Abstract
Background
The inflammatory response triggered by cardiac surgery with cardiopulmonary bypass (CPB) is a primary mechanism in the pathogenesis of postoperative myocardial infarction (PMI), a multifactorial disorder with significant inter-patient variability poorly predicted by clinical and procedural factors. We tested the hypothesis that candidate gene polymorphisms in inflammatory pathways contribute to risk of PMI after cardiac surgery.
Methods and Results
We genotyped 48 polymorphisms from 23 candidate genes in a prospective cohort of 434 patients undergoing elective cardiac surgery with CPB. PMI was defined as creatine kinase-MB isoenzyme level ≥10 × upper limit of normal at 24 hours postoperatively. A 2-step analysis strategy was used: marker selection, followed by model building. To minimize false-positive associations, we adjusted for multiple testing by permutation analysis, Bonferroni correction, and controlling the false discovery rate; 52 patients (12%) experienced PMI. After adjusting for multiple comparisons and clinical risk factors, 3 polymorphisms were found to be independent predictors of PMI (adjusted P < 0.05; false discovery rate < 10%). These gene variants encode the proinflammatory cytokine interleukin 6 (IL6 −572G > C; odds ratio [OR], 2.47), and 2 adhesion molecules: intercellular adhesion molecule-1 (ICAM1 Lys469Glu; OR, 1.88), and E-selectin (SELE 98G > T; OR, 0.16). The inclusion of genotypic information from these polymorphisms improved prediction models for PMI based on traditional risk factors alone (C-statistic 0.764 versus 0.703).
Conclusions
Functional genetic variants in cytokine and leukocyte–endothelial interaction pathways are independently associated with severity of myonecrosis after cardiac surgery. This may aid in preoperative identification of high-risk cardiac surgical patients and development of novel cardioprotective strategies.
Keywords: cardiopulmonary bypass, genetics, inflammation, myocardial infarction, single nucleotide polymorphisms
Despite substantial advances in surgical, cardioprotective, and anesthetic techniques, the incidence of perioperative myocardial infarction (PMI) after cardiac surgery remains at 7% to 15%1 and is associated with reduced long-term survival.2 PMI is a multifactorial disorder with significant inter-patient variability poorly predicted by clinical and procedural factors, suggesting a possible genetic component.
One of the primary mechanisms in the pathogenesis of perioperative myonecrosis is the complex acute inflammatory response to cardiac surgery with cardiopulmonary bypass (CPB). The extent of perioperative systemic inflammation and associated morbidity and mortality have been related to a variety of environmental stimuli including direct surgical trauma, bioincompatibility of the extracorporeal perfusion circuit, endotoxemia, and multi-organ system ischemia-reperfusion injury.3 However, increased evidence for heritability of the pro-inflammatory state suggests that individual genetic background also modulates the magnitude of postoperative systemic inflammatory response after cardiac surgery.4 Therefore, we tested the hypothesis that single nucleotide polymorphisms (SNPs) in candidate genes regulating inflammatory pathways are associated with the incidence of postoperative myocardial infarction in a cohort of patients undergoing cardiac surgery with CPB.
Methods
Patient Population
We studied prospectively collected DNA samples from a cohort of 434 patients undergoing elective cardiac surgery with CPB between September 1997 and May 2002, in whom serial perioperative serum levels of creatine kinase-MB isoenzyme (CK-MB) were measured. All patients were participants in the Perioperative Genetics and Safety Outcomes Study (PEGASUS), an ongoing Institutional Review Board-approved, prospective, longitudinal study at Duke University Medical Center, and provided informed consent. Exclusion criteria were history of renal failure, active liver disease, bleeding disorders, autoimmune diseases, or immunosuppressive therapy. Intraoperative anesthetic, perfusion, and cardioprotective management was standardized, with fentanyl/isoflurane anesthesia, nonpulsatile CPB (30°C to 32°C), crystalloid prime, pump flow rates > 2.4 L/min per m2, cold blood cardioplegia, α-stat blood gas management, heparin to maintain activated clotting times > 450 seconds, ε-aminocaproic acid infusion, and serial hematocrits kept ≥0.18 during CPB.
Measurement of CK-MB
Serum was collected for measurement of CK-MB levels at baseline and 4.5, 24, and 48 hours after aortic cross-clamp removal, and immediately frozen at − 80°C until analysis. CK-MB levels (mass assays) were determined using a forward immunometric assay at a core laboratory (Biosite Diagnostics, San Diego, Calif). The upper limit of normal (ULN) for CK-MB values at this laboratory is 5 ng/mL.
Definition of Myocardial Injury Phenotype
Recent receiver-operator characteristic analyses from several large cardiac surgery trials have identified a cutoff value of 10-times the ULN for postoperative CK-MB to result in optimal specificity (85%) and sensitivity (39%) for 6-month mortality.5 Based on these findings and the American College of Cardiology recommendations,6 PMI was defined as CK-MB serum concentration exceeding 50 ng/mL (ie, 10-times the ULN for the reference laboratory) at 24 hours postoperatively. This time point was chosen to exclude early enzyme peaks, previously associated with a washout phenomenon.7
Candidate Genes and Polymorphisms Selection
Twenty-three candidate genes involved in the pathogenesis of inflammation and myocardial ischemia-reperfusion injury were selected a priori based on previous transcription profiling in humans8,9 and animal models,10 pathway analysis,11 a review of linkage and association studies reported in the literature, and expert opinion. Forty-eight single nucleotide polymorphisms (SNPs) were subsequently selected in these process-specific candidate genes, based on literature review, genomic context,12 and predictive analyses13 with an emphasis on functionally important variants (Table 1).
TABLE 1.
Genetic Polymorphisms Evaluated in the Study
| Gene (Symbol) | Nucleotide Substitution | Genomic Context | SNPid* | Cytogenetic Locus | Minor Allele Frequency (unaffected) |
|---|---|---|---|---|---|
| Bactericidal/permeability-increasing protein (BPI) | 270T > C | Val16Ala | rs1341023 | 20q11.23-q12 | 0.4755 |
| Catalase (CAT) | −844C > T | 5“UTR | rs769214 | 11p13 | 0.3485 |
| −262C > T | 5”UTR | rs1001179 | 0.1915 | ||
| −21A > T | 5′UTR | rs17880664 | 0.372 | ||
| Cathepsin G (CTSG) | 2108A > G | Asp125Ser | SNP014† | 14q11.2 | 0.0812 |
| CD14 Antigen (CD14) | −260C > T | 5′UTR | rs2569190 | 5q31.1 | 0.4754 |
| C-reactive protein (CRP) | 1059G > C | Leu184Leu | rs1800947 | 1q21-q23 | 0.072 |
| 1846 C > T | 3′UTR | rs1205 | 0.3116 | ||
| Endothelial nitric oxide synthase (NOS3) | −786T > C | 5′UTR | rs2070744 | 7q36 | 0.3551 |
| 894G > T | Glu298Asp | rs1799983 | 0.3179 | ||
| 20628G > T | Intron | rs1799985 | 0.3121 | ||
| E-selectin (SELE) | 98G > T | 5′UTR | rs1805193 | 1q23-q25 | 0.1088 |
| 561A > C | Ser149Arg | rs5361 | 0.093 | ||
| 1839C > T | Leu575Phe | rs5355 | 0.0583 | ||
| Heat shock protein 70-homologous (HSPA1L) | 1478C > T | Thr493Met | rs2227956 | 6p21.3 | 0.179 |
| 1804G > A | Glu602Lys | rs2075800 | 0.2463 | ||
| Intercellular adhesion molecule-1 (ICAM1) | 778G > A | Gly241Arg | rs1799969 | 19p13.3- | 0.0891 |
| 1112C > T | Pro352Leu | rs1801714 | p13.2 | 0.0281 | |
| 1462A > G | Lys469Glu | rs5498 | 0.3631 | ||
| Interleukin 1-alpha (IL1A) | −889C > T | 5′UTR | rs1800587 | 2q14 | 0.3244 |
| 10876G > T | Ala114Ser | rs17561 | 0.2997 | ||
| Interleukin 1-beta (IL1B) | −511T > C | 5′UTR | rs16944 | 2q14 | 0.3546 |
| 5810G > A | Intron | rs1143633 | 0.3171 | ||
| 5887C > T | Phe105Phe | rs1143634 | 0.2123 | ||
| Interleukin 1-receptor antagonist (IL1RN) | 13760T > C | Ala39Ala | rs2229235 | 2q14.2 | 0.2767 |
| 16857T > C | Ser112Ser | rs315952 | 0.3023 | ||
| Interleukin 6 (IL6) | −597G > A | 5′UTR | rs1800797 | 7p21 | 0.4351 |
| −572G > C | 5′UTR | rs1800796 | 0.0447 | ||
| −174G > C | 5′UTR | rs1800795 | 0.4412 | ||
| Interleukin 8 (IL8) | −251A > T | 5′UTR | rs4073 | 4q12-q13 | 0.4671 |
| Interleukin 10 (IL10) | −819C > T | 5′UTR | rs3021097 | 1q31-q32 | 0.27 |
| −592C > A | 5′UTR | rs1800872 | 0.2644 | ||
| Lipopolysaccharide-binding protein (LBP) | 19983T > C | Pro97Pro | rs2232582 | 2q11.23-q12 | 0.1516 |
| 42711T > C | Phe436Leu | rs2232618 | 0.1017 | ||
| Platelet-endothelial cell adhesion molecule-1 (PECAM1) | 1688G > A | Ser563Asn | rs12953 | 17q23 | 0.3613 |
| P-selectin (SELP) | −1969A > G | 5′UTR | rs1800805 | 1q23-q25 | 0.4195 |
| 1087G > A | Ser331Asn | rs6131 | 0.1943 | ||
| 1902A > G | Asn603Asp | rs6127 | 0.3576 | ||
| 2013G > T | Val640Leu | rs6133 | 0.1393 | ||
| 2361A > C | Thr756Pro | rs6136 | 0.1179 | ||
| Superoxide dismutase 3 (SOD3) | 760C > G | Arg231Gly | rs1799895 | 4pter-q21 | 0.0066 |
| Toll-like receptor 4 (TLR4) | 896A > G | Asp299Gly | rs4986790 | 9q32-q33 | 0.0634 |
| 1196C > T | Thr399Ile | rs4986791 | 0.0888 | ||
| Thrombomodulin (THBD) | 1959C > T | Ala473Val | rs1042579 | 20p11.2 | 0.1804 |
| Tumor necrosis-alpha (TNFA) | −308G > A | 5′UTR | rs1800629 | 6p21.3 | 0.1628 |
| −238G > A | 5′UTR | rs361525 | 0.0418 | ||
| 1078G > A | Intron | rs1800610 | 0.0672 | ||
| Vascular cell adhesion molecule-1 (VCAM1) | −1594T > C | 5′UTR | rs1800821 | 1p32-p31 | 0.1705 |
From NCBI’s dbSNP public database (http://www.ncbi.nlm.nih.gov/SNP/).
Duke internal polymorphism ID number.
UTR indicates untranslated region.
Genotype Analysis
Genotyping was performed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry on a Sequenom™ system (Sequenom, San Diego, Calif) at a core facility (Agencourt Bioscience Corporation, Beverly, Mass). Primers used and polymorphism details can be found at http://anesthesia.duhs.duke.edu/pegasus/. Genotyping accuracy was validated at > 99% by scoring a panel of 6 SNPs in 100 randomly selected patients using an ABI 3700 capillary sequencer (Applied Biosystems, Foster City, Calif).
Statistical Analysis
Before evaluating the contribution of genetic factors, the relationship between traditional risk factors (Table 2) and PMI was explored by multivariable logistic regression (clinical model). We chose the most parsimonious set of significant factors by forward selection and used 100 bootstrap samples for model validation.
TABLE 2.
Demographic, Clinical, and Procedural Characteristics of the Study Population
| Characteristics | PMI (n = 52) | No PMI (n = 382) | P* |
|---|---|---|---|
| Age, y | 63.7 (11.1) | 62.1 (11.1) | 0.31 |
| Female, % | 32.7 | 34.2 | 0.88 |
| History of MI, % | 38.5 | 40.8 | 0.77 |
| Unstable angina, % | 63.0 | 61.7 | 0.87 |
| Congestive heart failure, % | 17.4 | 23.8 | 0.36 |
| Diabetes, % | 36.5 | 29.2 | 0.33 |
| Hypertension, % | 63.5 | 62.1 | 0.88 |
| History of stroke, % | 4.3 | 2.9 | 0.64 |
| Peripheral vascular disease, % | 15.2 | 13.3 | 0.82 |
| Body surface area, m2 | 2 (0.2) | 2 (0.2) | 0.56 |
| Body mass index, kg/m2 | 29 (5) | 29 (6) | 0.66 |
| LVEF, % | 53 (13) | 53 (12) | 0.76 |
| Preoperative hematocrit, % | 39.5 (4.3) | 38.8 (8.4) | 0.39 |
| Parsonnet risk score | 11.5 (9.7) | 8.6 (7.4) | 0.07 |
| No. of grafts | 3 (1.5) | 2.6 (1.3) | 0.07 |
| Duration of CPB, min | 154 (57) | 124 (54) | < 0.0001 |
| Duration of cross-clamp, min | 95 (44) | 69 (35) | < 0.0001 |
| Procedure | |||
| CABG, % | 76.9 | 81.3 | 0.22 |
| CABG + valve, % | 11.5 | 5.3 | |
| Valve, % | 11.5 | 13.4 | |
| Redo CABG, % | 13.0 | 3.7 | 0.02 |
| Intraoperative inotropes, % | 26.9 | 12.5 | 0.008 |
| Self-reported ethnicity: | |||
| African American | 11.5 | 10.3 | 0.34 |
| European American | 80.8 | 85.8 | |
| Native American | 7.7 | 3.2 | |
| Other | 0.0 | 0.8 | |
Values expressed as mean (SD) or as %.
Wilcoxon rank-sum (continuous variables); exact Pearson χ2 (categorical variables).
LVEF indicates left ventricular ejection fraction; CPB, cardiopulmonary bypass; CABG, coronary artery bypass grafting.
For each polymorphism, allele and genotype frequencies were calculated and Hardy-Weinberg equilibrium evaluated using an exact test among unaffected patients. The association between 48 candidate gene polymorphisms and incidence of PMI was tested using a 2-stage analysis approach: marker selection, followed by modeling of genotype–phenotype relationships.14 Allelic associations with incident PMI were first assessed using χ2 tests for each of the 48 polymorphisms, and a set of influential markers selected based on nominal P < 0.1. To avoid assumptions regarding the modes of inheritance, all analyses were performed using additive (homozygote major allele versus heterozygote versus homozygote minor allele), dominant (homozygote major allele versus heterozygote + homozygote minor allele), or recessive (homozygote major allele plus heterozygote versus homozygote minor allele) models for each polymorphism. Second, we performed multivariable logistic regression analyses to sequentially test main effects and interactions for all pairs followed by 3-way combinations of markers selected in the previous step on incidence of PMI (multi-locus genetic models).
Polymorphism combinations selected by logistic regression were finally entered into a model adjusting for traditional risk factors (clinicogenetic model). The genetic contribution to model fit was tested with a multiple-degree of freedom Wald χ2 test. In addition, to compare the efficacy of PMI risk prediction models based on clinical and genetic information versus models based on traditional risk factors alone, we computed Akaike’s Information Criterion, which adjusts for the number of terms in the multivariable models, and the C-statistic, representing the area under the receiver operator characteristic curve. Population stratification was investigated by genotyping a panel of 54 unlinked null SNPs and computing a scaling factor for adjusting the χ2 test statistic in 100 bootstrap samples.15 Furthermore, self-reported ethnicity was tested as a covariate in multiple logistic regression models.16
Because the analysis strategy used many separate tests of independence, we used several approaches to account for multiple comparisons. In the genetic model selection process, permutation testing (4000 samples) was used to adjust probability values in pair-wise SNP logistic regressions,17 and Bonferroni correction to adjust overall genetic model probability values. In addition, we used false discovery rate analysis of all candidate SNPs to estimate and control the proportion of errors among the rejected hypotheses.18
All statistical analyses were performed using SAS/Genetics version 9.1 (SAS Inc, Cary, NC). Continuous variables were described as mean ± standard deviation; categorical variables were described as percentages. Adjusted P < 0.05 (Bonferroni correction or permutation testing) were considered significant.
Statement of Responsibility
The authors had full access to the data and take full responsibility for their integrity. All authors have read and agree to the manuscript as written.
Results
Of the 434 patients with complete genotype–phenotype data, PMI developed in 52 (12%). Consistent with previous studies, duration of aortic cross-clamping, number of coronary grafts, and history of previous cardiac surgery were identified as independent predictors of postoperative myonecrosis in our population (Table 3).
TABLE 3.
Results of Multivariable Logistic Regression Models Using Clinical-Procedural Risk Factors and Genotypic Information
| Predictor | OR (95% CI) | Predictor P | Model P | Model C-statistic | Model AIC |
|---|---|---|---|---|---|
| Clinical Model | |||||
| AXC time (10 min) | 1.17 (1.09–1.26) | < 0.0001 | < 0.0001 | 0.703 | 367 |
| No. coronary grafts | 1.35 (1.09–1.66) | 0.005 | |||
| Previous cardiac surgery | 2.62 (1.11–6.19) | 0.027 | |||
| Multi-Locus Genetic Model | |||||
| IL6 –572G > C (rs1800796) | 2.75 (1.2–6.3) | 0.017 | 0.01* | 0.695 | 268 |
| ICAM1 K469E (rs5498) | 1.82 (1.17–2.85) | 0.009 | |||
| SELE 98G > T(rs1805193) | 0.19 (0.05–0.81) | 0.025 | |||
| Clinico-Genetic Model | |||||
| AXC time (10 minutes) | 1.16 (1.07–1.26) | 0.0002 | < 0.0001 | 0.764 | 251 |
| No. coronary grafts | 1.29 (1.002–1.65) | 0.048 | |||
| Previous cardiac surgery | 3.93 (1.21–12.81) | 0.023 | |||
| IL6 – 572G > C (rs1800796) | 2.47 (1.02–5.97) | 0.045 | |||
| ICAM1 K469E (rs5498) | 1.88 (1.17–3.04) | 0.009 | |||
| SELE 98G > T(rs1805193) | 0.16 (0.03–0.74) | 0.019 | |||
Bonferroni-adjusted for n = 136 independent tests (55 2-SNP models, 81 3-SNP models).
C-statistic indicates area under the receiver operator characteristic curve; AIC, Akaike’s Information Criterion.
Among the 48 candidate polymorphisms examined, 4 deviated from Hardy-Weinberg equilibrium in both the white unaffected and PMI groups, and were excluded from subsequent analyses. A set of 11 SNPs was identified based on nominal univariable P < 0.1 for association with incident PMI, in any mode of inheritance (Table 4).
TABLE 4.
Estimated Effects of Polymorphisms Selected in Univariable, Multivariable, and Risk Factor-Adjusted Logistic Regression Analyses of PMI
| Univariable* |
Multivariable†
|
Multivariable, Risk Factor-Adjusted‡ |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Polymorphism | Inheritance Mode | OR [95%CI] | P | OR [95%CI] | P | OR [95%CI] | P | FDR§ | |
| IL6 | −572G > C | Additive | 2.8 [1.27–5.83] | 0.0077 | 2.75 [1.2–6.3]¶ | 0.017 | 2.47 [1.02–5.97]e | 0.045 | + |
| ICAM1 | 1462A > G | Additive | 1.9 [1.25–2.91] | 0.003 | 1.82 [1.17–2.85]¶ | 0.009 | 1.88 [1.17–3.04]¶ | 0.009 | + |
| SELE | 98G > T | Dominant | 0.17 [0.03–0.57] | 0.0157 | 0.19 [0.05–0.81]¶ | 0.025 | 0.16 [0.03–0.74]¶ | 0.019 | + |
| CRP | 1846 C > T | Dominant | 1.89 [1.01–3.7[ | 0.0519 | 2.50 [1.27–5.24]|| | 0.011 | 2.25 [1.11–4.85] || | 0.03 | − |
| LBP | 19983T > C | Dominant | 2.28 [1.22–4.21] | 0.0089 | 2.37 [1.23–4.53] || | 0.009 | 2.81 [1.41–5.65] || | 0.003 | − |
| CAT | −844C > T | Additive | 0.67 [0.42–1.04] | 0.0830 | 0.57 [0.37–0.93] || | 0.031 | 0.55 [0.31–0.93] || | 0.03 | − |
| IL1B | −511T > C | Additive | 1.59 [0.93–2.75] | 0.0948 | |||||
| 5887C > T | Dominant | 0.56 [0.28–1.07] | 0.0898 | ||||||
| IL6 | −597G > A | Additive | 0.62 [0.37–1.01] | 0.0602 | |||||
| CD14 | −260C > T | Dominant | 0.56 [0.3–1.08] | 0.0754 | |||||
| SELE | 561A > C | Dominant | 0.10 [0.01–0.46] | 0.0231 | |||||
Univariable and †multivariable logistic regression tests for allelic association.
Multivariable logistic regression adjusted for duration of aortic cross-clamping, number of coronary grafts and redo-surgery.
False discovery rate controlled at 10% to adjust for multiple comparisons in univariate tests among 48 polymorphisms.
Primary multi-locus genetic model.
Alternate multi-locus genetic model.
After our conservative analysis strategy, 3 marker associations remained significant after full adjustment for multiple comparisons and traditional risk factors: the additive effect of −572G > C interleukin-6 (IL6) polymorphism, the additive effect of Lys469Glu intercellular adhesion molecule 1 (ICAM1) polymorphism, and the dominant effect of 98G> T E-selectin (SELE) polymorphism. Specifically, in multivariable logistic regression models evaluating all possible pairwise and subsequent 3-way SNP combinations, the IL6 −572G > C (odds ratio [OR], 2.47; 95% confidence interval [CI], 1.02 to 5.97; P = 0.045), the ICAM1 Lys469Glu (OR, 1.88; 95% CI, 1.17 to 3.04; P = 0.009), and the SELE 98G > T (OR, 0.16; 95% CI, 0.03 to 0.74; P = 0.019) polymorphisms were independent predictors of PMI. Collectively, these 3 SNPs resulted in a model with a Bonferroni-adjusted P = 0.01, and a C-statistic of 0.695 (Table 3). Importantly, these polymorphisms were also individually significant after controlling the false discovery rate at 10% (Table 4). In a model of PMI risk combining both genetic and clinical factors, the contribution of these 3 polymorphisms remained highly significant (P = 0.007), over and above the information provided by traditional risk factors alone. The C-statistic of the final clinicogenetic model based on the IL6, ICAM1, and SELE polymorphisms was 0.764 compared with 0.703 in the clinical-only model, suggesting a gain in discriminatory accuracy (Table 3).
In addition to the 3 polymorphisms significant in the false discovery rate analysis, 3 others were found to be associated with PMI in multivariable logistic regression models with Bonferroni-adjusted P < 0.05. These included the 1846C > T polymorphism in C-reactive protein (CRP), 19983T > C polymorphism in lipopolysaccharide-binding protein (LBP), and −844T > C polymorphism in the catalase (CAT) gene (Tables 4 and 5).
TABLE 5.
Genotype Frequencies for the 6 Polymorphisms Selected in Multivariable Logistic Regression Models
| Polymorphism | PMI | No PMI | |
|---|---|---|---|
| IL6 −572G > C (rs1800796) | GG | 0.784 | 0.911 |
| GC | 0.216 | 0.089 | |
| ICAM1 1462A > G (rs5498) | AA | 0.235 | 0.405 |
| AG | 0.490 | 0.464 | |
| GG | 0.275 | 0.131 | |
| SELE 98G > T (rs1805193) | GG | 0.959 | 0.799 |
| GT | 0.020 | 0.184 | |
| TT | 0.020 | 0.017 | |
| CRP 1846C > T (rs1205) | CC | 0.306 | 0.455 |
| CT | 0.633 | 0.467 | |
| TT | 0.061 | 0.078 | |
| LBP 19983T > C (rs2232582) | TT | 0.542 | 0.729 |
| CT | 0.417 | 0.239 | |
| CC | 0.042 | 0.032 | |
| CAT −844T > C (rs769214) | TT | 0.558 | 0.437 |
| CT | 0.365 | 0.429 | |
| CC | 0.077 | 0.134 |
In multivariable risk factor-adjusted analyses we found no evidence for an interaction between any of these genetic polymorphisms and self-reported race in explaining incident PMI. Moreover, in analysis of 54 unlinked markers, the mean (standard error) χ2 value over 100 bootstrap samples was 0.989 (0.02), suggesting that no cryptic population substructure was present in these data.
Discussion
Despite well-described associations between genetic variation and susceptibility to myocardial infarction among ambulatory populations, there is a paucity of data regarding the occurrence of similar relationships with perioperative myocardial injury in cardiac surgical patients. In this initial report from a prospective cohort study of patients undergoing cardiac surgery with CPB, we found 3 inflammatory polymorphisms to be associated with incident PMI, after adjustment for multiple comparisons. Both risk and protective alleles were identified. These findings add to previous data implicating plasma levels of several cytokines, cell adhesion molecules, and other inflammatory mediators as key determinants of risk of perioperative myocardial injury,3 suggesting that the products of these genes may represent important targets in preventing perioperative myonecrosis after cardiac surgery.
For interleukin-6 (IL6), the encoded protein is a major proinflammatory cytokine involved in the acute inflammatory response to CPB.3 Polymorphisms in the promoter of IL6 gene (−572G > C and −174G > C) have been associated with significantly higher postoperative plasma IL-6 levels19 and prolonged hospitalization after cardiac surgery with CPB.20
Intercellular adhesion molecule-1 (ICAM1) is an important adhesion molecule mediating the interaction between activated leukocytes (CD11b) and endothelial surfaces. The non-conservative Lys469Glu polymorphism in ICAM1 gene, located in an immunodominant epitope involved in integrin-mediated B-cell adhesion and neutrophil transmigration, has been associated with a variety of pro-inflammatory phenotypes like transplant rejection and vasculopathy, vascular restenosis, and multiple sclerosis.21 These findings are consistent with the observed relationship between the number of Glu469 alleles and incidence of PMI identified in the current study.
With regard to E-selectin, this endothelial membrane protein, also called ELAM1, is expressed by cytokine-stimulated endothelial cells and mediates accumulation/adhesion of leukocytes at sites of inflammation and endothelial damage, implicated in inflammatory injury after cardiac surgery with CPB.3 Genetic variants in E-selectin have been reported as risk factors for premature/severe coronary artery disease, and are associated with altered leukocyte binding and soluble E-selectin release,22,23 suggesting a similar functional role in modulating perioperative myonecrosis.
Three additional polymorphisms were also found to be associated with incident PMI at nominal significance levels. One of these, a 1846C > T polymorphism in the 3′-untranslated region of the C-reactive protein (CRP) gene, has been associated with altered plasma CRP levels and increased risk of cardiovascular events.24 The current data thus provide additional support to previous reports implicating CRP as a mediator of tissue damage in acute myocardial ischemia.25 We also found a 326T > C polymorphism in the lipopolysac-charide-binding protein (LBP) gene to be associated with incidence of PMI, an intriguing finding because endotoxin peaks at 4 to 24 hours after CPB, and has been implicated in modulating acute myocardial injury. Finally, the −844T > C polymorphism in catalase (CAT) gene associated with protection against PMI is located in the consensus sequence of several transcription factor binding sites; this suggests that allele-specific differential binding of transcription factors may influence gene expression levels and overall antioxidant activity, thus buffering the oxidative stress characteristic of myocardial ischemia-reperfusion injury. Although these latter results are intriguing, more data are needed to provide statistical support for an association.
The specific criteria for defining PMI in the setting of cardiac surgery are still subject to debate, because postoperative biomarker elevations can be caused by several nonischemic etiologies like surgical trauma (atrial cannulation, sewing needles) and manipulation of the heart. However, regardless of causation or the diagnostic cutoff used, it should be emphasized that the biomarker evidence of myonecrosis after cardiac surgery has been consistently associated with an increase in adverse clinical outcomes.6
When interpreting any genetic association study, several epidemiological limitations potentially leading to false-positive findings should be considered, including inadequate sample size, selection of control groups, multiple testing, and population stratification. With regard to these concerns, strengths of our study include a relatively large population of cardiac surgery patients and a prospective cohort design that reduces the selection bias inherent in case-control studies. It is possible that the SNPs identified as associated with PMI are in linkage disequilibrium with other functional (causal) variants not included in this study. A much larger study incorporating many more SNPs might be necessary to delineate this effect. Further, we found no race effect in multivariable regression models, and genomic control analysis revealed no evidence of population stratification in these data. Finally, we adjusted for multiple comparisons using several different techniques (permutation testing, Bonferroni correction, false discovery rate), and are presenting all data simultaneously rather than focusing on any one specific finding.
Conclusions
Genetic variants in cytokine and leukocyte–endothelial interaction pathways are independently associated with severity of myonecrosis after cardiac surgery. These initial findings suggest that genetic epidemiological studies can assist in evaluating perioperative morbidity and, if corroborated in other populations, provide insight into preoperative identification of high-risk cardiac surgical patients. Future clinical trials investigating efficacy of novel cardioprotective strategies on cardiac biomarker release may have to be conducted in genotype-stratified patient populations.
Acknowledgments
Supported in part by grants AG09663 and HL54316 (M.F.N.), AG17556 (D.A.S.), HL075273 (D.A.S., M.V.P.), M01-RR-30 (Duke General Clinical Research Center) from the NIH; 0256342U and 9951185U (J.P.M.), 9970128N (M.F.N.), 0120492U (M.V.P.) from the American Heart Association.
Perioperative Genetics and Safety Outcomes Study (PEGASUS) Investigative Team
Andrew Allen, PhD, Carmelo A. Milano, MD
Ellen Bennett, PhD, Eugene Moretti, MD
Chonna Campbell, BS, Richard W. Morris, PhD
Fiona Clements, MD, Mark F. Newman, MD
R. Duane Davis, MD, Dahlia M. Nielsen, PhD
Bonita Funk, RN, Margaret Pericak-Vance, PhD
Donald Glower, MD, Barbara Phillips-Bute, PhD
Katherine P. Grichnik, MD, Mihai V. Podgoreanu, MD
Hilary P. Grocott, MD, Debra A. Schwinn, MD
Roger L. Hall, AAS, Andrew D. Shaw, MD
Elizabeth Hauser, PhD, Michael P. Smith, MS
Steven E. Hill, MD, Peter K. Smith, MD
Robert Jones, MD, Mark Stafford-Smith, MD
Jerry Kirchner, BS, Madhav Swaminathan, MD
Daniel Laskowitz, MD, Jeffrey M. Taekman, MD
Andrew Lodge, MD, Jeffrey M. Vance, MD, PhD
James Lowe, MD, Ian J. Welsby, MD
Eden Martin, PhD, William D. White, MPH
Joseph P. Mathew, MD, Huntington F. Willard, PhD
G. Burkhard Mackensen, MD, Walter Wolfe, MD
Footnotes
Vivien Thomas Young Investigator Award Finalist, American Heart Association Scientific Sessions, Dallas, Texas, 2005.
Presented at the American Heart Association Scientific Sessions, Dallas, Tex, November 13–16, 2005.
Disclosures
None.
Sources of Funding
References
- 1.Mangano DT. Effects of acadesine on myocardial infarction, stroke, and death following surgery. A meta-analysis of the 5 international randomized trials. The Multicenter Study of Perioperative Ischemia (McSPI) Research Group. JAMA. 1997;277:325–332. doi: 10.1001/jama.277.4.325. [DOI] [PubMed] [Google Scholar]
- 2.Force T, Hibberd P, Weeks G, Kemper AJ, Bloomfield P, Tow D, Josa M, Khuri S, Parisi AF. Perioperative myocardial infarction after coronary artery bypass surgery. Clinical significance and approach to risk stratification. Circulation. 1990;82:903–912. doi: 10.1161/01.cir.82.3.903. [DOI] [PubMed] [Google Scholar]
- 3.Menasche P, Edmunds LHJ. Extracorporeal circulation: the inflammatory response. In: Cohn LH, Edmunds LHJ, editors. Cardiac surgery in the adult. 2. New York: McGraw-Hill; 2003. pp. 349–360. [Google Scholar]
- 4.Pankow JS, Folsom AR, Cushman M, Borecki IB, Hopkins PN, Eckfeldt JH, Tracy RP. Familial and genetic determinants of systemic markers of inflammation: the NHLBI family heart study. Atherosclerosis. 2001;154:681–689. doi: 10.1016/s0021-9150(00)00586-4. [DOI] [PubMed] [Google Scholar]
- 5.Klatte K, Chaitman BR, Theroux P, Gavard JA, Stocke K, Boyce S, Bartels C, Keller B, Jessel A. Increased mortality after coronary artery bypass graft surgery is associated with increased levels of postoperative creatine kinase-myocardial band isoenzyme release: results from the GUARDIAN trial. J Am Coll Cardiol. 2001;38:1070–1077. doi: 10.1016/s0735-1097(01)01481-4. [DOI] [PubMed] [Google Scholar]
- 6.Newby LK, Alpert JS, Ohman EM, Thygesen K, Califf RM. Changing the diagnosis of acute myocardial infarction: implications for practice and clinical investigations. Am Heart J. 2002;144:957–980. doi: 10.1067/mhj.2002.129778. [DOI] [PubMed] [Google Scholar]
- 7.Dahlin LG, Kagedal B, Nylander E, Olin C, Rutberg H, Svedjeholm R. Early identification of permanent myocardial damage after coronary surgery is aided by repeated measurements of CK-MB. Scand Cardiovasc J. 2002;36:35–40. doi: 10.1080/140174302317282366. [DOI] [PubMed] [Google Scholar]
- 8.Ruel M, Bianchi C, Khan TA, Xu S, Liddicoat JR, Voisine P, Araujo E, Lyon H, Kohane IS, Libermann TA, Sellke FW. Gene expression profile after cardiopulmonary bypass and cardioplegic arrest. J Thorac Cardiovasc Surg. 2003;126:1521–1530. doi: 10.1016/s0022-5223(03)00969-3. [DOI] [PubMed] [Google Scholar]
- 9.Tomic V, Russwurm S, Moller E, Claus RA, Blaess M, Brunkhorst F, Bruegel M, Bode K, Bloos F, Wippermann J, Wahlers T, Deigner HP, Thiery J, Reinhart K, Bauer M. Transcriptomic and proteomic patterns of systemic inflammation in on-pump and off-pump coronary artery bypass grafting. Circulation. 2005;112:2912–2920. doi: 10.1161/CIRCULATIONAHA.104.531152. [DOI] [PubMed] [Google Scholar]
- 10.Podgoreanu MV, Michelotti GA, Sato Y, Smith MP, Lin S, Morris RW, Grocott HP, Mathew JP, Schwinn DA. Differential cardiac gene expression during cardiopulmonary bypass: ischemia-independent upregulation of proinflammatory genes. J Thorac Cardiovasc Surg. 2005;130:330–339. doi: 10.1016/j.jtcvs.2004.11.052. [DOI] [PubMed] [Google Scholar]
- 11.Calvano SE, Xiao W, Richards DR, Felciano RM, Baker HV, Cho RJ, Chen RO, Brownstein BH, Cobb JP, Tschoeke SK, Miller-Graziano C, Moldawer LL, Mindrinos MN, Davis RW, Tompkins RG, Lowry SF. A network-based analysis of systemic inflammation in humans. Nature. 2005;437:1032–1037. doi: 10.1038/nature03985. [DOI] [PubMed] [Google Scholar]
- 12.Tabor HK, Risch NJ, Myers RM. Opinion: Candidate-gene approaches for studying complex genetic traits: practical considerations. Nat Rev Genet. 2002;3:391–397. doi: 10.1038/nrg796. [DOI] [PubMed] [Google Scholar]
- 13.Ng PC, Henikoff S. SIFT: Predicting amino acid changes that affect protein function. Nucleic Acids Res. 2003;31:3812–3814. doi: 10.1093/nar/gkg509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hoh J, Wille A, Zee R, Cheng S, Reynolds R, Lindpaintner K, Ott J. Selecting SNPs in two-stage analysis of disease association data: a model-free approach. Ann Hum Genet. 2000;64(Pt 5):413–417. doi: 10.1046/j.1469-1809.2000.6450413.x. [DOI] [PubMed] [Google Scholar]
- 15.Reich DE, Goldstein DB. Detecting association in a case-control study while correcting for population stratification. Genet Epidemiol. 2001;20:4–16. doi: 10.1002/1098-2272(200101)20:1<4::AID-GEPI2>3.0.CO;2-T. [DOI] [PubMed] [Google Scholar]
- 16.Tang H, Quertermous T, Rodriguez B, Kardia SL, Zhu X, Brown A, Pankow JS, Province MA, Hunt SC, Boerwinkle E, Schork NJ, Risch NJ. Genetic structure, self-identified race/ethnicity, and confounding in case-control association studies. Am J Hum Genet. 2005;76:268–275. doi: 10.1086/427888. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Good PI. Permutation tests: a practical guide to resampling methods for testing hypotheses. 2. New York: Springer-Verlag; 2000. [Google Scholar]
- 18.Devlin B, Roeder K, Wasserman L. Analysis of multilocus models of association. Genet Epidemiol. 2003;25:36–47. doi: 10.1002/gepi.10237. [DOI] [PubMed] [Google Scholar]
- 19.Brull DJ, Montgomery HE, Sanders J, Dhamrait S, Luong L, Rumley A, Lowe GD, Humphries SE. Interleukin-6 gene -174g > c and -572g > c promoter polymorphisms are strong predictors of plasma interleukin-6 levels after coronary artery bypass surgery. Arterioscler Thromb Vasc Biol. 2001;21:1458–1463. doi: 10.1161/hq0901.094280. [DOI] [PubMed] [Google Scholar]
- 20.Burzotta F, Iacoviello L, Di Castelnuovo A, Glieca F, Luciani N, Zamparelli R, Schiavello R, Donati MB, Maseri A, Possati G, Andreotti F. Relation of the -174 G/C polymorphism of interleukin-6 to interleukin-6 plasma levels and to length of hospitalization after surgical coronary revascularization. Am J Cardiol. 2001;88:1125–1128. doi: 10.1016/s0002-9149(01)02046-x. [DOI] [PubMed] [Google Scholar]
- 21.Borozdenkova S, Smith J, Marshall S, Yacoub M, Rose M. Identification of ICAM-1 polymorphism that is associated with protection from transplant associated vasculopathy after cardiac transplantation. Hum Immunol. 2001;62:247–255. doi: 10.1016/s0198-8859(01)00208-7. [DOI] [PubMed] [Google Scholar]
- 22.Wenzel K, Stahn R, Speer A, Denner K, Glaser C, Affeldt M, Moobed M, Scheer A, Baumann G, Felix SB. Functional characterization of atherosclerosis-associated Ser128Arg and Leu554Phe E-selectin mutations. Biol Chem. 1999;380:661–667. doi: 10.1515/BC.1999.082. [DOI] [PubMed] [Google Scholar]
- 23.Yoshida M, Takano Y, Sasaoka T, Izumi T, Kimura A. E-selectin polymorphism associated with myocardial infarction causes enhanced leukocyte-endothelial interactions under flow conditions. Arterioscler Thromb Vasc Biol. 2003;23:783–788. doi: 10.1161/01.ATV.0000067427.40133.59. [DOI] [PubMed] [Google Scholar]
- 24.Miller DT, Zee RY, Suk Danik J, Kozlowski P, Chasman DI, Lazarus R, Cook NR, Ridker PM, Kwiatkowski DJ. Association of common CRP gene variants with CRP levels and cardiovascular events. Ann Hum Genet. 2005;69(Pt 6):623–638. doi: 10.1111/j.1529-8817.2005.00210.x. [DOI] [PubMed] [Google Scholar]
- 25.Griselli M, Herbert J, Hutchinson WL, Taylor KM, Sohail M, Krausz T, Pepys MB. C-reactive protein and complement are important mediators of tissue damage in acute myocardial infarction. J Exp Med. 1999;190:1733–1740. doi: 10.1084/jem.190.12.1733. [DOI] [PMC free article] [PubMed] [Google Scholar]