LIVER TRANSPLANTATION 21:13–21, 2015
ORIGINAL ARTICLE
Adverse Cardiac Events After Orthotopic Liver Transplantation: A Cross-Sectional Study in 389 Consecutive Patients Ramona Nicolau-Raducu,1 Marina Gitman,1 Donald Ganier,1 George E. Loss,2 Ari J. Cohen,2 Hamang Patel,3 Nigel Girgrah,4 Krish Sekar,1 and Bobby Nossaman1 Departments of 1Anesthesiology, 2Surgery, 3Cardiology, and 4Gastroenterology and Hepatology, Ochsner Clinic Foundation, New Orleans, LA
Current American College of Cardiology/American Heart Association guidelines caution that preoperative noninvasive cardiac tests may have poor predictive value for detecting coronary artery disease in liver transplant candidates. The purpose of our study was to evaluate the role of clinical predictor variables for early and late cardiac morbidity and mortality and the predictive values of noninvasive cardiac tests for perioperative cardiac events in a high-risk liver transplant population. In all, 389 adult recipients were retrospectively analyzed for a median follow-up time of 3.4 years (range 5 2.3-4.4 years). Overall survival was 83%. During the first year after transplantation, cardiovascular morbidity and mortality rates were 15.2% and 2.8%. In patients who survived the first year, cardiovascular morbidity and mortality rates were 3.9% and 2%, with cardiovascular etiology as the third leading cause of death. Dobutamine stress echocardiography (DSE) and single-photon emission computed tomography had respective sensitivities of 9% and 57%, specificities of 98% and 75%, positive predictive values of 33% and 28%, and negative predictive values of 89% and 91% for predicting early cardiac events. A rate blood pressure product less than 12,000 with DSE was associated with an increased risk for postoperative atrial fibrillation. Correspondence analysis identified a statistical association between nonalcoholic steatohepatitis/cryptogenic cirrhosis and postoperative myocardial ischemia. Logistic regression identified 3 risk factors for postoperative acute coronary syndrome: age, history of coronary artery disease, and pretransplant requirement for vasopressors. Multivariable analysis showed statistical associations of the Model for End-Stage Liver Disease score and the development of acute kidney injury as risk factors for overall cardiac-related mortality. These findings may help in identifying high-risk patients and may lead to the development C 2014 AASLD. of better cardiac tests. Liver Transpl 21:13-21, 2015. V Received April 1, 2014; accepted September 8, 2014. Since the first successful orthotopic liver transplantation (OLT) in 1963,1 patient survival has significantly improved.2 However, current OLT recipients are older and have more comorbidities.2 Although late postoperative cardiac outcomes from cardiac events are the third most common cause of death, studies commenting on early cardiac morbidity and mortality are
scarce.2 Unfortunately, studies of early cardiac events do not uniformly report their data, and this makes comparisons difficult. Recent American College of Cardiology/American Heart Association guidelines for preoperative evaluations for cardiac disease in OLT recipients recommend noninvasive stress testing in patients with more than
Abbreviations: ACS, acute coronary syndrome; AKI, acute kidney injury; CAD, coronary artery disease; CVA, cerebrovascular accident; CVE, cardiovascular event; DSE, dobutamine stress echocardiography; ECG, electrocardiogram; HF, heart failure; HR, heart rate; LVEF, left ventricular ejection fraction; MELD, Model for End-Stage Liver Disease; MI, myocardial infarction; MSOF, multisystem organ failure; NASH, nonalcoholic steatohepatitis; NPV, negative predictive value; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis; PPV, positive predictive value; PSC, primary sclerosis cholangitis; PVD, peripheral vascular disease; RPP, rate blood pressure product at peak dobutamine stress echocardiography; SPECT, single-photon emission computed tomography. Potential conflict of interest: Nothing to report. Address reprint requests to Ramona Nicolau-Raducu, M.D., Ph.D., Department of Anesthesiology, Ochsner Medical Center, 1514 Jefferson Highway, New Orleans, LA 70121. Telephone: 504-842-3755; FAX: 504-842-2036; E-mail: [email protected] DOI 10.1002/lt.23997 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
14 NICOLAU-RADUCU ET AL.
3 cardiac risk factors but caution that these tests have low specificity and sensitivity for detecting coronary artery disease (CAD) in this pretransplant population.3 The purpose of this study was to investigate the role of reported predictor variables in the incidence of early and late cardiac morbidity and mortality in a high-risk group of OLT recipients. Additionally, the sensitivity and specificity of noninvasive stress testing were studied to clarify the predictive value of these tests for severe postoperative cardiac events.
PATIENTS AND METHODS After institutional board review approval and informed patient consent were waived, 389 consecutive adult liver transplants performed at the Ochsner Medical Center from January 2008 to December 2011 were retrospectively analyzed. Patients were followed for a median of 3.4 years (range 5 2.3-4.4 years). During the study period, there were 21 retransplants in this cohort (18 liver transplants and 3 liver-kidney transplants). Each patient is presented once. No patient was lost to follow-up. The following clinical predictors were recorded from patients’ medical records: baseline variables [age, sex, ethnicity, body mass index (kg/m2), etiology of liver disease, Model for End-Stage Liver Disease (MELD) score, and retransplantation status], clinical variables [pretransplant acute kidney injury (AKI) as defined by Kellum et al.,4 pretransplant renal replacement therapy, and presence of portopulmonary hypertension],5 and CAD risk variables [diabetes mellitus, systemic high blood pressure, tobacco use, dyslipidemia (one of the following: total cholesterol > 200 mg/dL, triglyceride level > 150 mg/dL, high-density lipoprotein level < 40 mg/dL, or low-density lipoprotein level > 130 mg/ dL), family history of premature CAD, personal history of CAD (coronary artery bypass grafting, coronary angioplasty and/or stenting, or unstable angina), and CAD equivalents such as cerebrovascular accident (CVA), transient ischemic attack, and peripheral vascular disease (PVD)]. Preoperative cardiac testing included electrocardiogram (ECG), transthoracic echocardiography, and noninvasive chemically induced stress tests. Dobutamine stress echocardiography (DSE) is used for the preoperative cardiac assessment of potential liver transplant candidates at the Ochsner Medical Center. Target DSE is defined as the attainment of greater than 85% of the maximum age-predicted heart rate (HR). In all, 278 pretransplant patients (71%) underwent DSE, but only 195 patients (70%) reached their predicted target HR. A protocol to discontinue beta-blockers before DSE was not established. The rate blood pressure product at peak dobutamine stress echocardiography (RPP) was calculated for each pretransplant candidate.6 Fortyseven patients (12%) underwent a single-photon emission computed tomography (SPECT) myocardial scan either for suboptimal HR during DSE (7 patients) or in accordance with cardiology recommendations, which
LIVER TRANSPLANTATION, January 2015
were based on each patient’s symptomatology, previous cardiac history, and assessment of risk factors. Eleven patients (3%) had an exercise stress test. Fifty-three patients (14%) did not undergo cardiac testing because they were less than 45 years of age and did not have any cardiac risk factors or because their critical condition precluded cardiac testing. Patients with ECG evidence of stress-induced ischemia were referred to cardiology for further evaluation. Twenty-one patients (5%) underwent left heart catheterization before transplantation. Among them, 7 were patients with stress-induced ischemia who had undergone previous cardiac interventions such as coronary artery stenting or coronary artery bypass grafting, and another 7 patients did not achieve target DSE. In a different group, 16 patients (4%) underwent coronary angiography after OLT for the postoperative evaluation of cardiac events. The decision for coronary lesion stenting was at the discretion of the cardiologist and was based on occlusion severity, lesion location, and a positive stress test. A review of medical records identified major perioperative cardiovascular events (CVEs), including acute coronary syndrome (ACS), new onset of heart failure (HF), arrhythmia, resuscitated cardiac arrest, CVA, and PVD. ACS included cases of unstable angina and myocardial infarction (MI). A diagnosis of MI was made on the basis of an elevation of all cardiac biomarkers [troponin > 0.026 ng/mL (normal, 0.0000.026 ng/mL), creatine phosphokinase MB > 6.5 ng/ mL (normal, 0.1-6.5 ng/mL), and MB > 5.0% (normal, 0.0%-5.0%)] and the presence of 1 or more of the following signs or symptoms: chest pain, ECG changes suggesting ischemia, and new wall motion abnormalities on ECG. In 1 case, the diagnosis of MI was made at autopsy. New onset of HF was defined by echocardiographic evidence of worsening left ventricular dysfunction. Severe HF was further characterized as a left ventricular ejection fraction (LVEF) 40%. Arrhythmia was defined as atrial fibrillation or symptomatic bradycardia. Causes of documented resuscitated cardiac arrest included asystole, symptomatic ventricular tachycardia, and ventricular fibrillation. Patients with postoperative sepsis or multisystem organ failure (MSOF) requiring cardiac resuscitation were not included in this study. Because the etiologies of early and late cardiovascular complication differ, we separated the results into early (less than 1 year) and late (more than 1 year and up to a median of 3.4 years) cardiovascular morbidity and mortality. These data were used to evaluate the accuracy of target DSE and SPECT imaging for predicting cardiac events during the first postoperative year. Multivariable analysis was used to determine the role of patient-specific risk factors in postoperative overall cardiac events and cardiac deaths.
Statistical Analysis Categorical variables were presented as percentages, and differences between the groups were assessed
LIVER TRANSPLANTATION, Vol. 21, No. 1, 2015
NICOLAU-RADUCU ET AL. 15
TABLE 1. Demographic and Clinical Characteristics (n 5 389) Age (years) Age 60 years, n (%) Male, n (%) Race, n (%) Caucasian Hispanic African American Body mass index (kg/m2) Body mass index 30 kg/m2, n (%) MELD score Retransplantation, n (%) AKI pretransplant, n (%) Pretransplant renal replacement therapy, n (%) Etiology, n (%) Viral hepatitis NASH/cryptogenic cirrhosis Alcohol cirrhosis Autoimmune hepatitis PBC/PSC Acetaminophen/idiosyncratic drug reaction Other (eg, alpha-1-antitrypsin deficiency and Budd-Chiari) Associated hepatocellular carcinoma Portopulmonary hypertension, n (%) Mild Moderate/severe
55 (50-60) 112 (29) 254 (65) 240 (62) 71 (18) 59 (15) 29 (26-33) 163 (42) 22 (20-27.5) 22 (6) 102 (26) 24 (6)
173 (44) 74 (19) 62 (16) 19 (5) 16 (4) 7 (2) 38 (10) 109 (28) 16 (4) 12 (3)
NOTE. The values are presented as medians (25th and 75th percentiles) or as numbers and percentages.
with Chi-Square (v2) tests. Continuous variables with skewed distributions were presented as medians and interquartile ranges (25%-75%), with differences between groups assessed with the Wilcoxon rank-sum test. Nominal logistic regression using a stepwise personality with a stopping rule P value threshold of 0.25 for probability to enter and a more stringent P value of 0.10 for probability to leave, conducted in a backward direction, was performed for the discovery of a predictive model identifying predictor variables statistically associated with cardiac morbidity/mortality and for adjustments for the potential influence of confounders. Odds ratios and 95% confidence intervals were calculated for cardiac morbidity/mortality. Misclassification rates calculated the proportion of observations allocated to the incorrect group and helped in interpreting whole-model R-squared values when categorical predictor variables were used in regression analysis. Misclassification rates showed the falsepositive rate. Ordinal logistic personality with the same stopping rule described previously was selected to analyze predictor variables associated with adverse events. C-statistics for both models were calculated to measure the strength of the associations. The alpha levels for the bivariate and logistic regression analyses
were set at 0.05. Robust statistical technique recursive partitioning was employed to explore the relationship of critical rate-pressure product cutoff values when they were plotted against atrial fibrillation. Recursive partitioning consolidated outcomes into similar groups on the basis of predictor variables.7,8 In recursive partitioning, a log-worth value greater than 2 was considered significant at the 1 Year Posttransplant) Among patients who survived the first year, there were 14 cases (3.9%) of cardiovascular morbidity. One patient with nonobstructive CAD had angina. Five patients suffered MI with a median time from transplantation to event of 24 months (range 5 20-30 months). Among them, 1 patient had a previously stented coronary lesion, and 3 patients had more than 3 CAD risk factors before transplantation. Other causes of cardiovascular morbidity included 2 cases of severe HF, 2 cases of atrial fibrillation, 2 CVAs (1 from a septic embolus), and 3 PVD complications (2 below-the-knee amputations and 1 renal artery stenosis). Sixty-six patients (17%) died during the follow-up time of 3.4 years (range 5 2.3-4.4 years), among whom a total of 34 patients (8.7%) died after 1 year post transplant, with 2% of these deaths qualified as being of cardiovascular origin. In patients who survived the first year, a cardiovascular etiology was the third highest cause of death (21%), and this was preceded by graft failure (29%) and malignancy (26%; Fig. 1). Specific cardiovascular causes were CVA (n 5 2), HF (n 5 3), and sudden death (n 5 2; 1 patient with aortic valve endocarditis who died within 1 month of discharge on parenteral antibiotics, and 1 dialysis-dependent, noncompliant patient with a history of severe HF). A correspondence analysis examining the role of liver disease etiology in the development of postopera-
NICOLAU-RADUCU ET AL. 17
tive myocardial ischemia revealed an association of nonalcoholic steatohepatitis (NASH)/cryptogenic cirrhosis with postoperative myocardial ischemia, whereas cholestatic liver disease primary biliary cirrhosis (PBC) and primary sclerosis cholangitis (PSC) were not associated (Fig. 2). Logistic regression identified 3 risk factors for postoperative ACS, and these results are summarized in Table 4. The whole model was statistically significant (v2 5 40.7, P 5 0.001), with the risk factors of age (v2 5 14.5, P < 0.001), history of CAD (v2 5 13.8, P < 0.001), and pretransplant vasopressor administration (v2 5 8.4, P 5 0.004) statistically associated with postoperative ACS (Table 4). A C-statistic of 0.82 was calculated for this group of risk factors associated with ACS (Fig. 3). A misclassification rate of 0.07 was observed. Multivariable analysis identified 2 risk factors, MELD and AKI, for total cardiac-related death. When logistic regression was used, the whole model (v2 5 10.6, P 5 0.005) was statistically significant. The risk factors AKI (v2 5 10.6, P 5 0.001) and MELD (v2 5 3.9, P 5 0.049) were statistically associated with total cardiac-related death (Table 4). A C-statistic of 0.71 was calculated for this group of risk factors associated with total cardiac-related death. A misclassification rate of 0.04 was observed. Logistic regression was used to examine the role of blood transfusion in total cardiac death. On statistical analysis, the whole model (v2 5 21.1, P < 0.001) was statistically significant. Recursive partitioning was employed to identify the critical number of transfused units that increased risk for total cardiac death. The risk of total cardiac death increased from 2.2% when
ORIGINAL ARTICLE
Adverse Cardiac Events After Orthotopic Liver Transplantation: A Cross-Sectional Study in 389 Consecutive Patients Ramona Nicolau-Raducu,1 Marina Gitman,1 Donald Ganier,1 George E. Loss,2 Ari J. Cohen,2 Hamang Patel,3 Nigel Girgrah,4 Krish Sekar,1 and Bobby Nossaman1 Departments of 1Anesthesiology, 2Surgery, 3Cardiology, and 4Gastroenterology and Hepatology, Ochsner Clinic Foundation, New Orleans, LA
Current American College of Cardiology/American Heart Association guidelines caution that preoperative noninvasive cardiac tests may have poor predictive value for detecting coronary artery disease in liver transplant candidates. The purpose of our study was to evaluate the role of clinical predictor variables for early and late cardiac morbidity and mortality and the predictive values of noninvasive cardiac tests for perioperative cardiac events in a high-risk liver transplant population. In all, 389 adult recipients were retrospectively analyzed for a median follow-up time of 3.4 years (range 5 2.3-4.4 years). Overall survival was 83%. During the first year after transplantation, cardiovascular morbidity and mortality rates were 15.2% and 2.8%. In patients who survived the first year, cardiovascular morbidity and mortality rates were 3.9% and 2%, with cardiovascular etiology as the third leading cause of death. Dobutamine stress echocardiography (DSE) and single-photon emission computed tomography had respective sensitivities of 9% and 57%, specificities of 98% and 75%, positive predictive values of 33% and 28%, and negative predictive values of 89% and 91% for predicting early cardiac events. A rate blood pressure product less than 12,000 with DSE was associated with an increased risk for postoperative atrial fibrillation. Correspondence analysis identified a statistical association between nonalcoholic steatohepatitis/cryptogenic cirrhosis and postoperative myocardial ischemia. Logistic regression identified 3 risk factors for postoperative acute coronary syndrome: age, history of coronary artery disease, and pretransplant requirement for vasopressors. Multivariable analysis showed statistical associations of the Model for End-Stage Liver Disease score and the development of acute kidney injury as risk factors for overall cardiac-related mortality. These findings may help in identifying high-risk patients and may lead to the development C 2014 AASLD. of better cardiac tests. Liver Transpl 21:13-21, 2015. V Received April 1, 2014; accepted September 8, 2014. Since the first successful orthotopic liver transplantation (OLT) in 1963,1 patient survival has significantly improved.2 However, current OLT recipients are older and have more comorbidities.2 Although late postoperative cardiac outcomes from cardiac events are the third most common cause of death, studies commenting on early cardiac morbidity and mortality are
scarce.2 Unfortunately, studies of early cardiac events do not uniformly report their data, and this makes comparisons difficult. Recent American College of Cardiology/American Heart Association guidelines for preoperative evaluations for cardiac disease in OLT recipients recommend noninvasive stress testing in patients with more than
Abbreviations: ACS, acute coronary syndrome; AKI, acute kidney injury; CAD, coronary artery disease; CVA, cerebrovascular accident; CVE, cardiovascular event; DSE, dobutamine stress echocardiography; ECG, electrocardiogram; HF, heart failure; HR, heart rate; LVEF, left ventricular ejection fraction; MELD, Model for End-Stage Liver Disease; MI, myocardial infarction; MSOF, multisystem organ failure; NASH, nonalcoholic steatohepatitis; NPV, negative predictive value; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis; PPV, positive predictive value; PSC, primary sclerosis cholangitis; PVD, peripheral vascular disease; RPP, rate blood pressure product at peak dobutamine stress echocardiography; SPECT, single-photon emission computed tomography. Potential conflict of interest: Nothing to report. Address reprint requests to Ramona Nicolau-Raducu, M.D., Ph.D., Department of Anesthesiology, Ochsner Medical Center, 1514 Jefferson Highway, New Orleans, LA 70121. Telephone: 504-842-3755; FAX: 504-842-2036; E-mail: [email protected] DOI 10.1002/lt.23997 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
14 NICOLAU-RADUCU ET AL.
3 cardiac risk factors but caution that these tests have low specificity and sensitivity for detecting coronary artery disease (CAD) in this pretransplant population.3 The purpose of this study was to investigate the role of reported predictor variables in the incidence of early and late cardiac morbidity and mortality in a high-risk group of OLT recipients. Additionally, the sensitivity and specificity of noninvasive stress testing were studied to clarify the predictive value of these tests for severe postoperative cardiac events.
PATIENTS AND METHODS After institutional board review approval and informed patient consent were waived, 389 consecutive adult liver transplants performed at the Ochsner Medical Center from January 2008 to December 2011 were retrospectively analyzed. Patients were followed for a median of 3.4 years (range 5 2.3-4.4 years). During the study period, there were 21 retransplants in this cohort (18 liver transplants and 3 liver-kidney transplants). Each patient is presented once. No patient was lost to follow-up. The following clinical predictors were recorded from patients’ medical records: baseline variables [age, sex, ethnicity, body mass index (kg/m2), etiology of liver disease, Model for End-Stage Liver Disease (MELD) score, and retransplantation status], clinical variables [pretransplant acute kidney injury (AKI) as defined by Kellum et al.,4 pretransplant renal replacement therapy, and presence of portopulmonary hypertension],5 and CAD risk variables [diabetes mellitus, systemic high blood pressure, tobacco use, dyslipidemia (one of the following: total cholesterol > 200 mg/dL, triglyceride level > 150 mg/dL, high-density lipoprotein level < 40 mg/dL, or low-density lipoprotein level > 130 mg/ dL), family history of premature CAD, personal history of CAD (coronary artery bypass grafting, coronary angioplasty and/or stenting, or unstable angina), and CAD equivalents such as cerebrovascular accident (CVA), transient ischemic attack, and peripheral vascular disease (PVD)]. Preoperative cardiac testing included electrocardiogram (ECG), transthoracic echocardiography, and noninvasive chemically induced stress tests. Dobutamine stress echocardiography (DSE) is used for the preoperative cardiac assessment of potential liver transplant candidates at the Ochsner Medical Center. Target DSE is defined as the attainment of greater than 85% of the maximum age-predicted heart rate (HR). In all, 278 pretransplant patients (71%) underwent DSE, but only 195 patients (70%) reached their predicted target HR. A protocol to discontinue beta-blockers before DSE was not established. The rate blood pressure product at peak dobutamine stress echocardiography (RPP) was calculated for each pretransplant candidate.6 Fortyseven patients (12%) underwent a single-photon emission computed tomography (SPECT) myocardial scan either for suboptimal HR during DSE (7 patients) or in accordance with cardiology recommendations, which
LIVER TRANSPLANTATION, January 2015
were based on each patient’s symptomatology, previous cardiac history, and assessment of risk factors. Eleven patients (3%) had an exercise stress test. Fifty-three patients (14%) did not undergo cardiac testing because they were less than 45 years of age and did not have any cardiac risk factors or because their critical condition precluded cardiac testing. Patients with ECG evidence of stress-induced ischemia were referred to cardiology for further evaluation. Twenty-one patients (5%) underwent left heart catheterization before transplantation. Among them, 7 were patients with stress-induced ischemia who had undergone previous cardiac interventions such as coronary artery stenting or coronary artery bypass grafting, and another 7 patients did not achieve target DSE. In a different group, 16 patients (4%) underwent coronary angiography after OLT for the postoperative evaluation of cardiac events. The decision for coronary lesion stenting was at the discretion of the cardiologist and was based on occlusion severity, lesion location, and a positive stress test. A review of medical records identified major perioperative cardiovascular events (CVEs), including acute coronary syndrome (ACS), new onset of heart failure (HF), arrhythmia, resuscitated cardiac arrest, CVA, and PVD. ACS included cases of unstable angina and myocardial infarction (MI). A diagnosis of MI was made on the basis of an elevation of all cardiac biomarkers [troponin > 0.026 ng/mL (normal, 0.0000.026 ng/mL), creatine phosphokinase MB > 6.5 ng/ mL (normal, 0.1-6.5 ng/mL), and MB > 5.0% (normal, 0.0%-5.0%)] and the presence of 1 or more of the following signs or symptoms: chest pain, ECG changes suggesting ischemia, and new wall motion abnormalities on ECG. In 1 case, the diagnosis of MI was made at autopsy. New onset of HF was defined by echocardiographic evidence of worsening left ventricular dysfunction. Severe HF was further characterized as a left ventricular ejection fraction (LVEF) 40%. Arrhythmia was defined as atrial fibrillation or symptomatic bradycardia. Causes of documented resuscitated cardiac arrest included asystole, symptomatic ventricular tachycardia, and ventricular fibrillation. Patients with postoperative sepsis or multisystem organ failure (MSOF) requiring cardiac resuscitation were not included in this study. Because the etiologies of early and late cardiovascular complication differ, we separated the results into early (less than 1 year) and late (more than 1 year and up to a median of 3.4 years) cardiovascular morbidity and mortality. These data were used to evaluate the accuracy of target DSE and SPECT imaging for predicting cardiac events during the first postoperative year. Multivariable analysis was used to determine the role of patient-specific risk factors in postoperative overall cardiac events and cardiac deaths.
Statistical Analysis Categorical variables were presented as percentages, and differences between the groups were assessed
LIVER TRANSPLANTATION, Vol. 21, No. 1, 2015
NICOLAU-RADUCU ET AL. 15
TABLE 1. Demographic and Clinical Characteristics (n 5 389) Age (years) Age 60 years, n (%) Male, n (%) Race, n (%) Caucasian Hispanic African American Body mass index (kg/m2) Body mass index 30 kg/m2, n (%) MELD score Retransplantation, n (%) AKI pretransplant, n (%) Pretransplant renal replacement therapy, n (%) Etiology, n (%) Viral hepatitis NASH/cryptogenic cirrhosis Alcohol cirrhosis Autoimmune hepatitis PBC/PSC Acetaminophen/idiosyncratic drug reaction Other (eg, alpha-1-antitrypsin deficiency and Budd-Chiari) Associated hepatocellular carcinoma Portopulmonary hypertension, n (%) Mild Moderate/severe
55 (50-60) 112 (29) 254 (65) 240 (62) 71 (18) 59 (15) 29 (26-33) 163 (42) 22 (20-27.5) 22 (6) 102 (26) 24 (6)
173 (44) 74 (19) 62 (16) 19 (5) 16 (4) 7 (2) 38 (10) 109 (28) 16 (4) 12 (3)
NOTE. The values are presented as medians (25th and 75th percentiles) or as numbers and percentages.
with Chi-Square (v2) tests. Continuous variables with skewed distributions were presented as medians and interquartile ranges (25%-75%), with differences between groups assessed with the Wilcoxon rank-sum test. Nominal logistic regression using a stepwise personality with a stopping rule P value threshold of 0.25 for probability to enter and a more stringent P value of 0.10 for probability to leave, conducted in a backward direction, was performed for the discovery of a predictive model identifying predictor variables statistically associated with cardiac morbidity/mortality and for adjustments for the potential influence of confounders. Odds ratios and 95% confidence intervals were calculated for cardiac morbidity/mortality. Misclassification rates calculated the proportion of observations allocated to the incorrect group and helped in interpreting whole-model R-squared values when categorical predictor variables were used in regression analysis. Misclassification rates showed the falsepositive rate. Ordinal logistic personality with the same stopping rule described previously was selected to analyze predictor variables associated with adverse events. C-statistics for both models were calculated to measure the strength of the associations. The alpha levels for the bivariate and logistic regression analyses
were set at 0.05. Robust statistical technique recursive partitioning was employed to explore the relationship of critical rate-pressure product cutoff values when they were plotted against atrial fibrillation. Recursive partitioning consolidated outcomes into similar groups on the basis of predictor variables.7,8 In recursive partitioning, a log-worth value greater than 2 was considered significant at the 1 Year Posttransplant) Among patients who survived the first year, there were 14 cases (3.9%) of cardiovascular morbidity. One patient with nonobstructive CAD had angina. Five patients suffered MI with a median time from transplantation to event of 24 months (range 5 20-30 months). Among them, 1 patient had a previously stented coronary lesion, and 3 patients had more than 3 CAD risk factors before transplantation. Other causes of cardiovascular morbidity included 2 cases of severe HF, 2 cases of atrial fibrillation, 2 CVAs (1 from a septic embolus), and 3 PVD complications (2 below-the-knee amputations and 1 renal artery stenosis). Sixty-six patients (17%) died during the follow-up time of 3.4 years (range 5 2.3-4.4 years), among whom a total of 34 patients (8.7%) died after 1 year post transplant, with 2% of these deaths qualified as being of cardiovascular origin. In patients who survived the first year, a cardiovascular etiology was the third highest cause of death (21%), and this was preceded by graft failure (29%) and malignancy (26%; Fig. 1). Specific cardiovascular causes were CVA (n 5 2), HF (n 5 3), and sudden death (n 5 2; 1 patient with aortic valve endocarditis who died within 1 month of discharge on parenteral antibiotics, and 1 dialysis-dependent, noncompliant patient with a history of severe HF). A correspondence analysis examining the role of liver disease etiology in the development of postopera-
NICOLAU-RADUCU ET AL. 17
tive myocardial ischemia revealed an association of nonalcoholic steatohepatitis (NASH)/cryptogenic cirrhosis with postoperative myocardial ischemia, whereas cholestatic liver disease primary biliary cirrhosis (PBC) and primary sclerosis cholangitis (PSC) were not associated (Fig. 2). Logistic regression identified 3 risk factors for postoperative ACS, and these results are summarized in Table 4. The whole model was statistically significant (v2 5 40.7, P 5 0.001), with the risk factors of age (v2 5 14.5, P < 0.001), history of CAD (v2 5 13.8, P < 0.001), and pretransplant vasopressor administration (v2 5 8.4, P 5 0.004) statistically associated with postoperative ACS (Table 4). A C-statistic of 0.82 was calculated for this group of risk factors associated with ACS (Fig. 3). A misclassification rate of 0.07 was observed. Multivariable analysis identified 2 risk factors, MELD and AKI, for total cardiac-related death. When logistic regression was used, the whole model (v2 5 10.6, P 5 0.005) was statistically significant. The risk factors AKI (v2 5 10.6, P 5 0.001) and MELD (v2 5 3.9, P 5 0.049) were statistically associated with total cardiac-related death (Table 4). A C-statistic of 0.71 was calculated for this group of risk factors associated with total cardiac-related death. A misclassification rate of 0.04 was observed. Logistic regression was used to examine the role of blood transfusion in total cardiac death. On statistical analysis, the whole model (v2 5 21.1, P < 0.001) was statistically significant. Recursive partitioning was employed to identify the critical number of transfused units that increased risk for total cardiac death. The risk of total cardiac death increased from 2.2% when
