Posttransplant cardiovascular (CV) complications are the leading cause of non–graft-related death in orthotopic liver transplant (OLT) patients. With a national shortage of donors, the selection of candidates who are likely to benefit most and live the longest after receiving a liver transplant is a priority. The disparity between organ availability and patient need has grown steadily over the past 15 years. According to Organ Procurement and Transplantation Network data as of October 2006, there were over 17,000 patients in the United States listed as candidates for liver transplantation. However, in 2005, only 7013 donors were available (www.unos.org).
Although the mortality rate for transplanted patients with established coronary artery disease (CAD) is prohibitively high at approximately 50% at 1 year,1 occult CAD with resultant posttransplant adverse CV events is not uncommon. As liver transplant techniques advance and older and higher risk patients are transplanted, adverse cardiac event rates are likely to increase. With the utilization of substantial resources and the shortage of available transplant organs, there is a need to decrease the frequency of cardiac events in the OLT population. To date, there is little data to guide either pre–liver transplant or post–liver transplant cardiac risk stratification. Traditional risk stratification using established algorithms such as the Framingham risk score (FRS) underestimate CV event rate risk and are poor predictors of risk in this population.2 Other risk stratification methods such as the Systematic Coronary Risk Evaluation Project and the Prospective Cardiovascular Münster Study have also been studied3 but have no better sensitivity and specificity than the FRS.4
Recognizing the problem of occult CAD in OLT candidates, the American Association for the Study of Liver Diseases has established guidelines for CV evaluation of patients who may be liver transplant candidates.5 These guidelines advise that prospective OLT patients should undergo an evaluation for CAD if they are >50 years old, are diabetic, are chronic smokers, or have a clinical or family history of heart disease. However, these guidelines have been made with only limited data, and there is no consensus regarding which noninvasive cardiac test may be best suited for this task.
Dobutamine stress echocardiography (DSE) has been touted as an effective screening test for CV risk in pre-OLT patients. Some studies have found DSE to have a high sensitivity and specificity,6, 7 whereas the results of other studies have been less impressive.8 At our institution, DSE has been the primary test for CV risk stratification of end-stage liver disease (ESLD) patients before potential OLT. We sought to retrospectively analyze the ability of clinical and DSE variables to predict intraoperative and perioperative cardiac events and cardiac mortality in our OLT patient experience up to 4 months post–liver transplant.
PATIENTS AND METHODS
A retrospective chart review was performed on all 284 patients who underwent OLT at our institution from June 1999 to August 2005. The study was approved by the Mayo Clinic Foundation Institutional Review Board. Of the 284 patients, 157 (55%) were evaluated by DSE prior to OLT. During this period, the selection criteria for stress testing prior to OLT included age > 45 years, a history of diabetes mellitus, a history of peripheral vascular disease, or the presence of more than 2 FRS cardiac risk factors. DSE was performed in our echocardiographic laboratory, which is accredited by the Intersocietal Commission for the Accreditation of Echocardiography Laboratories. All studies were interpreted by level III trained physician echocardiographers in accordance with previously established guidelines.9, 10
Patients' charts were reviewed from the time of pretransplant evaluation through their routine 4-month posttransplant follow-up. Follow-up was available in 100% of patients. Baseline variables included age, gender, ethnicity, etiology of liver failure, and history of hypertension, smoking, dyslipidemia, diabetes, or cardiac disease. Histories of beta blocker or statin use were obtained from the medical record. The Model for End-Stage Liver Disease (MELD) score at the time of transplant was calculated with established methodology.11, 12 Clinical variables that were recorded included height, weight, body mass index, heart rate, and resting blood pressure. Echo variables included the left ventricular (LV) ejection fraction and rate pressure product at peak DSE stress (heart rate × systolic blood pressure). DSE tests were reported as positive, negative, or inconclusive. A test was deemed positive for inducible ischemia if new regional wall motion abnormalities were seen with pharmacologic stress. A negative DSE was defined as reaching >85% of the maximal predicted heart rate [(220 − age) × 0.85]13 with a normal increase in the LV ejection fraction and no regional wall motion abnormalities. An inconclusive test showed no regional wall motion abnormalities but failed to reach 85% of the maximum predicted heart rate.
Adverse CV events were assessed from operative reports, hospital records, discharge summaries, and the 4-month posttransplant follow-up visit records. Primary endpoints were death from cardiac cause, new myocardial infarction (MI),14 new heart failure,15 and asystole or symptomatic ventricular tachycardia requiring urgent treatment. Adjudication of CV events was performed by 2 board-certified staff cardiologists.
Descriptive statistics were performed and expressed as means with standard deviations. Continuous measures were evaluated with receiver operating characteristic analysis (ROC). For each continuous measure, we smoothed the ROC curve and identified the cutoff level that yielded the highest likelihood ratio for prediction of an event. A multivariable prediction model was created with multiple logistic regressions. Confidence intervals for incidence were calculated with the exact binomial method. We compared the incidence of adverse CV between the levels of the predictive models, and the statistical significance was calculated with the Fisher exact test. Statistical computations were performed with SAS software version 9 (SAS Institute, Cary, NC).
A total of 284 OLTs were performed for ESLD from June 1999 to August 2005 at our institution. Of these, 157 patients underwent DSE prior to their OLT. Table 1 shows the baseline demographic, clinical, and CAD risk factors of the study population. The average age was 55 years, 71% of the patients were male, and 94% of the patients were white. Ten percent had a history of diabetes mellitus, 20% had a history of hypertension, and 58% had current or previous tobacco use. The most common causes for liver failure were hepatitis C (48%), hepatocellular carcinoma (13%), cryptogenic cirrhosis (10%), and alcoholic cirrhosis (10%). The mean MELD score at the time of OLT was 21 (range 7 to 44).
Table 1. Baseline Characteristics of the Orthotopic Liver Transplant Patients
Abbreviation: MELD, Model for End-Stage Liver Disease.
54.5 ± 7.5
History of diabetes
History of smoking
MELD score at time of transplant
20.9 ± 7.4
Etiology of liver failure
Cryptogenic idiopathic cirrhosis
Primary biliary cirrhosis
Primary sclerosing cholangitis
No patient who went on to OLT had a positive DSE. Ninety-nine patients (63%) had a negative DSE, whereas 58 patients (37%) had an inconclusive DSE because of the failure to reach 85% of their maximum predicted heart rate. Fifty percent of the inconclusive test patients (29 of 58) had taken a beta blocker within 12 hours of their DSE test. Nine of these were referred for coronary angiography, and none had significant CAD. The other 49 patients with an inconclusive DSE were reviewed by the liver transplant group and went on to OLT without further cardiac testing.
As shown in Table 2, 10% of the patients (16 of 157) with a negative or inconclusive DSE had an adverse CV event during surgery or within 4 months after their OLT. Nine of these 16 (56%) patients did not reach their target heart rate during their DSE, demonstrating chronotropic incompetence. Five of the 9 patients were on a beta blocker at the time of the DSE, which may have contributed to their blunted heart rate response. Of the remaining 4 patients, 3 (patients 9, 10, and 11) received maximal dobutamine and atropine but achieved only 81% to 82% of their maximal predicted heart rate, whereas 1 patient had his DSE stopped at 20 μg/kg/minute dobutamine because they developed a large LV outflow tract gradient. Seven of the 157 patients with a negative DSE had an adverse CV event for a false negative rate of 4%. In these patients, the peak rate pressure product ranged from 11,397 to 18,602 (mean 15,529).
Table 2. Beta Blocker Use, DSE, and Cardiovascular Event Data
Peak Dobutamine (μg/kg/minute)/Atropine (mg) Dose
% Maximum Predicted Heart Rate Achieved
Peak Rate Pressure Product
Abbreviations: A, asystole; C, cardiomyopathy; CK-MB, creatine kinase muscle-brain; DSE, dobutamine stress echocardiography; IO, intraoperative; MELD, Model for End-Stage Liver Disease; MI, myocardial infarction; NA, not applicable; PH, post-hospitalization; PO, postoperative; V, ventricular arrhythmia.
Developed hypotensive response (systolic blood pressure < 100 mm Hg) to stress despite fluid resuscitation.
Developed marked left ventricular outflow tract obstruction during stress.
In the 16 patients with adverse CV events, 9 (56%) occurred intraoperatively, including both deaths. Six of the 9 events occurred during or within minutes of donor liver reperfusion, including 2 patients with asystole and 5 with ventricular tachycardia or fibrillation. The remaining intraoperative events were a small MI and 2 episodes of ventricular tachycardia/fibrillation occurring before donor liver reperfusion. Five postoperative events occurred, included 2 MIs, 1 new heart failure, and 2 episodes of ventricular tachycardia. After discharge, 2 patients had small non–Q-wave MIs. Table 3 shows the available serum electrolyte and acid-base data closest to the adverse CV event.
Table 3. Serum Electrolyte and Acid-Base Data at the Time of Adverse Cardiovascular Events
Abbreviations: A, asystole; C, cardiomyopathy; IO, intraoperative; MI, myocardial infarction; NA, not applicable; PH, post-hospitalization; PO, postoperative; V, ventricular arrhythmia.
Two patients had CV events that led to cardiac death. The first patient (patient 1, Tables 2 and 3) had an inconclusive DSE and developed a serious ventricular arrhythmia during reperfusion of the transplanted organ from which he initially recovered. Within hours after leaving the operating room, they developed refractory pulseless electrical activity. The second patient (patient 2, Tables 2 and 3) had a negative DSE but was quite ill on admission with cachexia, chronic active hepatitis C, cryoglobulinemia, renal failure requiring hemodialysis, cutaneous vasculitis, and chronic hypotension. At the time of attempted OLT, epinephrine and norepinephrine were needed intraoperatively to maintain hemodynamic stability. During reperfusion of the transplanted liver, they became profoundly hypotensive and suffered a fatal cardiopulmonary arrest.
Of the 5 patients with MIs, 1 occurred intraoperatively, 2 were in the postoperative period, and 1 occurred after hospital discharge. All were documented by small increases in troponin and the creatine kinase muscle-brain relative index and were classified as small non–ST-elevation myocardial events. Thus, of the 16 adverse cardiac events, all but 2 (88%) occurred before hospital discharge
The risk of an adverse CV event in the OLT patients was most strongly associated with 3 variables: (1) the percentage of maximal predicted heart rate achieved during DSE, (2) the MELD score, and (3) the maximal peak rate pressure product during DSE. Table 4 summarizes the risk of an adverse cardiac event above and below cutoff values derived by ROC curves within 4 months after OLT among the study patients. The 4-month risk of adverse CV events was 22% among patients who failed to achieve 82% of the maximum predicted heart rate during DSE versus 6% among patients that achieved or exceeded this value, 17% among patients that did not exceed a peak rate pressure product during DSE of >16,333 versus 5% among patients who did, and 20% for patients with a MELD score of >24 points versus 6% among those with lesser scores.
Table 4. Risk of Cardiovascular Events Occurring Above and Below Cutoff Values
95% Confidence Interval
95% Confidence Interval
Abbreviation: MELD, Model for End-Stage Liver Disease.
Gender (positive if female)
Age (years, positive if >58)
MELD score (positive if >24
Systolic blood pressure at rest (positive if >110)
Peak rate pressure product (positive if >16,333)
Achieved maximum heart rate (positive if >135)
% Maximal heart rate achieved (positive if >0.824)
History of hypertension
History of smoking
Figure 1 depicts the Kaplan-Meier incidence of CV events in patients who achieved <82% of their maximum predicted heart rate during DSE versus those who exceeded this value. A multivariate model calculated from the maximum achieved heart rate (MAHR) and the MELD score (result = 3.78 + 0.07 MELD − 0.05 MAHR) showed a 47% risk among patients with a positive score (>0) versus only 6% risk among patients with a negative score (<0; P < 0.001).
In the present study examining preoperative DSE testing in our OLT patients, 3 variables predicted increased risk for serious adverse CV events up to their 4-month postoperative visit. These were inability to achieve >81% of the maximum predicted heart rate during DSE, a MELD score of >24 at the time of transplantation, and a peak rate pressure product during DSE of <16,332. A multivariable model combining the DSE maximum heart rate and the MELD score was the most powerful predictor of a serious CV event (47% versus 6%, P < 0.001). Because both variables are easily obtained as part of a pre-OLT evaluation, they could, if confirmed to have similar predictive value in other studies, have wide clinical applicability for CV risk stratification in the ESLD population. Our results further demonstrate that failure to achieve a peak rate pressure product > 16,333 during DSE in conjunction with an elevated MELD score or a submaximal heart rate response (<82%) also helps define patients who are at increased risk for serious CV events.
Our results agree with, and expand on, several previous studies that support DSE as a valuable risk stratification test for potential OLT candidates.7, 8, 16–18
As observed in other studies,6, 8, 17, 18 the results confirm the high negative predictive value of a normal DSE (96%). The excellent prognosis of this finding has also been shown in patients undergoing vascular surgery procedures,19–21 a group at high risk for CV events. Previous reports in other non-ESLD patient populations19, 20, 22, 23 also demonstrate that the maximal percentage of the predicted heart rate and peak rate pressure product during DSE are predictive of CV events. The results from both populations suggest that obtaining the highest possible heart rate (85%-100% of the predicted maximum) should be the major focus of preoperative DSE testing.
In the 37% of our ESLD patients who did not reach their target heart rate during DSE, 50% of them were on a beta blocker. This was in all cases prescribed to reduce the risk for variceal bleeding rather than for a CV indication. Because serious arrhythmias during DSE are rarely seen (none in the current study), withholding beta blockers before DSE appears safe and may reduce the number of inconclusive tests due to submaximal heart rates.
Some ESLD patients who are not on beta blockers do not reach their target heart during DSE even after maximal dobutamine and atropine doses (18% in this study). This is termed “chronotropic incompetence” and has been reported in other DSE studies on ESLD patients with a reported incidence as high as 50%.8 The 29 patients with chronotropic incompetence in this study had an average age of 51 years and an average MELD score of 25. In non-ESLD patients undergoing stress testing, chronotropic incompetence has been shown to be a strong and independent predictor of death, even after angiographic severity of CAD is accounted for.24 Additional investigation is needed to determine how ESLD affects DSE heart rate response, how it confers increased CV risk for arrhythmias and MI, and whether this finding is corrected after liver transplantation.
In our study, the most serious CV events were asystole and ventricular arrhythmias, which usually occurred intraoperatively and at the time of donor liver reperfusion. These happened without obvious evidence of myocardial ischemia, raising questions about the basic premise of preoperative DSE testing in ESLD patients. There are many factors not related to myocardial ischemia that could contribute to intraoperative arrhythmias, such as electrolyte or acid-base disturbances, electrocardiogram (ECG) QT prolongation, increased T-wave dispersion, or poor donor liver quality. An important unresolved issue from the current retrospective study is the strength of the relation between a preoperative DSE and the risk for intraoperative arrhythmias versus other possible factors. This can be answered only by a prospective study that includes both DSE and intraoperative variables.
The current results also show that a secondary goal during DSE should be the achievement of a peak rate pressure product of >16,500. This target is commonly reached in patients without ESLD undergoing DSE but is more difficult in pre-OLT patients because of their low systemic vascular resistance and low blood pressures. A low rate pressure product as a risk factor has previously been reported in the pre-OLT population and is more common in patients with MELD scores ≥ 25.16 Table 2 shows that 10 of 16 patients (62%) with an adverse CV event in the current study had a peak rate pressure product of <16,500, and only 3 of 16 patients (patients 3, 4, and 7) with an adverse event reached the target maximal heart rate and a rate pressure product > 16,500. Thus, the ideal DSE study would achieve both a maximal heart rate of 85% to 100% of the predicted rate and a rate pressure product of >16,500.
CV risk stratification with DSE in ESLD patients has several unique challenges. The marked liver dysfunction leads to metabolic abnormalities that result in generalized arterial vasodilatation, low systemic vascular resistance, and low blood pressures despite an increased cardiac output. The heart is often tachycardic, with a hyperdynamic left ventricle, increased ejection fraction, a low end-systolic volume, and near systolic cavity obliteration. Dobutamine can cause further LV systolic cavity obliteration, which sometimes results in a precipitous drop in blood pressure in mid-study, presumably because of neurocardiogenic reflexes; as was seen in 4 of our patients who had adverse CV events (patients 1, 8, 14, and 16, Table 2). Practical considerations to avoid systolic cavity obliteration and hypotension include the administration of intravenous fluids before the study and elevating the legs to increase venous return. If LV cavity obliteration and intracavitary gradients are confined to the apex and mid-ventricle, administering atropine often attenuates the reflex hypotension, and the blood pressure will recover so the DSE can be completed. If LV outflow tract obstruction is present, giving atropine is contraindicated because a faster heart rate and smaller LV cavity may result in worsening systolic anterior motion of the mitral valve and dangerously high LV outflow tract gradients. In these cases, the study must be aborted for safety (patients 12 and 13, Table 2).
This is a retrospective study whose conclusions need further verification using prospective methodology. Acid-base and electrolyte data nearest the time of the serious intraoperative CV events in this study were gathered (Table 3) but were sometimes incomplete, and no ECG data were available. Future studies need more data gathering during the critical surgical period along with an assessment of liver donor quality.
The absolute number of serious adverse CV events was relatively low in comparison with prior reports on OLT patient groups. This may reflect the “hard” CV event endpoints used in the current report or possibly elimination at our institution of patients with pulmonary hypertension or a prior history of CAD because of their excessive mortality after OLT.
The issue of what to do with inconclusive DSE studies is an important question. Coronary angiography may represent the best answer until more data are available. Yet, the 9 patients who underwent coronary angiography who had an inconclusive DSE in our study and the majority in the literature with8, 17, 18 or without a positive DSE usually did not have significant epicardial stenosis. Only 5 of 16 events in this study were MIs, and all were very small by cardiac enzymes; this suggested minor plaque rupture or demand ischemia as their etiology. There was also no evidence that severe ischemia with ECG changes precedes the serious ventricular arrhythmias seen during an operation or in the postoperative period. It will be up to future studies to better define the links between adverse CV events (especial intraoperative) and chronotropic incompetence, angiographic findings, or any of several potentially electrical destabilizing factors that may occur during the surgery. Whether an important aspect of stress testing for CV risk stratification is the rate-related shortening of the diastole and coronary perfusion time is unknown, but it is one of the more interesting possible relations between DSE findings and OLT patient outcomes.
The patients were not followed past their 4-month follow-up. How to avoid future CV events after this time is unknown but is more likely related to control of conventional risk factors rather than a preoperative risk stratification test.
In conclusion, the percentage of the maximum predicted heart rate and the rate pressure product achieved during DSE, together with the MELD score at the time of transplant, are associated with adverse CV events at the time of and up to 4 months after liver transplantation. We recommend that a preoperative DSE protocol strive to achieve 85% to 100% of the maximum predicted heart rate and a rate pressure product > 16,500. A prospective study to confirm the results of this study, as well as further investigation into the relationship between an adequate DSE and intraoperative CV events, appears warranted.
The authors express their appreciation to Ms. Christine A. Grumeretz for her assistance with information database retrieval and maintenance.