Mortality Benefit With Prasugrel in CABG Cohort
Mortality Benefit With Prasugrel in CABG Cohort
The study design and principal results from the pivotal phase III TRITON–TIMI 38 trial have been published. Briefly, 13,608 patients with ACS to be managed with percutaneous coronary intervention (PCI) were randomized to receive a loading dose (60 mg of prasugrel or 300 mg of clopidogrel) followed by a daily maintenance dose (10 mg of prasugrel or 75 mg of clopidogrel) for up to 15 months in combination with aspirin. The primary efficacy endpoint was the time of first occurrence of any element of the composite of cardiovascular (CV) death, nonfatal myocardial infarction (MI), or nonfatal stroke. Safety was assessed by using TIMI bleeding criteria and based on a fall in hemoglobin levels or intracranial hemorrhage versus actual measurement of chest tube output.
A supplemental case report form was developed to collect additional data from patients in TRITON–TIMI 38 who underwent CABG at any time point during the study. The data collection was designed to characterize the relationship of the withdrawal of thienopyridine before CABG to cumulative chest tube drainage and transfusions and to collect additional clinical risk factors for bleeding and adverse outcomes. An analysis plan was prospectively developed to include recognized risk-adjustment methods for CABG (EuroSCORE and STS scoring) mortality. Supplementary retrospective data collection was conducted by using chart review and was limited to the information captured during the patient's study participation.
This review was performed with the acknowledgment or approval of the ethics committee and regulatory board or as required by local regulations. The data were independently analyzed by a statistician from Duke University Medical Center. The corresponding author had full access to the data in the study.
The initial study population included the 485 patients who underwent CABG with or without concomitant cardiac procedures during their participation in the TRITON–TIMI 38 study from November 2004 to January of 2007. In 36 patients, a supplementary case report form could not be obtained. Two of the 36 (1 randomized to receive prasugrel but did not receive study drug; 1 randomized to receive clopidogrel and received drug before the procedure) failed to survive surgery and were included; the remaining 34 patients (11 in the prasugrel group and 23 in the clopidogrel group) with missing supplemental data were excluded, leaving 451 patients. Three additional patients were excluded due to an inability to determine the type of operative procedure performed, for a balance of 448 patients. Thus, the EuroSCORE and STS risk-adjusted predicted mortality was considered calculable for 446 of these 448 patients. The cohort of 448 included all deaths and all but 2 patients who were classified in the TRITON–TIMI 38 trial as having had TIMI major/minor bleeds (1 prasugrel-treated and 1 clopidogrel-treated patient).
The cohort of 448 patients was heterogeneous and included patients with major concomitant cardiac procedures in addition to CABG (n = 26), open-label use of antiplatelet therapy before the procedure (n = 20), and patients who did not receive study medication (n = 56). Therefore, the key population of interest was further refined to include only the cohort of 346 patients who underwent isolated CABG and received study drug before procedure.
Selection bias may be introduced due to the nonrandomized nature of the decision to perform CABG and the timing of study drug withdrawal; this could confound the comparison between prasugrel and clopidogrel. Thus, predicted probability of periprocedural mortality was used to adjust for potential imbalances and to validate comparisons. The predicted probability of mortality was calculated using EuroSCORE or STS score methods. For the purpose of calculating EuroSCORE and STS score based on risk factors, any missing values were considered as absence of that risk factor as recommended by the STS National Cardiac Database Committee. The primary safety endpoint was cumulative chest tube output in the first 12 post-operative hours after CABG. Key secondary safety endpoints included incidence of surgical re-exploration for bleeding, mortality within 30 days after CABG, occurrence of CV death, nonfatal MI, or nonfatal stroke within 30 days after CABG and total donor exposure. No adjustment was made for the possible selection bias in comparison of bleeding risk between prasugrel and clopidogrel due to the absence of a widely accepted measure for predicted risk of CABG-related bleeding.
Major surgical characteristics and mortality were analyzed for isolated CABG and for major cardiac procedures in addition to the CABG procedure (designated as CABG+). Statistical analyses and summaries are presented for isolated CABG procedures. Surgical characteristics, medical history, preoperative and postoperative (collected in hospital until discharge or on readmission within 30 days of CABG) state, hemodynamic and catheterization information, concomitant medication use during the periprocedural period, and the predicted probability of mortality at the time of CABG were summarized for each study drug. Statistical comparisons of these characteristics between prasugrel and clopidogrel were guided according to the data type: comparison of distributions of nominal data by using the Cochran-Mantel-Haenszel general association test, comparison of medians by using the Cochran-Mantel-Haenszel row mean score test for ordinal data, comparison of medians by using the Kruskal-Wallis test, and comparison of means by using the 2-sample t test or analysis of variance, as appropriate, for interval scale data.
Average cumulative chest tube blood loss at 12 h after CABG was compared by using the Kruskal-Wallis test. Kaplan-Meier methods were used for estimating the time profiles of cumulative hazard. A Cox proportional hazards model was used for estimating the unadjusted hazard ratio for prasugrel versus clopidogrel. Comparison between time profiles were conducted by using the log-rank test. Comparison of the risk of all-cause death, CV death, all-cause death through 30 days from CABG, and in-hospital death between prasugrel and clopidogrel was performed using logistic regression analysis, and the predicted probability of periprocedural mortality was included as a covariate. Separate analyses were performed by using the STS predicted mortality score and the EuroSCORE as covariates and produced similar results.
All analyses were conducted by using SAS version 9.1 (SAS Institute, Inc., Cary, North Carolina).
Methods
The study design and principal results from the pivotal phase III TRITON–TIMI 38 trial have been published. Briefly, 13,608 patients with ACS to be managed with percutaneous coronary intervention (PCI) were randomized to receive a loading dose (60 mg of prasugrel or 300 mg of clopidogrel) followed by a daily maintenance dose (10 mg of prasugrel or 75 mg of clopidogrel) for up to 15 months in combination with aspirin. The primary efficacy endpoint was the time of first occurrence of any element of the composite of cardiovascular (CV) death, nonfatal myocardial infarction (MI), or nonfatal stroke. Safety was assessed by using TIMI bleeding criteria and based on a fall in hemoglobin levels or intracranial hemorrhage versus actual measurement of chest tube output.
A supplemental case report form was developed to collect additional data from patients in TRITON–TIMI 38 who underwent CABG at any time point during the study. The data collection was designed to characterize the relationship of the withdrawal of thienopyridine before CABG to cumulative chest tube drainage and transfusions and to collect additional clinical risk factors for bleeding and adverse outcomes. An analysis plan was prospectively developed to include recognized risk-adjustment methods for CABG (EuroSCORE and STS scoring) mortality. Supplementary retrospective data collection was conducted by using chart review and was limited to the information captured during the patient's study participation.
This review was performed with the acknowledgment or approval of the ethics committee and regulatory board or as required by local regulations. The data were independently analyzed by a statistician from Duke University Medical Center. The corresponding author had full access to the data in the study.
Population
The initial study population included the 485 patients who underwent CABG with or without concomitant cardiac procedures during their participation in the TRITON–TIMI 38 study from November 2004 to January of 2007. In 36 patients, a supplementary case report form could not be obtained. Two of the 36 (1 randomized to receive prasugrel but did not receive study drug; 1 randomized to receive clopidogrel and received drug before the procedure) failed to survive surgery and were included; the remaining 34 patients (11 in the prasugrel group and 23 in the clopidogrel group) with missing supplemental data were excluded, leaving 451 patients. Three additional patients were excluded due to an inability to determine the type of operative procedure performed, for a balance of 448 patients. Thus, the EuroSCORE and STS risk-adjusted predicted mortality was considered calculable for 446 of these 448 patients. The cohort of 448 included all deaths and all but 2 patients who were classified in the TRITON–TIMI 38 trial as having had TIMI major/minor bleeds (1 prasugrel-treated and 1 clopidogrel-treated patient).
The cohort of 448 patients was heterogeneous and included patients with major concomitant cardiac procedures in addition to CABG (n = 26), open-label use of antiplatelet therapy before the procedure (n = 20), and patients who did not receive study medication (n = 56). Therefore, the key population of interest was further refined to include only the cohort of 346 patients who underwent isolated CABG and received study drug before procedure.
Safety and Mortality Analysis
Selection bias may be introduced due to the nonrandomized nature of the decision to perform CABG and the timing of study drug withdrawal; this could confound the comparison between prasugrel and clopidogrel. Thus, predicted probability of periprocedural mortality was used to adjust for potential imbalances and to validate comparisons. The predicted probability of mortality was calculated using EuroSCORE or STS score methods. For the purpose of calculating EuroSCORE and STS score based on risk factors, any missing values were considered as absence of that risk factor as recommended by the STS National Cardiac Database Committee. The primary safety endpoint was cumulative chest tube output in the first 12 post-operative hours after CABG. Key secondary safety endpoints included incidence of surgical re-exploration for bleeding, mortality within 30 days after CABG, occurrence of CV death, nonfatal MI, or nonfatal stroke within 30 days after CABG and total donor exposure. No adjustment was made for the possible selection bias in comparison of bleeding risk between prasugrel and clopidogrel due to the absence of a widely accepted measure for predicted risk of CABG-related bleeding.
Statistical Methods
Major surgical characteristics and mortality were analyzed for isolated CABG and for major cardiac procedures in addition to the CABG procedure (designated as CABG+). Statistical analyses and summaries are presented for isolated CABG procedures. Surgical characteristics, medical history, preoperative and postoperative (collected in hospital until discharge or on readmission within 30 days of CABG) state, hemodynamic and catheterization information, concomitant medication use during the periprocedural period, and the predicted probability of mortality at the time of CABG were summarized for each study drug. Statistical comparisons of these characteristics between prasugrel and clopidogrel were guided according to the data type: comparison of distributions of nominal data by using the Cochran-Mantel-Haenszel general association test, comparison of medians by using the Cochran-Mantel-Haenszel row mean score test for ordinal data, comparison of medians by using the Kruskal-Wallis test, and comparison of means by using the 2-sample t test or analysis of variance, as appropriate, for interval scale data.
Average cumulative chest tube blood loss at 12 h after CABG was compared by using the Kruskal-Wallis test. Kaplan-Meier methods were used for estimating the time profiles of cumulative hazard. A Cox proportional hazards model was used for estimating the unadjusted hazard ratio for prasugrel versus clopidogrel. Comparison between time profiles were conducted by using the log-rank test. Comparison of the risk of all-cause death, CV death, all-cause death through 30 days from CABG, and in-hospital death between prasugrel and clopidogrel was performed using logistic regression analysis, and the predicted probability of periprocedural mortality was included as a covariate. Separate analyses were performed by using the STS predicted mortality score and the EuroSCORE as covariates and produced similar results.
All analyses were conducted by using SAS version 9.1 (SAS Institute, Inc., Cary, North Carolina).
