Risk Factors for Ischemic Stroke After Acute Coronary Syndrome

Due to the sensitive nature of the data collected for this research, they are not made widely available. This study is based on the retrospective registry study MADDEC (Mass Data in Detection and Prevention of Serious Adverse Events in Cardiovascular Disease) recording mass data of all patients with cardiac diseasetreated in Tays Heart Hospital, which is a sole provider of specialized and tertiary health care for seriously ill patients in the region of Pirkanmaa (Finland).²⁵ Data for the registry are collected from multiple sources, including electronic health records, written patient records, and the prospectively updated KARDIO registry, which records structured data from all patients. The data of the KARDIO registry are maintained by treating nurses and physicians using an online tool. As a result, the MADDEC database comprises high‐quality phenotype information of all patients treated for ACS.²⁵ The study design was approved by the scientific monitoring board of Pirkanmaa hospital district. Due the nature of the study, informed consent was not required. The study complies with the Declaration of Helsinki on the ethical principles for medical research. This study design and baseline phenotype data collection have been thoroughly described previously.^{1, 25}

All patients undergoing coronary angiography for ACS between January 1, 2007, and December 31, 2018, were followed up for incident strokes during and after hospitalization. For the purpose of this study, only the first ACS for each patient was selected as the baseline/index event even if multiple events were recorded. A total of 10 314 patients were treated for ACS during this time period. Additionally, patients living permanently outside the hospital district of Pirkanmaa were excluded from the study due to no access to written medical records for the verification of all baseline data and end points during the follow‐up. As a result, we analyzed the data of 8049 patients treated for ACS. We also collected data and analyzed the impact of the primary treatment modality of ACS (PCI, coronary artery bypass grafting [CABG], and conservative treatment) in early‐onset IS risk.

CHA₂DS₂‐VASc points were calculated on the basis of available risk factors (ie, age, sex, history of HF), hypertension, prior stroke, transient ischemic attack or thromboembolism history, vascular disease history (prior MI, peripheral artery disease, or aortic plaque), and type 2 diabetes data.^{9, 11, 26} For early‐onset IS, the first 3 CHA₂DS₂‐VASc score categories were combined to 1 reference group for statistical analysis due to a low number of events. Killip classification (I–IV) for AMI (class I for no sign of congestion, class II for S3 [ventricular gallop] and basal rales on auscultation, class III for acute pulmonary edema, class IV for cardiogenic shock) was used as a covariate considering the development of HF in statistical modeling.²⁷

This contemporary retrospective registry study brings forth the most common risk factors for IS after ACS, which also predict overall high cardiovascular risk. Our results are based on an unprecedented series of consecutive patients in a limited geographic area in which there is no other cardiac care for patients with ACS. Our findings stress the importance of AF/AFL, high CHA₂DS₂‐VASc score despite atrial arrhythmias, and Killip classification in IS risk evaluation. Our results are in line with previous literature, and they deepen the understanding of different relevant factors in IS risk prevention. To the best of our knowledge, there are no prior studies disclosing the role CHA₂DS₂‐VASc score with or without AF/AFL has on IS risk after ACS. Additionally, it is important to note that previous research tends to focus on solely MI, leaving out unstable angina pectoris, which undermines the widespread comprehension of IS risk factors in the ACS population.^{2, 3, 4, 5, 6}

A previous retrospective meta‐analysis by Witt et al³ based on studies published in 1978 to 2004 suggests the most substantial risk factors for IS after MI to be advanced age, diabetes, hypertension, previous stroke or MI, anterior location of index MI, atrial fibrillation, HF, and non‐White race in general. Although these results are based on somewhat outdated research, focusing on in‐hospital strokes and not accounting for unstable angina pectoris, our findings are very similar. The only striking difference was that Witt et al found Killip class to have no impact on IS risk, even though HF was a strong indicator for stroke. Also, our study was conducted in Finland, with most patients being White, and therefore no stand can be taken on what impact race and ethnicity has on stroke risk.

Our other major findings relating to increased risk of especially late‐onset (after 30 days) stroke after ACS include previous stroke, peripheral artery disease, history of kidney failure, and severity of coronary artery disease, which corresponds with recent literature.^{2, 4, 6, 28} Female sex, on the other hand, has previously been often linked to higher risk of stroke, yet our results suggest there to be no difference between sexes in early‐onset IS and the risk to the opposite in late‐onset IS risk, although it was not evident in multivariable analyses.^{2, 6, 21} A prospective trial by Sampson et al⁸ with 14 703 patients also found female sex not to be a risk factor for early‐ (<45 days) or late‐onset (>45 days) stroke.

High CHA₂DS₂‐VASc score has been proven to increase the risk of stroke despite AF/AFL in earlier literature, and we demonstrated similar and strong impact also after ACS in both early‐ and late‐onset IS.^{9, 10, 11} Notably, patients with low scores (1–3) have only a minor risk of early‐onset IS (0.4% occurrence rate), and, conversely, patients with ≥6 points have a >6‐fold risk in comparison. Similar magnitude difference was observed in the risk of late‐onset IS when comparing patients with only 1 CHA₂DS₂‐VASc point and patients with ≥6 points. These findings can be deemed significant and, especially combined with serum creatinine levels, could be valuable in IS risk evaluation for patients with ACS in the future.

A prospective VALIANT (Valsartan in Acute Myocardial Infarction) study by Szummer et al⁴ demonstrated Killip class III to have a 1.5 times higher in‐hospital stroke risk after MI compared with class I, whereas class II had a similar risk as that of patients without HF. Similarly, we found the Killip class III to have an impact in early‐onset IS (>2‐fold risk). When considering long‐term IS risk, patients with Killip II or III had also notably higher risk of IS, but the magnitude was attenuated (SDH, 1.26 [95% CI, 1.02–1.56]). In turn, the class IV was not associated with the risk of IS in either early‐ or late‐onset IS in multivariable analyses. This might be explained by the high death rate of patients with Killip class IV HF and the resulting survival bias.

Low LVEF has been linked to increased in‐hospital stroke risk in a retrospective registry study by Hachet et al,² whereas a similar research by Podolecki et al⁶ found low LVEF (<35%) not to be a risk factor for stroke in the long term. Our results suggest the role of low LVEF increasing IS risk, especially in the short term, since higher LVEF was the strongest protective factor in early‐onset IS. Thus, it could be concluded that low LVEF is a protective factor for IS risk, specifically during the first month after ACS. More comprehensive studies regarding the role of LVEF and the cause of such stroke is still needed, especially since the number of early‐onset cardioembolic strokes was too low for analysis in our study. Additionally, the effect of the proportion of IS due to cardioembolism without AF/AFL being considerably higher compared with cardioembolism with AF/AFL in early‐onset IS and vice versa in late‐onset IS has been thoroughly demonstrated with multiple time points in our previously published research based on the same data.¹

Treatment of ACS by CABG compared with PCI has previously been linked to increased IS risk, and we had similar findings.^{29, 30} However, when AF/AFL status and the extent of coronary artery disease (factors also influencing the selection of the treatment modality itself) were included in the risk analysis, the association did not persist. Similarly, Head et al³¹ demonstrated in their research how the long‐term prognosis was significantly better when patients with multivessel disease were treated by CABG instead of PCI. Therefore, the risk of IS depending on the treatment modality of ACS is multifactorial and treatment by PCI instead of CABG does not always lead to better results and, depending on the patient, is not necessarily even preferable.

One clear limitation in our study was the deficiencies of specific information on, in particular, anticoagulant medication at the time of the incident stroke events. This is partly due to the fact that many patients died of stroke before reaching the hospital or the information was not reliably extractable or unambiguous despite having access to all written medical records from specialized health care at the time of the event (patients unable to give an accurate account of medical treatments or medication poorly described by treating neurologists). For this reason, it is hard to draw conclusions on the efficiency of these medications in IS prevention. However, we can estimate that over two‐thirds of patients who suffered IS were not anticoagulated at the time of the event and could therefore perhaps benefit from more aggressive anticoagulative therapy if correctly identified beforehand. Further research in this regard is, without fail, still needed.

Another disadvantage is the fact that even though the phenotype data was collected by treating physicians in the prospectively updated KARDIO registry, a majority of the data were collected retrospectively, and our results are based on post hoc analyses of the MADDEC database. For this reason, despite the strong observations, our results could benefit from independent replication. Furthermore, we also included only patients who underwent invasive evaluation (ie, coronary angiography) for ACS and thus left out conservatively diagnosed and treated patients with ACS.¹ However, in our study center, only 8.4% of patients are ruled out of invasive evaluation due to poor general condition and functional status, and have a very high overall death rate, which reduces the clinical impact of possible secondary preventive measures in any case.³² An additional factor to consider is the causal relationship between serious risk factors and death. For example, risk factors such as cardiogenic shock (Killip class IV), which is almost always due to systolic HF, was seen to associate less significantly with the risk of IS in the analyses. This is likely to be caused by 2 phenomena: the overall high death rate related to cardiogenic shock, leaving fewer patients alive to suffer an IS, and strong selection bias, leaving only the most otherwise fit to recover and survive.

Our research establishes the most common risk (and protective) factors for early‐ (0–30 days after ACS) and late‐onset (31 days to 14 years after ACS) IS and demonstrates the significant role of AF/AFL, high CHA₂DS₂‐VASc score despite AF/AFL, and Killip classification II to III in IS risk evaluation. These findings enable improving secondary prevention of this serious complication not just in patients with MI but all patients with ACS alike in the future.

This research was supported by Business Finland research funding (grant No. 4197/31/2015) as part of a collaboration between Tays Heart Hospital, University of Tampere, VTT Technical Research Centre of Finland Ltd, Politecnico di Milano, GE Healthcare Finland Ltd, Fimlab Laboratories Ltd, and Bittium Medanalytics Ltd; the Academy of Finland: grants (322098 and 286284 for Dr Lehtimäki), Competitive State Research Financing of the Expert Responsibility area of Tampere University Hospitals (Grant X51001 for Dr Lehtimäki, Z60104 for Dr Hernesniemi); Finnish Foundation for Cardiovascular Research (Dr Lehtimäki); Tampere Tuberculosis Foundation (Dr Lehtimäki, NOK); Emil Aaltonen Foundation (Dr Lehtimäki and NOK); Diabetes Research Foundation of Finnish Diabetes Association (Dr Lehtimäki); EU Horizon 2020 (Grant 755320 for TAXINOMISIS and Grant 848146 for To Aition); and Tampere University Hospital Supporting Foundation (Dr Lehtimäki). In addition, this research was supported by Competitive State Research Financing of the Expert Responsibility Area of Tampere University Hospital, Tampere University Hospital support association, Business Finland research funding (Grant 4197/31/2015), Finnish Foundation for Cardiovascular Research, Tampere University Kalle Kaihari Trust, and Aarne Koskelo Trust, independent and impartial research foundations.

None.

This research was made possible by the cooperation of the administration, medical registries, and the Tampere University Hospital and Tays Heart Hospital, University of Tampere.