Author + information
- Received March 29, 2018
- Revision received October 15, 2018
- Accepted October 19, 2018
- Published online January 16, 2019.
- Anisiia T. Doytchinova, MDa,b,∗ (, )
- Thomas D. Feigenbaum, MSa,
- Roja C. Pondicherry-Harish, MDa,
- Peter Sepanski, PhDa,
- Deborah Green-Hess, BSa,
- Harvey Feigenbaum, MDa and
- Stephen G. Sawada, MDa
- aKrannert Institute, Indiana University School of Medicine and IU Health, Indianapolis, Indiana
- bDivision of Cardiovascular Health and Disease, University of Cincinnati Medical Center, Cincinnati, Ohio
- ↵∗Address for correspondence:
Dr. Anisiia Doytchinova, University of Cincinnati Medical Center, 231 Albert Sabin Way, MLC 0542, Cincinnati, Ohio 45267.
Objectives This study determined the test performance of dobutamine stress echocardiography (DSE) in end-stage liver disease (ESLD).
Background The reported sensitivity of DSE in ESLD has been variable.
Methods Data from 633 ESLD patients who had coronary angiography within 6 months after DSE was analyzed.
Results The prevalence of coronary arterial disease (CAD) (≥70% stenosis by quantitative angiography) was 12% (74 of 633 patients). DSE sensitivity was 24% (17 of 72 patients), and specificity was 90% (503 of 559 patients). The positive and negative predictive values were 23% (17 of 73 patients) and 90% (503 of 558 patients), respectively. Stratifying the cohort into low-, intermediate-, and high-risk CAD groups yielded sensitivities of 0%, 21%, and 32%, respectively. Independent predictors of an accurate ischemic DSE result included left ventricular internal dimension at end-diastole (LVIDd) >4.8 cm and assigning ischemia based on tardokinesis or lack of low-to-peak dose hyperkinesis (p < 0.05 for all). DSE sensitivity was 38% in LVIDd >4.8 cm versus 13% with LVIDd ≤4.8 cm (p = 0.013). The sensitivity was 67% when tardokinesis or lack of hyperkinesis was considered abnormal versus 15% (p < 0.001) for readings that did not consider tardokinesis or lack of hyperkinesis abnormal. There was a higher frequency of cardiac events in patients with significant CAD who had abnormal (45%) versus normal (18%) DSE (p = 0.01).
Conclusions The sensitivity of DSE in ESLD was low. DSE sensitivity was higher for those with larger cavity dimension and when tardokinesis or lack of hyperkinesis was considered abnormal. An abnormal DSE in those with significant CAD was associated with worse outcome.
- dobutamine stress echocardiogram
- end-stage liver disease
- left ventricular internal dimension at end-diastole
- rate−pressure product
The reported prevalence of coronary arterial disease (CAD) in patients with end-stage liver disease (ESLD) undergoing transplantation evaluation ranges from 18% to 27% (1). Liver transplantation guidelines advise using stress echocardiography as an initial screening test in patients with risk factors (2). Dobutamine stress echocardiography (DSE) is commonly used because of the inability of ESLD patients to adequately exercise. The reported sensitivity of DSE in ESLD has varied widely (from 13% to 100%), with early small studies reporting higher sensitivity (3,4). More recent studies have reported substantially lower values (5) with 1 meta-analysis reporting a sensitivity of 32% (6).
In the general population, DSE sensitivity is reduced in individuals with smaller left ventricular cavity size, baseline hyperkinesis, and reduced wall stress during testing (7–9). The sensitivity of DSE in the ESLD population may be lower than desired because of baseline hyperkinesis and peripheral vasodilation, which predisposes these subjects to declines in blood pressure and wall stress with further pre-load and afterload reductions that occur with dobutamine.
Our aim was to determine DSE sensitivity and specificity in a large ESLD cohort, identify factors that influence diagnostic performance, and to assess its prognostic value in patients with significant CAD.
This study was approved by the Indiana University Institutional Review Board. We retrospectively analyzed data from 633 ESLD patients who underwent angiography within 6 months after DSE between July 19, 2006 and February 1,2017. Patients were referred for angiography in cases of positive or equivocal DSE or with a negative DSE in the presence of ≥1 risk factors from a practice model that showed a decrease in post-operative myocardial infarction and all-cause mortality after liver transplantation (10). Based on a query of our echo database, we estimated that approximately 85% of all liver transplantation candidates who underwent DSE during the study period also underwent angiography.
The dobutamine protocol was previously described, and standard endpoints were used. Target heart rate was 85% of the age-predicted maximum. Beta-blockers were withheld for 3 days (11). Because of higher than expected incidences of hypotension and failure to achieve target heart rate, pre-treatment of patients using 0.5 to 1 mg of atropine was initiated in 2009, along with dosing of additional atropine in the early stages (≤20 μg/kg/min) of the test.
Images were acquired in parasternal long- and short-axis and apical 4- and 2-chamber views at baseline, low dose, peak dose, and early recovery. In 2015, an apical long-axis view was added to the imaging protocol. DSE examinations were visually assessed by experienced Indiana University faculty. Studies with no stress-induced wall motion abnormalities were considered nonischemic. Studies with new or worsening wall motion abnormality, reduced global ventricular systolic function compared with baseline, or cavity dilatation at peak stress were designated as ischemic. One investigator uniformly applied the 2007 American Society of Echocardiography guidelines for interpretation of DSE examinations and noted the presence of delayed onset or termination of contraction (tardokinesis) and lack of low-to-peak dose hyperkinesis (either regional or global) as abnormal. These interpretations were also considered ischemic (12) and are subsequently referred to as guideline-directed interpretations. Hyperkinesis at a peak dose relative to a low dose was the expected response in patients without obstructive disease. Visually assessed improvement in global and regional contractility from low-to-peak dose and reduction in end-systolic volume from low-to-peak dose were considered the expected normal responses. Lack of improvement in global or regional contractility from low-to-peak dose and/or lack of reduction in end-systolic volume from low-to-peak dose were considered evidence of obstructive disease (13).
Studies with fixed wall motion abnormalities (11 studies) were considered true negative in cases in which no ischemia was expected based on angiography or excluded if clear correlation was difficult. In our laboratory, the interobserver reproducibility of DSE interpretation among experienced investigators is 89%, without consideration of lack of hyperkinesis or tardokinesis (11). The interobserver agreement between the investigator who used guideline-directed interpretation and an investigator who used standard criteria for assigning ischemia was investigated before study enrollment, and found to be 85%. Electrocardiography (ECG) was considered positive for ischemia in cases in which ≥1 mm horizontal or downsloping ST-segment depressions were observed 80 ms after the J-point in any 2 consecutive leads during stress or recovery.
Because of the extended study period and equipment improvement overtime, we compared DSE sensitivity before and after January 1, 2015. This period corresponds to the introduction of the Vivid E-95 (General Electric, Waukesha, Wisconsin) and Epiq (Phillips, Bothell, Washington) machines in our laboratory.
Angiograms were visually assessed by an experienced interventional cardiologist, and quantitative coronary analysis was subsequently performed in all patients with visual stenosis ≥40% by a blinded investigator. Significant CAD was assigned at the ≥70% and ≥50% stenosis levels for all vessels, except for the left main artery, where lesions ≥50% were considered significant.
Statistical analysis was performed using SPSS version 23 (IBM, Armonk, New York) or KeLabs (Indianapolis, Indiana). Continuous variables were expressed as mean ± SD, and categorical variables were expressed as numbers and percentages. Analysis of variance and the chi-square test were used to analyze differences in means and proportions, respectively, among the groups.
The population was divided into low, intermediate, and high probability groups for CAD to determine if the performance of DSE varied according to the probability of disease. We used combined historical thresholds and a method described by Kotler and Diamond (14). Multivariate logistic regression analysis was performed on all available clinical variables to determine independent predictors of significant CAD. Significant CAD was defined as the combined group of patients with known CAD before their transplantation evaluation and those with ≥50% stenosis by quantitative coronary angiography. Clinical characteristics found in the analysis and used to define the risk groups included: age; family history of CAD; smoking, presence of nonalcoholic steatohepatitis; history of hyperlipidemia, diabetes mellitus, and hypertension; therapy with statins, aspirin, beta-blockers, and angiotensin-converting enzyme inhibitors; and ejection fraction. Patients with known CAD or a probability of >54% were assigned to the high-risk group; those with a probability of 10% to 54% were assigned to the intermediate-risk group, and those with <10% probability were assigned to the low-risk group. These thresholds were derived from the Kotler and Diamond method and based on the previously described predictors of CAD (14).
Logistic regression was also performed to determine variables that were independently associated with an ischemic DSE result in patients with CAD, defined as ≥70% stenosis of any vessel. A threshold value of left ventricular diastolic dimension (LVIDd) for analysis was determined based on the cutoff value between the middle and top tertiles of the population and based on receiver-operator characteristic curve analysis. Values of p < 0.05 were considered significant.
Follow-up by medical record review was obtained in patients with significant CAD for early revascularization procedures and cardiac events, including admission for angina that required late revascularization, left-sided heart failure, myocardial infarction, witnessed cardiac arrest, and cardiac death. Kaplan-Meier analysis with the log-rank test was used to compare cardiac event-free survival in those with abnormal and normal DSE.
There were 633 patients (age 59 ± 7 years) with baseline characteristics listed in Table 1. Fifty-five subjects had repeat DSE and angiographic examinations. The prevalence of CAD, as defined by ≥70% and ≥50% angiographic stenosis in any vessel, was 12% (74 of 633 patients) and 17% (110 of 633 patients), respectively.
Diagnostic performance of DSE
The stress testing results are shown in Table 2. Ejection fraction was 65 ± 7%, and the peak rate−pressure product was 18,087 ± 4,312. In the group that received atropine during rest and stress, 85.0% achieved the target heart rate versus 73.8% of those who received atropine only during stress (p = 0.05) (Table 2). DSE and angiographic results are shown in Table 3. The sensitivity for ≥70% angiographic stenosis of any vessel was 24%. The specificity was 90%. The sensitivity of DSE remained low for multivessel disease, when a ≥50% stenosis threshold was used, or when repeated DSE and angiograms were included. The combined sensitivity for a positive stress ECG or echocardiogram was 28% (20 of 72 patients) with a specificity of 86% (480 of 559 patients). The diagnostic performance of DSE according to the estimated probability of CAD is shown in Table 4. The prevalence of significant CAD, as defined by ≥70% stenosis, was 3%, 11%, and 29% in the low-, intermediate-, and high-risk groups, respectively. DSE sensitivity was 0%, 21%, and 32%, respectively.
Because of the extended duration of the study, and the potential impact of changes in the imaging and stress testing protocols over time, DSE sensitivity in the first two-thirds of the study population was compared with the sensitivity in the latter third of subjects enrolled. The latter third of subjects were enrolled beginning in 2015. DSE sensitivity was 16.3% (8 of 49 patients) before January 1, 2015 versus 39.1% (9 of 23 patients) after January 1, 2015 (p = 0.03).
Variables associated with DSE ischemia in patients with CAD
Variables included in the analysis for predictors of an accurate ischemic result in those with ≥70% stenosis are shown in Table 5. The value of 4.8 cm for LVIDd chosen for analysis represented the optimal threshold value for this parameter based on receiver-operator characteristic curve analysis, with a sensitivity of 0.53, specificity of 0.60, and an AUC of 0.60. The independent predictors included statin use, LVIDd >4.8 cm, and guideline-directed interpretations. Guideline-directed interpretations had the highest odds ratio [OR]. Peak rate−pressure product, achievement of 85% of the age-predicted maximal heart rate, and atropine pre-treatment were not predictors.
Guideline-directed interpretation improved sensitivity from 15% (9 of 60 patients) to 67% (8 of 12 patients) (p < 0.001) with a borderline significant decrease in specificity from 94% (408 of 436 patients) to 78% (95 of 122 patients) (p = 0.05). The sensitivity of DSE in those with LVIDd >4.8 cm was 38% (12 of 32 patients) versus 13% (5 of 40 patients) in those with LVIDd ≤4.8 cm (p = 0.013) and the specificity was 89% (170 of 190 patients) versus 90% (333 of 369 patients) (p = 0.77) (Figure 1).
Follow-up in patients with significant CAD
Follow-up was obtained for 78 patients with CAD for 19.8 ± 21.9 months. Fifty-six patients had normal DSE and 22 (15 ischemic, 3 mixed, 4 fixed defects) had abnormal DSE. Multivessel disease was present in 36% of patients (20 of 56 patients) with normal DSE and in 45% of patients (10 of 22 patients) with abnormal DSE (p = 0.42). The frequency of revascularization because significant CAD was found was similar in patients with normal DSE (79%; 44 of 56 patients) and abnormal DSE (82%; 18 of 22 patients) (p = 0.75). In those who were revascularized, 2 patients underwent coronary bypass grafting, and the remaining 60 patients underwent coronary stenting. The frequency of complete revascularization was similar in patients with normal DSE (70%; 39 of 56 patients) and abnormal DSE (63%; 14 of 22 patients) (p = 0.61). The reasons for no or incomplete revascularization included chronic total occlusion of a vessel with collateral flow supply, anatomy not amenable to intervention, smaller vessel supplying a limited area of myocardium, and co-morbidities that made a patient too high risk for transplantation.
Liver transplantation was limited to a minority of those with normal DSE (23%; 13 of 56 patients) and abnormal DSE (18%; 4 of 22 patients) DSE (p = 0.62). Cardiac events occurred in 18% of patients (10 of 56 patients) with normal DSE and in 45% of patients (10 of 22 patients) with abnormal DSE (p = 0.01). Five patients (2 normal with DSE, 3 with abnormal DSE) had events following liver transplantation. Early revascularization was performed in 85% (17 of 20 patients) of those who had subsequent events. There were 2 cardiac deaths, 5 infarctions, 4 heart failure admissions, and 3 admissions for angina that required late revascularization in the 10 patients with normal DSE and events. There were 2 cardiac deaths, 2 cardiac arrests (with successful resuscitation), 5 infarctions, 4 heart failure admissions, and 3 late revascularizations for angina in the 10 patients with abnormal DSE and events. Two patients with abnormal DSE and late revascularization died of complications related to the revascularization procedure but were not coded as cardiac deaths. There were 3 patients with abnormal DSE and stable ESLD who died of unknown causes. Cardiac event-free survival was significantly shorter in those with abnormal DSE compared with those with normal DSE (11.2 ± 12.6 months vs. 23.2 ± 23.8 months; p = 0.001).
In 633 ESLD patients with a low prevalence of CAD, DSE sensitivity was low (24%) and specificity was high (90%). The sensitivity varied according to the probability of disease and ranged up to 32% in the high-risk group. DSE sensitivity was significantly higher in subjects with LVIDd >4.8 cm, when tardokinesis or lack of hyperkinesis from low-to-peak dose were considered markers of ischemia, and in studies performed after January 1, 2015. Sensitivity was not improved by achievement of the target heart rate. Patients with CAD and an abnormal DSE had significantly worse outcome than those with CAD and a normal DSE.
Diagnostic performance of DSE
The reported sensitivity of DSE for CAD in pre-liver transplantation patients has varied widely from 13% to 100%. Early small studies reported higher sensitivity (75% to 100%) but also used chest pain or dyspnea during stress as a marker of test positivity (3,4). A more recent meta-analysis reported an average sensitivity of 32% (6). The study of Harinstein et al. (5), which showed a sensitivity of 13%, and our results suggest that DSE has low sensitivity in ESLD. Reasons for the lower than desired sensitivity are discussed in the following.
Effect of test verification bias on diagnostic performance
The sensitivity of stress echocardiography has been primarily investigated in populations subjected to test verification bias, in which most of the patients referred for angiography had an ischemic result. Roger et al. (15) showed that when adjusted for verification bias, the sensitivity of exercise echocardiography was 32% for women and 42% for men. In our population, the prevalence of obstructive CAD was lower than that in the general population, which was consistent with a reduced influence of test verification bias on the performance of angiography. Sensitivity increased to 32% in the group with the highest prevalence of CAD. Even this group had a lower prevalence of disease (29%) compared with previous studies that adjusted for referral bias (15,16). The study by Harinstein et al. (5) also had a low prevalence of disease.
Circulatory effects of ESLD and DSE sensitivity
ESLD patients typically have hyperdynamic left ventricular function, systemic vasodilation, and decreased central blood volume with reduced left ventricular cavity size (17). Pre-load and afterload decrease with dobutamine, which results in further reductions in cavity dimension and declines in blood pressure and wall stress. These changes can result in early termination of stress due to hypotension and inadequate increases in myocardial oxygen consumption to induce ischemia. Detection of ischemia may also be more difficult in a population with hyperdynamic function and primarily single-vessel disease, due to tethering of a small ischemic segment by adjacent vigorously moving, nonischemic myocardium. Previous studies in non-ESLD populations have shown that baseline hyperdynamic systolic function, small cavity size, and decreases in wall stress with DSE are associated with lower sensitivity (7,9). In our study, LVIDd >4.8 cm nearly tripled DSE sensitivity, which underscored the difficulty of induction and/or detection of ischemia in ESLD patients with smaller cavity dimensions.
Heart rate or rate−pressure product as indicators of adequate stress
During the study period, our laboratory transitioned to using atropine before and during the early stages of stress in an attempt to get more subjects to the target heart rate. Atropine pre-treatment showed a trend towards a higher proportion of subjects who achieved the target heart rate, but achievement of the target heart rate did not predict an ischemic result. Results of previous studies also suggested that achievement of the target heart rate was not an indicator of adequate stress in ESLD. A meta-analysis that showed a sensitivity of 32% for DSE included only those who achieved the target heart rate (6).
Peak rate−pressure product may be a better measure of myocardial oxygen consumption and adequacy of stress with DSE than heart rate. Krivokapich et al. (18) showed that rate−pressure product and myocardial oxygen consumption were closely related in subjects who underwent DSE (18). Rallidis et al. (19) demonstrated that in patients with CAD who underwent both DSE and exercise echocardiography, the extent of ischemic wall motion abnormalities and rate−pressure product was less with DSE than exercise despite achievement of higher heart rates with dobutamine. The rate−pressure product required to elicit ischemia was correlated with coronary stenosis severity by Garot et al. (20). Approximately one-third of patients with 70% to 80% stenosis required rate−pressure products >18,000 to demonstrate ischemia. The average peak rate−pressure product in our population was 18,087. Reported peak rate−pressure products in ESLD patients with negative DSE and adverse outcomes ranged from 11,397 to 18,602, which suggested that similar to our study, this level of stress might be suboptimal (21).
Impact of accounting for lack of low-to-peak dose hyperkinesis or tardokinesis
Recommendations for performance and interpretation of stress echocardiography published in 2007 suggested that tardokinesis and the lack of hyperkinesis were possible markers of ischemia. Since that publication, quantitative methods demonstrated that assessment of abnormalities in the timing of contraction might improve the sensitivity of stress echocardiography (22). Lack of hyperkinesis was correlated with the functional significance of CAD but was also described in patients with normal coronaries (23), which indicated that lack of hyperkinesis was not a specific sign of obstructive CAD. Despite these limitations, we demonstrated, in a population with baseline hyperdynamic function and lower disease prevalence, that designating ischemia based on the lack of low-to-peak dose hyperkinesis or tardokinesis added a significant increase in sensitivity, with only a modest decrease in specificity.
Change in diagnostic performance over time
Improvement in the sensitivity of DSE after 2015 corresponded to introduction of the latest generation of machines into the stress laboratory. Improved image quality, higher frame rates that allowed for better detection of tardokinesis, and the expanding use of contrast agents for endocardial border detection were possible contributors to improved sensitivity in studies performed after January 1, 2015. It was also during this time that we introduced a fifth view (the apical long-axis view), initially for the purposes of strain acquisition. This might have improved detection of left circumflex ischemia.
Outcomes in patients with significant CAD
Patients with CAD were at risk for cardiac events, and those with abnormal DSE had significantly worse outcomes than those with normal DSE. The association of abnormal DSE with poorer outcomes was consistent with the findings of numerous studies in patients without ESLD, but our findings were at variance with the results of recent studies that showed an inability of DSE to predict outcomes in liver transplantation patients and a low rate of ischemia and significant CAD in this population. Patel et al. (24) reported a 0.4% rate of DSE ischemia, and 0% rate of angiography in 460 patients who had transplantation. Snipelisky et al. (25) reported a 1.4% frequency of moderate to severe CAD in 2,010 liver transplantation patients, and ischemia with DSE occurred in only 1 of the 5 patients who subsequently had cardiac deaths. The ESLD cohort in our study had a higher risk than the populations in these previous studies that confined enrollment to those who underwent liver transplantation rather than all patients considered as potential candidates for transplantation. In our study, 11.5% of patients had ischemia, and 12% of patients had significant CAD. In our study, and that by Snipelisky et al. (25), cardiac risk was not reduced by coronary stents. The prevalence of diabetes was high in both studies; 55% of patients in our study and 45% in CAD patients in the study by Snipelisky et al (25). This suggested the presence of diffuse disease in both populations that increased cardiac risk that was not substantially improved by stenting of vessels with significant stenosis.
Our study had several limitations. The study was retrospective, and no standard criteria for interpretation of the examination were established before the study. However, there remained reasonable agreement between guideline-directed and standard interpretation for assessment of ischemia. We did not apply quantitative techniques for assessment of ischemia, and the prognostic significance of DSE was investigated only in those with significant CAD. We also did not test the performance of clinical predictors combined with DSE results for identification of CAD.
DSE sensitivity was low in an ESLD population that was less affected by test verification bias. Subjects in this study achieved a relatively low level of stress with a mean rate−pressure product of 18,000. DSE sensitivity was higher in those with LVIDd >4.8 cm, when the lack of low-to-peak dose hyperkinesis or tardokinesis was taken into consideration, and in studies performed with the latest technology and imaging protocols. Achievement of the target heart rate was not associated with improved sensitivity. Current guidelines should be modified to reflect the limitations of DSE. For diagnostic purposes, the test might be most useful in ESLD patients with larger cavity dimensions and with the highest probability of CAD. Quantitative methods might improve DSE sensitivity by enabling detection of subtle abnormalities in the timing and magnitude of contraction during stress that are more difficult to identify by visual assessment. Dopamine or norepinephrine might be alternatives to dobutamine as stress agents in ESLD patients to avoid hypotension and an inadequate rate−pressure product. Finally, although the sensitivity of DSE is suboptimal, the test had prognostic value in ESLD patients with CAD.
COMPETENCY IN MEDICAL KNOWLEDGE: Overall DSE sensitivity at an average rate−pressure product of 18,000 is low in pre-liver transplantation patients, but it is improved in patients with larger cavity dimensions, when lack of hyperkinesis and temporal abnormalities in contraction are considered markers of ischemia, and when the latest technology and imaging protocols are used. In patients with CAD, abnormal DSE is associated with worse outcomes. Our results raised the question of whether DSE should be routinely used as a diagnostic test for CAD in ESLD or be limited to certain higher risk subpopulations and those with larger cavity dimensions. Achievement of the target heart rate appeared to be an inadequate measure of sufficient stress, at least in ESLD patients. The rate−pressure product might be a better parameter for judging adequacy of stress.
TRANSLATIONAL OUTLOOK: Achievement of the target heart rate appears to be an inadequate measure of sufficient stress, at least in ESLD patients. The rate−pressure product may be a better parameter for judging adequacy of stress.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary arterial disease
- dobutamine stress echocardiography
- end-stage liver disease
- left ventricular internal dimension at end-diastole
- model for end-stage liver disease
- negative predictive value
- positive predictive value
- Received March 29, 2018.
- Revision received October 15, 2018.
- Accepted October 19, 2018.
- 2019 American College of Cardiology Foundation
- Garg A.,
- Armstrong W.F.
- Smart S.C.,
- Knickelbine T.,
- Malik F.,
- Sagar K.B.
- Yuda S.,
- Khoury V.,
- Marwick T.H.
- Sawada S.G.,
- Segar D.S.,
- Ryan T.,
- et al.
- Senior R.,
- Lahiri A.
- Roger V.L.,
- Pellikka P.A.,
- Bell M.R.,
- Chow C.W.,
- Bailey K.R.,
- Seward J.B.
- Rallidis L.,
- Cokkinos P.,
- Tousoulis D.,
- Nihoyannopoulos P.
- Garot J.,
- Hoffer E.P.,
- Monin J.L.,
- Duval A.M.,
- Pierard L.A.,
- Gueret P.
- Yip G.,
- Khandheria B.,
- Belohlavek M.,
- et al.
- Patel K.K.,
- Young L.,
- Carey W.,
- et al.
- Snipelisky D.,
- Levy M.,
- Shapiro B.