|Home | About | Journals | Submit | Contact Us | Français|
The objective was to determine the association of four clinical risk scores and coronary plaque burden as detected by computed tomography (CT) with the outcome of acute coronary syndrome (ACS) in patients with acute chest pain. The hypothesis was that the combination of risk scores and plaque burden improved the discriminatory capacity for the diagnosis of ACS.
The study was a subanalysis of the Rule Out Myocardial Infarction Using Computer-Assisted Tomography (ROMICAT) trial—a prospective observational cohort study. The authors enrolled patients presenting to the emergency department (ED) with a chief complaint of acute chest pain, inconclusive initial evaluation (negative biomarkers, nondiagnostic electrocardiogram [ECG]), and no history of coronary artery disease (CAD). Patients underwent contrast-enhanced 64-multidetector-row cardiac CT and received standard clinical care (serial ECG, cardiac biomarkers, and subsequent diagnostic testing, such as exercise treadmill testing, nuclear stress perfusion imaging, and/or invasive coronary angiography), as deemed clinically appropriate. The clinical providers were blinded to CT results. The chest pain score was calculated and the results were dichotomized to ≥10 (high-risk) and <10 (low-risk). Three risk scores were calculated, Goldman, Sanchis, and Thrombolysis in Myocardial Infarction (TIMI), and each patient was assigned to a low-, intermediate-, or high-risk category. Because of the low number of subjects in the high-risk group, the intermediate- and high-risk groups were combined into one. CT images were evaluated for the presence of plaque in 17 coronary segments. Plaque burden was stratified into none, intermediate, and high (zero, one to four, and more than four segments with plaque). An outcome panel of two physicians (blinded to CT findings) established the primary outcome of ACS (defined as either an acute myocardial infarction or unstable angina) during the index hospitalization (from the presentation to the ED to the discharge from the hospital). Logistic regression modeling was performed to examine the association of risk scores and coronary plaque burden to the outcome of ACS. Unadjusted models were individually fitted for the coronary plaque burden and for Goldman, Sanchis, TIMI, and chest pain scores. In adjusted analyses, the authors tested whether the association between risk scores and ACS persisted after controlling for the coronary plaque burden. The prognostic discriminatory capacity of the risk scores and plaque burden for ACS was assessed using c-statistics. The differences in area under the receiver-operating characteristic curve (AUC) and c-statistics were tested by performing the −2 log likelihood ratio test of nested models. A p value <0.05 was considered statistically significant.
Among 368 subjects, 31 (8%) subjects were diagnosed with ACS. Goldman (AUC = 0.61), Sanchis (AUC = 0.71), and TIMI (AUC = 0.63) had modest discriminatory capacity for the diagnosis of ACS. Plaque burden was the strongest predictor of ACS (AUC = 0.86; p < 0.05 for all comparisons with individual risk scores). The combination of plaque burden and risk scores improved prediction of ACS (plaque + Goldman AUC = 0.88, plaque + Sanchis AUC = 0.90, plaque + TIMI AUC = 0.88; p < 0.01 for all comparisons with coronary plaque burden alone).
Risk scores (Goldman, Sanchis, TIMI) have modest discriminatory capacity and coronary plaque burden has good discriminatory capacity for the diagnosis of ACS in patients with acute chest pain. The combined information of risk scores and plaque burden significantly improves the discriminatory capacity for the diagnosis of ACS.
Early triage of patients presenting to the emergency department (ED) with the chief complaint of chest pain suggestive of acute coronary syndrome (ACS) remains a diagnostic challenge. Elements of the chest pain history are associated with increased or decreased likelihoods of ACS.1 Sensitive troponin assays and initial electrocardiogram (ECG) provide important information in the initial decision.2–4 However, clinical history, a single set of cardiac biomarkers, and the initial ECG alone or in combination cannot identify the group of patients who can be safely discharged without further diagnostic testing. The percentage of patients who present to the ED with ACS and are inappropriately discharged is low (around 2%).5 However, discharged subjects have a two to three times increased risk of dying.5 Therefore, the standard of care for subjects with acute chest pain typically includes serial ECG and cardiac biomarkers followed by a stress test with or without imaging.6–9 The standard work-up decreases the risk of diagnostic errors. However, the complete “rule-out myocardial infarction” protocol is time-consuming and costly.
Clinical risk scores (e.g., Thrombolysis in Myocardial Infarction [TIMI],10 Goldman,11 and Sanchis12) and triage algorithms are available to predict adverse cardiac events and facilitate the triage decision.10–14 However, the performance of these tools is not optimal.15 Risk scores used for the prediction of future cardiovascular events (e.g., Framingham Risk Score) and risk factors for coronary artery disease (CAD) have only limited value for the confirmation or exclusion of ACS in the ED.16,17
Coronary computed tomography angiography (CTA) permits detection of coronary stenosis and atherosclerotic plaque with high sensitivity, specificity, and negative predictive value.18–22 In patients who present to the ED with acute chest pain, exclusion of a significant coronary stenosis and plaque by coronary CTA has a high negative predictive value for ACS.23–32 In contrast, patients with a significant coronary stenosis detected by computed tomography (CT) have a high likelihood of ACS. However, there is a significant proportion of patients in whom coronary atherosclerotic plaque is present, without significant stenosis. In the Rule Out Myocardial Infarction Using Computer-Assisted Tomography (ROMICAT) trial, we observed approximately one-third (113 of 368) of patients with nonobstructive plaque.28 The quantitative assessment of coronary plaque and the combination of coronary plaque burden with clinical risk scores could possibly improve the accuracy for the diagnosis as well as the exclusion of ACS.
We determined the association of four dichotomized (low- vs. intermediate- to high-risk categories) clinical risk scores and of the coronary atherosclerotic plaque burden as detected by CT with the outcome of ACS in patients who presented to the ED with the chief complaint of chest pain and inconclusive initial evaluation (negative cardiac biomarkers, nondiagnostic ECG changes), and no history of CAD. We hypothesized that the combination of risk scores and plaque burden improves the discriminatory capacity for the diagnosis of ACS.
This was a planned secondary analysis of data from the ROMICAT trial.28 ROMICAT was a prospective observational cohort study in patients with acute chest pain, but inconclusive initial ED assessment. The primary goal of the study was to determine the diagnostic accuracy of CT findings to rule out ACS during the index hospitalization and to establish the safety of cardiac CT findings to guide the discharge decision in an unbiased fashion. The institutional review board approved the study. All patients provided written informed consent.
The study participants were enrolled in the ED of a tertiary care academic medical center. The study population consisted of patients whose clinical presentation was concerning for ACS (chief complaint of acute chest pain lasting >5 minutes during the past 24 hours), but who had normal initial troponin and an initial ECG without evidence of myocardial ischemia. These patients were awaiting hospital admission to rule out ACS.
All eligible patients who agreed to participate underwent contrast-enhanced coronary CTA prior to admission to the hospital floor, where all patients received standard clinical care. Standard clinical care typically included serial ECG, cardiac biomarkers, and subsequent diagnostic testing, such as exercise treadmill testing, nuclear stress perfusion imaging, and/or invasive coronary angiography, as deemed clinically indicated. The detailed study protocol was described previously.28
We prospectively collected data on demographics, risk factor profile, and clinical course of the enrolled subjects. Medical records were reviewed to obtain data about all diagnostic tests performed during each patient’s index hospitalization. Presence of risk factors was established from actual measurements obtained during the hospitalization (i.e., hypertension, hypercholesterolemia, and diabetes mellitus). Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg as measured during index hospitalization or being on antihypertensive treatment. Hyperlipidemia was defined as total cholesterol ≥200 mg/dL or being on lipid-lowering treatment. Diabetes mellitus was defined as a fasting plasma glucose ≥126 mg/dL or treatment for diabetes. Participants were considered to be smokers if they smoked at least one cigarette per day for the past year or if they were currently smoking.
Characteristics of chest pain at presentation were prospectively recorded. We used 18 different characteristics of chest pain as proposed by Geleijnse et al.33 Chest pain score was calculated based on the characteristics of the chest pain at the time of presentation (location, radiation, character, severity, influence of nitroglycerine, and positioning-associated symptoms).33 The chest pain score could vary between –4 and 20 and was categorized in low (<10) and high (≥10).12,33
Goldman score was calculated using clinical characteristics of each patient’s history:11 known unstable CAD, physical examination, systolic blood pressure <110 mm Hg, presence of bilateral rales, and ECG (suspected ischemia or infarction).
Low, intermediate, and high Goldman risk score was assigned based on the algorithm.11 Patients with infarction on ECG or suspected ischemia on ECG and two or more risk factors were deemed to have high risk. Patients with suspected ischemia on ECG and fewer than two risk factors, or no ischemia on ECG and two or more risk factors, were deemed to have moderate risk. Finally, patients with no ischemia on ECG and fewer than two risk factors were deemed to have low risk.
The Sanchis score combines clinical characteristics of chest pain (18 components) and clinical history data.12 One point was assigned for any of the following: chest pain score ≥10, two or more episodes of chest pain in the past 24 hours, age ≥67 years, insulin-dependent diabetes mellitus, and prior percutaneous coronary intervention. The Sanchis score could range from 0 to 5 points. Patients were categorized into low (0 to 1 points), intermediate (2 points), and high (≥3 points) risk groups.
The TIMI risk score10 is based on seven risk factors, and one point was added for the presence of each risk factor: age >65 years, documented prior coronary stenosis (>50%), three or more conventional cardiac risk factors (hypertension, diabetes mellitus, hypercholesterolemia, family history of premature CAD, and smoking), use of aspirin in the preceding 7 days, two or more anginal events in the past 24 hours, ST-segment deviation >1 mm, and elevated cardiac biomarkers. The TIMI score could range from 0 to 7. Patients were categorized as low (0 to 2), intermediate (3 to 4), and high (5 to 7) risk based on TIMI score.
A CT scan was performed before the admission to the hospital floor. All subjects were scanned using a 64-multidetector-row CT scanner (Sensation 64, Siemens Medical Solutions, Forchheim, Germany). The scan protocol included the premedication with 0.6 mg of sublingual nitroglycerin and an intravenous beta-blocker (metoprolol, 5–20 mg) for all subjects with a heart rate >60 beats/min. Image acquisitions were performed during an inspiration breath hold.
Axial images were reconstructed with a slice thickness of 0.75 mm and increment of 0.4 mm using a retrospectively ECG-gated reconstruction with a temporal resolution of 165 milliseconds. Images were initially reconstructed at 65% of the cardiac cycle with additional reconstructions at 5% increments to obtain motion-free images. The best data set was used for further analysis.
Two independent observers evaluated CT images. A consensus reading was used in the cases with differences. In case no consensus could be reached, a third expert reader made the final diagnosis. The presence of coronary atherosclerotic plaque was evaluated on an offline workstation (Leonardo, Siemens Medical Solutions, Forchheim, Germany) using axial, multiplanar reformatted and maximum intensity projection images. Noncalcified plaque was detected as any discernible structure that could be assigned to the coronary artery wall that had the CT attenuation below the contrast-enhanced coronary lumen but above the surrounding connective tissue and epicardial fat in at least two independent planes within one coronary segment.21,22 Calcified plaque was defined as any structure with a CT attenuation of >130 HU that could be visually separated from the contrast-enhanced coronary lumen.21,22 The presence of calcified and noncalcified coronary plaque was reported in all 17 coronary segments.
Acute coronary syndrome was defined as either acute myocardial infarction (ST-elevation myocardial infarction, non–ST-elevation myocardial infarction) or unstable angina pectoris (UAP) according to the American Heart Association/American College of Cardiology/European Society of Cardiology guidelines.7,9 UAP was defined as clinical symptoms suggestive of ACS (typical chest discomfort or equivalent with an unstable pattern of chest pain, i.e., at rest, new onset, or crescendo angina) in the absence of elevated troponin, coinciding with appropriate objective evidence of myocardial ischemia in stress perfusion imaging or coronary angiography demonstrating a >50% coronary stenosis.
To establish the diagnosis of ACS, an outcome panel of two experienced physicians reviewed patient data forms containing prospectively collected information.28 The reviewers were blinded to the findings of CT. Disagreement was resolved by consensus, which included an additional senior cardiologist.
Continuous variables are reported as mean ± standard deviation (SD) or median and interquartile range (IQR), depending on their distribution. Nominal variables were reported as frequency and percentages. Chi-square test was used to compare proportions between groups. The coronary plaque burden was categorized into patients with no plaque (median), one to four, and more than four coronary segments (upper quartile) containing plaque. Because of the low number of subjects in the high-risk group, the intermediate-risk and high-risk groups were combined into one (“intermediate- to high-risk group”). We performed logistic regression modeling to examine the association of risk scores and CT coronary plaque burden to the outcome of ACS. Due to the occurrence of monotone likelihood (zero events in the “no plaque” group), the Firth’s bias-reducing penalized likelihood method was used in all logistic regression models to generate odds ratio (OR) estimates and standard errors.34,35 Unadjusted models were individually fitted for Goldman, Sanchis, TIMI, and chest pain scores, as well as the categorized coronary plaque burden, which were treated as categorical variables, and the no-plaque group served as a reference. In adjusted analyses, we then tested whether the association between risk scores and ACS persisted after controlling for the categorized coronary plaque burden. For continuous data, we used the same analytical approach. The prognostic discriminatory capacity of the risk scores for ACS and plaque burden was assessed using c-statistics, which is equivalent to the area under the receiver-operating characteristics curve (AUC).36 The asymptotic 95% confidence intervals (CIs) for the AUCs were estimated using a nonparametric approach, which is closely related to the jackknife technique as proposed by DeLong et al.37 and differences between AUC and the c-statistics were tested by performing the −2 log likelihood ratio test of nested models. Finally, we calculated the diagnostic accuracy (sensitivity, specificity, positive predictive value, and negative predictive value) for the strata of the coronary plaque burden, as well as for the combination of strata of the coronary plaque burden and one risk score using two-by-two tables. We derived 95% CIs for sensitivity, specificity, positive predictive value, and negative predictive value from the binomial distribution. All performed tests were two-sided and performed with SAS (version 9.2, SAS Institute Inc., Cary, NC). A p-value <0.05 was considered statistically significant.
The baseline characteristics of the studied subjects are summarized in Table 1. ACS was diagnosed in 31 subjects: eight patients had myocardial infarction, 18 patients had >50% stenosis on invasive coronary angiography, five patients had positive nuclear myocardial perfusion imaging scan, and one patient had positive exercise stress test without imaging. The subjects who were diagnosed with ACS were older and more frequently had hypertension and hyperlipidemia.
Patients with ACS had more than four segments with coronary plaque more frequently (Table 2). The median number of coronary segments with atherosclerotic plaque was higher in subjects with ACS compared to those without ACS (8, IQR = 4 to 10 vs. 0, IQR = 0 to 3; p < 0.001). Coronary plaque burden was a good predictor of ACS as shown by the receiver operating curve and the AUC of 0.86 (95% CI = 0.82 to 0.90; Figure 1). The odds of ACS increased significantly for patients with coronary plaque compared with those with no evidence of coronary atherosclerosis (Table 3).
Goldman, Sanchis, and TIMI score were significantly higher in subjects with ACS (Table 2). The categorized chest pain score did not differ between those with and without ACS (p = 0.33). The Goldman, Sanchis, and TIMI scores were moderate predictors of ACS, as shown by the receiver operating curves and the AUC (Figure 1). The AUC was 0.61 (95% CI = 0.53 to 0.70) for Goldman, 0.71 (95% CI = 0.62 to 0.80) for Sanchis, and 0.63 (95% CI = 0.55 to 0.71) for TIMI score. The risk for ACS increased significantly between the low-risk group and the intermediate to high-risk group for Goldman, Sanchis, and TIMI scores (Table 3). The Goldman, Sanchis, and TIMI scores predicted ACS independently of the coronary plaque burden (Table 3).
The AUC for the prediction of ACS by the categorized coronary plaque burden was greater compared to risk scores (Goldman, p < 0.001; Sanchis, p = 0.001; TIMI, p < 0.001; Figure 1). Further, we observed that the combination of risk scores and coronary plaque burden provided an incremental value for the prediction of ACS (Figure 2). The AUC was improved from 0.86 (95% CI = 0.82 to 0.90) for the categorized plaque burden to 0.88 (95% CI = 0.84 to 0.93; p = 0.001) for the combination of the plaque burden and Goldman score. Similarly, the combination of the plaque burden and TIMI score improved the AUC to 0.88 (95% CI = 0.84 to 0.92; p = 0.004). The greatest improvement was observed with the combination of the plaque burden and Sanchis score (AUC 0.90, 95% CI = 0.85 to 0.94; p < 0.001).
When risk scores and plaque burden were used in their raw form (as continuous variables), we observed that plaque burden had better discriminatory capacity for the diagnosis of ACS compared to the Goldman, TIMI, or Sanchis scores (p < 0.001 for all comparisons). Further, the combination of risk scores and plaque burden was incremental for the diagnosis of ACS (plaque burden and TIMI 0.89, 95% CI = 0.83 to 0.94, p = 0.01; plaque burden and Sanchis 0.90, 95% CI = 0.85 to 0.96, p < 0.001; plaque burden and Goldman 0.89, 95% CI = 0.84 to 0.94, p < 0.01) compared to plaque burden alone. The AUC values for continuous variables were similar to the AUC values derived from categorized variables.
The diagnostic performance of the coronary plaque burden combined with individual risk scores is summarized in Table 4. The combination of the coronary plaque burden with the risk scores improved specificity. The addition of intermediate to high risk Sanchis score to the plaque burden of more than segments provided the best results. Specificity improved from 83% to 99%, and positive predictive value improved from 29% to 80%. Negative predictive value decreased from 97% to 95%. The combination of the plaque burden and risk scores resulted in decreased sensitivity.
We demonstrated that the Goldman, Sanchis, and TIMI risk scores had a modest discriminatory capacity for the diagnosis of ACS in patients who presented to the ED with chief complaint of chest pain and inconclusive initial evaluation. Chest pain characteristics alone did not predict ACS. The coronary plaque burden was the best predictor of ACS in this population. The combination of Goldman, Sanchis, and TIMI scores and plaque burden improved the discriminatory capacity of the model for ACS.
Certain characteristics of chest pain increase or decrease the likelihood of ACS and help the triage of subjects with suspected ACS.1 Chest pain score developed by Geleijnse et al.33 predicted development of ACS in patients with chest pain and nondiagnostic ECG. Further improvement can be achieved through the combination of chest pain history and other clinical factors.13,14 The Sanchis risk score combines clinical characteristics of chest pain and history data and predicts major cardiac adverse events at 14 days and at 1 year.12
Traditional cardiovascular risk factors can be used to assess risk of cardiovascular events. The Framingham risk score is the most widely used algorithm for the prediction of cardiovascular events over a 10-year period.38 However, the Framingham risk score was not designed to predict ACS. The absence of traditional risk factors decreases the likelihood of ACS, but traditional risk factors were not helpful for the confirmation or exclusion of ACS in the ED population.16,17
In our study, chest pain characteristics in isolation did not predict ACS. The combination of chest pain history and additional clinical data such as in the Sanchis or TIMI risk scores improved the prediction of ACS in our study.
Absence of coronary stenosis and plaque accurately predicts absence of ACS in patients who presented to the ED with acute chest pain.23–32,39–41 Hollander et al.42 used coronary CTA as an additional tool to TIMI risk score. They showed that according to TIMI score in low-risk subjects, CT can guide rapid and safe discharge of subjects with chest pain. Furthermore, patients with negative CT had no major adverse cardiac events during follow-up for as long as 2 years.39,43 Presence and extent of coronary stenosis and plaque burden assessed by CT are strong and independent predictors of future cardiovascular events.44–46
Various risk scores and triage algorithms have been developed to facilitate the triage of patients with suspected ACS.10–12,14 The focus has been mostly on the prediction of major adverse cardiac events (e.g., death, myocardial infarction, and need for revascularization). There is less information on the diagnosis of ACS itself. Furthermore, risk scores were tested in patients with a relatively high pretest likelihood of ACS.
The TIMI and Sanchis scores applied to the ED population with chest pain and a lower pretest likelihood of ACS predicted major adverse cardiac events.12,47–49 Manini et al.15 showed in a similar population that TIMI, Goldman, and Sanchis scores used alone had inadequate sensitivity to detect ACS. The major problem of the risk scores is that patients categorized as low risk had clinically consequential rates of ACS (7.8% to 9.0%).15 Therefore, further evaluation and testing is necessary resulting in longer hospital stays.
In our study, we found that the Goldman, Sanchis, and TIMI scores were modest predictors of ACS in patients with acute chest pain and inconclusive initial evaluation. We confirmed nonnegligible rates of ACS in subjects in low-risk categories for Goldman (6%), Sanchis (5%), and TIMI (6%) scores. These findings confirm previous experiences that clinical evaluation, initial ECG, and biomarkers combined with clinical scores are not sufficient for early safe discharge from the ED.
Studies have found that coronary plaque burden is higher in patients with ACS.23,28 We also found that plaque burden was higher in patients with ACS. The extent of coronary plaque significantly improved the prediction of ACS based on the combination of risk factors or clinical estimates.23 We demonstrated that coronary plaque burden had good discriminatory capacity for the diagnosis of ACS. However, specificity and positive predictive value of coronary plaque burden alone for the prediction of ACS were low. In other words, many patients with significant plaque burden did not have ACS.
The combination of Goldman, Sanchis, and TIMI risk scores with coronary plaque burden improved specificity and positive predictive value of ACS prediction. Therefore, patients with both high plaque burden and intermediate to high risk scores had increased likelihood of ACS. We may speculate that patients with high scores and high plaque burden would benefit from an early intensive medical and possibly invasive management.
In contrast, low coronary plaque burden and low clinical scores had high negative predictive value for the exclusion of ACS. Patients with small amounts of plaque and low scores could possibly wait for the second set of biomarkers and ECG and then be discharged. The combination of risk scores, clinical characteristics, and information on coronary plaque burden may help the triage of patients with chest pain, streamline their care, and eventually lead to cost savings.
This study was performed in a single tertiary care academic center. Approximately 200,000 patients were evaluated in our ED from January 2005 through December 2007. Among those patients, 4.2% presented with the chief complaint of chest pain. We applied relatively strict inclusion and exclusion criteria. Therefore, the study population may have a risk of ACS different compared to the average population presenting to the ED. We observed approximately 8% prevalence of ACS in our population. Previous studies reported the prevalence of ACS between 5 and 20% in a broad population of patients with chest pain.5,47–49 The risk of ACS in the studied population may affect the results of the study, as various scores may perform differently in subjects with low, intermediate, and high risk. Furthermore, patients with previous history of CAD were excluded. Patients with known CAD represent an important subgroup of patients with acute chest pain. Therefore, the results may not be fully generalizable.
There was no formal sample size calculation for the evaluation of risk scores. Sample size was estimated to allow for the assessment of coronary CTA for ruling out ACS.28
The TIMI, Goldman, and Sanchis scores were not designed for the evaluation of the ED population. Only limited experiences with the performance of these scores in ED patients with acute chest pain have been reported.12,47–49 Further, the ranges of values in our score were affected by the inclusion and exclusion criteria of the ROMICAT trial.
The use of CT in the ED is limited by the availability of both technology and expert readers 24 hours per day and 7 days per week. The radiation dose of a single coronary CTA scan is significant. Nevertheless, the radiation dose of nuclear stress perfusion imaging that is used very often in the evaluation of patients with chest pain is similar, and newer CT scanner technologies result in significant reduction of the radiation dose.
Available risk scores (Goldman, Sanchis, Thrombolysis in Myocardial Infarction) have modest and coronary plaque burden has good discriminatory capacity for the diagnosis of acute coronary syndrome in patients with acute chest pain. The combination of the coronary plaque burden and individual risk scores significantly improves the discriminatory capacity for the diagnosis of acute coronary syndrome.
This work was supported by the National Institutes of Health (RO1 HL080053) and in part supported by Siemens Medical Solutions and General Electrics Healthcare.
Drs. Ferencik, Rogers, Truong, and Shapiro were supported by the National Institutes of Health grant (T32 HL076136). Dr. Hoffmann has received research grants from Siemens Medical Solutions and General Electric Healthcare. Dr. Nagurney is funded by Biosite for a biomarker research study.
The rest of the authors have no disclosures or potential conflicts of interest to report.