Discussion
We performed a comparative validation of GRACE and CHA2DS2VASc risk scores to predict ischaemic stroke after ACS. The main clinical finding of our study was that the discriminative ability of GRACE risk score to predict the primary end point was similar to CHA2DS2VASc, even in patients with AF.
Although the GRACE risk score was validated to quantify the risk of mortality and reinfarction in the acute phase17 or in the follow-up,11 ,12 little is known about its usefulness in predicting post-ACS stroke.18 Our study provides new evidence in this setting.
Stroke, a relatively rare complication after ACS, is associated with high mortality.5 ,18 It may occur due to a multitude of reasons, such as atherosclerotic disease and thromboembolic events. Stroke represents one of the major causes of morbidity among hospital survivors of ACS. The identification of predictors for stroke in patients with ACS may help to optimise the treatment of high-risk patients. This could prevent fatal postdischarge consequences.
The incidence of stroke post-ACS has been estimated in several trials.19 This is highest in the first 5 days,1 and subsequently it is reduced.18 ,20 Most of studies had assessed the incidence of stroke after ACS during the in-hospital phase, at 30 days or in the first year. In recent years, there have been some studies about the trends in the incidence of post-ACS stroke. Brammas et al published a reduction in the incidence of mortality in patients with post-ACS stroke, in the past few years, due to the improved treatment based on evidence.19
Clinical practice guidelines recommended risk stratification in ACS,21 ,22 in the acute phase and after hospital discharge, to improve the prognosis of these patients. To date, we have several risk scores to estimate the risk of thrombotic and thromboembolic events in ACS and AF, respectively. These risk scores have been developed and validated in the past years.
The 6-month GRACE11 risk score is a simple tool for predicting mortality in patients with ACS. It was described in 2004. This is derived from the largest multinational registry and includes the complete spectrum of patients with ACS, based on independent predictors of outcome. The 6-month GRACE risk score was developed and validated with more than 20 000 patients, between 1999 and 2003, who were included in the GRACE registry (14 countries, 94 hospitals). This score allows us to calculate the probability of mortality during the first 6 months after discharge in all ACS spectrum (c-index higher than 0.70). The application of the GRACE risk score at admission was recommended by the clinical practice guidelines for risk stratification in ACS. The 6-month GRACE risk score was validated entirely across the wide range of current patients with ACS12 and demonstrated superiority over other risk scores. Its predictive value was also further validated over 6 months (even in the 5-year follow-up).13
CHA2DS2VASc9 is a simple risk stratification schema to determine thromboembolic risk in patients with non-valvular AF. It was validated in 2009; 5333 ambulant and hospitalised patients with AF were enrolled from 2003 to 2004. The CHADS2 is commonly used to assess risk of stroke, but CHA2DS2VASc scores have better discrimination of stroke risk, particularly in low-risk patients. Clinical practice guidelines recommended calculation of CHA2DS2VASc risk scores to determine whether patients with non-valvular AF need antithrombotic therapies for the prevention of stroke and systemic embolisation.
To date, no risk scores specifically estimate the risk of stroke after ACS. However, GRACE and CHA2DS2VASc risk scores could be useful in this setting. Our trial demonstrated that the accuracy of GRACE risk score was similar to the CHA2DS2VASc score to predict post-ACS stroke. This is important because with only one tool, which was initially designed to assess thrombotic risk, we can predict thromboembolic risk after ACS. We can identify patients who will benefit from a more potent antithrombotic treatment at discharge. Although our results are striking, they are in line with other studies previously published, such as Barra et al.23 Also this observational retrospective single-centre cohort study, with fewer patients (n=1.711 patients and post-ACS stroke rate of 4.3%) and a shorter follow-up (median 17.4±8.7 months), shows great predictive ability of the 6-month GRACE risk score (c-index 0.782±0.019). In our cohort, we showed there was no difference between GRACE and CHA2DS2VASc risk scores to predict the risk of stroke after ACS. We have identified the GRACE risk score as a new independent predictor of stroke post-ACS.
The stroke has been considered a complication after ACS; its frequency is especially increased in the first few months.19 One of the possible reasons that the GRACE risk score could predict the risk of stroke post-ACS is because many variables were previously demonstrated as independent predictors of post-ACS stroke. Thus, in recent years, predictors of post-ACS stroke have been identified in several trials: elderly, female sex, heart failure, coronary heart disease, AF, prior stroke, diabetes mellitus, timely revascularisation therapy, secondary prevention therapies and renal function.4 ,6 ,18 ,2 ,24–29
Advanced age, a variable with great weightage in GRACE risk score, has proven to be a powerful risk factor for stroke in this population.2 ,28 Several reports have shown that impaired renal function is also an efficient predictor of stroke and systemic embolism, and inhospital mortality of these patients. This was recently validated by Piccini et al in the ROCKET AF and ATRIAL Study Cohorts.22 Both elevated heart rate (more than 100 bpm)28 ,30 as well as a high Killip class17 at admission have demonstrated increased mortality, reinfarction as well as the greater possibility of post-ACS stroke. Also, the ST-segment changes on index ECG, specially STEMI, predicts more events.31 Likewise, the value of GRACE at admission, GRACE risk score and higher markers of myocardial damage are predictors to stroke post-ACS.17 ,31 Percutaneous coronary intervention during hospitalisation predicts a decreased risk of ischaemic stroke;30 Van De Graaff et al6 showed a significant relationship between timely revascularisation therapy and risk of inhospital ischaemic stroke. The patients with previous coronary heart disease and especially anterior myocardial infarction showed an increase in post-ACS stroke.3 Heart failure and reduced left ventricular ejection fraction4 ,18 ,32 had stronger association with mortality and also have been associated with increased risk of stroke after ACS. Because of this, although the GRACE score was not designed to determine the risk of stroke, it has good discriminative ability and a good correlation with CHA2DS2VASc to predict the primary end point. GRACE risk score included some variables which have not been included in the CHA2DS2VASc score. The high-risk GRACE score patients are usually older patients with renal failure and more Killip; therefore, these patients could be at higher risk of stroke.
Clinical implications
In recent years, there has been increasing interest to show the risk of stroke after ACS, and the risk factors associated with it. The GRACE risk score could eventually contribute to a better risk stratification and help us make decisions about interventions to reduce stroke after ACS in high-risk patients. Nowadays, the estimation of cardiovascular risk and individualisation has become a priority. GRACE risk score is used in routine clinical practice for risk stratification to optimise the treatment. Our study shows a new utility to predict stroke in the follow-up post-ACS.
Limitations
These data must be interpreted in the context of this study’s limitations. It is a retrospective analysis of clinical single centre data. Our small-sized sample should be considered the main limitation of this study. In fact, the relatively low absolute number of ischaemic stroke events during follow-up reinforces the need for more studies with larger cohorts of patients to confirm the usefulness of GRACE risk score. However, irrespective of its potential future clinical validation, our study has supported the applicability of GRACE risk score in patients with ACS prior to discharge. Moreover, as many patients who died during follow-up were not autopsied or previously observed at the emergency department, it is very hard, if not impossible in some cases, to know whether stroke was the cause of death. Therefore, the true incidence of stroke was probably underestimated.
Conclusions
The GRACE model is a clinical risk score that facilitates the identification of individual patients who are at high risk of stroke after ACS discharge. New therapeutic interventions that have the potential to limit preventable post-ACS stroke may have the greatest impact on this vulnerable population.