Unstable angina (UA) is classically defined as ischemic discomfort that occurs at rest (or with minimal exertion), occurs with a crescendo pattern, or is severe and of new onset. If these symptoms are accompanied by release of cardiac biomarkers of necrosis (e.g., creatine kinase MB [CK-MB] or cardiac-specific troponin) then the diagnosis of non–ST-segment elevation myocardial infarction (NSTEMI) is made. Both entities typically share a common pathobiologic basis, development of a severe but nonocclusive coronary artery thrombus superimposed on a recently disrupted vulnerable plaque. Thus, treatments for UA and NSTEMI are identical and consist of a combination of anti-ischemic and antithrombotic therapies, and potentially coronary revascularization. Nonetheless, among patients presenting with UA-NSTEMI, there is substantial heterogeneity in the risk of death and major cardiac ischemic events over time. Short-term mortality is lower in patients with UA (1.7% at 30 days) as compared with those with NSTEMI or ST-segment elevation myocardial infarction (STEMI; 5.1% at 30 days for each). In contrast, long-term risk of death or cardiovascular complications is higher in patients with UA or NSTEMI than in patients with STEMI. This higher risk has been associated with the increased age and burden of other diseases in patients with UA and NSTEMI.
Risk stratification, aimed at providing a more accurate estimate of a patient’s prognosis, is pivotal in the clinical management of patients with UA-NSTEMI. Such information is important to the patient and family and also allows for more effective triage and allocation of clinical resources. Clinical trials have demonstrated the efficacy of multiple pharmacologic agents, including aspirin, adenosine diphosphate (ADP) receptor blockers, glycoprotein (GP) IIb/IIIa inhibitors, unfractionated heparin, low-molecular-weight heparin (LMWH), and direct thrombin inhibitors, as well as an early invasive strategy of management. However, many of these treatment options are associated with bleeding, which in turn has been associated with significant morbidity and mortality. Therefore, risk stratification is an integral part of clinical decision making and may guide the use of more aggressive therapy in those who are likely to derive the greatest benefit.
Data from observational studies and clinical trials have documented the prognostic usefulness of individual factors for risk stratification. Demographic and historical features, as well as information collected during the initial evaluation, including physical examination findings and electrocardiographic changes, have been used in simple risk stratification schema. Elevations in various cardiac biomarkers have now been proven useful in identifying high-risk populations. The TIMI Risk Score for UA-NSTEMI and the GRACE Risk Score are two assessment tools that integrate both clinical and biomarker data. Of note, these scores and others have been designed to predict different types of events over different time periods following an acute coronary syndrome (ACS). For example, the TIMI Risk Score incorporates seven clinical predictors and evaluates the risk of mortality, new or recurrent MI, or severe recurrent ischemia requiring urgent revascularization over the 14-day period following a UA-NSTEMI. The GRACE Risk Score, on the other hand, applies nine variables to assess the mortality risk over 6 months following STEMI, NSTEMI, or UA. Whereas both of these risk scores incorporate features on presentation, dynamic risk stratification may also provide incremental information. By integrating a combination of baseline, discharge, and follow-up data, clinicians may be able to assess a patient’s risk for recurrent events more completely and accurately.
Demographic and Historical Risk Factors
Age
Increasing age has been shown to be a risk factor across the spectrum of ACS. For the sake of simplicity, age is often treated as a dichotomous variable (e.g., younger than 64 vs. 65 years and older). However, in an analysis based on data from the TIMI III Registry, age was treated as a continuous variable, and each decade was found to confer a relative risk of 1.43 ( P < .001) for the composite of death or MI over 1 year. Moreover, in the PURSUIT trial, use of cubic spline functions , has revealed that the univariate relationship between age as a continuous variable and mortality was curvilinear. In both UA and NSTEMI patients, an inflection point is evident at approximately 65 years, thus supporting the use of 65 years as a cut point when a binary approach is desired. An alternative approach has been to treat age as a categorical variable, with the clinician assigning increasing weight for each decade above a certain threshold. , The increase in risk with age may be steeper in patients with NSTEMI than in patients with UA. Thus, clinicians should bear in mind that advanced age likely conveys increased prognostic importance when shifting from UA to NSTEMI to STEMI.
Gender
The impact of gender on outcomes in ACS is complex. Because women with ACS tend to have more traditional risk factors, crude univariate associations showing a harmful or protective effect of female gender are likely confounded. For example, Hochman and colleagues have found that in TIMI IIIB, a clinical trial involving patients with UA-NSTEMI, and in GUSTO IIb, a clinical trial that enrolled patients across the spectrum of ACS, women presenting with an ACS were older and were more likely to have hypertension, diabetes, and hyperlipidemia. , Studies have revealed that women are also more likely to present with atypical features and thus may seek medical attention more slowly and may not receive appropriate care after presentation. , In angiographic studies, women presenting with presumed ACS tend to have less severe epicardial coronary artery disease (CAD) than their male counterparts. To account for these gender-specific differences in ACS presentation, appropriately adjusted analyses, including age and baseline cardiovascular risk factors, are important.
In a multivariate model used to test the association of 50 baseline variables with the end point of death or MI following a non–ST-segment elevation ACS, gender was not found to be related to cardiovascular outcomes. Similarly, in the development of the GRACE Risk Score (used to evaluate the mortality risk following ACS) and the TIMI Risk Score (used to evaluate the mortality, new or recurrent MI, or severe recurrent ischemia requiring urgent revascularization risk following UA and NSTEMI), gender was not an independent predictor of poor outcomes. , Likewise, in a contemporary cohort of patients presenting with UA/NSTEMI, women and men were found to be at similar adjusted risk for subsequent cardiovascular death, myocardial infarction, or recurrent ischemia. There have been some studies in which female gender was associated with a statistically significant protective effect, although this may be restricted to certain subsets of UA patients. Thus, the totality of the data suggests that female gender, after adjusting for other established predictors, conveys neither harm nor protection following a non–ST-segment elevation ACS.
Diabetes
Both in patients at risk and with known CAD, diabetes has emerged as a potent risk indicator, and is of increasing importance because of the rise in the incidence of diabetes. Over the past 2 decades, the worldwide prevalence has increased significantly, with 30 million cases in 1985 and 177 million in 2000; it is estimated that more than 260 million individuals will have diabetes by 2030. In the United States, approximately 7% of the population has diabetes. The metabolic syndrome is even more common, defined as the presence of at least three of the following :
- •
Waist circumference more than 102 cm in men and more than 88 cm in women
- •
Serum triglyceride level of at least 150 mg/dL
- •
High-density lipoprotein cholesterol level of less than 40 mg/dL in men and 50 mg/dL in women
- •
Blood pressure of at least 130/85 mm Hg
- •
Serum glucose level of at least 110 mg/dL
As is the case for type 2 diabetes, insulin resistance is thought to be the underlying cause of the metabolic syndrome. In the United States, the age-adjusted prevalence of the metabolic syndrome is 34% for men and 35% for women, using data from the National Health and Nutrition Examination Survey (NHANES) III. The high prevalence of the metabolic syndrome also illustrates the tendency for other traditional cardiovascular risk factors to be seen in patients with glucose intolerance or overt diabetes. For example, half of diabetic patients have concomitant hypertension and one third have concomitant hyperlipidemia.
Pathophysiologically, diabetes results in increased oxidative stress and the development of advanced glycation end products, which may be proatherogenic. Hemostatic sequelae include heightened platelet aggregation, , increased levels of fibrinogen and plasminogen activator inhibitor (PAI) type 1, , upregulated cell surface adhesion molecules, , and impaired endothelial function.
In population-based studies, it has been noted that the risk of first or recurrent MI in diabetic patients without a prior MI was approximately equal to the risk in nondiabetic patients with a prior MI. In two trials of patients with STEMI—TAMI and GUSTO-I —patients with diabetes were found to have nearly twice the risk of death as their nondiabetic counterparts, despite similar rates of infarct-related artery patency. , In multiple clinical trials in UA-NSTEMI, including GUSTO IIb, PRISM-PLUS, FRISC II, TACTICS-TIMI 18, and GUSTO IV-ACS, diabetics were found to have 1.5- to 2.0-fold higher rates of death and cardiac ischemic events. The independent prognostic significance of diabetes was demonstrated using data from the OASIS registry, which showed that diabetes was an independent risk factor for mortality in non–ST-segment elevation ACS (adjusted [adj] risk ratio [RR], 1.57; 95% confidence interval [CI], 1.38 to 1.81). Similarly, in an analysis that included 62,036 subjects across 11 TIMI trials, diabetes was independently associated with higher 30-day mortality (adj odds ratio [OR], 1.78; 95% CI, 1.24 to 2.56) and 1-year mortality (adj hazard ratio [HR], 1.65; 95% CI, 1.30 to 2.10) after UA-STEMI. Moreover, the risk of death among diabetics who presented with UA-NSTEMI at 1 year approached that of nondiabetic patients who presented with STEMI, whereas the nondiabetic UA-NSTEMI patients continued on a low-risk trajectory ( Fig. 18-1 ).
Smoking
The smoker’s paradox in ACS has been described previously. Current smokers tend to have lower rates of death and ischemic events than nonsmokers. This paradoxical beneficial effect appears to be explained largely by the fact that smokers present at an earlier age than nonsmokers and hence have fewer other comorbidities and less extensive CAD. Thus, among patients with ACS, although current smokers have lower event rates than nonsmokers in univariate analyses, current smoking in multivariable analyses is not a significant independent prognostic factor. ,
Peripheral Arterial Disease
Patients with peripheral arterial disease (PAD) frequently have significant CAD and it is not surprising that PAD is a risk factor for death and ischemic complications in patients with UA-NSTEMI. However, in a multivariable analysis in OPUS-TIMI 16 that adjusted for other traditional risk factors, PAD remained an independent risk factor for death (adj OR, 1.44; P = .0045) as well as a composite of cardiac ischemic events (adj OR, 1.21; P = .0035); similar results were also found in the PURSUIT study. ,
Prior Aspirin Use
Multiple studies have confirmed that patients with prior aspirin use are at increased risk. This may be caused by the presence of aspirin-resistant platelet-rich thrombi or the greater likelihood of severe CAD in patients who present with UA-NSTEMI despite taking aspirin. , There are also some studies suggesting that patients exhibit differing degrees of aspirin responsiveness, and aspirin resistance may be associated with an increased risk of death and cardiovascular complications. ,
Acute Presentation
The tempo of the acute presentation, specific physical findings, electrocardiographic changes, and biochemical evidence of myocardial necrosis have all been shown to convey important prognostic information.
Severity of Angina
In 1989, Braunwald differentiated between primary angina (caused by plaque rupture and a reduction in myocardial blood supply), secondary angina (caused by non–cardiac-induced mismatch), and postinfarction angina. He further differentiated between new-onset, crescendo, and rest angina. The importance of these distinctions has been supported in several studies, in which multiple episodes of angina in the preceding 24 hours, angina at rest, and postinfarction angina each have been shown to convey a worse prognosis.
Physical Examination
Physical findings indicative of severe left ventricular contractile dysfunction, such as the presence of an S 3 gallop, rales, a mitral regurgitation murmur, hypotension, and tachycardia are more commonly seen in the setting of STEMI rather than in UA-NSTEMI, but also confer adverse prognosis in the latter syndrome. In patients with STEMI, the significance of these physical findings was noted over 40 years ago by Killip and Kimball, and remain important components of contemporary integrated risk scores. , Although rarer, when they are found in UA-NSTEMI they suggest significant underlying CAD and are associated with mortality rates in excess of 60%.
Electrocardiogram
The admission electrocardiogram (ECG) is one of the most useful and powerful predictors of adverse outcomes in ACS. ST-segment depression on the presenting ECG indicates severe acute ischemia and is correlated with a worse in-hospital prognosis. , The presence of ST-segment depression is also associated with greater complexity of the culprit lesion and hence a greater likelihood of requiring revascularization. ST-segment depression is also indicative of more extensive CAD , and is associated with worse outcomes at 6 months and at 1, 4, and 10 years.
Importantly, ST-segment deviation of as little as 0.05-mV conveys a higher rate of adverse events. In the TIMI III Registry, patients with 0.05-mV ST-segment depression had an approximately twofold higher risk of death or MI at 30 days and at 1 year. , Moreover, there appears to be a gradient of increasing risk with the increasing degree of ST depression. Among patients with non–ST-segment elevation ACS, the 4-year survival for patients with 0.05-, 0.10-, or 0.20-mV or more ST-segment depression was 82%, 77%, and 53%, respectively ( P < .0001).
In contrast to ST-segment depressions, T-wave inversions in general have not been shown to be associated with a worse prognosis. However, deep (≥0.20-mV) precordial T-wave inversions are suggestive of LAD disease and are associated with a worse prognosis. ,
Detection of Necrosis
Another important predictor of outcome is the detection of myocyte necrosis. Patients with documented biochemical evidence of myocyte necrosis have higher mortality rates than patients without elevations. Furthermore, there is a quantitative relationship between the magnitude of CK-MB elevation and the risk of death ( Fig. 18-2 ).
Cardiac-specific troponins, with their superior sensitivity and specificity, have emerged as the biomarkers of choice for detecting myocyte necrosis. As with CK-MB, there is a clear relationship between the magnitude of troponin level elevation and mortality ( Fig. 18-3 ). , With greater clinical sensitivity, troponins have enabled the detection of microinfarctions in approximately 30% of patients who otherwise would have been diagnosed as having UA. These patients, with an elevated troponin but negative CK-MB level, have been shown to be at a three- to fourfold greater risk of dying compared with patients with negative troponin and CK-MB levels ( Fig. 18-4 ).
There had been debate about what the appropriate cut point(s) for troponin assays should be. Consensus panels recommended that a single cut point be adopted based on the 99th percentile in a cohort of healthy individuals and a coefficient of variation less than 10%. , However, work from trials of UA-NSTEMI has supported the prognostic importance of low-level troponin elevations, even below such cut points ( Fig. 18-5 ). Moreover, with the advent of ultrasensitive assays capable of detecting troponin at picogram per milliliter levels, troponin may be evolving from a semiquantitative variable (undetectable or a measurable level) to a true continuous variable that is detectable in all individuals. The prognostic significance in ACS of troponin levels below the 99th percentile remains to be defined.
Myoglobin is a small cytosolic protein found in myocardial and skeletal muscle and is one of the earliest markers to be released in the circulation following a MI. Although not specific for myocardial injury, myoglobin is a sensitive marker, especially in the first 4 to 8 hours after the onset of necrosis. Heart-type fatty acid binding protein (H-FABP) is another cytosolic protein released from the cardiomyocyte in response to myocardial injury. In a study that included patients across the spectrum of ACS, elevated levels of H-FABP were significantly associated with major cardiac events independent of other established clinical risk predictors and biomarkers. Whether either of these markers will be of value in the setting of ultrasensitive troponin assays remains to be determined.
Integrated Approaches
Although the prognostic information associated with each of the variables described is useful, focusing on a single variable does not permit the clinician to use all the information at his or her disposal. For example, a patient may have negative cardiac biomarkers, but be older, with multiple cardiac risk factors, have had a prior MI, and be presenting with severe angina with ST-segment depressions, despite being on an aspirin regimen. Clearly, this patient is at high risk for death or cardiac ischemic events over the ensuing days and weeks, despite not having an elevated CK-MB or troponin level. Thus, relying on one predictor while ignoring others may lead to misclassification of a patient’s risk.
The need for an integrated approach was recognized more than a decade ago with the Braunwald classification of unstable angina. Although typically used to grade the severity of the acute presentation, the classification system actually contains four axes—severity of acute symptoms, clinical circumstances, intensity of medical treatment, and electrocardiographic changes. The acute presentation (Class I, II, or III) was categorized as new-onset or crescendo angina without rest pain, angina at rest, but not within the preceding 48 hours, and angina at rest within 48 hours. The clinical circumstances (A, B, or C) were divided into secondary angina caused by an extracardiac condition that intensified myocardial ischemia, primary angina presumably caused by plaque rupture, and postinfarction angina. The intensity of medical treatment (denoted with subscripts 1, 2, or 3) ranged from angina occurring in the setting of no treatment, during treatment for chronic angina, and despite maximal anti-ischemic therapy. Finally, patients were divided into those with and without transient ST-T–wave changes during pain. Prospective validation of the Braunwald classification system confirmed the usefulness of such an approach. ,
The completion of several recent clinical trials in which a wealth of baseline clinical, electrocardiographic, and serum marker data was gathered offered the opportunity to develop modern integrated approaches to prognostication in UA-NSTEMI. Using these data, several risk scores have been developed. One example is the TIMI Risk Score for UA-NSTEMI, which was designed to provide clinicians with a prognostic tool with high discriminatory ability using baseline variables that are part of the routine medical evaluation.
TIMI Risk Score for Unstable Angina and Non–ST-Segment Elevation Myocardial Infarction
Developing a Model
The TIMI Risk Score for UA-NSTEMI was developed in a derivation cohort consisting of 1957 patients who were randomized to the unfractionated heparin (UFH) arm of the TIMI 11B trial. Potential predictor variables were selected from baseline characteristics that could be readily identified at presentation and that had previously been reported to be important variables in predicting outcome ( Table 18-1 ). Using multivariable logistic regression, seven independent, statistically significant predictors of the composite end point at 14 days were identified: age, 65 years or older; three or more risk factors for CAD; prior coronary artery stenosis 50% or higher; severe anginal symptoms (two or more anginal events in prior 24 hours); use of aspirin in last 7 days, ST-segment deviation, 0.05-mV or more; and an elevated serum cardiac marker level (CK-MB or cardiac-specific troponin). The final model demonstrated excellent calibration of the model predictions to the observed event rates (Hosmer-Lemeshow statistic, 3.56 df8 ; P = .89) as well as good overall predictive capacity of the model (C-statistic, 0.65). Because the magnitudes of the prognostic significance (i.e., the odds ratios) for each independent predictor variable were similar, the TIMI Risk Score for UA-NSTEMI was constructed as the simple arithmetic sum of the number of predictors. Thus, the risk score is calculated by assigning 1 point for each variable that is present ( Table 18-2 ).
| Characteristic | Univariate Analysis | Multivariable Analysis | ||
|---|---|---|---|---|
| OR (95% CI) | P Value | OR (95% CI) | P Value | |
| Age ≥65 years | 1.60 (1.25-2.04) | <.001 | 1.75 (1.35-2.25) | <.001 |
| Three or more risk factors for CAD * | 1.45 (1.10-1.91) | .009 | 1.54 (1.16-2.06) | .003 |
| Prior coronary stenosis ≥50% | 1.73 (1.34-2.23) | <.001 | 1.70 (1.30-2.21) | <.001 |
| Prior MI | 1.27 (0.99-1.63) | .06 | ||
| Prior CABG | 1.35 (0.97-1.88) | .07 | ||
| Prior PTCA | 1.62 (1.16-2.26) | .004 | ||
| ST deviation ≥0.05 mV | 1.40 (1.06-1.85) | .02 | 1.51 (1.13-2.02) | .005 |
| Severe anginal symptoms (≥two anginal events in prior 24 hours) | 1.57 (1.24-2.00) | <.001 | 1.53 (1.20-1.96) | .001 |
| Use of aspirin in last 7 days | 1.86 (1.26-2.73) | .002 | 1.74 (1.17-2.59) | .006 |
| Use of IV UFH within 24 hours of enrollment | 1.18 (0.92-1.51) | .19 | ||
| Elevated serum cardiac markers (CK-MB or troponin) | 1.42 (1.12-1.80) | .004 | 1.56 (1.21-1.99) | <.001 |
| Prior history of CHF | 0.90 (0.53-1.53) | .70 | ||
* Risk factors included family history of CAD, hypertension, hypercholesterolemia, diabetes, or being a current smoker.
| Charactersitic | Points |
|---|---|
| Historical | |
| Age ≥65 years | 1 |
| Three or more risk factors for CAD | 1 |
| Known CAD (stenosis ≥50%) | 1 |
| Aspirin use in past 7 days | 1 |
| Presentation | |
| Recent (≤24 hr) severe angina | 1 |
| ST deviation ≥0.5 mm | 1 |
| ↑ Cardiac markers | 1 |
| Risk Score = Total Points | (0-7) |
Of note, the application of the C-statistic in the setting of prognostication has been debated. , The C-statistic (area under the receiver operating characteristic curve) is useful in the setting of diagnostic testing, where the sensitivity and specificity of tests are important in discriminating diseased versus nondiseased patients. However, the C-statistic may not be the ideal parameter to assess models or variables that aim to predict future risk or separate subjects into distinct risk groups. For example, the C-statistic is altered minimally by accepted risk factors such as hypertension and cholesterol levels, although these variables are clearly important in classifying an individual’s risk of cardiovascular disease. Thus, it has been suggested that prognostic models should be evaluated based on their calibration and ability to reclassify individuals, in addition to their ability to alter the C-statistic.
Clinical Usefulness of the Model
Application of the TIMI Risk Score for UA-NSTEMI to patients in the UFH derivation cohort revealed that the score has several features desirable in risk stratification. First, there was a progressive, significant pattern of increasing event rates for the composite end point of death, MI, and urgent revascularization ( P < .001 by χ 2 for trend; Fig. 18-6 ) with increasing TIMI Risk Score. Second, this pattern was also seen for each individual component of the composite end point (P < .001 by χ 2 for trend for each component). Third, the TIMI Risk Score categorized patients into a wide range of risk. Patients with a score of 0 or 1 had less than a 5% rate of death, MI, or urgent revascularization, whereas patients with a score of 6 or 7 had more than a 40% rate of these events.
