Background

Acute myocardial infarction (AMI) affects 4% to 14% of patients hospitalized in the intensive care unit (ICU) for non-cardiac disease [1,2,3] (vs. ≃2 per 1000 admissions for non-cardiac causes in general wards [4]) and is independently associated with increased mortality [1]. In this setting, myocardial ischemia caused by obstructive coronary artery disease (OCAD), designated as a type 1 myocardial infarction [5], is one of the main underlying mechanisms [6]. Decisions regarding use of reperfusion therapy by percutaneous coronary intervention (PCI) in association with dual antiplatelet therapy [6, 7] should be based on the balance of benefit versus risk for each patient, including the risks of ischemia, bleeding, and death [8]. In these patients, inflammation, a prothrombotic state, and endothelial dysfunction [9,10,11,12,13] may increase the ischemic risk [14,15,16], while platelet disorders [17] and coagulopathy [18] may increase the risk of bleeding [19]. Thus, assessment of the clinical risk is challenging [20], which may explain, in part, why half of these patients do not receive standard of care management as defined in European guidelines [6, 7]. We therefore conducted a multicenter retrospective study in patients with AMI with OCAD during an ICU stay for non-cardiac disease, to assess the incidence and predictive factors of poor outcome, using a composite endpoint of severe ischemic event (AMI recurrence, ischemic stroke), major bleeding, or all-cause death.

Material and methods

Selection of patients

In this multicenter retrospective study in three medical ICUs and one mixed medical-surgical ICU of four university teaching hospitals (Paris, France), all consecutive adult patients who had an AMI with OCAD during their ICU stay between January 2012 and December 2018 were considered for inclusion. Diagnosis of AMI with OCAD was based on (i) an elevated cardiac troponin value greater than the 99th percentile of the upper reference level with an increase and/or decrease in troponin values (analytical characteristics of cardiac troponin assays in the different centers are provided in Additional file 1: Table S1) with at least one sign suggestive of myocardial ischemia (typical chest pain, electrocardiogram [ECG] changes, or significant left ventricular (LV) systolic dysfunction [LV ejection fraction [LVEF] ≤ 45%] on echocardiography [2]) [2], and (ii) a coronary angiography showing OCAD (detailed definition in Additional file 1: Table S2) [5]. Exclusion criteria were cardiac disease (myocardial infarction, myocarditis, cardiac rhythm disorders, cardiogenic shock, or cardiogenic pulmonary edema) as a principal diagnosis on ICU admission and cardiac surgery, PCI, or coronary artery bypass grafting within the month prior to ICU admission.

ICU patients who had coronary angiography were identified by the investigator of each participating center, either from hospital medical reports, using the function “research for file in which the word ‘coronary angiography’ occurs” of Microsoft Windows®, or through a search using the following International Classification of Diseases (10th revision) codes: I21 (‘AMI’), I22 (‘subsequent AMI’), R93.1 (‘abnormal findings on diagnostic imaging of heart and coronary circulation’), Z13.6 (‘special screening examination for cardiovascular disorders’), I25.1 (‘atherosclerotic heart disease’). The medical records (including clinical observations, hospitalization reports, as well as electrocardiogram, biological and radiological examinations) of each identified patient were reviewed by the investigators to first verify the inclusion criteria, and second to collect the data. The presence of typical chest pain up to 7 days prior to the day of troponin elevation was noted (it was considered not to be present in patients under sedation). All ECGs performed on the day of troponin elevation were systematically reviewed. This observational, non-interventional analysis of medical records was approved by the Institutional Review Board of the French Society of Intensive Care (CE SRLF 20-76). As per French law, no informed consent was required for this type of study.

Collection of data

Patient demographics, past medical history, prior antithrombotic treatments, admission category (medical, scheduled surgery, emergency surgery), the principal diagnosis, and the Simplified Acute Physiologic Score II (SAPSII [21]) were recorded on ICU admission. Sepsis and septic shock were defined in accordance with the Sepsis-3 definition [22], and sites of infection were recorded. At the onset of the AMI, the thrombolysis in myocardial infarction (TIMI) risk score [23], the sequential organ failure assessment (SOFA) score [24], the presence of cardiogenic shock, routine blood test results, and details regarding the management of organ failure and of the AMI were collected. Early mechanical reperfusion therapy was defined as coronary reperfusion by PCI or coronary artery bypass graft (CABG) surgery within the first 24 h for ST elevation AMI (STEMI), and within 72 h for non-ST elevation AMI [6, 7].

Outcomes

The primary endpoint was the occurrence of a composite clinical outcome, including a severe ischemic event (AMI recurrence, stroke), major bleeding (according to the Bleeding Academic Research Consortium, BARC [25]), or death from any cause from the day of AMI (day-1) until hospital discharge. Secondary outcomes were the occurrence of individual components of the primary outcome from the day of AMI until hospital discharge (Detailed definitions in Additional file 1: Table S2) and were not mutually exclusive.

Statistical analysis

Categorical variables are given as numbers (percentages) and quantitative variables as medians (interquartile ranges [IQR]). Associations with the composite primary outcome were tested using standard Cox models. Potential predictive factors were chosen according to their clinical relevance and their statistical significance (p ≤ 0.05) in the primary outcome univariate analysis. To avoid overfitting, we considered that we could enter a maximum of four variables in our primary outcome model (in view of the 48 events observed) [26]. A multivariable model was built for the primary outcome only, as the number of events was judged too low to avoid overfitting for the other outcomes. Associations with secondary outcomes (individual components of the composite primary outcome) were tested using univariate cause-specific Fine-Gray models for the first occurrence of severe ischemic event or the first occurrence of major bleeding event (accounting for the competing risk of death), and by standard Cox models for all-cause in-hospital mortality. The proportional hazard assumption was assessed through inspection of Schoenfeld residuals. Sensitivity analyses accounting for time and center effect were performed. No power calculation was necessary in view of the methodology used. Hazard ratios (HRs) were estimated and are reported with their 95% confidence intervals (CIs). The level of significance was set a priori at 0.05. Statistical analyses were performed with R software 3.6.0 version for Mac (Foundation for statistical Computing, Vienna, Austria).

Results

Population characteristics

During the 7-year study period, 637 adult patients with an AMI had coronary angiography (2.2% of the patients admitted, Fig. 1). Among this population, 96 patients (median age of 69 years [60–78]; 74 men and 22 women, 59 [61%] admitted for sepsis/septic shock) met the study inclusion criteria (Fig. 1, main characteristics in Table 1). The AMI occurred on the day of ICU admission in 83% of the patients (min–max: 0–10 days). On the day of the AMI, the median TIMI and SOFA scores were 4 [3–5] and 8 [3–11], respectively, and the median cardiac troponin peak value was 131 (44–303) times the upper reference limit (URL). Typical chest pain, ECG modifications, and significant LV systolic dysfunction were observed in 28 (29%), 87 (91%), and 50 (52%) patients, respectively (Table 2). Coronary angiography was performed a median (IQR) of 1 (0–6) day after AMI and revealed one vessel-disease, two vessel-disease, and three vessel-disease in 52 (54%), 18 (19%), and 20 (21%) patients, respectively (left main artery, n = 8; left anterior descending artery, n = 54; left circumflex artery, n = 44; right coronary artery, n = 52).

Fig. 1
figure 1

Study flowchart. AMI: acute myocardial infarction; ICU: intensive care unit. a Within the first 24 h for ST elevation AMI and 72 h for non-ST elevation AMI [6, 7]

Full size image
Table 1 Baseline characteristics according to occurrence of composite outcome components
Full size table
Table 2 Symptoms, laboratory and electrocardiogram findings, and sequential organ failure assessment scores, on the day of the acute myocardial infarction according to occurrence of composite outcome components
Full size table

Management

On the day of the AMI, antiplatelet therapy, dual antiplatelet therapy, or therapeutic anticoagulation was administered in 95 (99%), 61 (63%), and 68 (71%) patients, respectively (Table 3). Early mechanical reperfusion therapy was performed in 47 (49%) patients (Fig. 1), including PCI in 45 patients (drug-eluting stent, n = 31; bare metal stent, n = 13; missing data, n = 1) and CABG in two patients. Delayed mechanical reperfusion therapy was performed in 27 (28%) patients, including PCI in 21 patients (drug-eluting stent, n = 14; bare metal stent, n = 7) and CABG in six patients. The reasons for delayed mechanical reperfusion were uncontrolled sepsis (n = 16), hemodynamic instability (n = 2), active bleeding (n = 3), triple vessel disease (n = 4), complex procedure (n = 1), and unknown (n = 1). Organ failure management on the day of AMI onset included catecholamines in 48 (51%) patients and invasive mechanical ventilation in 56 (58%, Table 3).

Table 3 Management of myocardial infarction and of organ dysfunction on the day of the myocardial infarction according to occurrence of composite outcome components
Full size table

Composite clinical outcome and associated patient factors

The composite clinical outcome occurred in 48 (50%) patients (cumulative incidence curve in Additional file 1: Fig. S1). Patients with a composite clinical outcome more frequently had a history of arterial hypertension (79% vs. 56%, p = 0.03) and diagnosis of abdominal disease on ICU admission (23% vs 2%, p = 0.005), higher SAPS II (50 [43–73] vs. 40 [27–49], p < 0.001), higher SOFA global and SOFA cardiovascular scores (respectively, 9 [5–11] vs. 4 [2–10], p = 0.004; 4 [0–4] vs. 0 [0–4], p = 0.016), and lower hematocrit levels (30 [26–35]% vs. 37 [32–43]%, p = 0.001) (Additional file 1, Tables S3 and S4). The occurrence of the composite clinical outcome was similar in patients with and without sepsis/septic shock (Additional file 1: Table S5). AMI management in terms of antithrombotic medication and early mechanical reperfusion was not significantly different between the groups with and without the composite clinical outcome, but a greater proportion of patients with the composite outcome received catecholamines and invasive mechanical ventilation on the day of AMI (Additional file 1: Table S6).

A history of arterial hypertension, diagnosis of abdominal disease on ICU admission, SOFA score, and peak cardiac troponin were entered in the multivariable model, and history of arterial hypertension (HR 2.05, 95% CI 1.01–4.16, p = 0.047) and high SOFA score (HR 1.07, 95% CI 1.00–1.15, p = 0.042) were identified as independently associated with an increased risk of the composite outcome (Table 4). There was no significant time (Additional file 1: Table S7) or center (Additional file 1: Table S8) effect on the occurrence of the composite outcome.

Table 4 Univariate and multivariable analyses of factors associated with in-hospital outcome a
Full size table

Components of the composite clinical outcome

A severe ischemic event occurred in 17 (18%) patients (median 5 [3–7] days from AMI onset), including 9 recurrent AMIs and 8 strokes. One severe ischemic event (recurrent AMI related to early stent thrombosis) was fatal. Baseline clinical (Table 1) and AMI characteristics (Table 2) were similar in patients with and without a severe ischemic event.

Major bleeding occurred in 26 (27%) patients (median 4 [3–13] days from AMI onset) including 40 major extracranial bleeding events (23 gastrointestinal, 1 hemoptysis, 1 epistaxis, 1 thigh hematoma, 5 urologic, 1 pericardial effusion, 8 surgical site) and 1 intracranial bleeding event; no episode of major bleeding was fatal. A blood transfusion (median number of red blood cell units 2 [1–5]) was required for 34 of the major bleeding events. The 41 major bleeding events were classified as follows: BARC 3a, n = 24; BARC 3b, n = 14; BARC 3c, n = 1; BARC 4, n = 2. Patients with a major bleeding event had more frequently been admitted to the ICU for abdominal disease, and had a lower hematocrit and less often received dual antiplatelet therapy on the day of AMI. Six patients had both a severe ischemic event and a major bleeding event.

Twenty-six (27%) patients died in the hospital. The causes of death were refractory cardiogenic shock (n = 7), cardiac arrest of cardiogenic origin (or suspected) (n = 5), multiple organ failure (n = 7), and end-of-life decision (n = 7). SAPS II, SOFA global, and SOFA cardiovascular scores were higher in non-survivors.

The occurrence of each component of the composite clinical outcome was similar in patients with and without sepsis or septic shock (Additional file 1: Table S5).

Discussion

In this retrospective multicenter study in patients with AMI with OCAD during an ICU stay for non-cardiac disease, the incidence of the composite in-hospital outcome, including severe ischemic event (18%), major bleeding (27%), and mortality (27%), was high (50%). A history of arterial hypertension and a high SOFA score were independently associated with a risk of poor outcome.

Ischemic risk

Our results confirm that risk of ischemia is a major concern in patients with an AMI during an ICU stay for non-cardiac disease (majority with sepsis). Smilowitz et al. reported that AMI was independently associated with increased mortality in a retrospective analysis of a large nationwide cohort of patients with sepsis [1]. The ischemic risk in this setting appears to be greater than that reported in patients with AMI in cardiology wards (< 5%) [27]. The pathophysiological mechanisms behind the increased ischemic risk are complex in this context. Myocardial infarction may be a marker of the severity of non-cardiac disease, such as septic shock, which itself is associated with a high thrombotic risk because of hemodynamic collapse, sepsis-induced coagulopathy with deregulated immunothrombosis, and endothelial dysfunction [28,29,30]. Several infectious agents and inflammatory diseases are associated with an increased risk of AMI, probably related to the overall burden of systemic inflammation [5, 31,32,33] that could lead to coronary plaque instability and thrombus formation. In an observational study, Del Pace et al. showed that occurrence of an infectious or inflammatory event may facilitate the development of coronary stent thrombosis [33]. In addition, tachycardia and blood pressure changes in critically ill patients can precipitate plaque rupture and coronary thrombosis [34]. Furthermore, the bioavailability of enteral drugs, such as antiplatelet agents, can be significantly altered in critically ill patients, leading to an increased risk of thrombosis [35, 36]. Finally, early invasive reperfusion was not performed in half our patients, leading to a potential increase in the risk of severe ischemic complications.

Bleeding risk

Episodes of major bleeding were also frequent, occurring in 27% of our patients. In medical ICU patients, Strauss et al. reported a similar incidence of major bleeding (20%) [19]. Similar to ischemic risk, bleeding risk in this setting appears to be greater than that reported in patients with AMI in cardiology wards (≃ 5%) [37]. The dysregulated infection-inflammation immune response may produce antithrombotic states with thrombocytopenia, decreased clotting factors, and increased fibrinolysis, which predispose to bleeding complications [38,39,40,41]. Several studies have developed predictive instruments for the estimation of bleeding risk in patients with AMI in cardiology wards [42]. Subherwal et al. reported that a lower baseline hematocrit was an independent predictor of bleeding events [42]; similarly, in our univariate analysis, patients with bleeding events had lower baseline hematocrit values. We found a counter-intuitive association between dual antiplatelet administration and a lower risk of bleeding events. This may be due to indication bias, or to a high alpha risk given the large number of statistical comparisons in this exploratory observational study.

Prediction of the composite clinical outcome

Our results are consistent with those from several studies in cardiology patients, which have reported that a history of arterial hypertension is associated with ischemic [23, 43] and bleeding [44] risk. However, no other factors commonly used to stratify cardiovascular risk in cardiology patients [43, 45], such as TIMI risk score, ECG abnormalities and baseline cardiovascular characteristics, were predictive of clinical outcome in our cohort. Thus, the approach used to estimate this risk in cardiology patients [43, 45] may not be relevant in patients admitted to the ICU for a non-cardiac condition. Several authors have reported that invasive reperfusion therapy is performed in only 30–50% of critically ill patients with AMI during an ICU stay for a non-cardiac cause [3, 36], similar to our findings. These observations suggest there is an urgent need for bedside risk stratification tools to determine which patients may benefit most from antithrombotic medications and early invasive reperfusion strategy.

Limitations

Our study has several limitations. First, this was a retrospective study with inherently associated bias, some missing data, and possible associated errors in data abstraction. Second, all our statistical results should be interpreted with caution in this exploratory retrospective study because of (i) the large number of statistical comparisons and the not adjusted 95% confidence intervals and p-values for multiple testing resulting in high alpha risk, and (ii) the relatively small number of patients limited power in all analyses. Third, severe ischemic events (or major bleeding events) may have been more relevant as a primary outcome, but we did not consider this option, because of their low incidence. Instead, we used a composite outcome that reflects the net clinical benefit of antithrombotic medication and invasive reperfusion strategy. Fourth, we did not assess the relationship between AMI and the occurrence of the composite outcome. Fifth, the study was observational, leading to potential indication biases. Specific treatments for AMI, including antithrombotic therapy and reperfusion therapy, may have influenced the occurrence of adverse events. Sixth, patients without coronary angiography to confirm AMI were not included, leading to potential selection bias. Indeed, many coronary angiographies are delayed or not performed in the acute phase of septic shock because of fear of stent thrombosis due to the pro-thrombotic state of the patients. Seventh, as the study was conducted in France, our findings may not be applicable elsewhere. Finally, the last patient was included in December 2018. However, to the best of our knowledge, no trial results or guidelines have been published since then that could have modified the usual management of these patients.

Conclusion

Patients having an AMI with OCAD during an ICU stay for non-cardiac disease are at high risk of poor clinical outcome, including development of severe ischemic events or major bleeding, and death. A history of arterial hypertension and a high SOFA score at the time of the AMI were the only factors associated with occurrence of the composite outcome, albeit the relatively small sample size. Further studies are needed to determine how to better stratify bedside cardiovascular risk, a preliminary requirement for establishing an appropriate anti-thrombotic and coronary reperfusion strategy in this context.