More than 1.6 million patients are hospitalized in North America each year with acute coronary syndrome (ACS) (acute myocardial infarction [MI] or unstable angina [UA]) (1). Evidence-based practice guidelines, developed jointly by the American Heart Association and American College of Cardiology, recommend that β-blocker (BB) therapy be initiated in all patients without a contraindication (2), and administration of BB after MI is an accepted performance measure of high-quality health care (4). These recommendations are based on results of older trials with less diverse cohorts, whereas subsequent larger studies in other racial groups (5) as well as meta-analyses (6) have questioned the consistency of BB benefit across the spectrum of MI patients, with the potential for hazard observed in at least 1 study (5). It has been hypothesized that a portion of this variability might be due to genetic variation, which is of particular interest, given the intersection of race with genetics and the disparities in ACS outcomes in the United States (7). Identifying genetic predictors of BB-related outcomes in ACS patients could allow more targeted prognostic risk estimates and, hopefully, support more appropriate use of therapy to reduce the public health burden of ACS and heterogeneity of outcomes.

Support for genetic variation contributing to differences in BB response is available in the case of systolic heart failure, where although BBs are a cornerstone of standard therapy due to their overall mortality benefit (8), substantial heterogeneity of individual responses—related in part to genetic sequence variants in the adrenergic pathway—have been documented (11). For example, functional genetic variants within genes encoding the β-adrenergic receptor 1 (ADRB1 R389G; rs1801253), the β-adrenergic receptor 2 (ADRB2 G16R; rs1042713 and ADRB2 Q27E; rs1042714), the alpha-2c receptor (ADRA2C 322-325 deletion; rs2234888), and—more recently—G-protein receptor kinase 5 (GRK5 L41Q; rs17098707), a critical regulator of β-receptor function, are associated with BB response in heart failure (13). These data raise the question as to whether functional adrenergic genetic variants might influence BB-related outcomes after ACS. Consistent with this notion is our previous work, from an exploratory analysis of a small ACS cohort, which suggested an association of ADRB2 genotypes with survival in patients treated with BB (19). However, this study was inadequately powered to accurately define the importance of each variant within racial subgroups. That most of the candidate variants differ in allele frequency in Caucasians, as compared with African Americans, underscores the importance of defining race-specific associations between genetic variants and outcomes. To address this existing gap in knowledge, we report the results of a large, prospective, longitudinal analysis of the effect of genetic variants on mortality in patients who suffered an ACS and were treated with BB, stratified by race. This analysis represents the first comprehensive study of previously reported functional adrenergic pathway genetic variants in an ACS cohort.

Between March 2001 and December 2008, 8,037 patients with ACS (7,517 acute MI, 520 UA) from 31 U.S. hospitals were prospectively screened and enrolled into 3 consecutive observational cohort studies: 1) the INFORM (INvestigation oF Outcomes from acute coronary syndRoMes) study (n = 1,199) from March 2001 to October 2002; 2) the PREMIER (Prospective Registry Evaluating outcomes after Myocardial Infarction: Events and Recovery) study (n = 2,498) from January 2003 to June 2004; and 3) the TRIUMPH (Translational Research Investigating Underlying disparities in acute Myocardial infarction Patients' Health status) study (n = 4,340) from April 2005 to December 2008, as previously described (19). The MI patients were identified by an elevated troponin blood test and either diagnostic electrocardiogram changes or ischemic symptoms. The UA patients had a normal troponin but presented with electrocardiogram changes (left bundle branch block 4%, ST-segment elevations 9%, ST-segment depression 12%, T-wave inversions 22%) and cardiac symptoms, defined as new-onset angina (<2 months) of at least Canadian Cardiovascular Society Classification class III, prolonged (>20 min) rest angina, recent (<2 months) worsening of angina, or angina that occurred within 2 weeks of a previous MI (23). To further increase the specificity of the UA diagnosis, those patients with a diagnostic study that excluded obstructive coronary disease (e.g., coronary angiography, nuclear or echocardiographic stress testing) (n = 125) or confirmed an alternative explanation for their presentation (e.g., esophagogastroduodenoscopy) were excluded. Three physicians adjudicated the charts of those patients with diagnostic uncertainty (n = 45) and attained consensus on the final diagnosis.

Within these combined cohorts, a subset of 3,373 patients were approached for and consented to genetic testing and were genotyped for adrenergic pathway polymorphisms. The genetics cohort was representative of the entire cohort (24). Of these, 2,946 (87%) were discharged alive on BB and were included in the present analyses. We further excluded 94 patients who were either transferred to other hospitals or who left against medical advice because their use of BB at discharge could not be definitively ascertained. Finally, given the large frequency differences for several genotypes of interest across race, we restricted the analyses to self-identified Caucasian and African-American patients, yielding a final sample size of 2,673 patients (2,469 MI; 204 UA; 2,072 Caucasians; 601 African Americans) from 22 centers.

Each patient was prospectively interviewed during hospital stay to ascertain their sociodemographic (including self-identified race), economic, and health status characteristics. Detailed chart abstractions were performed to obtain medical history of patients, laboratory results, disease severity, and the processes of inpatient care. All 3 studies received institutional review board approval at all participating sites, and written informed consent was obtained from each participant.

The Social Security Administration Death Master File was queried to determine vital status of patients as of December 30, 2009, and was available for all patients in this study.

Deoxyribonucleic acid (DNA) was isolated and purified from whole blood with the Qiagen QIAamp DNA purification kit (Quiagen, Germantown, Maryland). The DNA segments containing the region of interest were amplified with the polymerase chain reaction (PCR). The PCR primers were designed with Primer3 online software (25), and pyrosequencing primers were designed with the Pyrosequencing SNP Primer Design software (version 1.01, QIAGEN, Hilden, Germany). Before use, PCR primer sequences were screened across the human genome with the National Center for Biotechnology Information Basic Local Alignment Search Tool program to ensure their specificity for the gene of interest. The PCR was carried out with Amplitaq Gold PCR master mix (ABI, Foster City, California), 1 pmol of each primer (IDT, Coralville, Iowa), and 1 ng of DNA. The PCR primers and conditions are listed in (6). Pyrosequencing was performed with the PSQ HS 96A system with MA v2.0 software (Qiagen, Kungsgatan, Germany), as previously described (26). Data were automatically transferred from the PSQ HS 96A to a Microsoft Access (Richmond, Washington) database for permanent storage and merging with the clinical datasets through SAS (version 9.1, SAS, Cary, North Carolina).

Baseline and follow-up characteristics were compared by genotype. Categorical data are reported as frequencies, and differences between groups were compared with chi-square or Fisher exact tests, as appropriate. Continuous data are reported as mean ± SD, and differences between groups were tested with 1-way analysis of variance. Hardy-Weinberg equilibrium was assessed with chi-square test or Monte Carlo permutation with 10,000 iterations, as appropriate. Kaplan-Meier estimates and Cox proportional hazards models were used to describe the effect of genotype on patient survival. A dominant model was used for ADRA2C 322-325 deletion and GRK5 L41Q genotypes due to the low minor allele frequencies of these variants. For all other genotypes, we used an additive model. Follow-up began at the time of discharge from the index hospital stay. Stratified proportional hazards models were used, adjusting for study and site. To estimate the independent contribution of genotype, after adjusting for potential confounders and other clinical predictors, the proportional hazards models included covariates that were thought to be clinically important (age, sex, type of ACS, hypertension, diabetes, heart failure, chronic obstructive pulmonary disease, coronary angiography, and coronary revascularization) as well as those that differed significantly by genotype (6). To mitigate the possibility of “over fitting,” we also tested models adjusted for age and sex only, for the variants that had significant associations (minimally adjusted model) (6); these were consistent with the results of the primary analysis.

Analyses were performed separately in Caucasians and African Americans to minimize the risk of false-positive findings due to population stratification. We further assessed the issue of population stratification with a subgroup of 492 TRIUMPH subjects (382 Caucasian; 110 African-American) who were genotyped as part of a separate project with a custom array containing 3,351 single nucleotide polymorphisms and 100 ancestry informative markers (27). With these single nucleotide polymorphisms, we estimated 10 principal components with Eigenstrat and repeated our analyses in this subgroup to assess whether population stratification impacted our findings. Including principal components in the model did not meaningfully alter the findings, providing support for our categorization of race. For primary effects, p values <0.05 were considered statistically significant (29). Analyses were performed with SAS (version 9.2, SAS) and R (version 2.11.1) (30).

Baseline characteristics of the cohort, stratified by race, are shown in (Table 1). Types of BBs prescribed at discharge are shown in (Table 2). Of the cohort, 91.6% was discharged on either metoprolol (82%; mean dose 73 ± 58 mg/day) or carvedilol (9.6%; mean dose 18 ± 14 mg/day); dose distribution is shown in (Table 3). The 2-year mortality rate was 7.5% for Caucasians (n = 152) and 16.7% for African Americans (n = 96). We genotyped the study subjects for functionally significant polymorphisms within the adrenergic pathway that have been previously linked to outcomes after BB use, specifically variants within the genes encoding β-adrenergic receptors 1 (ADRB1) and 2 (ADRB2), alpha-2c adrenergic receptor (ADRA2C), and G-protein receptor kinase 5 (GRK5). For all variants, genotype call rates were >96%. Variants, genotype frequencies in Caucasians and African Americans, and the test of Hardy-Weinberg equilibrium are shown in (Table 4). Consistent with prior reports, GRK5 L41 and ADRA2C deletion allele frequencies were much higher among African Americans as compared with Caucasians (15). The frequencies of the other genotypes examined were similar to those previously reported (31).

Among the 2,072 Caucasian subjects, 2-year mortality differences across genotypes were assessed with Kaplan-Meier estimates ((Figure 51_gr1)A) and proportional hazards regression modeling (Figure 51_gr5) with adjustment for a priori selected baseline characteristics and clinical characteristics that differed significantly between genotype groups (6). The ADRA2C 322-325 insertion/deletion (I/D) genotype was significantly associated with 2-year mortality among Caucasians in the adjusted analyses (hazard ratio [HR]: 0.46; confidence interval [CI]: 0.21 to 0.99; p = 0.047) with ADRA2C ID and ADRA2C DD (i.e., deletion carriers) subjects having greater survival compared with ADRA2C II homozygous individuals. By contrast, none of the other genetic variants tested were significantly associated with 2-year mortality among Caucasians in either adjusted or unadjusted analyses (all p > 0.3).

In similarly constructed models, the association of each polymorphism with mortality was tested in a total of 601 African Americans ((Figure 51_gr1)B and Figure 51_gr5). Although neither the ADRB1 nor ADRA2C variants were associated with mortality (p > 0.5), ADRB2 variants were. The ADRB2 G16R variant was significantly associated with mortality among BB-treated African Americans in both unadjusted and adjusted analyses. Compared with African-American ADRB2 G16 allele homozygous individuals treated with BB, ADRB2 R16 allele carriers had greater mortality in both unadjusted (RG vs. GG HR: 1.74; CI: 0.96 to 3.16; RR vs. GG HR: 2.25; CI: 1.21 to 4.21; p = 0.039) (Figure 51_gr3B) and fully adjusted analyses (RG vs. GG HR: 2.10; CI: 1.14 to 3.86; RR vs. GG HR: 2.65; CI: 1.38 to 5.08; p = 0.013) (Figure 51_gr5). Notably, patients homozygous for the R allele had the greatest risk, and heterozygous patients had an intermediate risk, consistent with a gene–dose effect.

Because our pilot data suggested that both of the ADRB2 variants (ADRB2 G16R and ADRB2 Q27E)—which are in linkage disequilibrium—might contribute to BB-related outcomes, we retested the mortality association of the compound genotype that most strongly correlated with risk in this earlier study. Among African Americans treated with BB, the results were similar to our pilot data, whereas this was not the case in Caucasians (Figure 51_gr4). In adjusted analyses among African Americans, the compound genotype was associated with a 6-fold risk of mortality in the high-risk group and approximately a 4-fold risk in the intermediate group (Figure 51_gr5) (p = 0.048).

In African Americans, the GRK5 Q41L variant had a borderline-significant association with mortality in adjusted analyses. Although not achieving statistical significance, the GRK5 L41 African-American patients treated with BB trended toward lower mortality as compared with homozygous GRK5 Q41 patients (adjusted HR: 0.68; CI: 0.45 to 1.04; p = 0.07). We also tested for but found no significant gene-gene interaction between ADRB2 and GRK5 and outcomes in African-American ACS patients (interaction p value between ADRB2 G16R and GRK5 Q41L = 0.86).

Because each of the 2 sequence variants that showed a significant survival association did so in only 1 racial group, we performed formal interaction tests to further assess the significance of the racial distinction. The race × genotype interaction approached significance for both ADRA2C I/D (p = 0.053) and ADRB2 G16R (p = 0.096).

In accordance with consensus guidelines and quality performance measures, it is recommended that all patients without contraindications be discharged on BB therapy after MI (4). However, the consistent efficacy of BB in all groups has been questioned, and this variability might be related in part to genetic heterogeneity (5). Our data suggest that variability in post-ACS outcomes among BB-treated patients is associated with genetic variation in the adrenergic pathway. Importantly, we identified different prognostically important genes in African Americans, as compared with Caucasians. Specifically, the ADRA2C I/D genotype was significantly associated with 2-year mortality in post-ACS Caucasians treated with BB, whereas the ADRB2 G16R and GRK5 Q41L genotypes were associated with 2-year mortality in post-ACS African Americans treated with BB. These findings suggest that race-specific adrenergic-pathway variants might be useful in defining prognosis after BB therapy for Caucasians and African-American ACS patients.

It is important to note that the race specificity of our findings is not due to differences in allele frequency. For example, although the ADRA2C D allele is much more common in African Americans, there was no risk relationship between this gene and mortality in these patients, whereas it was associated with risk in Caucasians (race × genotype interaction p = 0.053). This contrasts with our pre-study hypothesis that this allele likely would have similar effects regardless of race. Our data suggest something more complex: that the allele effects might be specific to the Caucasian genetic environment. This is a novel observation requiring replication. If verified in other populations, then mechanistic studies are needed to understand whether ancestry-specific linkage or other genetic modifiers might explain the observed differences in allelic associations across race.

Beyond the prognostic difference between races, the association of the ADRA2C I/D variant with 2-year mortality in Caucasian patients discharged on BB is the first report of this finding in ACS patients. ADRA2C D creates a hypo-functioning form of this presynaptic inhibitory receptor, which has been previously associated with increased catecholamine release in vitro (35) and in vivo (36). Consistent with this, the ADRA2C D allele has also been previously associated with increased risk of heart failure (35). Our findings could be consistent with this line of evidence, considering that patients with the ADRA2C D allele might have a higher adrenergic state and thus display enhanced benefit from BB (37), potentially resulting in the lower mortality we observed when comparing genotype groups treated with BB. Conversely, our findings contrast with recent observations in heart failure patients where the BB bucindolol provided survival benefit in ADRA2C II homozygous individuals but not in deletion carriers, possibly due to enhanced sympatholysis in ADRA2C D carriers (18). This inconsistency might be attributable to differences between the pathophysiologic role of adrenergic signaling in ACS patients as compared with heart failure. For example, it has been proposed that sympatholysis might be particularly detrimental in heart failure patients (38). Alternatively, the inconsistency could be due to pharmacological differences in the specific agents used (no bucindolol in the present study [Table 2] vs. 100% bucindolol use in BEST [Beta-Blocker Evaluation of Survival Trial]), especially given that bucindolol is unique among BB agents for causing sympatholysis. Our findings warrant further investigation of the role of genetic variation in ADRA2C and the possible variability by type of BB agent, in Caucasian ACS patients.

Beta-adrenergic receptor signaling is down-regulated by G-protein receptor kinases (GRKs) (39). The GRK 5 phosphorylates, uncouples, and internalizes β1-adrenergic receptors, and the GRK5 L41 variant has greater activity than the GRK5 Q41 variant for all of these functions (15). We previously described a pharmacogenomic effect of GRK5 in subjects with ACS (15). The GRK5 Q41L polymorphism was previously found to be a determinant of β-blocker responsiveness in heart failure among African Americans, with a protective β-blocker–mimetic effect seen in GRK5 L41 allele carriers (13). Similarly, we found a strong trend toward improved outcomes in post-ACS African-American patients carrying this allele. Our findings are also consistent with our previous observations in heart failure patients in terms of the magnitude of effect: roughly one-half the mortality risk as compared with the GRK5 Q41 allele homozygous patients. The borderline statistical significance, with a similar effect size, likely reflects suboptimal power for this particular analysis as well as our inability to contrast BB-treated patients with those not treated with BB (due to the near universal compliance with BB as a performance measure of quality in ACS).

In this study, the ADRB2 G16R genotype was significantly associated with mortality in African-American ACS patients, such that patients homozygous for the R allele had greatest risk and heterozygous patients had an intermediate risk, consistent with a gene–dose effect and our previous findings (19). There was an even greater stratification in risk when both functional G16R and Q27E ADRB2 variants were accounted for, showing approximately a 4-fold and 6-fold risk in the intermediate- and high-risk groups, respectively. However, unlike our pilot data, in which the association was found in a much smaller cohort that included both African-American and Caucasian patients, our current analysis suggests that the association is specific to African Americans. Due to this distinction, the ADRB2 association is not a strict validation and requires confirmation in additional, adequately sized, cohort(s) of African-American ACS patients. Nonetheless, these data extend previous reports and are the first description, to our knowledge, of an African-American–specific effect of ADRB2 genotype in ACS patients. Conversely, the lack of association in Caucasians is strong, suggesting that additional inquiry into this association is unlikely to reveal a strong effect of ADRB2 genotype among Caucasian ACS patients treated with BB.

Our study should be interpreted in the context of several potential limitations. First, due to the high level of adherence to guidelines recommendations, only a small number of subjects were not treated with BB. This precluded a formal analysis to assess the interaction of genotype with BB treatment. Thus, we cannot specifically comment on BB efficacy, and have focused solely on the genetic association with outcomes in BB-treated patients. In light of these data, however, it might be worthwhile to consider a randomized trial of BB in the genetic subgroups that show poor outcomes with BB therapy in order to exclude the possibility of harm from BB treatment. Alternatively, new study approaches such as using quantified drug exposure metrics (as opposed to simply present vs. absent) or identification of a non-U.S. cohort of ACS patients where BB treatment is less common might show greater variability in BB use (i.e., more subjects off BB) and allow drug × genotype interaction analyses. Despite this inherent limitation, our findings are important, because they identify genetic subgroups with worse outcomes despite BB treatment and in whom more aggressive interventions to improve mortality might be warranted. An additional limitation is that we had insufficient sample size to differentiate between BB agents, and thus agent-specific effects are not obtainable from this study. A third limitation is that the current study includes 735 subjects previously described in our preliminary analysis testing the hypothesis that ADRB1 and ADRB2 polymorphisms would be associated with mortality in BB-treated patients (19). However, in the current study, we have increased the total cohort by more than 4-fold (including a quadrupling of the African-American population), enabling us to examine race-specific associations between genetic variation and mortality. We have also included additional published variants in our analysis (GRK5 and ADRA2C), enabling a more complete analysis of genes within the adrenergic pathway, and have performed analyses that provide evidence that population stratification is not confounding our findings. Finally, we were not able to account for changes in medications over time. However, classifying patients by discharge medication status is a well-recognized and often-used approach, because most patients remain on their discharge regimen after hospital stay (40).

In summary, our findings represent the first comprehensive analysis of established functional genetic variants within adrenergic pathway genes in an ACS cohort and show that several of these variants are associated with mortality among BB-treated patients and that these associations vary by race. Among Caucasian patients that have suffered an ACS and are treated with BB, ADRA2C II homozygous individuals had increased mortality as compared with ADRA2C ID and ADRA2C DD subjects. In African-American ACS patients treated with BB, GRK5 L41 allele carriers trended toward having lower mortality compared with GRK5 Q41 homozygous patients, whereas ADRB2 16R allele carriers had significantly increased mortality in a gene–dose response manner. Further study is needed to replicate our findings in African-American patients; randomized trials of post-ACS BB treatment in race-specific, genotype-defined subgroups might be warranted to insure the benefits of BB therapy in these high-risk individuals.