Non-ST-elevation acute coronary syndrome (NSTE-ACS) is a heterogeneous clinical entity in which prospective risk stratification is essential to identify patients at high risk of adverse events1,2 and to offer them, if appropriate, an early invasive strategy to improve short and long-term prognosis.3,4

Current clinical guidelines from the European Society of Cardiology5 and the American Heart Association/American College of Cardiology6 recommend the use of risk scores. Both recommend the use of the Global Registry of Acute Coronary Events (GRACE) score at admission to predict the risk of in-hospital mortality. The GRACE score is derived from eight clinical, electrocardiographic and laboratory variables: five semiquantitative (age, systolic blood pressure, heart rate, Killip class and serum creatinine level) and three dichotomous (cardiac arrest during presentation, ST-segment deviation and elevated cardiac enzymes), and ranges from 1 to 372 points.2 It has good accuracy for predicting in-hospital death in this cohort of patients.

Patients with NSTE-ACS and a GRACE score ≤108 points at admission are classified as low-risk, defined as <1% risk of in-hospital death,2 and as suitable for initially conservative management.5,6 However, risk scores, including GRACE, are not intended to determine the severity of coronary artery disease (CAD) and, if an initial conservative approach is adopted in accordance with the calculated risk, this may result in failure to identify patients with severe CAD requiring revascularization, such as left main and/or three-vessel disease (LM/3VD).7

Several studies have evaluated the performance of clinical risk scores for detecting and predicting the severity of CAD.

In patients with NSTE-ACS, the TIMI risk score has a modest ability to predict obstructive CAD, defined as the presence of at least one ≥50% stenosis,8,9 while a TIMI score >4 shows a modest10 to good11 performance in detecting three-vessel CAD. In one study, sensitivity of 53% and specificity of 83% were demonstrated for this diagnosis (area under the curve [AUC] 0.71; 95% CI 0.61-0.81, p<0.001).11

Variability in accuracy for prediction of CAD using the GRACE score is greater due to the different cut-off points studied. Its performance is good for the detection of multivessel disease (AUC 0.72; 95% CI 0.64-0.80, p=0.001)12 but only modest for three-vessel disease with a cut-off of >11911 or LM/3VD with a cut-off of >117.13 The GRACE score has better predictive ability for three-vessel disease when the cut-off is >133,10 and higher sensitivity but lower specificity than the TIMI risk score.11

This study aimed to determine the prevalence of LM/3VD in non-ST-elevation myocardial infarction (NSTEMI) patients with a GRACE score ≤108 at admission, to assess its impact on clinical outcomes, and to determine its predictors.

The major findings of this study are that patients admitted with NSTEMI and a low-risk GRACE score (≤108) have a low risk of in-hospital adverse events despite almost one fifth of them having high-risk coronary anatomy, but that these patients had not only a worse long-term prognosis, with higher one-year mortality, but also higher in-hospital morbidity, as shown by increased length of hospital stay, and a trend for higher absolute values of mortality and MACCE. The findings indicate that these patients are not in fact at such low risk and should be identified promptly.

Another important finding was that simple clinical variables appeared as independent predictors of LM/3VD in this population: two of them included in the GRACE score (age and heart rate), male gender, and important known comorbidities (heart failure and PAD).

The relationship between the GRACE risk score and the severity and complexity of CAD has been studied in several publications,8–13,16,17 however very few stratified patients within the three GRACE risk categories and, to our knowledge, none analyzed predictors of LM/3VD disease in low-risk patients.

In the largest study, Beigel et al.16 assessed 923 consecutive patients with moderate-high risk NSTE-ACS enrolled in the Acute Coronary Syndrome Israeli Survey and found that the major predictors of high-risk coronary anatomy (defined as >50% stenosis in the left main, >70% in the proximal LAD and/or two- or three-vessel disease involving the LAD) were GRACE score >140 (high-risk) (OR 1.88, 95% CI 1.29-2.75, p<0.001), PAD (OR 1.88, 95% CI 1.62-5.8, p<0.001) and chronic renal failure (OR 1.7, 95% CI 1.02-2.80, p=0.03). Likewise, Mahmood et al. found that patients with a GRACE score >133 had a higher likelihood of LM/3VD (OR 3.41, 95% CI 2.16-5.36, p<0.01).10 Other groups have set out to correlate the severity of CAD in NSTE-ACS patients with their GRACE score. GRACE >119 had only moderate accuracy for prediction of three-vessel disease, with sensitivity of 80% and specificity of 55% (AUC 0.68; 95% CI 0.58-0.78, p=0.001),11 while GRACE >117 had modest accuracy for predicting LM/3VD, with sensitivity of 66% and specificity of 59% (AUC 0.66; 95% CI 0.58-0.74, p=0.01).13

By contrast, in a study by Barbosa et al.8 the GRACE score was unable to identify the presence of LM/3VD (AUC 0.59; 95% CI 0.48-0.70, p=0.13), and there were no statistical differences between the three GRACE tertiles regarding the prevalence of LM/3VD (25%, 33% and 37%, respectively; p=0.56). Santos et al.9 also showed that the GRACE score had only moderate predictive ability even for the presence of at least one-vessel disease (AUC 0.62; 95% CI 0.573-0.673, p<0.001).

Two studies assessed the severity of CAD according to the number of diseased vessels stratified by GRACE risk categories. In the study by Cakar et al.,12 356 patients with NSTE-ACS were analyzed retrospectively, 39.6% of them with a low-risk GRACE score. The prevalence of LM/3VD was 15.5%, with no in-hospital mortality and a low rate of reinfarction or revascularization (2.1%). These findings were similar to our study. However, they did not assess potential predictors of severe CAD in this population. In a retrospective study of 95 patients with NSTE-ACS, another group reported that 40% of the population had a low-risk GRACE score, with a prevalence of 39.4% for three-vessel disease and 5.2% for left main disease.17 No report on outcomes was available.

Our search of the literature revealed no studies regarding determination of predictors of LM/3VD in low-risk GRACE populations.

Many simple predictors of worse prognosis are correlated with severe CAD. Clinical variables like increasing age and high resting heart rate are well-known risk factors.18 Both these variables are features of the GRACE score and in our study were also found to be predictors of LM/3VD, which is in line with the literature.19 This could mean that, even for low-risk patients, extra care should be taken regarding tachycardic and elderly patients.

PAD is associated with increased risk for cardiovascular death20 and higher in-hospital mortality in ACS,21 and there are several reports associating its presence with greater severity and complexity of CAD.19,22 In our study, despite the low prevalence of PAD in both groups, which was significantly less than in modern cohorts of ACS patients,23 it was an independent predictor of LM/3VD, as it also was in the study by Beigel et al.16 and in a meta-analysis.19 An ankle-brachial index of <0.9 is known to be a predictor of PAD24 and of multivessel CAD in patients admitted with ACS,25 so this simple measure from the physical examination could potentially improve detection of this important comorbidity.

In heart failure, long-term prognosis is directly related to the angiographic extent and severity of CAD,26,27 in patients with both reduced and preserved systolic function.28 In the setting of ACS, development of acute heart failure is considered a powerful predictor of LM/3VD,19 but less is known about the predictive ability of a previous history of chronic heart failure. However, in our study neither the proportion of patients in Killip class II-IV nor the incidence of new-onset heart failure during hospitalization differed between the groups. There were also no differences in the incidence of left ventricular dysfunction. These findings are not surprising, since the population only included patients with a low-risk GRACE score, which means that most patients would be in Killip class I. Another finding is that a previous history of heart failure was an independent predictor of LM/3VD. Nonetheless, this result should be interpreted with caution, since it reached only a borderline significant p-value and the number of patients affected was low.

We also found that male gender was an independent predictor for the presence of LM/3VD, which is in line with other studies that show that men with myocardial infarction have more severe CAD.29–32

In our study, there were no differences between groups in in-hospital adverse events. Although patients with LM/3VD had non-significantly more MACCE and higher in-hospital mortality, the rates were low, as would be accurately predicted by the GRACE score.2 However, a significant interaction was found with the secondary endpoint of urgent referral for CABG during hospitalization, which was more frequent in the LM/3VD group. Indeed, one reason for the need to be able to predict severe CAD using clinical risk scores is to improve assessment of patients who would most likely be indicated for CABG. This is important because it could have therapeutic implications for the timing and duration of dual antiplatelet therapy.

Patients admitted for NSTEMI with a low-risk GRACE score of ≤108 at admission had few in-hospital adverse events and low mortality, even though almost one in five had LM/3VD. Nonetheless, despite this apparently good in-hospital prognosis for hard clinical endpoints, these patients had higher absolute values of in-hospital mortality and MACCE, and LM/3VD was associated with longer hospital stay and higher one-year mortality, and was also an independent predictor for CABG during hospitalization. Simple clinical variables are independent predictors of this high-risk coronary anatomy, like older age, higher resting heart rate, male gender, previous diagnosis of heart failure and PAD, and could help the clinician to identify patients who would benefit from an early invasive strategy.

The authors have no conflicts of interest to declare.