Associations of early creatine phosphokinase-MB (CK-MB) elevation and new Q waves and their association with cardiovascular death (CVD) after coronary artery bypass grafting (CABG) have been reported, but this association has not been studied in a large population of patients with diabetes mellitus. In this study, we examine the association of periprocedural CK-MB elevations and new Q waves with CVD in the Future Revascularization Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multivessel Disease trial. Cox proportional hazards regression was used to assess the relation of CK-MB elevations and new Q waves in the first 24 hours after procedure and their relation to CVD; logistic regression was used to assess odds ratios of these variables. Hazard ratios, 95% confidence intervals, and p values associated with Wald chi-square test are reported. CK-MB elevation in first 24 hours after procedure was independently associated with CVD. CVD hazard increased by 6% (p <0.001) with each multiple of CK-MB above the upper reference limit (URL); odds of new post-CABG Q waves increased by a factor of 1.08 (p <0.001); at 7× CK-MB URL, HR was >2. CK-MB URL multiples of 7, 12, and 15 were associated with new Q-wave odds ratios of 9, 16, and 27 times, respectively (p ≤0.001, C-statistic >0.70). New Q waves were independently associated with survival in the multivariate model only when CK-MB was excluded (p = 0.01). In conclusion, independent associations included (1) CVD and early post-CABG CK-MB elevation; (2) new Q waves with early post-CABG CK-MB elevation; (3) CVD with new Q waves only when CK-MB elevation is excluded from analysis.
The Future Revascularization Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multivessel Disease (FREEDOM) trial demonstrated that patients with diabetes and multivessel coronary artery disease (CAD) have better survival when coronary artery bypass grafting (CABG) rather than percutaneous intervention is used for revascularization.
Patients with diabetes and multivessel CAD have a poor prognosis because of their aggressive vascular disease. However, because the power to detect a survival difference calculated during trial design was <80%, the primary end point of FREEDOM was a composite outcome combining death and critical events known to be associated with reduced survival, specifically, nonfatal myocardial infarction (MI) and nonfatal stroke. A consideration in developing FREEDOM was how to best define MI early after CABG. The likely association of late cardiovascular death (CVD) with enzyme elevations and/or new Q waves in FREEDOM patients was a significant consideration in choosing study end points. Published data demonstrate that even small creatine phosphokinase-MB (CK-MB) elevations early after CABG are associated with worsened survival. With the conclusion of the FREEDOM trial, we can retrospectively examine the associations of CK-MB elevations and new Q waves with post-CABG survival in the FREEDOM patients.
Methods
This article is based on the data collected from FREEDOM randomized clinical trial conducted from 2005 to 2010 at 140 international centers (funded by the National Heart, Lung, and Blood Institute and others; FREEDOM ClinicalTrials.gov number, NCT00086450 ). The design and results of the FREEDOM trial have been reported in detail. The study enrolled 1,900 patients with diabetes and angiographically confirmed multivessel CAD with stenosis of >70% in ≥2 major epicardial vessels involving at least 2 separate coronary artery territories but without left main coronary stenosis.
Per protocol (which required a normal CK/CK-MB level before randomization), determinations of CK-MB level were performed after index procedure, within 6 to 8 hours and 18 to 24 hours or at hospital discharge, whichever occurred first. If CK was above the upper limit of normal, CK and CK-MB were evaluated every 8 hours until they were on a downward trend or until hospital discharge. As there was no core laboratory, these determinations were performed in local on-site laboratories and were used as an indicator of myocardial damage. Values were normalized by dividing the laboratory result by the local laboratory-specified upper reference limit (URL). The maximum enzyme value (out of multiple measurements within 24 hours after index procedure) of each of the normalized results was selected for each subject for analysis. A similar approach was used for troponin level analyses.
Per protocol, a postprocedural electrocardiogram was required within 48 hours of the procedure. All new Q waves (in at least ≥2 contiguous leads compared to baseline) meeting criteria for significance identified within 48 hours of the index procedure were included in analyses. An electrocardiography core laboratory was used to adjudicate new Q waves.
The actual index procedure was used in all analyses. Time to event was calculated from the time of the index procedure. Death is qualified as cardiovascular (CVD) based on the adjudication by an independent clinical events committee.
We used Cox proportional hazards regression to examine the relation between CVD mortality and postprocedural enzyme elevations and new Q waves. This was done with and without the baseline and perioperative covariates, which include key patient characteristics ( Table 1 ). We report hazard ratios (HRs) along with 95% confidence intervals and p values associated with the Wald chi-square test. Statistical significance was considered with a 2-sided critical value of 0.05. The proportional hazards assumption was evaluated using a supremum test for nonproportionality based on Martingale residuals. Stepwise regression (with p = 0.15 as entry significance level and p = 0.05 as stay significance level) was used to inform selection of final models. These models were fitted using SAS PROC PHREG. Variable selection for final models was also guided by bootstrap bagging (1,000 random sample data sets) to assess reliability (percentage of random sample data sets for which the variable was selected) for inclusion. Kaplan–Meier plots (with log-rank and Wilcoxon significance tests) were used to compare survival patterns between subjects with new Q waves or elevated enzyme level and the rest of the cohort. Logistic regression was used to assess odds ratios for post-CABG new Q waves and CK-MB elevations, with C-statistics used as a measure of goodness of fit. Imputation for missing data was not performed. All analyses were conducted using SAS software version 9.3 (SAS Institute) ( Figure 1 ).
