VENTRICULAR dysfunction (VnD) has been reported after cardiac surgery in 10–20%1–3of patients and has been associated with increased hospital stay and long-term mortality after primary coronary artery bypass graft (CABG) surgery with cardiopulmonary bypass (CPB).2B-type natriuretic peptide (BNP) is a member of the natriuretic peptide family, which includes three structurally homologous peptides: A-type (ANP), BNP, and C-type natriuretic peptides.4ANP and BNP are primarily released from atrial and ventricular myocytes in response to increased cardiac atrial and ventricular wall stress.4C-type natriuretic peptide is released from vascular endothelial cells and other tissues.4Natriuretic peptides bind to two guanylyl cyclase-coupled effector receptors (natriuretic peptide receptors A and B) and a clearance receptor (natriuretic peptide receptor C). In addition to compensatory natriuretic, diuretic, and vasorelaxant properties, natriuretic peptides also have mitigating paracrine influences on myocardial ischemia–reperfusion injury and cardiovascular remodeling.4–14
Circulating plasma BNP levels are used to assess the severity of heart failure and myocardial ischemia and to predict adverse outcomes in patients with acute coronary syndromes or undergoing cardiovascular surgery.1,2,15–18Natriuretic peptide system gene variants have been associated with cardiovascular disease states such as hypertension,19–21stroke,22,23and myocardial infarction.24However, to date, the association between natriuretic peptide system gene variants and development of VnD after cardiac surgery with CPB has not been examined. Knowledge of such genotypic predictors may enhance our understanding of the molecular mechanisms underlying postoperative VnD. We thus hypothesized that variants within the natriuretic peptide precursor protein (NPPA, NPPB, NPPC ), receptor (NPR1, NPR2, NPR3 ), and precursor converting enzyme (CORIN ) genes are independently associated with the risk of postoperative VnD in patients of European descent undergoing primary CABG surgery with CPB.
Materials and Methods
Between August 2001 and June 2005, 1,164 men and women aged 20 to 89 yr scheduled for isolated primary CABG surgery with CPB at Brigham and Women’s Hospital, Boston, Massachusetts, and Texas Heart Institute, St. Luke’s Episcopal Hospital, Houston, Texas, were prospectively enrolled into an ongoing parent study known as the CABG Genomics Study.##Respective Institutional Review Board approval and subject written informed consent were obtained. The overall objective of the CABG Genomics Study is to identify associations between genotypic variants and adverse perioperative outcomes. CABG Genomics Study exclusion criteria include a preoperative hematocrit less than 25% or transfusion of leukocyte-rich blood products within 30 days before surgery. Enrolled subjects were prospectively excluded from analysis for this study if they underwent emergency surgery, received a preoperative inotrope, intraaortic balloon pump, or ventricular assist device support, had severe preoperative renal dysfunction (preoperative hemodialysis or preoperative serum creatinine greater than 3 mg/dl) or were missing preoperative or peak postoperative plasma BNP or cardiac troponin I (cTnI) measurements. To avoid confounding due to population stratification, subjects were further prospectively excluded from analysis if they reported having a parent or grandparent of non-European descent or if assessment of genomic control single nucleotide polymorphisms (SNPs) indicated non-European ethnicity.25
Data and Blood Collection
Data were collected for each enrolled subject during primary hospitalization using a detailed prospectively designed study case report form that includes: (1) preoperative demographic characteristics, comorbidities, and medications; (2) surgical characteristics; and (3) postoperative in-hospital events. Data were subject to automated range and logic checking as well as additional manual audit of a proportion of records.
Plasma samples obtained preoperatively and on postoperative days 1–5 were stored in vapor-phase liquid nitrogen until analysis. Plasma BNP and cTnI were measured at all time points using immunoassays conducted at a single reference laboratory (Biosite, San Diego, CA). DNA was extracted from white blood cells using standard procedures. Genotyping was performed using Sequenom MassARRAY® iPLEX (Sequenom, San Diego, CA) at Helmholtz Zentrum München (Neuherberg, Germany). Raw genotyping signals were analyzed with SpectroTyper 3.4 software (Sequenom) and were manually curated.
Seven candidate genes related to the natriuretic peptide system were selected a priori : NPPA, NPPB, NPPC, NPR1, NPR2, NPR3, and CORIN . SNPs were selected for genotyping to achieve approximately 2,500 base pair spacing across each gene region. SNPs were preferentially selected if they were assessed in previously conducted cardiovascular studies, were nonsynonymous coding SNPs, were within exon regions, were within 50 base pairs of intron-exon boundaries, were in promoter regions, or if they “tagged” haplotype blocks (as per the Caucasian HapMap resource).***Genotyping was performed for 155 SNPs. SNPs were excluded from analysis for one or more of the following reasons: genotyping rate less than 90% (3 SNPs), observed minor allele frequency less than 1% (13 SNPs), or not being in Hardy Weinberg equilibrium in control subjects (P ≤ 0.001; 1 SNP). No SNPs required exclusion for differential missingness greater than 10% between case and control groups. A total of 71 additional genomic control SNPs were also genotyped to exclude subjects with non-European ethnicities. To account for potential stratification by Northern versus Southern European ancestry, 6 SNPs within the lactase gene (rs182549, rs2322659, rs3754689, rs3769005, rs4954490, rs4988235) were also genotyped.25,26Subjects were also excluded if they were missing more than 10% of their genotyping data.
Postoperative VnD was defined as a new requirement for two or more inotropes or new placement of an intraaortic balloon pump or ventricular assist device either during the intraoperative period after the patient separated from CPB or postoperatively in the intensive care unit. Inotrope support was defined as continuous infusion of amrinone, milrinone, dobutamine, dopamine (> 5 μg · kg–1· min−1), epinephrine, isoproterenol, norepinephrine, or vasopressin. Dyspnea score was derived using the New York Heart Association classification (1 = no dyspnea, 2 = mild impairment of daily functioning, 3 = substantial functional impairment when not at rest, 4 = functional impairment at rest).27Urgent CABG surgery was defined as occurring within the same hospitalization as the diagnosis of an acute coronary event or coronary artery disease. Stenosis of more than 50% of the left anterior descending left circumflex or right coronary arteries or their major branches were quantified on the basis of cardiac catheterization data and scored as regions of coronary arterial disease (1, 2, or 3 regions total). Stenosis of more than 50% of the left main coronary artery was counted as two regions of significant disease. Peak postoperative plasma BNP and cTnI was assessed if a subject had at least three postoperative day 1–5 measures or, for peak postoperative cTnI, at least postoperative day 1 and 2 measures.
Categorical and continuous patient and clinical characteristics were compared between case and control groups with chi-square, Fisher exact, or Wilcoxon rank sum test as indicated using SAS (version 9.1.3, SAS Institute, Cary, NC). Continuous plasma BNP and cTnI data were log10transformed to normalize data distribution before analysis. Multivariable logistic regression models of nongenetic clinical and biomarker predictors of VnD were also created using SAS. One model was created using preoperative and intraoperative covariates, and the other model also included peak postoperative cTnI to adjust in addition for the effects of postoperative myocardial injury. The two multivariable clinical prediction models were created using combined forward and backward regression modeling with covariate entry into the models determined by univariate association with VnD (P < 0.15; table 1) and exit from the models determined by P value of 0.05. Preoperative cTnI greather than 0.1 μg/l was kept in the model that includes preoperative and intraoperative characteristics because it has been previously associated with VnD in a larger study group.2CPB time greater than 120 min was maintained in the model that includes postoperative peak cTnI concentration, as this covariate independently predicts VnD in the multivariable model that includes only preoperative and intraoperative covariates. Demographic characteristics of age, gender, institution, and body mass index greater than 30 kg/m2were locked into both multivariable models regardless of significance. Simultaneous inclusion of preoperative myocardial infarction in the model with peak postoperative cTnI did not improve the predictive ability of the model.
Genetic statistical analyses were conducted with PLINK (version 1.01).28Hardy Weinberg equilibrium was evaluated using Fisher exact tests. Potential confounding of associations between natriuretic peptide SNP frequencies and VnD by subjects’ Northern versus Southern European ancestries was determined by investigating association between lactase gene SNPs and VnD. Individual SNP associations with VnD were estimated using logistic regression for additive (each copy of the minor allele has an equivalent additional additive predictive value, i.e. , 0, 1, 2), dominant (1 or 2 copies of the minor allele vs. 0 copies of the minor allele), and recessive (2 copies of the minor allele vs. 0 or 1 copies of the minor allele) genetic models, and corresponding odds ratios (OR) and 95% confidence intervals (CI) were calculated for each SNP according to the role of the SNP’s minor allele in each genetic model. SNPs associated with VnD in single covariate logistic assessment were then individually entered into the multivariable clinical logistic regression models for predicting VnD, and multivariate adjusted ORs and 95% CIs were determined for each SNP with regards to developing VnD.
Empirical P values were determined for all SNP associations based on permuting case/control status (15,000 repetitions) and adjusting for family-wise type 1 error within each gene. Point-wise permutation analyses were conducted for each SNP to adjust for potentially random occurrence of the SNP’s minor allele in VnD cases versus controls. Permutation analyses adjusting for family-wise type 1 error were conducted to adjust for the probability of making false-positive discoveries when assessing multiple SNPs within a gene region for association with the VnD outcome. NPPA and NPPB were assessed as one genetic region, as these two genes lie in close proximity on chromosome 1.
Linkage disequilibrium blocks and associated haplotypes for NPPA /NPPB (fig. 1) and NPR3 (fig. 2) gene regions were derived in Haploview (version 4.1; The Broad Institute, Cambridge, MA) using default criteria.29,30Omnibus tests were conducted to assess if variance in haplotype frequencies within each linkage disequilibrium block had an overall significant association with the VnD phenotype. Haplotypes within linkage disequilibrium blocks that had omnibus test P values < 0.05 were then assessed with Pearson chi-square tests for association with VnD. These univariate haplotype results were consistent with observed SNP associations; therefore, additional permutation analyses and multivariable adjustments for clinical covariates were not conducted.
Of the 1,164 subjects enrolled into the source cohort during the study period, 407 were prospectively excluded from analysis for one or more of the following exclusion criteria: non-European ethnicity (n = 176), no CABG surgery performed (n = 6), history of cardiac surgery (n = 4), concurrent valve surgery performed (n = 51), emergency surgery (n = 4), CPB not used (n = 39), aortic cross-clamp not used (n = 68), preoperative ventricular assist device (n = 1), preoperative intraaortic balloon pump (n = 27), preoperative inotropes (n = 4), no preoperative or peak postoperative BNP or cTnI measurements (n = 72), preoperative hemodialysis (n = 1), preoperative serum creatinine greater than 3 mg/dl (n = 4), or no genotyping performed (n = 83). In addition to these exclusions 60 subjects were excluded for having less than 90% success for their individual genotyping.
Perioperative patient characteristics for 697 subjects included in the study analysis are shown in table 1, stratified by occurrence of postoperative VnD (n = 76). None of the subjects who developed the VnD phenotype underwent ventricular assist device placement. Twelve VnD subjects received an intraaortic balloon pump intraoperatively after separating from CPB. An additional seven VnD subjects underwent intraaortic balloon pump insertion postoperatively in the intensive care unit. Forty- seven VnD subjects (6.7%) required at least 2 inotropes intraoperatively, and 43 subjects (6.2%) required at least 2 inotropes postoperatively.
Patients who developed postoperative VnD were significantly more likely to have a more than 30 pack-year history of smoking, renal insufficiency, recent myocardial infarction, lower preoperative left ventricular ejection fraction, higher preoperative plasma BNP concentration, moderate to severe mitral regurgitation, or to be taking an angiotensin-converting enzyme inhibitor or diuretic. Subjects who developed VnD had significantly longer CPB and aortic cross-clamp times and higher peak postoperative cTnI concentrations. Multivariable logistic regression models of patient and clinical predictors of postoperative VnD in the 697 study subjects are shown in table 2and are consistent with our previous report from a larger, multiethnic cohort, that preoperative left ventricular ejection fraction, BNP, and CPB time greater than 120 min are independent clinical predictors of VnD after CABG.2
Natriuretic Peptide System SNP Associations with Ventricular Dysfunction
The 139 SNPs that passed quality control criteria are listed in Supplemental Digital Content along with corresponding P values for univariate associations with VnD as assessed with additive, dominant, and recessive genetic models (see appendix, Supplemental Digital Content 1, which shows 139 single nucleotide polymorphisms genotyped within 7 natriuretic peptide system gene regions in 697 patients of European descent with and without ventricular dysfunction after primary coronary artery bypass graft surgery, http://links.lww.com/A819). The genotyping rate for these 139 SNPs was 98.8%. The minor alleles of 13 NPPA/NPPB gene region SNPs were univariately associated with a decreased incidence of VnD after CABG surgery, and the minor alleles of seven NPR3 gene SNPs were univariately associated with an increased incidence of VnD after CABG surgery (asymptotic P < 0.05; table 3). SNPs in the NPPC , NPR1, and CORIN gene regions were not associated with VnD. One NPR3 SNP rs11740580 and one NPR2 region SNP rs3808864 were univariately associated with decreased VnD, but these associations did not survive statistical adjustments for multiple comparisons.
After adjusting for clinical predictors of VnD using either multivariable model shown in table 2, the minor alleles of eleven NPPA/NPPB gene SNPs and the minor alleles of six NPR3 gene SNPs remained associated with the occurrence of postoperative VnD (asymptotic P < 0.05). After additional point-wise permutation adjustment for subject case-control status and family-wise permutation adjustment for total number of SNPs assessed in the gene region, the minor alleles of seven NPPA/NPPB SNPs and four NPR3 SNPs remained associated with occurrence of postoperative VnD using either multivariable model shown in table 2(family-wise empirical P < 0.05). Table 4presents the gene association results adjusted for the clinical model containing preoperative cTnI greater than 0.1 μg/l, but the asymptotic, point-wise, and family-wise significant associations (P < 0.05) were unchanged when adjustments were made for the second clinical model.
The logistic regression model including demographic, clinical, and nongenetic preoperative biomarker variables (table 2) predicted 15.9% of the variability (r2) in occurrence of postoperative VnD, and the area under the receiver operating characteristic curve related to this model was 0.781. When the SNPs most significantly associated with VnD from the NPPA/NPPB region (rs549596; additive model) and the NPR3 region (rs700923; recessive model) were both added to the clinical model, the proportion of explained variability (r2) for the model increased to 21.9% (P < 0.0001) and the area under the receiver operating characteristic curve for the model increased to 0.828.
Natriuretic Peptide System Haplotype Associations with Ventricular Dysfunction
The seven NPPA/NPPB SNPs that remained associated with a decreased risk of VnD after statistical adjustment for clinical predictors and multiple comparisons are contained within the linkage disequilibrium blocks 2, 3, and 4 (chr1:11,832, 122–11, 851, 301) of the NPPA/NPPB gene region (fig. 1). Omnibus tests for global association between haplotype variation within each NPPA/NPPB linkage disequilibrium block and the VnD phenotype were significant for blocks 1, 3, and 4 (P < 0.05; block 2 P = 0.055). Within these blocks, analyses of individual haplotype associations with occurrence of VnD were consistent with the NPPA/NPPB region SNP associations.
The four NPR3 SNPs that remained associated with an increased risk of VnD after statistical adjustment for clinical predictors and multiple comparisons are contained within block 3 of the NPR3 gene (chr5:32, 769, 725–32, 781,040; fig. 2). Omnibus tests for global association between haplotype variability within NPR3 linkage disequilibrium blocks and occurrence of the VnD phenotype showed significant associations for blocks 3 and 5 (P < 0.05). Within these blocks, analyses of individual haplotype associations were consistent with the associations observed for the NPR3 SNPs.
We have identified novel regions of genetic variation within the NPPA , NPPB , and NPR3 genes that are associated with VnD after primary CABG surgery. After adjusting for environmental covariates and multiple comparisons, the minor alleles of the NPPA/NPPB SNPs identified in this study associate with a decreased risk of postoperative VnD, and the minor alleles of the NPR3 SNPs identified in this study associate with an increased risk of postoperative VnD. Furthermore, simultaneous addition of the NPPA/NPPB SNP rs549596 and the NPR3 rs700923 into the clinical multivariable logistic regression model for predicting postoperative VnD improved the predictive ability of the model.
Although NPPA and NPPB SNPs have not been assessed previously in relation to adverse perioperative cardiovascular outcomes, several NPPA/NPPB gene variants have been examined for association with ambulatory cardiovascular diseases.19,22–24,31–35In the only study examining NPPA/NPPB variants in ambulatory heart failure patients, the NPPA nonsynonymous coding polymorphism rs5065 was found to be associated with elevated plasma BNP and amino-terminal BNP (NT-proBNP) levels in patients who were New York Heart Association class III-IV, but not in patients who were New York Heart Association class I–II.31rs5065 has also been associated with nonfatal myocardial infarction24and coronary artery disease24and, along with the rare nonsynonymous coding polymorphism rs5063, has been associated with stroke,22hypertension,19and left ventricular mass in hypertensive patients.34We did not identify any association between rs5065 or rs5063 and VnD after CABG surgery.
In addition to the above studies of NPPA/NPPB gene variants in relation to cardiovascular disease, the NPPA/NPPB promoter SNPs rs198388 and rs198389 have been associated with elevated circulating NT-proBNP concentrations in ambulatory diabetics with nephropathy,33and these same two SNPs along with NPPA/NPPB SNPs rs632793, rs6668352, 6676300, and rs198375 have been associated with elevated plasma BNP levels in a large ambulatory Japanese cohort.32The strong overlap between these NPPA/NPPB SNPs and the NPPA/NPPB SNPs associated with VnD in the current study supports the need for further studies regarding how genetic variation in the region containing these SNPs may interact with the natriuretic peptide system and influence postoperative VnD. The NPPB promoter region has multiple cis regulatory elements known to be gene regulators, and multiple physiologic stimuli including mechanical stretch, ischemic injury and hypoxia, and inflammatory mediators are known to stimulate regulation of the NPPB gene.36Furthermore, processing of natriuretic peptide preprohormones to more active fragments is complex and at present incompletely understood.13Autocrine and paracrine effects of natriuretic peptides on myocardial ischemia–reperfusion injury and remodeling are also being identified.13
In comparison with the NPPA/NPPB gene region, less is known about the association between SNPs within the natriuretic peptide receptor C (NPR3 ) gene and adverse cardiovascular outcomes. Although natriuretic peptide receptor C was initially identified as a natriuretic peptide clearance receptor, animal studies indicate that this receptor may locally modulate the physiologic effects of natriuretic peptides and may influence microvascular permeability and cardiac sarcolemmal Na+-K+pump activity.12,37,38The major allele of the NPR3 promoter SNP rs9716700 has been associated with increased family history of hypertension, as well as lower ANP levels and higher systolic blood pressure in obese hypertensive patients,20,21and a different study found a univariate association between the major allele of rs9716700 and higher plasma ANP and BNP levels in ambulatory patients without heart failure but not in patients with heart failure.31While rs9716700 was not assessed in the current study, we did not find associations between other SNPs in the NPR3 promoter region and VnD after CABG. Furthermore, the NPR3 SNPs that we did find to be associated with VnD are within a linkage disequilibrium block that does not include the NPR3 promoter region and is demarcated from the promoter region by a strong recombination point. Therefore, the NPR3 SNPs associated with VnD in this study are unlikely to be related to rs9716700.
The NPR3 SNPs (rs700923, rs16890196, rs765199, rs700926) that we identified as significantly associated with decreased postoperative VnD lie within currently assumed noncoding intronic sequences, and there are several potential mechanisms that could explain their association with the study’s clinical VnD outcome and warrant further study. These SNPs may be in linkage disequilibrium with promoter SNPs that have not yet been identified or that were not genotyped in this study. Furthermore, these intronic SNPs may have promoter functions that have not yet been identified. Finally, intronic variants may potentially affect receptor function through alternative splicing mechanisms. For example, recent reports have demonstrated associations between intronic SNPs unrelated to the NPR3 gene and alternatively spliced mRNA transcripts that alter function of ultimately translated protein.39
There are several potential limitations to this study. First, the current cardiac surgical literature does not include a standardized outcome definition for ventricular dysfunction after cardiac surgery. Many primary CABG patients at both study institutions do not undergo perioperative monitoring with transesophageal echocardiography or pulmonary artery catheters. We elected to define VnD after CABG surgery as a need for two or more inotropes or new intraaortic balloon pump or ventricular assist device support to best ensure that we were not including patients with normal ventricular function. It is not standard organizational or surgeon-based practice at either study institution to separate from CPB on prophylactic inotropes. Furthermore, we have previously reported that patients with our definition of VnD experience significantly prolonged hospital stays and increased up to 5-yr all-cause mortality after primary CABG surgery.2Second, we limited analysis to people of European descent only, as we had an insufficient number of non-European subjects to allow for statistical accommodation of genetic admixture. Thus, further studies are warranted to replicate our findings in non-European cardiac surgical patient populations. Third, our study’s sample and effect sizes allowed adjustment for multiple comparisons within each gene, but our findings should be viewed as exploratory until further validation studies are conducted in other European-descent cardiac surgical populations. Fourth, we have identified regions in NPPA/NPPB and NPR3 that contain genetic variants independently associated with VnD after CABG surgery, but there is considerable linkage disequilibrium within these regions. Consequently, we cannot conclude that the SNPs identified in this study are directly related to occurrence of VnD after CABG surgery. Finally, our findings do not identify mechanistic pathways that link identified SNP associations to occurrence of postoperative VnD. The natriuretic peptide system and its regulation are complex.36Further investigation is required to determine how the NPPA/NPPB and NPR3 SNPs associations identified in this study relate to development of postoperative VnD.
VnD after CABG surgery with CPB is associated with increased postoperative hospital stay and mortality. Genetic variation within defined regions of the NPPA/NPPB and NPR3 natriuretic peptide system genes is associated with VnD after CABG surgery and improves ability to predict the VnD outcome beyond what can be predicted using clinical covariates alone. Further investigation of these regions is warranted, as knowledge of such genotypic predictors may enhance our understanding of the molecular mechanisms underlying postoperative VnD and heart failure.