Comparison of Coronary Artery Calcium Scoring with Dobutamine Stress Echo for Detection of Coronary Artery Disease Before Liver Transplantation

Background

Major adverse cardiovascular events have a significant impact on survival after orthotopic liver transplantation (OLT). While cardiovascular mortality after liver transplantation has declined, it remains a leading cause of early and late mortality after transplantation [1]. The presence of pre-existing coronary artery disease (CAD) is one such predictor of early cardiovascular mortality [1]. Therefore, it is essential to stratify the risk for ischemic CAD in patients listed for OLT. The current American Association for the Study of Liver Disease guidelines recommend that patients with traditional risk factors for CAD, including diabetes, smoking, and obesity [2], should undergo noninvasive testing with stress echocardiography, and subsequently with coronary angiography (CAG) if CAD cannot be definitively excluded (class IB recommendation) [3]. Dobutamine stress echocardiography (DSE) is a widely used screening test for preoperative risk assessment before OLT. Existing literature suggests that the sensitivity of DSE for CAD is low in liver transplant patients, ranging from 13% to 41% [4–6]. Furthermore, patients with cirrhosis have hyperdynamic ventricles with poor chronotropic response as well as vasodilated circulatory systems with a low systemic vascular resistance [7]. Hence, they may not meet the target heart rate or generate an appropriate heart rate response to dobutamine required for the pharmacologic stress test to be considered diagnostic. As a result, many patients with cirrhosis have a nondiagnostic DSE test and subsequently need invasive CAG for risk stratification, which has its own associated risks of bleeding and contrast-induced nephropathy [8].

In patients with end-stage liver disease, the coronary artery calcium score (CACS) has been shown to have a positive correlation with many cardiovascular risk factors, such as family history of cardiovascular disease, systolic/diastolic blood pressure, and metabolic syndrome [9]. Additionally, some studies, including a study by Agatston et al, which defined a “high” CACS as ≥400 AS, found that that a high CACS is associated with higher rates of early postoperative cardiac complications after liver transplantation, including nonfatal myocardial infarction, arrhythmia, heart block, and cardiac death [10]. Patients with higher CACSs are also more likely to have significant CAD (defined as > 50% occlusive disease) [11].

We postulated that the CACS may be a useful tool for cardiovascular risk stratification prior to OLT and be a good alternative test in many patients when DSE is nondiagnostic or inconclusive. The CACS offers several advantages, including that it is noninvasive, does not require contrast administration, and does not require patients to reach a target heart rate [12]. While the sensitivity and specificity of CACS and DSE have been assessed independently in OLT candidates, they have not been directly compared against each other in the same sample of patients for their ability to detect CAD against the criterion standard test, CAG. The goal of this study was to compare the diagnostic accuracy of DSE with CACS for the detection of significant CAD in patients undergoing OLT.

Material and Methods

STATISTICAL ANALYSIS:

Continuous data were reported as medians (interquartile range) and means (standard deviation) based on the distribution of the data. Categorical data were reported as a count (proportion). We used the Wilcoxon rank-sum test and t test to compare medians and means, respectively. Pearson’s chi-squared test was used for comparing categorical data. Patients missing either DSE or CACS were excluded. The sensitivity and specificity of the CACS was calculated at various CACS cutoffs (≥100, 200, 300, and 400 Agatston score [AS]), and the cutoff with the highest area under the receiver operating characteristic (ROC) curve (AUC) was used. For the head-to-head comparison, the sensitivity and specificity of DSE and CACS were compared using CAG as the criterion standard test among patients who had both a diagnostic DSE and CT scan that could be used for the CACS. For the direct comparison between the 2 tests, the sensitivity and specificity of DSE and CACS were compared using a 2×2 contingency table stratified by the actual CAD outcome. The sensitivity and specificity of DSE and CACS were statistically significantly different if the confidence interval did not include 0. Additional secondary analyses were conducted to evaluate the diagnostic accuracy of DSE and CACS compared separately with CAG for all patients who had a diagnostic result for either test. All sensitivity/specificity analyses were done using STATA SE 13 (STATA Corp, College Station, TX, USA). ROC analysis was performed using CACSs for their ability to detect presence of obstructive CAD using SPSS version 25 (IBM Corp, Armonk, NY, USA). No adjustments were performed for multiplicity. A P value <0.05 was considered statistically significant.

Results

DSE VS CACS: HEAD-TO-HEAD COMPARISON:

We compared the sensitivity, specificity, positive predictive value, and negative predictive values of DSE and CACS for CAD in patients who had both studies (n=36). A cutoff CACS of ≥100 was used for comparison. The CACS has a sensitivity of 71.4% for the detection of obstructive CAD, while DSE had a sensitivity of 57.1%. The specificity of DSE was significantly higher than that of the CACS when a cutoff of ≥100 AS in the CACS was used (89.7% and 62.1%, respectively). However, when the cutoff for the CACS was increased to ≥400 AS, the specificity increased to 86.2%, while the sensitivity decreased to 57.1% (Table 2).

DSE AND CACS SUBSETS (ALL PATIENTS WITH VALID RESULT FOR EITHER TEST):

The sensitivity and specificity of DSE and the CACS were calculated separately for all patients in the sample with a valid result for either test. In the DSE subset (n=73), the sensitivity and specificity for detecting obstructive CAD were 30.8% and 85%, respectively. In the CACS subset (n=133), the sensitivity was 80.0% at a cutoff of ≥100 AS and 55.0% at a cutoff of ≥400 AS. The specificity of the CACS was 62.8% at ≥100 AS, which improved to 79.6% at the higher cutoff of ≥400 AS (Table 3). Because the sensitivity and specificity were not calculated on the same patients for each test in a head-to-head fashion, it was not possible to determine whether this difference was statistically significant.

The DSE and CACS subsets were also used to calculate the positive and negative predictive values for each of the tests. Using a threshold of ≥100 AS, the positive predictive value of the CACS was low at 28.2%, but the specificity was high at 93.4%. The positive predictive value of DSE was low at 28.2%, but the negative predictive value was high at 83.9%. However, the assessment of the predictive value of DSE did not include all of the patients owing to the high number of nondiagnostic studies.

ROC analysis showed that a CACS ≥100 AS was a robust predictor of the presence of obstructive CAD (AUC 0.76±0.06; 95% CI 0.66–0.87, P<0.001) (Figure 3).

Discussion

LIMITATIONS:

A major limitation of our study is the potential bias introduced when considering that patients referred for CAG may have been at a high risk for CAD. It is important to note that the inclusion criteria for the study was all patients undergoing OLT evaluation who had a CAG prior to the transplant. This included patients who either had a nondiagnostic DSE or other indications for CAG as listed above. These patients could have had a higher risk of obstructive CAD than a general cohort of patients undergoing preoperative cardiac evaluation for OLT. The physicians who referred these patients to either CAG or DSE likely had a reasonable clinical suspicion that these patients had CAD, which may have confounded our results. This is a possible explanation for why in the head-to-head comparison, we found a higher sensitivity of DSE for obstructive CAD (57.1%) than other previous studies (Table 4). The disparity between the sensitivity of DSE of our study in the head-to-head comparison and that described in previous literature may have also been due to random error related to the small sample size, since our subset analysis of all patients who had either study found a sensitivity of 30.8%. Furthermore, our comparison of DSE and CACS subsets (all patients with a valid result for either test) compared test results in different patients. This analysis may have been affected by unknown confounders since the test subjects were different patients with different baseline demographics and comorbidities.

Another limitation of this study is that the CT scans used for the calcium scoring were not originally intended for calcium scoring. Hence, the calcium scores may have been overestimated due to motion artifact. Additionally, the Agatston method was originally developed using electrocardiography-gated CT scans [12]. Our CT scans were non-gated. However, more recent research has shown that the CACS on non-gated CT scans correlated well with the Agatson score on gated CT scans [22].

Furthermore, there were still many patients with obstructive CAD who had calcium scores less than the various thresholds for what would be considered a “positive” result. Although the average CACS for patients without CAD (69.4 AS) and with obstructive CAD (769.8) were mostly divergent, there were still outliers in each category. For example, even some patients with obstructive CAD had calcium scores of 0. These patients with clinically significant CAD would be missed if calcium scoring alone was used to risk stratify patients and determine which patients should go on to invasive CAG. Because of this, it is important to consider the presence of clinical risk factors such as age, diabetes, hypertension, obesity, and family history of CAD when interpreting the CACS and using it to guide clinical decision making. For instance, patients with high risk based on clinical risk stratification may be preferentially evaluated using CAG, while patients with low to moderate risk could be further screened with the CACS (with DSE/CAG as the next step).

Conclusions

In conclusion, the CACS has comparable sensitivity for the detection of obstructive CAD compared with DSE at all AS thresholds investigated in this study. ROC analysis showed that the CACS was a satisfactory test for detecting the presence of underlying obstructive CAD. DSE was nondiagnostic in an unacceptably high number of patients, restricting its utility as a screening test for CAD. Therefore, we conclude that calcium scoring may be a suitable alternative method for preoperative cardiac risk stratification prior to liver transplantation, at least in patients with low to moderate risk or when DSE is nondiagnostic. Until more prospective research is available, this retrospective study provides some insight into the role of the CACS as an alternative to DSE for preoperative cardiac risk assessment prior to liver transplantation.