Author + information
- Received May 6, 2004
- Revision received March 27, 2005
- Accepted March 29, 2005
- Published online September 6, 2005.
- Khim Leng Tong, MD⁎,
- Sanjiv Kaul, MD, FACC⁎,
- Xin-Qun Wang, MS†,
- Diana Rinkevich, MD, FACC⁎,
- Saul Kalvaitis, MD⁎,
- Todd Belcik, RDCS⁎,
- Wolfgang Lepper, MD⁎,
- William A. Foster, MD⁎ and
- Kevin Wei, MD, FACC⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Kevin Wei, Cardiovascular Division, OHSU, UHN62, SW Sam Jackson Park Road, Portland, Oregon 97239.
Objectives We hypothesized that regional function (RF) and myocardial perfusion (MP) are superior to the Thrombolysis In Myocardial Infarction (TIMI) score for diagnosis and prognostication in patients presenting to the emergency department (ED) with chest pain (CP) and a nondiagnostic electrocardiogram.
Background Rapid diagnosis and prognostication is difficult in patients presenting to the ED with suspected cardiac CP.
Methods Contrast echocardiography was performed to assess RF and MP on 957 patients presenting to the ED with suspected cardiac CP and a nondiagnostic electrocardiogram. A modified TIMI (mTIMI) score was calculated from six immediately available variables. A full TIMI score also was derived after troponin levels were able to be accessed adequately. Follow-up was performed for early (within 24 h), intermediate (30 day), and late primary (death and myocardial infarction) or secondary (unstable angina and revascularization) events.
Results The mTIMI score was unable to discriminate between intermediate- compared to high-risk patients at any follow-up time point, whereas only 2 of 523 patients with normal RF had an early primary event. Regional function provided incremental prognostic value over mTIMI scores for predicting intermediate and late events. In patients with abnormal RF, MP further classified patients into intermediate- and high-risk groups. The full TIMI score could not improve upon these results at any follow-up time point.
Conclusions Contrast echocardiography can rapidly and accurately provide short-, intermediate-, and long-term prognostic information in patients presenting to the ED with suspected cardiac CP even before serum cardiac markers are known. Integrating contrast echocardiography into the ED evaluation of CP may improve the risk stratification of such patients.
Chest pain (CP) is a common complaint in patients seeking care from an emergency department (ED). The electrocardiogram (ECG) is effective in helping to diagnose 30% to 50% of patients with acute myocardial infarction (AMI) (1–3), and early determinations of serum cardiac enzymes frequently are negative (4,5). The unreliability of these parameters results, on one hand, in the inadvertent discharge of approximately 5% of patients with ongoing AMI and, on the other hand, unnecessary hospital admissions (6).
Apart from the identification of ischemia, the accurate risk stratification of these patients also is important. Those patients who are at intermediate or high risk for an adverse outcome may require admission to a coronary care or telemetry unit, treatment with potent antiplatelet agents (7–9), or early referral for cardiac catheterization (10). The presence of elevated levels of troponin identifies those patients at increased risk and is a component of the Thrombolysis In Myocardial Infarction (TIMI) risk score (11). It may, however, not be elevated or immediately available at the time of patient presentation. Thus, complete risk stratification and initiation of therapy may be unnecessarily delayed. We hypothesized that an assessment of left ventricular (LV) regional function (RF) and myocardial perfusion (MP) would be superior to the TIMI score in patients presenting to an ED with suspected cardiac CP and no ST-segment elevation on the ECG.
This study was approved by the Human Investigation Committee at the University of Virginia. Patients >30 years of age who presented to the ED with a complaint of CP of >30 min in duration that was not easily attributable to an obvious noncardiac cause and who did not exhibit ST-segment elevation on the ECG were enrolled in the study. All patients had their contrast echocardiography (CE) study completed within 12 h of the onset of their symptoms. All patients provided written informed consent.
Determination of TIMI risk score
The TIMI risk score was calculated as previously defined (11). A score of 1 was assigned to each of the following seven variables when present: age >65 years, more than three coronary disease risk factors, known coronary luminal diameter narrowing of >50%, ST-segment deviation on ECG, two or more angina events in the previous 24 h, use of aspirin in the previous seven days, and elevated levels of troponin. The TIMI risk score was calculated as the sum of all scores.
We also wanted to evaluate the prognostic utility of a score derived from variables that are available immediately at the time of ED presentation. Because of inherent delays in receiving laboratory results and because troponin elevation may be delayed for many hours after the initial presentation, a modified TIMI risk score (mTIMI) that excluded this variable was derived (maximum score of 6).
An infusion of 3 ml of Optison (GE Healthcare, Princeton, New Jersey) diluted in 60 ml of saline was administered intravenously at a rate of 3 ml · min−1using a syringe pump (model AS40A, Baxter, Deerfield, Illinois). Imaging was performed using a Sonos 5500 system (Phillips Ultrasound, Andover, Massachusetts). Regional function data were acquired using low mechanical index harmonic imaging and contrast for LV border delineation.
For MP, intermittent high mechanical index ultraharmonic imaging was performed using transmit/receive frequencies of 1.3/3.6 MHz, respectively, with ultrasound transmission gated to the ECG at end-systole. Images were acquired digitally at pulsing intervals of 1, 2, 3, 4, and 5 cycles. Compression was set at 75%, and all gain settings were optimized at the beginning of the study and then held constant. If a signal from myocardial regions was discernable despite optimization of ultraharmonic image settings, intermittent harmonic power Doppler was performed to achieve better tissue signal suppression. The transmit focus initially was set at the level of the mitral valve but was moved to the apex when an apical defect was observed to exclude the apical destruction artifact (Video A; see the online version of this article for all videos). Off-axis images were acquired as needed. Regional function and MP data were acquired digitally onto magneto-optical disks.
Regional function and MP studies were interpreted separately by experienced observers who were blinded to all data: both were scored visually using a 14-segment model as normal, abnormal, or not interpretable (12). The segments were then grouped into anteroapical, lateral, or inferoposterior territories. Studies were classified as normal if RF or MP in the majority of segments within each perfusion territory were normal. Studies were called abnormal if RF or MP was abnormal in one or more territories, even if all territories were not visualized. If a study could not be classified as previously mentioned, it was deemed not assessable.
All patients had a complete history and physical examination as well as a 12-lead ECG. Blood was drawn for troponin I at the time of ED presentation and repeated twice at 6-h intervals. Troponin I levels were measured using a direct chemiluminometric immunoassay (Bayer, Tarrytown, New York). Contrast echocardiography was performed as soon as possible after the patient’s presentation to the ED. Information regarding RF or MP was not shared with the ED physician, and the decision to admit or discharge was based solely on routine clinical, laboratory, and ECG criteria.
Follow-up was obtained by questionnaire or telephone interview with the patient, patient’s family, physician, or a combination. All reported events were confirmed by review of medical charts or other documents (for example, death certificates). Primary end points included all-cause mortality and AMI, which was defined by an abnormally elevated troponin I level (>0.6 ng · ml−1). Secondary events included revascularization (either coronary artery bypass surgery or percutaneous coronary intervention) and unstable angina, defined as ischemic CP of >30 min duration associated with dynamic ECG changes and/or mild troponin elevation (0.08 to 0.6 ng · ml−1) and requiring hospitalization and/or revascularization. Secondary events occurring before a primary event were ignored. If a patient had more than one event, only the first was considered for analysis. Cumulative patient outcomes were determined at three time points: early (within the first 24 h), intermediate (up to 30 days), and late (>30 days).
The explanatory variables were the same for all types of models. To test whether RF or RF plus MP had the ability to discriminate between subjects with different TIMI risk scores, logistic regression models (13,14) were applied for cumulative events that had accrued at early and intermediate time points, and Cox proportional hazards regression models (14,15) were used for cumulative events at the late time point. Adjusted survival probabilities for late events also were estimated based on the Cox proportional hazards survival model (16). A Wald chi-square test was used to assess the significance of each predictor and an odds ratio (OR) or a hazard ratio was used to quantify its effect. A log likelihood ratio test was applied to determine whether RF and/or MP provided additional information to the TIMI scores in predicting events. The area under the receiver-operating characteristic curve (or C-index) was used to quantify the discriminatory power of RF or combined RF and MP (14). C-indices of <0.69, 0.70 to 0.79, and >0.79 indicated poor, good, and excellent discrimination, respectively.
Between October 2000 and January 2003, 1,236 patients were recruited. Of these, 86 (7%) had no follow-up, and 151 had RF or MP studies of inadequate quality (12%). Another 42 (3%) developed ST-segment elevation on the ECG after being admitted to the ED and were excluded from analysis. Therefore, a total of 957 patients are presented in this report, of whom 153 (16%) had CE performed during ongoing CP. No significant differences were found in demographics between patients who were excluded and the study group.
A little more than one half (52%) of the 957 patients were men; the median age was 60 years (range, 32 to 92 years), 27% were smokers, 28% had diabetes, 53% had hypercholesterolemia, 47% had a family history of coronary artery disease, and 66% had essential hypertension.
On the basis of their mTIMI or TIMI scores, patients were categorized as either low (score ≤2), intermediate (score of 3 or 4), or high (score ≥5) risk. In 485 of the 957 patients (51%), both RF and MP were normal (Video A); in 164 (17%), RF was abnormal but MP was normal (Video B); in 270 (28%), both RF and MP were abnormal (Video C); and in 38 (4%), RF was normal but MP was abnormal.
The incidence of early events based on the mTIMI score is shown in Table 1.Acute myocardial infarction was the most common event in all groups. Although patients with a low mTIMI score had the lowest incidence of primary events, 24 (4.1%) of these still had an AMI. Patients with an intermediate score had a similar AMI event rate as those with a high-risk score (11% vs. 8.9%, p = 0.71). Thus, the mTIMI score was unable to discriminate between these groups.
Table 2depicts early events in patients with different RF and MP findings. The presence of normal RF identified a population at low risk of a primary event (n = 2, 0.4%). Furthermore, in patients with abnormal RF, those with normal MP had a significantly lower event rate than those with abnormal MP (9.8% vs. 16%; p < 0.05). A significant interaction was observed between the mTIMI score and RF. In patients with a low mTIMI score (≤2), those with normal RF were significantly less likely to have an event compared with those with abnormal RF (OR 0.11; 95% confidence interval 0.05 to 0.25; p < 0.001). Likewise, patients with an intermediate mTIMI score (3 or 4) who had normal RF were also significantly less likely to have an event (OR 0.14, 95% confidence interval 0.07 to 0.30; p < 0.001). In patients with the highest mTIMI scores (≥5), however, RF did not provide additional information. After serum cardiac markers were included into the final TIMI score, approximately 5% of patients in the low- and intermediate-mTIMI risk groups were reassigned to higher-risk groups. Similar to previous reports (11), the complete TIMI score adequately risk stratified patients into low, intermediate, and high risk for early events. However, because of a primary event rate of >2% even in patients with a low score, use of the TIMI score alone would not allow discharge from the ED.
Figure 1illustrates the predictive value of tests performed in a hierarchical order. By adding RF to the mTIMI score, the C-index increased from 0.68 to 0.79, and the model provided 127% additional information compared with the mTIMI score alone (p < 0.001). The addition of MP provided 148% additional information compared with mTIMI score alone (p < 0.001). A complete TIMI score did not provide any more information compared with that already available from the combination of the mTIMI score and the CE information (C-index of 0.77 vs. 0.80).
The incidence of events at 30-day follow-up in low-, intermediate-, and high-risk patients based on the mTIMI and TIMI scores are shown in Table 3,and those in the different CE groups are shown in Table 4.Similar to the data for early events, RF provided 85% additional prognostic information (p < 0.001), whereas RF plus MP provided 99% incremental predictive benefit (p < 0.001) compared with the mTIMI score alone (Fig. 2).
Patients were followed for up to two years (median, 8.2 months). The incidence of late events in various mTIMI and TIMI groups is shown in Table 5,whereas Table 6depicts the results derived from different RF and MP subgroups. The predictive ability of tests performed in a hierarchical order based on the C-index derived from individual Cox regression models are shown in Figure 3.The mTIMI score had the weakest predictive ability and was superseded by the addition of RF. Myocardial perfusion provided further incremental value to RF. The complete TIMI score provided greater prognostic information compared with that already available from the CE study. Unlike the setting of early events, where troponin markedly improved the ability of the mTIMI score for predicting events (0.77 vs. 0.68), it had only a marginal effect on the mTIMI score for predicting late events (0.74 vs. 0.69).
Figures 4 and 5⇓show the incremental benefit of RF and MP over clinical variables alone in patients with low (≤2) and intermediate (3 to 4) mTIMI scores, respectively. Those with both abnormal RF and MP had a significantly worse two-year event-free survival compared with those with abnormal RF but preserved MP (p < 0.01). In patients with a high (≥5) mTIMI score, however, CE did not provide any further risk stratification.
Patients presenting to the ED with suspected cardiac CP but no ST-segment elevation on the ECG pose a challenge. Those patients at high risk for developing events and thus requiring immediate hospitalization need to be separated from those who are at low risk and can be safely discharged. The new finding in this study is that before obtaining troponin levels, the use of a composite score calculated from immediately assessable clinical variables (mTIMI) does not adequately risk stratify all patients with CP and that the assessment of RF and MP on CE, combined with these clinical variables, provides superior diagnostic and prognostic information when compared with even the complete TIMI score.
In the U.S., more than 6 million people present annually to EDs with symptoms that are suggestive of myocardial ischemia (13). The principal tools that are used for diagnosis and prognostication include history and physical examination, ECG, cardiac serum markers, and stress testing. The ECG is nondiagnostic in a significant number of patients with CP (2,3), and the kinetics of biochemical cardiac marker release make them insensitive for myocardial injury at the time of the patient’s presentation to the ED (4,5). Consequently, many patients with a nondiagnostic ECG are admitted to the hospital unnecessarily to “rule out” an acute coronary syndrome. Most of these patients do not have acute myocardial ischemia or even coronary artery disease. Thus, the current “rule out” strategy wastes billions of dollars annually.
The patients admitted for “rule out” are very heterogeneous and represent a wide risk range for developing a serious cardiac event. However, the assessment of risk is essential for triage to an adequate location for care (regular ward vs. step-down unit vs. coronary care unit) and to determine those who could benefit the most from aggressive therapy. Although single variables (such as ECG abnormalities or elevated serum cardiac markers) (17–21) have prognostic value, the composite TIMI score has been shown to effectively categorize risk of cardiovascular events and has been proposed for use in clinical decision making (11).
In this study, we found that patients with a low mTIMI score could not be discharged from the ED because of a substantial event rate within 24 h. In comparison, RF on CE has an excellent negative predictive value—only 0.4% of patients had an early primary event. In those with abnormal RF, an assessment of MP provided additional information. Contrast echocardiography also can provide additional information in patients with either a low or intermediate mTIMI score. Those patients with normal RF had a significantly lower risk of an early event compared with those with abnormal RF. The complete TIMI score that also includes troponin was not found to provide incremental information compared with the combination of the mTIMI score and CE.
Intermediate and late events
Previous studies have shown that clinical variables, such as the history, physical examination, and ECG, are suboptimal for identifying patients at high risk for late cardiovascular events that develop >24 h after presentation (22). Because the incidence of late events may be as high as that of early ones (22), early identification of those at risk for future events is imperative for planning further follow-up or diagnostic studies. Our results indicate that CE provides superior information to the mTIMI score in these patients and that combining clinical variables with CE provides better risk stratification than clinical variables plus troponin (the complete TIMI score) for predicting late events. A previous study also has reported nonsuperiority of troponin over RF in predicting events (23).
Although the performance of CE may be feasible during the day in most medical centers, the logistics of nighttime performance and interpretation of images is not clear. It is imperative for studies to be of high quality and all myocardial segments to be well delineated. Even with state-of-the-art systems and tissue harmonic imaging, endocardial borders are visualized inadequately in a sizeable minority of patients that is reduced markedly when CE is performed.
The evaluation of RF abnormalities on echocardiography is one of the most subjective and difficult skills to master. In this context, CE has been shown to improve visualization of endocardial borders, image quality, reader confidence, and observer agreement (24,25). For rapid interpretation at night or on weekends, digitally acquired CE images could be accessed online for RF interpretation (26,27). Patients who have normal resting RF could be safely discharged from the ED and those with abnormal RF could undergo MP imaging.
Because RF and MP studies were assessed independently, a small number of patients with artifacts were classified incorrectly as having perfusion defects (those with normal RF at rest). In clinical practice, when RF and MP are assessed simultaneously, such errors will likely be minimized.
Clinical implications and conclusions
Contrast echocardiography (including RF and MP data) may be useful in patients presenting to the ED with suspected cardiac CP but no ST-segment elevation on the ECG because it provides incremental diagnostic and prognostic value over clinical variables. Furthermore, in patients with a low or intermediate mTIMI score (≤4), it effectively identifies those who will develop events later.
Current guidelines recommend observation for 4 to 8 h followed by stress testing in patients with negative serum cardiac markers (28). Because the likelihood of significant ischemia or infarction is exceedingly low in patients with normal resting RF, the observation time could be shortened in patients with this finding, and immediate stress echocardiography could be performed. However, the incidence of positive stress echocardiograms in this population would be low.
Patients with abnormal RF but normal MP were found to have an intermediate risk of events. The discordance between RF and MP could be the result of stunning after the resolution of acute ischemia, the presence of abundant collateral flow, or represent nonischemic cardiomyopathy. These patients could follow current guidelines and undergo stress echocardiography if serum cardiac markers remained normal. The incidence of positive stress echocardiograms in this population would be relatively high. Alternatively, early cardiac catheterization could be considered if concurrent clinical variables suggested that the patient belonged to a high-risk cohort.
Patients with both abnormal RF and MP are at high risk for early events and could be triaged even before cardiac serum marker results become available. Implementation of this approach in the ED may help streamline the care of patients presenting with suspected cardiac CP and reduce cost significantly. Although the cost effectiveness of this approach needs to be evaluated, echocardiography (even contrast-enhanced) is relatively inexpensive compared with other noninvasive imaging modalities. Further studies are needed to assess the value of such an approach in an ED.
Please see the online version of this article for the accompanying videos.
normal regional left ventricular function and myocardial perfusion with an apical microbubble destruction artifact that disappears when the focus is moved to the apex
abnormal regional left ventricular function but normal myocardial perfusion
both regional left ventricular function and myocardial perfusion are normal
Supported in part by grants (RO1-HL66034) from the National Institutes of Health, Bethesda, Maryland, and American Society of Echocardiography, Durham, North Carolina. The ultrasound contrast agent was provided by GE Healthcare (Princeton, New Jersey) and the ultrasound equipment was provided by Philips Ultrasound (Andover, Massachusetts).
Drs. Tong and Kalvaitis were recipients of Research Fellowship Awards from the National Medical Research Council of Singapore and POINT Biomedical Corporation (San Carlos, California), respectively. Dr. Wei was the recipient of a Mentored Clinical Scientist Development Award (K08-HL03909) from the National Institutes of Health. Current address of Dr. Tong: Changi General Hospital, Singapore. Drs. Kaul and Wei are currently located at the Cardiovascular Division, OHSU, Portland, Oregon.
Presented in part at the Samuel Levine Young Investigator Award Competition at the 77th Annual Scientific Session of the American Heart Association, New Orleans, Louisiana.
- Abbreviations and Acronyms
- acute myocardial infarction
- contrast echocardiography
- chest pain
- emergency department
- left ventricular
- myocardial perfusion
- modified Thrombolysis In Myocardial Infarction
- regional function
- Thrombolysis In Myocardial Infarction
- Received May 6, 2004.
- Revision received March 27, 2005.
- Accepted March 29, 2005.
- American College of Cardiology Foundation
- Sabia P.,
- Afrookteh A.,
- Touchstone D.A.,
- Keller M.W.,
- Esquivel L.,
- Kaul S.
- Short D.
- McCord J.,
- Nowak R.M.,
- Hudson M.P.,
- et al.
- (PRISM-PLUS) Study Investigators
- Cannon C.P.,
- Weintraub W.S.,
- Demopoulos L.A.,
- et al.
- Dawson D.,
- Rinkevich D.,
- Belcik T.,
- et al.
- Strussman B.J.
- Collet D.
- Harrell F.E.
- Collet D.
- Cohen M.,
- Hawkins L.,
- Greenberg S.,
- Fuster V.
- Luscher M.S.,
- Thygesen K.,
- Ravkilde J.,
- Heickendorff L.,
- TRIM Study Group
- Trippi J.A.,
- Lee K.S.,
- Kopp G.,
- Nelson D.,
- Kovacs R.
- Trippi J.A.,
- Lee K.S.,
- Kopp G.,
- Nelson D.R.,
- Yee K.G.,
- Cordell W.H.
- Braunwald E.,
- Antman E.M.,
- Beasley J.W.,
- et al.