observe a main effect of either NPPA T2238C or NPPA G664A on 6-month change in blood pressure after randomization to treatment;
in all genotype groups there was a similar reduction in both systolic (range, −5.5 to −6.4 mm Hg) and diastolic (range, −3.4
to −3.6 mm Hg) blood pressure. However, there was evidence of pharmacogenetic associations with NPPA T2238C for both systolic and diastolic blood pressure. For systolic blood pressure change, all genotype groups experienced a greater reduction in pressure with chlorthalidone compared with the other treatments;
however, minor C allele carriers had more significant reductions when randomized to chlorthalidone vs the other treatments. For diastolic blood pressure change, there was a similar association: minor C allele carriers as a group had greater reductions in diastolic blood pressure when randomized to chlorthalidone vs the other drugs. Those with the more common TT genotype had almost the same level of reduction irrespective of randomization assignment. There was no similar pharmacogenetic association observed for change in systolic or diastolic blood pressure with NPPA G664A.
In models showing evidence of pharmacogenetic associations,
we further tested whether there were significant 3-way interactions with sex, race, or diabetes status. We found no evidence of 3-way interactions (data not shown). When the ethnicity × treatment parameter was added to the models for stroke and combined CVD, the result was that the genotype × treatment association did not disappear (stroke: chlorthalidone vs amlodipine, P = .004; chlorthalidone vs amlodipine plus lisinopril, P = .03; combined CVD: chlorthalidone vs amlodipine, P = .01).
We also noted that the ethnicity × treatment interaction on stroke and combined CVD previously reported did not disappear with the genotype × treatment parameter in the model (stroke:
chlorthalidone vs lisinopril, P = .009;
combined CVD: chlorthalidone vs lisinopril, P = .02).
These data provide evidence of a pharmacogenetic association of the NPPA T2238C variant with CHD, stroke,
all-cause mortality, combined CHD, and combined CVD when comparing the chlorthalidone group with the amlodipine group and for stroke when comparing the chlorthalidone group with those receiving amlodipine or lisinopril. The association was consistent for all outcomes: those with at least 1 copy of the minor C allele had lower risk of disease and/or death when assigned to chlorthalidone compared with those assigned to amlodipine (and the amlodipine group plus the lisinopril group for stroke), while those in the chlorthalidone group with the TT genotype had higher risk of disease and/or death than those assigned to amlodipine. When comparing chlorthalidone with lisinopril or doxazosin, no pharmacogenetic association was detected for clinical outcomes.
We also observed a pharmacogenetic association of NPPA T2238 on change in systolic and diastolic blood pressure 6 months after treatment randomization in a similar direction: generally,
minor C allele carriers had greater reductions in blood pressure when randomized to chlorthalidone vs either lisinopril or doxazosin relative to those with the common TT genotype. We found no evidence of a pharmacogenetic association for NPPA G664A for either clinical outcomes or blood pressure change.
After Bonferroni correction for multiple comparisons (72 genotype ×
treatment interaction tests), none of the findings reached statistical significance (P < .0007), though some of the blood pressure findings approached statistical significance.
However, the consistency of the association with NPPA T2238C for 5 of the 7 clinical outcome classifications and with blood pressure change lends some credibility to the findings.
ALLHAT reported that those randomized to the diuretic had more favorable outcomes for some CVD events than those randomized to the other study drugs.19,23,24
We replicate those findings in the GenHAT subpopulation overall, particularly for heart failure and combined CVD. However, our findings indicate that NPPA T2238C interacts with the type of antihypertensive treatment to modify the risk of some CVD outcomes; ie, those patients with hypertension and a C allele at the NPPA T2238C locus experienced a more favorable outcome when randomized to the diuretic, while those with the more common TT genotype had a better outcome when randomized to a calcium channel blocker. This was true for all 5 outcomes, showing evidence of a pharmacogenetic association.
It was also true for end-stage renal disease, though the interaction P value was not below the .05 threshold (P = .07). For heart failure, all genotype groups for the NPPA T2238C locus had lower rates when randomized to diuretic vs calcium channel blocker, though there was a slightly stronger benefit observed for the minor C allele carriers. Regarding blood pressure lowering, while all genotype groups experienced greater reduction in systolic pressure after 6 months when randomized to chlorthalidone, the minor C allele carriers in particular benefited more from chlorthalidone treatment vs lisinopril or doxazosin than the common TT homozygotes. The same was true for diastolic blood pressure when comparing chlorthalidone with lisinopril.
With regard to racial differences in treatment effect previously reported for ALLHAT, our results show that racial differences in allele frequencies do not account for the observed genotype × treatment interaction, nor is it likely that the genotype × treatment interactions reported herein account for the observed racial differences reported by ALLHAT.
Given the diuretic action of ANP, it is possible that individuals with a “higher-risk” genotype would have more favorable outcomes when taking a diuretic compared with their counterparts taking other classes of antihypertensive medications. Mechanistically, it is possible that those with the minor C allele of NPPA T2238C have impaired ANP production, the effect of which is reduced by treatment with diuretic. Our study did not collect serum ANP measurements, which would have allowed for an assessment of how gene polymorphisms relate to ANP levels and may have provided a firmer mechanistic pathway with which to explain our results. The effect of specific NPPA polymorphisms on ANP levels has not been fully described;
though there is some evidence that the NPPA G664A minor allele is associated with decreased plasma ANP levels, the same has not been found for the NPPA T2238C variant.14,16
We report evidence of a pharmacogenetic association for NPPA T2238C but not for NPPA G664A.
According to National Center for Biotechnology Information build 36.2
data, these variants are estimated to be 1580 base pairs apart on chromosome 1 (NPPA T2238C = SNP database ID rs5065, NPPA G664A = rs5063). NPPA T2238C is located within a stop codon in exon 3, with the T-to-C transition leading to an extension of 2 additional arginine amino acids on the atrial natriuretic peptide. NPPA G664A is a G-to-A transition in exon 1 of the ANP gene.
This transition results in a valine-to-methionine substitution on the pro-ANP peptide at amino acid position 7. It is likely that these variants yield functional differences. In addition, the minor allele was less common for NPPA G664A than for NPPA T2238C (5.4% vs 23.4%); therefore, power to detect a pharmacogenetic association of similar magnitude was reduced for NPPA G664A vs NPPA T2238C.
In general, the pharmacogenetic association of NPPA T2238C with clinical outcomes was restricted to comparisons of diuretic vs calcium channel blocker, not diuretic vs ACE inhibitor or α-adrenergic blocker, while the opposite was true for the blood pressure findings, where more pharmacogenetic associations were observed for diuretic vs ACE inhibitor or α-adrenergic blocker.
The α-adrenergic blocker comparisons allowed a shorter follow-up time because of the discontinuation of the doxazosin treatment arm;
thus, fewer clinical events leading to reduced power to detect an association. It is possible that the pharmacogenetic associations with long-term clinical outcomes and short-term blood pressure change are mediated through different pathways. Angiotensin-converting enzyme inhibitors prevent the angiotensin I to angiotensin II conversion but also tend to reduce aldosterone levels,23 thereby likely reducing sodium and fluid in the body, as diuretics do. Perhaps this similarity between the pharmacologic effect of the ACE inhibitor and the diuretic had the effect of diminishing any pharmacogenetic effect on clinical outcomes;
however, it is not clear why the same effect would not be observed for blood pressure change.
Because this study included only hypertensive participants aged 55 years or older, the generalizability of these findings to younger,
healthier populations is unknown. Since ALLHAT was not originally designed as a pharmacogenetic study, GenHAT may be considered a post hoc subgroup analysis of a randomized controlled trial. As such, it is not an ideal pharmacogenetic study design, which would likely be prospective and crossover, among other factors.25
Additionally, since only 2 NPPA variants were typed by the GenHAT study, we cannot view this study as a comprehensive evaluation of the pharmacogenetic effects of variability within the NPPA gene. However,
this research encourages further study of potential pharmacogenetic effects of the NPPA gene in other populations and for variants other than NPPA T2238C and NPPA G664A.
This study demonstrates the importance (and sometimes paradoxical findings) of pharmacogenetic research; for example, while minor NPPA T2238C allele carriers (as well as the entire study population viewed as a whole) may have had more favorable outcomes when randomized to a diuretic (chlorthalidone), participants with the most common genotype (TT) responded better when assigned to a calcium channel blocker (amlodipine) for some clinical outcomes. These findings also encourage further research in the area of molecular biology to uncover the physiologic effects of NPPA T2238C and other closely linked variants to determine what effect, if any,
they are having on circulating ANP. This could help better describe the underlying mechanisms of the interactions reported here.
This study illustrates not only the logistical demands (ie,
tens of thousands of participants, long-term follow-up, multiple outcome phenotypes, subgroup analysis, etc) but also the potential fruits (clinically useful genetic markers) of the pharmacogenetic study of complex disease phenotypes such as hypertension and its sequelae.
Although this research may not have immediate clinical implications,
it moves us one step closer toward the ultimate goal of providing individualized treatment guided by genetic information. Given the potentially large paradigm shift pharmacogenetic knowledge may insinuate into clinical practice, it is essential that findings such as those presented here be replicated, that gene × gene and gene × environment interactions be thoroughly explored,
and that cost-benefit analyses be conducted before pre-prescription genetic testing is warranted for hypertension.
In conclusion, this study found evidence of a pharmacogenetic association of the NPPA T2238C variant with CHD, stroke, all-cause mortality, combined CHD, and combined CVD with an average follow-up of 4.9 years, and change in systolic and diastolic blood pressure 6 months after randomization to treatment. Further research is needed to determine the optimal approach for personalizing antihypertensive medication treatment regimens according to genotype information and for achieving the best possible clinical outcomes.
Corresponding Author: Donna K. Arnett,
PhD, University of Alabama at Birmingham, RPHB 220E, 1530 Third Ave S, Birmingham, AL 55294-0022 (email@example.com).
Author Contributions: Drs Lynch, Davis,
and Ford had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Boerwinkle,
Acquisition of data: Boerwinkle, Davis,
Eckfeldt, Leiendecker-Foster, Arnett.
Analysis and interpretation of data:
Lynch, Boerwinkle, Davis, Ford, Leiendecker-Foster, Arnett.
Drafting of the manuscript: Lynch,
Critical revision of the manuscript for important intellectual content: Lynch, Boerwinkle, Davis, Ford, Eckfeldt,
Statistical analysis: Lynch, Davis,
Obtained funding: Boerwinkle, Davis,
Ford, Eckfeldt, Leiendecker-Foster, Arnett.
Administrative, technical, or material support:
Davis, Ford, Eckfeldt, Leiendecker-Foster, Arnett.
Study supervision: Eckfeldt, Leiendecker-Foster,
Financial Disclosures: None reported.
Funding/Support: This study was supported in part by grant HL63082 (GenHAT) from the National Heart, Lung, and Blood Institute and grant N01-HC-35130 (ALLHAT) from the National Institutes of Health.
Role of the Sponsor: The National Heart,
Lung, and Blood Institute had a contributing role in the analysis and interpretation of the ALLHAT data and in the review and approval of this GenHAT article.
Additional Contributions: Steven A.
Claas, MS, University of Alabama at Birmingham, provided editorial assistance and was compensated through grant HL63082 (GenHAT).