Acadesine (AICA riboside) (5-amino-1-[beta-D-ribofuranosyl]imidazole-4-carboxamide) is a purine nucleoside analog belonging to a new class of agents generally termed adenosine regulating agents (ARAs) that increase the availability of adenosine locally in ischemic tissues. The effects of acadesine on the incidence of fatal and nonfatal myocardial infarction (MI) an on the incidence of all adverse cardiovascular outcomes (cardiac death, MI, congestive heart failure, life-threatening dysrhythmia, or cerebrovascular accident) was investigated in patients undergoing coronary artery bypass graft (CABG) surgery.
In 20 medical centers in the United States participating in the Multicenter Study of Perioperative Ischemia (McSPI), 633 patients undergoing CABG surgery were randomized in a double-blind fashion to receive either placebo (n = 212), low-dose acadesine (0.05 mg.kg-1.min-1, n = 214), or high-dose acadesine (0.1 mg.kg-1.min-1, n = 207) by intravenous infusion starting 15 min before anesthetic induction and continuing for 7 h, as well as added to the cardioplegic solution (final concentration of 5 micrograms/ml for those patients receiving acadesine). Anesthesia was standardized, and perioperative hemodynamics were to be strictly controlled. Twelve-lead electrocardiograms (ECGs), CK-MB isoenzyme concentrations, and autopsy were used to assess the occurrence of MI.
There was a similar incidence of adverse events in the acadesine groups and the placebo group, with the exception that serum uric acid transiently increased in the high-dose acadesine group. The incidence of perioperative MI, using the prespecified MI criterion (EGF Q wave, CK-MB elevation, or autopsy evidence), was not different between groups (24% versus 26% versus 21% [P = 0.574]), nor was the incidence of all cardiovascular outcomes (30% versus 30% versus 22% [P = 0.151]). After completion of the study, a post hoc analysis also was performed using the more specific definition of MI (ECG Q wave and CK-MB elevation, or autopsy evidence), and the incidence of MI was lower (P = 0.018, alpha = 0.017, corrected for multiple comparisons), as were adverse cardiovascular outcomes (P = 0.002) and CVA (P = 0.02) for patients treated with 0.1 mg.kg-1.min-1 acadesine. In patients with Q-wave infarction, the high-dose acadesine group had a lower peak median CK-MB (P = 0.042) and area under the CK-MB curve (P = 0.021). No differences were found in the incidence or characteristics of MI (Holter or transesophageal echocardiography).
The results of this trial did not demonstrate a statistically significant difference between acadesine and placebo using the prespecified criterion for MI. Of interest are the results of the post hoc analysis, using the more specific criterion for MI, which indicate that acadesine may reduce the incidence of larger Q-wave infarctions after coronary artery bypass surgery. A second trial is underway to evaluate this contention.
Methods: In 20 medical centers in the United States participating in the Multicenter Study of Perioperative Ischemia (McSPI), 633 patients undergoing CABG surgery were randomized in a double-blind fashion to receive either placebo (n = 212), low-dose acadesine (0.05 mg *symbol* kg sup -1 *symbol* min sup -1, n = 214), or high-dose acadesine (0.1 mg *symbol* kg sup -1 *symbol* min sup -1, n = 207) by intravenous infusion starting 15 min before anesthetic induction and continuing for 7 h, as well as added to the cardioplegic solution (final concentration of 5 micro gram/ml for those patients receiving acadesine). Anesthesia was standardized, and perioperative hemodynamics were to be strictly controlled. Twelve-lead electrocardiograms (ECGs), CK-MB isoenzyme concentrations, and autopsy were used to assess the occurrence of MI.
Results: There was a similar incidence of adverse events in the acadesine groups and the placebo group, with the exception that serum uric acid transiently increased in the high-dose acadesine group. The incidence of perioperative MI, using the prespecified MI criterion (ECG Q wave, CK-MB elevation, or autopsy evidence), was not different between groups (24% versus 26% versus 21% [P = 0.574]), nor was the incidence of all cardiovascular outcomes (30% versus 30% versus 22% [P = 0.151]). After completion of the study, a post hoc analysis also was performed using the more specific definition of MI (ECG Q wave and CK-MB elevation, or autopsy evidence), and the incidence of MI was lower (P = 0.018, alpha = 0.017, corrected for multiple comparisons), as were adverse cardiovascular outcomes (P = 0.002) and CVA (P = 0.02) for patients treated with 0.1 mg *symbol* kg sup -1 *symbol* min sup -1 acadesine. In patients with Q-wave infarction, the high-dose acadesine group had a lower peak median CK-MB (P = 0.042) and area under the CK-MB curve (P = 0.021). No differences were found in the incidence or characteristics of MI (Holter or transesophageal echocardiography).
Conclusions: The results of this trial did not demonstrate a statistically significant difference between acadesine and placebo using the prespecified criterion for MI. Of interest are the results of the post hoc analysis, using the more specific criterion for MI, which indicate that acadesine may reduce the incidence of larger Q-wave infarctions after coronary artery bypass surgery. A second trial is underway to evaluate this contention.
Key words: Heart: acadesine; adenosine; adenosine-regulating agent; myocardial infarction; myocardial injury; myocardial ischemia; perioperative cardiac morbidity; ventricular function. Surgery: coronary artery bypass graft.
MORE than 600,000 patients undergo coronary artery bypass graft (CABG) surgery annually in the United States, Canada, Western Europe, and Japan. [1-3]Over the past 5 yr, morbidity and mortality after CABG have increased, primarily because of changes in patient demographics; patients are older, sicker, or have had previous CABG surgery or acutely failed angioplasty. As a result, mortality now ranges between 0.5 and 11%, and morbidity between 1 and 28%, problems that are likely to get worse with the increasing use of nonsurgical approaches and the continued aging of the population. [1-9]
Initial studies evaluating outcomes after CABG surgery, such as the Coronary Artery Surgery Study (CASS) [10,11]and the Veterans Administration and European cooperative studies, identified mostly irreversible predictors of adverse outcomes, such as age, multivessel disease, and poor ventricular function. More recently, approaches have been directed at identifying potentially reversible perioperative predictors of outcome, such as myocardial ischemia and ventricular dysfunction after cardiopulmonary bypass, that can be evaluated using electrocardiographic and echocardiographic measures. [14-19]Few large-scale therapeutic trials, however, have been directed at reducing ischemia and ventricular dysfunction.
Recently, the potential of adenosine-regulating agents (ARAs) in mitigating ischemia and ventricular dysfunction has been investigated. One such agent, acadesine, a purine nucleoside analog, raises adenosine levels in tissues under ischemic conditions, but not in nonischemic myocardium. [20,21]Ischemia and other forms of stress to a tissue (defined by net ATP catabolism) stimulate the production of adenosine. Acadesine selectively augments these extracellular adenosine levels, but does not increase basal adenosine levels or those in nonischemic or nonstressed tissues. The mechanism by which acadesine regulates adenosine levels during ATP catabolism has not been established fully. [21,22]Studies of experimentally induced myocardial ischemia in dogs, rabbits, and rats [20,23-28]demonstrate that acadesine: 1) increases adenosine concentrations in blood draining an ischemic area [20,29]; 2) improves left ventricular wall motion after intermittent ischemia [28,30]; 3) enhances collateral endocardial blood flow to the ischemic myocardium without producing coronary steal [20,28]; 4) attenuates the frequency of ventricular dysrhythmias [20,31]; 5) reduces granulocyte accumulation [20,32]; 6) attenuates myocardial stunning ; 7) preserves myocardial function after cardiac arrest and cold cardioplegia [33-36]; and 8) has antithrombotic effects. [23,37]The cardioprotective properties of acadesine are prevented if adenosine receptors are blocked or the formation of adenosine is suppressed, indicating that the ability of acadesine to enhance ischemia-induced adenosine levels is the basis of its beneficial effects.
Because of its potential usefulness in patients undergoing CABG surgery, our group conducted a preliminary study in 116 patients undergoing CABG surgery, randomizing patients to receive either placebo or a low (0.05 mg *symbol* kg sup -1 *symbol* min sup -1) or high (0.10 mg *symbol* kg sup -1 *symbol* min sup -1) dose of acadesine in a double-blind fashion. The results of that trial demonstrated that acadesine could be safely administered intravenously and in cardioplegia over a 7-h perioperative period, and that acadesine may limit the incidence of myocardial infarction (MI), the extent of myocardial injury (as measured by CK-MB release), and the severity of perioperative ischemia. That preliminary study, however, was not powered to examine the effects of acadesine on the incidence of perioperative MI.
This trial was designed to investigate these findings in a larger cohort of patients. The specific aims of the trial were to assess the effects of acadesine on the incidence of perioperative MI and other adverse cardiovascular outcomes (cardiac death, congestive heart failure, life-threatening dysrhythmia, and cerebrovascular accident) and to assess the safety of acadesine.
Materials and Methods
General Study Design
This study was conducted in 20 Multicenter Study of Perioperative Ischemia Research Group medical centers in the United States. The study was randomized, placebo controlled, and double blinded. After institutional approval and informed consent was obtained, 677 patients were screened for enrollment in this study between June, 1991, and April, 1992. Patients scheduled for CABG surgery were randomly allocated within each center to treatment with placebo (normal saline) or one of two doses of acadesine administered intravenously (0.05 or 0.1 mg *symbol* kg sup -1 *symbol* min sup -1 for 7 h) and in the cardioplegia solution (at a final concentration of 5 micro gram/ml), according to a previously prepared randomization code. We stratified patients after enrollment into one of two groups: 1) high-risk and 2) all other patients. High-risk patients had one or more of the following characteristics: age of > 70 years, previous CABG surgery, acutely failed PTCA (without an evolving MI), unstable angina, or left ventricular dysfunction (ejection fraction < 30%). Unstable angina was defined as severe precordial chest pain that was nonsurgical, lasting 30 min or longer, unresponsive to standard therapeutic maneuvers (nitroglycerin and rest), and associated with transient changes in the ST segment or T wave without the development of Q waves or diagnostic enzyme abnormalities.
The method of randomization used was block randomization within each center for each risk group. That is, within each center, the patients were stratified into risk group (high risk or other) and then randomly allocated to one of three treatment assignments (placebo, low dose, or high dose). For each stratum, a blocking factor of 6 was used. SAS (Cary, NC) software's uniform density function random number generator was used to construct the randomization table.
To be included in the study, patients were required to have stenoses of at least two major coronary arteries (greater or equal to 70% of luminal cross section) or a left main coronary artery stenosis (greater or equal to 50%). Patients were excluded from the study if they had any of the following conditions: concurrent valve replacement, significant valvular disease, suspected or definite MI within 7 days before surgery, cardiogenic shock, uric acid nephropathy, renal insufficiency, hepatic dysfunction, or esophageal disease precluding positioning of the transesophageal echocardiography (TEE) probe.
An independent Safety and Data Monitoring Panel was responsible for decisions about safe conduct and continuation of the trial.
Before Surgery. The investigator ascertained a history of prior MI, angioplasty, hypertension, diabetes mellitus, unstable angina, and preoperative cardiac medications. The number of significant coronary artery stenoses were recorded. Dipyridamole was discontinued at least 48 h before surgery and for at least 24 h after surgery. Nicotinic acid and theophylline were discontinued 24 h before surgery and for 48 h after surgery. Adenosine and pentoxifylline were discontinued 12 h before surgery and for 48 h after surgery. Adenosine was not allowed during the period indicated, because its presence would complicate the analysis for any effect of acadesine; the other medications were excluded because they could modulate the effects of adenosine. All other cardiovascular medications, including nitrates, beta blockers, and calcium-channel blockers, were continued until the time of operation. Patients received lorazepam 1-4 mg PO before surgery.
Prebypass. On entry into the operating room, patients were monitored with the usual cardiovascular and pulmonary monitors. Study drug, acadesine or placebo, was administered intravenously starting approximately 15 min before induction of anesthesia and continued for a total of 7 h. The duration of 7 h for drug administration included the intraoperative (prebypass and postbypass) and the immediate postoperative (into the intensive care unit) periods for most patients. The cardioplegia solution used for myocardial protection during cardiopulmonary bypass contained acadesine (at a fixed concentration of 5 micro gram/ml) or placebo (Sterile Water for Injection, USP, Bethesda, MD), as appropriate to match the randomization to acadesine or placebo for intravenous administration. Anesthesia was induced using fentanyl (up to 50 micro gram/kg), midazolam (up to 0.2 mg/kg), and sodium thiopental (up to 7 mg/kg). Maintenance of anesthesia was accomplished using a continuous intravenous infusion of fentanyl (up to 0.25 micro gram *symbol* kg sup -1 *symbol* min sup -1) and midazolam (up to 0.5 micro gram *symbol* kg sup -1 *symbol* min sup -1). Vecuronium was used for muscle relaxation. No other anesthetic agents were used. Throughout the prebypass period, systolic blood pressure and heart rate were maintained within 20% of the preoperative baseline values, using prescribed anesthetic changes and administration of prescribed cardiovascular agents. Prophylactic use of cardiovascular agents with potential antiischemic properties (nitrates and calcium-channel blockers) were specifically excluded because they might confound the interpretation of the results.
Bypass. The conduct of cardiopulmonary bypass was not controlled. Typically, bypass was conducted using a membrane oxygenator and an arterial filter, with hemodilution and moderate systemic hypothermia. However, neither the type nor the method of administration of cardioplegia, nor the surgical procedure, were controlled. Usually, distal anastomoses were performed first during continuous cross clamping of the aorta, followed by proximal vein grafting during partial aortic occlusion. Eighty-five percent (541 of 633) of patients received internal mammary artery (IMA) grafts to either the left anterior descending or the first diagonal coronary artery, in addition to saphenous vein grafts. Twelve percent (74 of 633) of patients received vein grafts only and 3% (18/633) received only IMA grafts. During bypass, blood pressure was maintained between 40 and 80 mmHg, using prescribed anesthetic and cardiovascular medications.
Postbypass. Anesthesia was continued after bypass. Systolic blood pressure was maintained between 90 and 130 mmHg and heart rate at < 110 beats/min, with prescribed cardiovascular medications. The use of inotropes and vasodilators was not controlled (except for antiischemic medications). Treatment of clinically detected ischemia was not controlled, but the medications used were recorded.
Postoperative. Morphine sulfate was used for pain and midazolam for anxiety.
General Clinical Care. Routine clinical monitors usually included a seven-lead electrocardiogram (ECG) and radial-artery and pulmonary-artery pressure measurements. Arterial blood oxygen saturation was to be continuously monitored and all patients were to receive 100% inspired oxygen to maintain PaO2greater than 70 mmHg. Ventilation was to be controlled to maintain PaCO2between 35 and 45 mmHg.
All outcomes were ascertained and validated by investigators at the coordinating center who were blinded to patient identity and group assignment.
Myocardial Infarction. Twelve-lead ECGs were obtained at the time of screening; on arrival in the intensive care unit (ICU); on postoperative days 1, 2, 3, and 5; and at hospital discharge. Serum for CK-MB was obtained at 1, 4, 8, 12, 16, 20, 24, 28, 32, 36, 42, 48, and 60 h after aortic cross-clamp removal. The period of MI risk was defined as from the time of anesthetic induction until hospital discharge. In the protocol, MI was diagnosed using the following criteria: 1) new Q waves on postoperative 12-lead ECG, using the Minnesota Code criteria ; or 2) CK-MB elevation (elevation of CK-MB concentration to greater or equal to 100 ng/ml at any time after surgery, greater or equal to 70 ng/ml at any time 12 h after surgery, or greater or equal to 12 ng/ml more than 24 h after surgery); or 3) acute MI diagnosed at autopsy. Twelve-lead ECGs were coded centrally by investigators of the McSPI analysis group using the Minnesota Code criteria, as modified by Chaitman, which included five categories, A through E, with A representing the strongest criteria for a Q wave MI and E the weakest. The diagnosis of a Q-wave MI by Minnesota Coding (1-1 to 1-3) was defined as a new, persistent postoperative Q wave with code category A, B, or C in any of three locations (anterior, anterolateral, or posterior). However, Minnesota Code 1-2-8 was not included in the definition of a new Q wave. Two separate cardiologists (MH and KZ) independently evaluated the set of ECGs to validate the coding and to make a final diagnosis as to whether a new Q wave was present. If there were disagreements with regard to coding or diagnosis between the two cardiologists, a third cardiologist (MG), also blinded, reviewed the set of ECGs and coding, with the majority prevailing.
Determinations of CK-MB concentrations were performed centrally by SmithKlineBeecham Clinical Laboratories (Van Nuys, CA) using an immunoenzymetric assay (Hybritech Tandem E CK-MB II). Autopsy diagnosis of MI was made by the pathologist at the respective institution. This diagnosis was confirmed by two blinded investigators (DM and UJ), and if disagreement occurred, by a third investigator (MH), with the majority prevailing.
To be included in analysis for myocardial injury (maximum CK-MB and area under the CK-MB curve; AUC), patients were required to have no more than two nonconsecutive missing CK-MB values (at least one of which was after 24 h for the time period of 1-42 h after cross-clamp removal) and no consecutive missing values.
Adverse Cardiac Outcomes. An adverse cardiovascular outcome was defined as the occurrence of cardiac death, MI, congestive heart failure, life-threatening dysrhythmia, or cerebrovascular accident. As per protocol, the definition of MI was made on the basis of the or criteria (Q wave or CK-MB, or autopsy). Cardiac death was defined as death from a primary cardiac cause. Other deaths were analyzed for safety purposes.
Congestive heart failure was defined as: 1) severe worsening of left ventricular function requiring an intraaortic balloon pump or left ventricular assist device for a cardiac index of less than 1.5 l *symbol* min sup -1 *symbol* M sup -2; or 2) cardiogenic shock with a cardiac index of less than 1.5 l *symbol* min sup -1 *symbol* M sup -2 and pulmonary capillary wedge pressure of more than 20 mmHg for greater than 1 h.
Life-threatening dysrhythmia was defined as: 1) ventricular dysrhythmia requiring cardioversion or 2) dysrhythmia requiring insertion of a pacemaker that was present at hospital discharge.
Cerebrovascular accident (CVA) was diagnosed if signs or symptoms of a significant focal neurologic defect persisted for greater than 24 h. The diagnosis was made by the neurologist at the respective institution. If nonfocal deficits existed on neurologic evaluation, results from a CT or MRI scan that were consistent with a new cerebral infarct or hemorrhage were required to make the diagnosis of CVA.
Validation of cardiac death, congestive heart failure, life-threatening dysrhythmias, and CVA was performed by two independent, blinded investigators (DM and UJ). If conflicts occurred, they were resolved by a third investigator (MH for cardiac death, heart failure, and dysrhythmias; SG for CVA), with the majority prevailing.
Electrocardiography. A total of 617 patients were monitored with continuous three-channel Holter electrocardiography (series 8500; Marquette Electronics) placed for a minimum of 8 h before surgery (baseline) and from the time of anesthetic induction, continuously throughout the intraoperative period, until 48 h after anesthetic induction. A total of 417 patients (placebo: 135; low dose: 149; and high dose: 143) were evaluable for the incidence of ischemia in each of the five monitoring periods. The frequency response met the American Heart Association specification for ST changes, the cutoff limit being 0.05 Hz for low frequency and 80 Hz for high frequency. Three bipolar leads, CC5, modified CM5, and ML, were used. Each ECG recording was scanned by readers using an ECG analysis system (series 8000; Marquette Electronics). All abnormal QRS complexes (e.g., ventricular ectopic beats and conduction abnormalities) were excluded. A continuous three-lead ST-segment trend was generated. Possible episodes of ischemia were reviewed and verified by two investigators (UJ and KZ) who were blinded to patient identity, treatment group, and outcome. The baseline ST segment level was defined as the average ST segment during a stable period (usually 1 h) preceding each episode. An ECG ischemic episode was defined as reversible ST depression of 0.1 mV or greater from baseline at J + 60 ms, or greater than 0.2 mV ST elevation at the J point lasting for at least 1 min. [42,43]The characteristics of each ECG ischemic episode were the magnitude, duration, and severity (AUC) of ST segment change, as well as average ischemic time (minutes of ischemia per hour monitored).
Transesophageal Echocardiography. Echocardiographic measurements were performed at 11 centers, totaling 394 patients (placebo: 128; low dose: 135; and high dose: 131). Immediately after tracheal intubation, a gastroscope tipped with either a 3.5-MHz or a 5-MHz phased-array transducer was introduced into the esophagus. The transducer was positioned and maintained at the level of the midpapillary muscles to obtain a short-axis view of the left ventricle. Echocardiographic data were recorded continuously onto videotapes during the prebypass period, from completion of tracheal intubation to the onset of cardiopulmonary bypass, and during the postbypass period from aortic side-clamp removal until skin closure.
The real-time video tape was edited to obtain samples for analysis. Echocardiographic samples of 60-s duration were obtained every 15 min throughout the prebypass period. Additionally, samples were obtained at other prespecified times to detect whether anesthetic or surgically imposed stresses had any immediate effect on regional wall motion. During the prebypass period, samples were obtained immediately after tracheal intubation and at 4 min before and 1 and 6 min after each of the following surgical events: skin incision, sternotomy, pericardiotomy, and aortic and right atrial cannulation.
The short-axis, cross-sectional image was divided into four segments using the papillary muscles as guides. This floating-reference system compensated for rotational movements of the heart. [17,44]A segment was considered suitable for wall motion analysis if 70% of its entire endocardial outline was visible continuously throughout systole and diastole. All samples were visually analyzed by an investigator (JL) blinded to patient identity, treatment group, and outcome. The wall motion of each of the four segments was graded as follows: 0 = normal, 1 = mild hypokinesis, 2 = severe hypokinesis with myocardial thickening, 3 = akinesis, and 4 = dyskinesis. Myocardial thickening was estimated by visual inspection in real time and slow motion. Each patient's best prebypass wall motion score for each segment was used as his or her baseline score. A "TEE episode" indicative of ischemia was defined by regional wall motion worsening 2 grades or more and lasting 1 min or greater.
Concordance analyses had been performed in our laboratory to determine interexamination variability. Using a random selection and blinded reanalysis at approximately 6 months from the first reading, the interexamination concordance was 90% for ECG and 98% for TEE.
Hemodynamics. Baseline values for blood pressure and heart rate were determined on the day before surgery by averaging three readings obtained at least 5 min apart. For research purposes, systolic, diastolic, mean arterial, and pulmonary artery pressures, as well as heart rate, were recorded throughout the prebypass, bypass, and postbypass periods and stored in an automated anesthesia recording system (ARKIVE, Diatek, San Diego, CA). Cardiac output and pulmonary capillary wedge pressure were obtained at prescribed times: before study drug infusion, after intubation, before sternotomy, at 15 and 30 min after cross-clamp removal postbypass, on chest closure, at ICU admission, and at 1, 2, 4, 8, 12, 24, and 48 h in the ICU. Hemodynamic readings were derived from the ARKIVE raw data set every 60 s throughout the intraoperative period for each patient. The data were reviewed by independent investigators to eliminate artifacts. Distributions for heart rate and blood pressure and characteristics of hemodynamic episodes were determined for the prebypass and postbypass periods. Hemodynamic episodes were defined as those occurring for more than 5 min and exceeding the plus/minus 20% prebypass limits, or the 110 beats/min or 90-130 mmHg postbypass limits. The onset time, offset time, and maximum value for each episode were recorded.
Left Ventricular Function. Postoperative left ventricular function, as measured by ejection fraction, was compared with the preoperative baseline ejection fraction. The preoperative ejection fraction was determined by ventriculography performed within 6 months of CABG surgery. Postoperative ejection fraction was determined by radionuclear ventriculography, performed within 14 days of surgery. To be evaluable, a patient had to have ejection fraction values for both the preoperative and postoperative measurements.
Data and Statistical Analyses
All research data were transmitted to the McSPI analysis group in San Francisco, CA, and analyses were performed by investigators blinded to patient identification, risk group, study drug, and clinical care. We estimated that a sample size of 200 patients per group was required to give a 90% chance of detecting a 50% reduction in the incidence of MI with a level of significance at 5% based on a placebo MI rate of 22%. [38,39,45]There were a total of 40 strata defined by combinations of 20 centers and two risk groups per center.
The Cochran-Mantel-Haenszel (CMH) test was used to compare primary efficacy variable response across the treatment groups. This analysis was stratified on center and risk groups. When the data were too sparse to permit valid use of the CMH test, the strata were combined, and the incidences were compared using either the chi-square analysis or the Fisher's exact test. For two group comparisons (e.g., high dose vs. placebo), Bonferroni alpha adjustments for multiple comparisons would be alpha = 0.0167 (0.05 of 3) for three possible comparisons (H vs. P, L vs. P, and H vs. L). Logistic regression was used to test for a treatment by stratum interaction. When variables were continuous, rank transform ANOVA methods were used. The ANOVA model contained effects for treatment and stratum, and a treatment by stratum interaction. The test for a treatment by stratum interaction was not significant for any of the efficacy variables. All analyses were performed using SAS software.
For the analysis of the AUC and maximum CK-MB data, the data window was 1-42 h postbypass, which encompassed 13 measurements of CK-MB levels. A trapezoidal rule was used to calculate the AUCs. Areas under the CK-MB curves and maximum CK-MB values were compared across the treatment groups using ANOVA.
After completion of the study, it became clear from analysis of the comprehensive CK-MB data collected in this study that sizable increases in CK-MB concentration occurred after myocardial revascularization in all study groups. The criterion for MI (Q wave or CK-MB, or autopsy evidence) defined in the protocol was sensitive but may have reflected the sizable increases in CK-MB concentration, and may not be specific. Therefore, we elected to also perform a post hoc analysis using a presumably more specific and less sensitive criterion for MI (Q wave and CK-MB, or autopsy evidence), perhaps mitigating CK-MB elevations not associated with ischemic injury. In addition, an analysis for the incidence of adverse cardiovascular outcome also was performed using this criterion.
A total of 677 patients were assessed for possible enrollment in this study between June, 1991, and April, 1992. Of these, 44 patients failed to complete screening procedures and did not enter the study. The remaining 633 patients were randomized and received a study drug. Eighteen patients who were assigned at the study site into the incorrect risk group were placed in the correct risk group before data analysis by an individual blinded to study group and patient outcomes. These reassignments did not affect any of the conclusions. The distributions of patients and mean durations of infusion were similar in the three study groups (Table 1). Study group patients also had a similar cardiac medical history, with a considerable number having unstable angina, previous MI, and hypertension (Table 2), and with more than one-half stratified to high-risk treatment (Table 3). Cardiac catheterization findings (Table 4) and cardioplegia type (Table 5) were also similar.
Overall, 97% of patients received cardioplegia containing crystalloid, 70% containing blood, 4% containing colloid, and 3% containing blood alone. The prebypass, bypass, aortic cross-clamp, and postbypass surgical time intervals were not different between groups (prebypass: 129 plus/minus 44, 128 plus/minus 44, and 123 plus/minus 43 min; bypass: 108 plus/minus 44, 107 plus/minus 44, and 102 plus/minus 43 min; aortic cross-clamp: 59 plus/minus 22, 57 plus/minus 22, and 57 plus/minus 23 min; and postbypass: 43 plus/minus 15, 43 plus/minus 15, and 40 plus/minus 14 min [mean plus/minus SD]; for placebo, low-dose, and high-dose acadesine, respectively).
Myocardial Infarction. Using the prespecified criterion for MI (Q wave or CK-MB, or autopsy), the incidence of MI was not different across groups (Table 6). Similarly, for high-risk patients, the incidence of MI was not different (29.5, 25, and 21.6% [P = 0.426] for placebo, low-dose, and high-dose acadesine, respectively). There did not appear to be a relationship between the incidence of MI and the number of vessels grafted, or the type of cardioplegia used during bypass. There was no effect of study site on the results. There were six patients with MI who died, and these were included as cardiac deaths (see results below).
Post hoc analysis, using the more specific criterion for MI (Q wave and CK-MB, or autopsy), demonstrated that the high-dose acadesine group had the lowest incidence of MI, with the differences being nearly significant (P = 0.018 using Bonferroni three-comparison correction [alpha = 0.0167; Table 6]). For high-risk patients, the incidence of MI was 5.4, 2.8, and 2.8% (P = 0.61) for the placebo, low-dose, and high-dose acadesine groups, respectively.
Adverse Cardiovascular Outcomes. The incidence of all adverse outcomes combined were not statistically different between the three study groups (Table 7). Post hoc analysis, using the more specific MI criterion, demonstrated a lower incidence of all adverse outcomes combined in the high-dose acadesine group, and the differences were statistically significant (Table 7). The incidence of CVA was significantly reduced with the high-dose acadesine treatment. The incidence of the other individual adverse cardiovascular outcomes (cardiac death, failure, and dysrhythmia) was not significantly different across groups.
Myocardial Injury. The numbers of patients analyzed for the CK-MB AUC data were 203 of 207, 208 of 214, and 211 of 212 for the high-dose, low-dose, and placebo groups, respectively. Figure 1shows the mean plus/minus SD serum concentration of CK-MB at each sampling time period for all patients for each treatment group. The concentrations have a similar pattern for the three groups and peak at approximately 50 ng/ml by 8 h after removal of the cross clamp. For patients diagnosed with a Q wave MI, the mean plus/minus SD CK-MB level is shown in Figure 2. Both maximum CK-MB and AUC were significantly lower in patients with Q-wave MI in the high-dose acadesine versus placebo group (Table 8).
Myocardial Ischemia. A total of 617 patients (placebo, 204; low dose, 209; and high dose, 204) were assessed for the presence of ischemia by Holter monitoring. A total of 427 (placebo, 135; low dose, 149; and high dose, 143) patients were evaluable for the incidence of ischemia in each of the 5 monitoring intervals. There were no differences in the incidence of ischemic episodes, as measured by ECG ST segment changes, during any perioperative period between the placebo, low-dose, and high-dose acadesine groups. Similarly, no differences were found in the quantitative measures of ischemia, including maximum ST change, duration, area under the ST curve, and ischemic minutes per hour of quality monitoring time. There were no differences in the incidence of ischemic episodes, as measured by regional wall motion abnormality (RWMA), or maximum change in wall motion score, mean duration of ischemia, and mean ischemic minutes per hour monitoring.
Left Ventricular Function.
Measurements of preoperative and postoperative left ventricular function were available on 368 patients (placebo, 113; low dose, 133; and high dose, 122). Comparison of the preoperative ejection fractions across patient groups revealed no differences: placebo, 56.4 plus/minus 14.9%; low dose, 53.9 plus/minus 13.8%; and high dose, 55.8 plus/minus 14.4% (mean plus/minus SD). In addition, the postoperative ejection fractions were not different between groups: placebo, 54.7 plus/minus 14.9%; low dose, 53.9 plus/minus 13.8%; and high dose, 56.7 plus/minus 13.3%. In patients with decreased preoperative ejection fractions (< 45%), postoperative ejection fraction increased in all groups with the mean numerical increase in ejection fraction as follows: placebo, 4.3 plus/minus 13.5%; low dose, 5.5 plus/minus 13.3%; and high dose, 8.9 plus/minus 8.3%, with none of the differences being significant.
Other Measures. There were no statistically significant differences among any parameter assessing difficulty weaning from bypass, including the overall difficulty coming off bypass (as assessed by clinician), the use of vasopressors, the return to bypass, the use of intraaortic balloon pump, or the use of temporary pacemakers. There were no differences in the use of concomitant medications during surgery and throughout the postoperative recovery period.
Adverse Experiences. A total of 13 deaths (placebo, 7; low dose, 5; and high dose, 1) occurred in the study. There was a trend toward reduction in the incidence of patient deaths when the placebo and high-dose groups were compared (P = 0.068, Fisher's exact test). The causes of the 13 deaths were primary cardiac etiology (n = 7), respiratory failure (n = 2), vascular catastrophe (n = 2), stroke (n = 1), and hepatorenal failure (n = 1). Adverse events occurred in 96% of patients in the placebo group, 92% of patients in the low-dose group, and 91% of patients in the high-dose group (P = 0.073). There were no significant differences between groups for heart-rate or rhythm disorders, with the most frequently encountered rhythm disturbance being atrial fibrillation (placebo, 32%; low dose, 30%; and high dose, 26%). Similarly, bradycardia occurred equally (placebo, 7%; low dose, 3%; and high dose, 5%), as did first- or second-degree AV block (placebo, 1%; low dose, 1%; and high dose, 1%) or third-degree AV block (placebo, 5%; low dose, 3%; and high dose, 1%). The incidence of serious adverse events was similar across the three treatment groups (placebo, 13%; low dose, 12%; and high dose, 13%).
Hemodynamics. The proportions of time that hemodynamic measurements (blood pressure and heart rate) deviated in excess of 20% from protocol-defined control values were similar between the three treatment groups (placebo, low dose, and high dose, respectively): 1) before bypass: systolic hypertension (8, 6, and 8%; P = 0.30); systolic hypotension (12, 14, and 15%; P = 0.37); tachycardia (2, 2, and 3%; P = 0.48); and bradycardia (48, 51, and 39%; P = 0.03, ANOVA); and 2) after bypass: systolic hypertension (12, 9, and 8%; P = 0.07); systolic hypotension (14, 17, and 19%; P = 0.07); tachycardia (4, 3, and 2%; P = 0.68); and bradycardia (96, 97, and 98%; P = 0.68).
Laboratory Evaluations. With the exception of increased serum uric acid concentrations after the end of infusion, which resolved by the fifth postoperative day, there were no differences between groups in abnormalities of clinical laboratory parameters. Figure 3shows serum uric acid results over time, with the normal range limit shown, and demonstrates a dose-dependent increase in uric acid concentration that peaks at the end of infusion. At discharge, the incidence of patients with uric acid concentrations greater than the upper limit of laboratory reference range was similar for the placebo and high-dose groups: placebo, 8% (18 of 212); high dose, 9% (18 of 207); and low dose, 15% (33 of 214). Absolute glucose concentrations were similar, as were the numbers of patients with glucose concentrations outside of the laboratory's normal range and considered clinically significant by the individual investigator (placebo, 26; low dose, 21; and high dose, 25).
Acadesine Plasma Concentrations.
The mean plus/minus SD plasma concentration of acadesine in the sample obtained at the end of bypass was 2.34 plus/minus 1.23 and 7.24 plus/minus 17.70 micro gram/ml in the low- and high-dose groups, respectively, with the difference being statistically significant (P < 0.001). The mean plus/minus SD concentration of acadesine in the first liter of cardioplegia solution was approximately 6 micro gram/ml (low dose, 6.16 plus/minus 5.48 and high dose, 6.13 plus/minus 5.32 micro gram/ml), close to the 5-micro gram/ml target concentration.
The results of this study indicate that administration of acadesine to patients undergoing CABG surgery is safe. Hyperuricemia did occur, but resolved within 5 days with no clinically significant sequelae. Regarding efficacy, using the prespecified criterion for MI (ECG Q-wave or CK-MB elevation, or autopsy evidence), acadesine had no effect on the incidence of MI (24 vs. 26 vs. 21%, P = 0.574 for the placebo, low-dose, and high-dose acadesine groups, respectively), or on adverse cardiovascular outcomes (30 vs. 30 vs. 22%). No differences were found in the incidence or characteristics of myocardial ischemia (Holter or TEE). In contrast, the post hoc analysis results, using the more specific criterion for MI (ECG Q-wave and CK-MB elevation, or autopsy evidence), demonstrated that high-dose acadesine may reduce the incidence of MI (5.2 to 1.4%, P = 0.018, alpha = 0.017, corrected for multiple comparisons) and did reduce all cardiovascular outcomes (14.2 to 5.3%, P = 0.002) and CVA (4.2 to 0.5%, P = 0.02). In patients with Q-wave infarction, the high-dose acadesine group had a lower peak median CK-MB (P = 0.042) and AUC (P = 0.021).
Despite advances in surgical and anesthetic care, perioperative morbidity and mortality associated with CABG surgery have been increasing, primarily because of changes in patient demographics, with patients now being older and sicker, or having had previous CABG surgery or acutely failed angioplasty. Perioperative infarction rates range from 1 to 28% and are associated with increased long-term morbidity and mortality. [1,47]
Therapies directed at reducing perioperative myocardial ischemia have been advocated for both prophylaxis and acute intervention, [48-55]with several smaller trials indicating that nitrates [48-50]and beta blockers [51,52]may have beneficial effects. However, prophylactic use of such therapies to treat ischemia has not gained widespread acceptance because of the lack of confirmation of these preliminary results. Regarding myocardial protection, a number of approaches have been investigated, including administration of cardioplegia at differing temperatures, in both retrograde and antegrade fashion, and with and without the use of blood. [56-59]However, there has been no clear-cut consensus regarding the superiority of individual techniques in large-scale multicenter outcome trials. Before this study, there had been no large-scale outcome trials specifically aimed at preventing MI in patients undergoing CABG surgery. 
Recently, there has been substantial interest in the antiischemic properties of adenosine-regulating agents in ischemic models. [20,23-28,60]One such agent, acadesine, has demonstrated both event specificity and site specificity, whereby local extracellular adenosine levels are increased only in areas of ischemia where adenosine triphosphate (precursor of adenosine) is depleted. In normally metabolizing tissues, adenosine levels are not affected. Therefore, it is likely that acadesine would be of most clinical benefit in cardiovascular conditions in which it could prevent the progression of ischemia to cardiac events associated with long-term morbidity and mortality, in which ischemia is common, such as in high-risk patients undergoing CABG surgery. Studies conducted in animals, using cardiac bypass models, have demonstrated improvement in ventricular function and reduction of ischemia without untoward peripheral hemodynamic or cardiac conduction effects. [24-27]These results were confirmed in our preliminary study in 116 patients undergoing CABG surgery. 
Myocardial Infarction and Injury
The diagnosis of MI in the perioperative period is more complicated than in the usual clinical setting with the ambulatory patient. At present, there is no consensus on perioperative MI criteria, especially for CABG surgery. Q waves are relatively accurate markers of MI, but after CABG surgery the presence of Q waves may not always be associated with MI; similarly, the absence of Q waves may be associated with MI. [61,62]For example, specificity is limited by the presence of inferior wall Q waves, which may not be associated with necrotic tissue, but may be associated with other changes, such as pericarditis or mechanical manipulation of the heart during surgery. Sensitivity is limited by the occurrence of nontransmural MI. Similarly, elevation of CK-MB is also nonspecific in the setting of CABG surgery, at times reflecting ischemia, injury, or mechanical manipulation after CABG surgery. [63-66]In contrast, CK-MB diagnosis criteria can be set to be highly sensitive, with a nearly zero false-negative rate. However, no clear consensus exists regarding the criteria that should be applied to CK-MB for the diagnosis of MI after CABG surgery. Therefore, the sensitivity and specificity of criteria for MI after CABG surgery depend on the criteria used (i.e., the specific combination chosen, including ECG, CK-MB, autopsy, or other diagnostic tests).
As our results demonstrate, the efficacy of acadesine in reducing MI may depend on the criteria that are chosen. In our study, statistical significance was not achieved using the prespecified criterion for MI (Q wave or CK, or autopsy; P = 0.574). However, recognizing the inherent difficulties in defining MI in patients undergoing CABG surgery, and finding (in this study) that a substantial number of patients (in all study groups) had sizable CK-MB release, a post hoc analysis was performed using the more specific criterion for MI (Q wave and CK, or autopsy). This post hoc analysis demonstrated that high-dose acadesine may be associated with a lower incidence of MI (P = 0.018, alpha = 0.017, Bonferroni three-comparison correction), with a relative risk for placebo/high dose of 3.7 for MI. The reasons for these differences in findings using the prespecified versus post hoc criterion may be attributable to the effect of acadesine on CK-MB release in this population. Significant decreases in CK-MB levels were found for patients who developed Q waves, with reductions in the high-dose group of approximately 32% for the mean maximum CK-MB, and 41% for the mean and 80% for the median AUC. These results indicate that high-dose acadesine may affect the extent of myocardial injury associated with MI; however, this observation requires confirmation in additional trials.
These results also may help to explain the MI efficacy results. Even using a high threshold, CK-MB was elevated in a considerable number of patients in each study group (placebo, 21.3%; low dose, 21.6%; and high dose, 16.7%). Although such CK-MB elevation may be indicative of MI, it may also be attributed to myocardial injury or ischemia, or mechanical manipulation of the heart, thereby making diagnosis of MI, when based on the or criterion, difficult. The high specificity associated with the and criterion may mitigate contributions of CK-MB not associated with MI. [63-66]However, without validation of CK-MB threshold, the issue remains unresolved and challenging.
Adverse Cardiovascular Outcomes
The incidence of all adverse outcomes combined were not statistically different between the three study groups. The incidence was significantly lower (P = 0.002) using the post hoc and MI criterion, with a relative risk for placebo/high dose of 2.7 for adverse cardiac outcomes. These findings reflect primarily the combined effects of acadesine on MI (discussed above) and cerebrovascular accident. In the high-dose group, the incidence of cerebrovascular accident was significantly reduced: nine occurrences in the placebo group versus one occurrence in the high-dose group (P = 0.02). There are a number of possible explanations, including acadesine reducing the occurrence of embolic phenomena or acadesine having a primary CNS effect. In laboratory models (in vitro and in vivo), acadesine has been shown to limit platelet aggregation and to have antioxidant properties, but these results have not been established in patients. Furthermore, acadesine does not appear to cross the blood-brain barrier effectively. Thus, the reasons for these CVA findings remain unknown, and the results await validation by future trials.
Regarding the effect of acadesine on other cardiovascular outcomes, no differences were found; however, no conclusion can be drawn regarding cardiac death because there were too few outcomes. Finally, low-dose acadesine (0.05 mg *symbol* kg sup -1 *symbol* min sup -1) in association with administration in the cardioplegic solution (5 micro gram/ml) did not appear to have an effect on any measure of infarction or cardiovascular outcome.
In contrast to our preliminary results, we did not demonstrate an effect of either low- or high-dose acadesine on the incidence or severity of myocardial ischemia before or after bypass, as assessed by either ECG ST-segment changes or TEE regional wall motion abnormality (RWMA). These results were surprising in light of the preliminary results, as well as the other infarction and injury findings of this study. There are several possible explanations for the results. First, the incidence of ischemia may not be affected by acadesine. Acadesine affects adenosine levels only locally in ischemic tissue ; therefore, although acadesine may decrease the consequences of ischemia, it may not decrease the incidence of ischemia. Second, the current study may have had insufficient power to detect differences in ischemia before bypass because of the low incidence of prebypass ischemia found in this study. After bypass, acadesine may have little effect on ischemia, because the damage may have been done at a time of cross clamping, and subsequent ischemia without substantial damage (injury or infarction) may be relatively minor and not affected by acadesine. Third, ischemia associated with MI may not be measurable in this study. The criteria for ECG ischemia was that it had to be a reversible process; therefore, ST changes that persisted would not be included in the calculations. However, such irreversible changes may be truly ischemic and associated with permanent damage, and by excluding such changes, more serious precursor ischemic events would be eliminated from analysis. Finally, in this study, after bypass, approximately 30% of the ECG and TEE data were not included in the analyses because they were not interpretable (e.g., paced rhythm or dysrhythmia). If adverse outcomes were higher in the group with noninterpretable ischemia data, an inherent error may exist.
Acadesine, when given to patients undergoing CABG surgery at a dose of 0.1 mg *symbol* kg sup -1 *symbol* min sup -1 for 7 h in association with administration in cardioplegic solution at a concentration of 5 micro gram/ml, is safe. Systemic blood pressure and heart rate, as well as ventricular function as measured by ventriculography, were no different in the acadesine-treated patients compared with the placebo-treated patients. These results demonstrate that, unlike adenosine, acadesine does not lower systemic vascular resistance or blood pressure, or produce conduction blockade or bradyarrhythmias. The overall adverse event profile was essentially indistinguishable from placebo. The only exception to this was transient hyperuricemia, which was maximal at the end of infusion and resolved during hospital stay without clinical sequelae. Patients with a history of gout had similar results, and the increases in uric acid were mitigated in patients receiving allopurinol therapy before surgery. This dose-dependent rise in uric acid was not unexpected, because acadesine is metabolized to uric acid. It is possible, however, that, in a larger patient population, selected patients may develop clinically significant hyperuricemia. However, based on the current results, it appears that this potential adverse reaction would be uncommon.
Study Design Considerations
The first consideration is that our a priori null hypothesis was that the incidence of MI would be equal in the three groups (high dose, low dose, and placebo) based on a definition of MI that included Q-wave and non-Q-wave infarcts. Therefore, comparisons between two groups are presented with the associated alpha levels corrected for multiple comparisons. It is to be noted that we did not prespecify use of corrections for multiple comparisons, nor use of the conservative Bonferroni correction. In this report, we have chosen to present our results using multiple-comparison Bonferroni corrections, realizing that our presentation of results is conservative. (For example, had we selected a less conservative correction, the post hoc analysis for MI would have demonstrated that statistical significance was achieved for the high-dose acadesine group versus the placebo group.) The second consideration is that two primary outcomes were prespecified: MI and adverse cardiovascular outcomes. Adverse cardiovascular outcomes, however, included MI; therefore, the second primary outcome was not independent of the first primary outcome. We did not prespecify a correction for choosing nonindependent primary outcomes. The third consideration is the effect of study site on treatment effects. Using the study site as a covariate, we found no interaction between treatment effect and study site. The fourth consideration is that the prospective design of the trial using MI as a primary outcome was made difficult by the lack of consensus regarding the criteria for defining MI after CABG. When we designed our trial, our preliminary data indicated that acadesine would be effective for prevention of both Q-wave and non-Q-wave infarcts; we chose a sensitive and less specific criterion for MI (Q wave or elevated CK-MB, or autopsy evidence). However, we realize that a second definition may be equally applicable; namely, the more specific (and less sensitive) criterion (Q wave and elevated CK-MB, or autopsy evidence). Therefore, our a priori study design using sensitive criteria for MI placed substantial emphasis on the postbypass CK-MB values, which were elevated in a considerable number of patients in each of the three study groups (placebo, 21.3%; low dose, 21.6%; and high dose, 16.7%). Therefore, CK-MB release occurs commonly in CABG patients, and choice of threshold is difficult, making prespecified definition of MI difficult. The fifth consideration is that the diagnosis of cerebrovascular accident was made using neurologic confirmation, with encephalopathy requiring confirmation of a focal defect using CT or MRI. However, an even more rigorous determination of cerebrovascular accident may also include a single neurologist at each center performing all neurologic examinations, both before and after surgery.
This study has established that acadesine, infused intravenously for up to 7 h, is safe, with an adverse event profile similar to placebo, with the exception of transient hyperuricemia. Low-dose acadesine (0.05 mg *symbol* kg sup -1 *symbol* h sup -1 intravenously and 5 micro gram/ml in cardioplegic solution) was not effective in preventing MI or adverse outcomes compared with placebo. The incidence of MI and all cardiovascular outcomes using high-dose acadesine (0.1 mg *symbol* kg sup -1 *symbol* min sup -1 intravenously and 5 micro gram/ml in cardioplegic solution) were not statistically different from the placebo results using the prespecified criteria (Q wave or CK-MB, or autopsy), perhaps because of the lack of specificity of CK-MB after bypass. Post hoc analysis using the more specific criteria for MI (Q wave and CK-MB, or autopsy) demonstrated that acadesine reduced all cardiovascular outcomes and may reduce the incidence of MI. Acadesine treatment significantly reduced the incidence of CVA. We believe that one cannot ignore the potentially useful findings resulting from post hoc analyses, especially when the results of the study itself call into question the a priori choice of outcome definition. Furthermore, the contrasting results of such analyses provide valuable information to other researchers regarding CK release and outcome definition for this population of patients, which we believe are critical data necessary for more rigorous design of future outcome trials. Finally, it should be recognized that this study was powered to detect a 50% reduction in the placebo MI (Q wave and non-Q wave) rate of 22% (preliminary study data) If the and criteria (large-Q-wave MI) is used in future clinical trials, the sample size would have to be increased substantially, given that the placebo rate for such infarctions is approximately 5%, as demonstrated in this trial.
In conclusion, the results of this trial demonstrated that acadesine was not effective in preventing perioperative MI, using the prespecified criterion for MI. Of note, however, are the results of the post hoc analysis, using the more specific criterion for MI, indicating that acadesine may reduce the incidence of larger Q-wave infarctions after coronary artery bypass surgery. A second trial is underway to evaluate this contention.