The PlA2 polymorphism of platelet glycoprotein IIIa has been identified as a prothrombotic risk factor in a number of cardiovascular settings. The aim of this study was to determine whether the PlA2 polymorphism of platelet glycoprotein IIIa and degree of platelet activation were associated with more severe myocardial injury as indicated by troponin I release following cardiopulmonary bypass.
The PlA2 genotype was determined in 66 patients undergoing elective coronary artery bypass grafting requiring cardiopulmonary bypass. Troponin I concentrations and the percentage of circulating, activated (CD62P+) platelets were measured at predetermined intervals perioperatively.
Forty-six patients were Pl(A1,A1), and 20 were Pl(A1,A2) or Pl(A2,A2). Patients with at least one PlA2 allele had significantly greater postoperative troponin I concentrations than PlA1 homozygotes (P = 0.006, analysis of variance). Peak troponin I concentrations also correlated significantly with the increase in circulating, activated platelets (P = 0.02, Spearman rank correlation).
The PlA2 allele of platelet glycoprotein IIIa is associated with higher troponin I concentrations following cardiopulmonary bypass surgery, suggesting that this platelet polymorphism contributes to perioperative myocardial injury.
ENHANCED platelet activation and prococagulant activity may contribute to the pathophysiology of arterial vascular events. PlA2, a polymorphism of the glycoprotein IIIa constituent of the platelet integrin receptor glycoprotein IIb/IIIa, has been proposed as a risk factor for myocardial infarction. 1Although still somewhat controversial, 2studies of allelic frequencies in patients with coronary artery disease with and without coronary thrombosis suggest that the PlA2polymorphism may predispose to increased thrombogenicity. 3,4Zotz et al. 5studied patients after coronary artery bypass grafting (CABG) surgery, beginning 30 days after surgery and extending to 1 yr, and demonstrated that the PlA2polymorphism increases the risk of cardiac events and death over this period. We hypothesized that procoagulant alterations in platelet function produced by the PlA2polymorphism would also exacerbate the degree of myocardial injury that is directly and acutely associated with CABG surgery. This pilot study examined whether the PlA2polymorphism is associated with increased myocardial injury as indicated by higher postoperative concentrations of the cardiac-specific protein, troponin I (cTpnI). In addition, since enhanced platelet activation is one proposed mechanism for PlA2-associated vascular injury, 6the percentage of circulating CD62P+ platelets was also measured perioperatively to determine whether the platelet activation response to cardiopulmonary bypass (CPB) correlated with PlAgenotype or postoperative cTpnI concentrations.
Material and Methods
Patient Selection and Conduct of Cardiopulmonary Bypass
Power analysis suggested that at least 17 PlA2+and 17 PlA1/A1patients would need to be studied to detect a significant cTpnI difference of 5 ng/ml. Given a PlA2allelic frequency of 28–30%, a minimum of 60 patients was required. After we obtained Human Investigation Committee approval and informed consent, 66 consecutive nonpregnant adults undergoing elective CABG at Yale–New Haven Hospital who were enrolled in the Multicenter Study of Perioperative Ischemia Research Group's prospective study of post-CPB outcomes were studied. All patients underwent CPB using a standardized membrane oxygenator, roller pump, and cardiotomy suction setup. 7
Determination of PlAGenotypes
Blood drawn preoperatively into 5 mm EDTA was spotted onto sterile filter paper and dried. 8Disks cut out of the blood spot were placed in polymerase chain reaction tubes with 20 μl methanol. After drying overnight, two separate polymerase chain reactions were performed for PlA1and PlA2, as reported by Skogen et al . 9Amplification products were electrophoresed onto a 2:1 Nusieve:Seakem agarose (FMC Bioproducts, Rockland, ME) gel in TBE buffer. Primers for β-actin were included as a control. 10
Five blood samples were drawn into fixative (1% paraformaldehyde final concentration) at the start of surgery, prior to and following aortic cross-clamp release, on arrival in the intensive care unit, and in the morning of postoperative day 1. Platelet activation was examined by flow cytometry as previously reported 11using monoclonal antibodies to CD41 to identify platelets and CD62P to assess their activation status.
Serial blood samples were drawn at the following time points: (1) start of surgery, (2) on arrival in the intensive care unit, (3) 6 h after termination of CPB, and (4) in the morning of postoperative day 1. Samples were centrifuged immediately, and serum aliquots were stored at −70°C. cTpnI concentrations were determined on the Axsym® (Abbott, Abbott Park, IL) EIA system according to the manufacturer's instructions. 12
Twelve-lead electrocardiograms were obtained preoperatively, on admission to the intensive care unit, between postoperative days 2 and 4, and in addition as clinically appropriate. All electrocardiograms were read and agreed on by two cardiologists blinded to the patient's clinical course and cardiac enzyme concentrations. Cardiac events occurring by electrocardiogram criteria were classified according to Greenson et al. 13as one of the following: (1) persistent new Q waves (at least one third the QRS height) occurring in at least two leads, (2) persistent ST segment changes greater than 0.2 mV in two or more leads, and (3) new-onset bundle branch block.
Genotype was categorized by the presence or absence of the PlA2allele as in previous studies. 1,3Troponin concentrations for individual time points were normally distributed, and PlA1homozygotes were accordingly compared with PlA2heterozygotes–homozygotes by two-way analysis of variance for troponin concentrations and time. Peak troponin concentrations did not meet the criteria for normality and were therefore analyzed using the nonparametric Mann–Whitney U test. The two genotypes were also compared for preoperative demographic characteristics 14and intraoperative events by nonparametric tests as follows: (1) for categorical variables (previous myocardial infarction, occurrence of perioperative electrocardiogram changes), the populations were compared using the Fisher exact test, and (2) for continuous variables (age, duration of bypass), the Mann–Whitney U test was used. Correlations between two continuous variables were made using the Spearman rank test. All statistical analysis was performed using Graphpad Prism® software (San Diego, CA).
Patients were largely white (93%), and the PlAgenotype distribution—46 patients (70%) PlA1,A1, 16 patients (24%) PlA1,A2, and 4 patients (6%) PlA2,A2—was similar to those in previously reported studies. 15Clinical characteristics relevant to risk of perioperative myocardial injury were comparable between PlA1homozygotes versus PlA1,A2and PlA2,A2patients (table 1). No patients received long-acting antiplatelet drugs, i.e. , GPIIb/IIIa antagonists, with the exception of aspirin within 1 week of elective surgery. All other platelet-active agents were stopped at least 24 h prior to surgery.
Troponin concentrations and PlA2Studies
Figure 1shows the cTpnI concentrations at each time point with patients stratified by PlAgenotype; patients carrying at least one PlA2allele had significantly greater cTpnI concentrations than PlA1homozygotes by analysis of variance (P = 0.0005). The four PLA2homozygotes were not outliers and were therefore included with PlA2heterozygotes for all statistical analysis, as has been done in previous studies. 1,3One PlA1,A2patient had postoperative cTpnI concentrations that met the criteria for outliers by the Grubb method. 16Assay verification and this patient's clinical course suggested that these values were real, albeit extreme. When this patient's values were modified (divided by SD of the population), 16PlA2heterozygotes–homozygotes were still significantly higher than PlA1homozygotes by analysis of variance (P = 0.006). Peak cTpnI concentrations were compared in patients with the PlA2polymorphism versus PlA1homozygotes using the Mann–Whitney test, and PlA2heterozygotes–homozygotes were significantly higher by this nonparametric test (P = 0.018). The peak cTpnIs for all patients are divided into four quartiles in figure 2, and each quartile is stratified by PlAgenotype. PlA2+patients represented only 20% and 12% of the first and second quartiles containing the lowest cTpnI concentrations, respectively; by contrast, PlA2+patients made up 39% and 50% of the third and fourth quartiles containing the highest cTpnI concentrations, respectively. A Fisher exact test was used to compare the proportion of each genotype in the 50% of patients with the lowest versus the highest troponins. The PlA2allele conferred a relative risk of 2.3 for having peak postoperative cTpnI concentrations in the highest 50th percentile (95% confidence interval, 1.06–5.2;P = 0.02). Evaluation of the individual time points (Mann–Whitney U test) revealed that the 6-h post-CPB time point differed significantly between the PlAgenotypes (P = 0.02), but the difference on postoperative day 1 did not reach the level of statistical significance (P = 0.11).
The percentage of circulating activated platelets in all patients increased over the course of CPB, as previously detailed. 11Peak platelet activation correlated significantly with peak cTpnI concentrations by the Spearman rank test (P = 0.024). However, the two PlAallelic groups did not differ significantly with respect to peak platelet activation (P = 0.37 by Mann–Whitney test).
With respect to postoperative electrocardiogram changes, one patient (PlA1,A1) had a baseline left bundle branch block that obviated postoperative electrocardiogram interpretation. Of the 65 patients with assessable electrocardiograms, 12 demonstrated electrocardiogram evidence of a new post-CPB cardiac event; one new Q-wave (PlA1,A1), two new rhythm disturbances (one PlA1,A1and one PlA2+, each with new bundle branch block), and the remaining 9 (6 PlA1,A1and 3 PlA2+) showed persistent ST segment changes. A total of 8 of the 45 PlA1homozygotes (15%) and 4 of the 20 PlA2+patients (20%) had electrocardiogram evidence of a cardiac event; this difference was not statistically significant by Fisher exact test (P = 0.99).
This pilot study has demonstrated in a modest number of patients undergoing CABG that the presence of the PlA2allele is associated with higher cTpnI concentrations following CPB as compared with PlA1homozygotes. We found that the peak increase in perioperative platelet activation correlated significantly with peak cTpnI concentrations. However, patients who were heterozygous–homozygous for the PlA2allele did not differ significantly from PlA1homozygotes with respect to this single marker of platelet activation (α-granule release-specific). Studies have suggested that the PlA2allele confers an increased risk for vascular occlusive events, but the mechanism responsible for this platelet hypercoagulability remains controversial. 1–4
Weiss et al. 1demonstrated a greater prevalence of the PlA2allele in coronary thrombosis patients, and other studies 3,17have supported this finding. Glycoprotein IIIa is the β3-subunit common to the platelet fibrinogen receptor glycoprotein IIb/IIIa (αIIbβ3integrin) and the platelet–endothelial vitronectin receptor (αvβ3integrin). The PlA2polymorphism results from a Leu 33→ Pro substitution in the glycoprotein IIIa amino terminus. 18The association of the PlA2allele with coronary artery disease has been controversial. Some studies 17have found a greater degree of coronary stenosis in PlA2+patients, but a number of other studies have not confirmed this polymorphism as a risk factor for atherosclerosis. 2,19,20Instead of predisposing to greater atherosclerotic disease, other investigations have found that the PlA2allele increases the risk of thrombosis in established atherosclerotic coronary arteries. 3,4,21Zotz et al. 5have demonstrated that the PlA2polymorphism increases the chance of cardiac events or death in the period 2–12 months after CABG surgery. We have now demonstrated that this platelet polymorphism increases the risk of myocardial injury that is directly and temporally associated with the operative procedure itself.
The precise mechanistic link between the PlA2polymorphism and increased risk of thrombosis is still unclear. Our study suggests that heightened platelet activation is associated with greater myocardial injury perioperatively in the CABG patient. However, although peak increases in platelet activation in all patients correlated with peak cTpnI concentrations, those patients carrying the PlA2allele did not demonstrate significantly greater platelet activation increases than did PlA1homozygotes. Analogous to our findings, Carter et al. 15found that while both the PlA2allele and platelet activation markers (in their case, platelet factor 4 and β-thromboglobulin) were independently correlated with the occurrence of stroke and poststroke mortality, PlA2+stroke patients did not demonstrate higher concentrations of platelet activation markers than PlA1homozygotes. Meiklejohn et al. 22similarly did not find a greater percent CD62P+ platelets in PlA2+patients with stroke, although stroke patients as a whole showed a greater circulating percent CD62P+ platelets than in healthy controls.
Some in vitro studies of PlA2+homozygous individuals have demonstrated increased platelet aggregability, 23,24increased fibrinogen binding, 25and enhanced platelet adhesion and clot retraction 26in PlA2+platelets. These alterations could account for the allele's arterial prothrombotic effects and, by creating a more robust platelet plug, would not necessarily increase the percentage of circulating CD62P+ platelets. The only in vitro study to show a difference in CD62P expression according to PlAgenotype was the study by Michelson et al. , 6which found a lower threshold for platelet activation and CD62P expression only in patients homozygous for the PlA2allele. In the current study, only four patients were PlA2homozygotes, too few to credibly examine the relation between PlA2homozygosity and the platelet CD62P response to CPB. It is possible that heterozygosity or homozygosity for the PlA2allele and CPB-induced increases in circulating activated platelets represent independent pathophysiologies enhancing platelet function by different pathways, each uniquely contributing to the myocardial injury associated with CPB. Alternatively, the relation between the PlA2allele and platelet activation may be too complex to demonstrate in a study of this size.
Assessing the extent of myocardial injury in the post-CPB patient is difficult. 27–29Perioperative myocardial infarction diagnosed by electrocardiogram criteria is clearly associated with a poor long-term outcome. 30However, new Q waves are not particularly sensitive markers of myocardial infarction 31; indeed, most myocardial infarctions associated with non-CPB surgery do not result in Q waves. 32Technetium-99 m pyrophosphate scans revealed a 21% incidence of acute myocardial infarction in a population of post-CPB patients with only a 3% incidence of new Q waves. 33Given these limitations, many investigators have turned to myocardial-specific enzyme leakage, and cTpnI in particular, as a more reliable marker of cardiac injury in the peri-CPB patient. 34One problem impeding application of cTpnI measurements is the difficulty defining the range of “normal” for the post-CPB patients. This task is hampered by the disparity between commercial cTpnI assays. The three most widely used commercial cTpnI assays report upper limits of normal (defined in non-CPB patients) that span a 20-fold range. A recent study of perioperative myocardial injury in CPB patients found that samples tested with two different assays differed by a factor of 20, 35underscoring the difficulties faced when attempting to establish widely applicable standards for post-CPB myocardial injury. We chose not to attempt to define a cTpnI threshold for myocardial infarction in our patients, but instead analyzed peak cTpnI concentrations as a continuous variable using nonparametric statistics. 14,36
The risks associated with elevated cTpnI concentrations after CPB have not been clearly established, but in the acute coronary care setting, even modest cTpnI elevations have predictive value for complications attributable to myocardial injury. 37Indeed, a recent study of post-CPB myocardial injury found that the degree of CK-MB elevation independently predicted adverse outcome. 38The current study's finding that cTpnI concentrations correlate with both a genetic variant predisposing to enhanced in vitro platelet function and the increase in circulating activated platelets suggests that heightened platelet function in the peri-CPB patient imposes myocardial risk.
One limitation of this study is the timing of samples for cTpnI determination, which were drawn in the early post-CPB period. Some investigators 13,39,40have demonstrated delayed cTpnI leakage after CPB, peaking as late as 48 h postoperatively, well after the last sample taken in the current study (18–24 h after CPB). By contrast, other investigators 14,35,36,41have demonstrated peak post-CPB cTpnI concentrations in the first postoperative day, which would have been captured in the current study. Thus, it is possible that some PlA1/A1patients might have developed late (48-h) cTpnI increases, but we know of no specific rationale for why that might occur more frequently than in PlA2+subjects.
In summary, this pilot study has demonstrated a significant link between the PlA2allele and the concentrations of cTpnI associated with CPB, suggesting that the PlA2polymorphism may be a risk factor for greater perioperative myocardial injury. In addition, the peak increase in platelet activation correlated with peak cTpnI concentrations, independent of PlAgenotype. Larger studies will be needed to confirm these findings and to further explore the role of platelet procoagulant function in the myocardial injury associated with CPB.