Importance of myocardial ischemia detected with protocol-based measurements of high-sensitivity troponin, ECG and echocardiography in critically ill patients without acute coronary syndrome—a prospective study

1. Introduction

Cardiac troponins (cTn) are regulatory proteins and part of the cardiomyocyte contractile apparatus. They are expressed almost exclusively in the heart and have been established as the preferred biomarker in diagnosing myocardial ischemia (MI) and injury as they have high myocardial tissue specificity and clinical sensitivity [1, 2]. Detection of reversible increase of cardiac biomarkers (preferably cTn) together with clinical signs of ischemia is the cornerstone of diagnosing myocardial infarction. cTn elevation without clinical signs of ischemia is myocardial injury, which could be acute or chronic [1, 2].

Critically ill patients are at higher risk for MI due to their older age, underlying coronary artery disease, and other comorbidities. The heart is affected by intrinsic and extrinsic sympathetic stimulation, various other toxic mediators and stressors – anaemia, tachycardia, hypotension. MI in the critically ill may not always be apparent, and establishing a diagnosis is challenging [3]. Patient communication is limited and ischemic symptoms are often absent. ECG changes may not be detected or are non-specific, and elevated cTn levels, although common, are difficult to interpret in the context of critical illness and other comorbidities [3, 4, 5, 6]. Echocardiography and coronary angiography are not routinely performed in all patients with suspected MI. There are currently no diagnostic criteria or treatment strategies for detecting MI in the critically ill. New laboratory methods to detect low levels of cTn, i.e., high sensitivity troponin T (hsTnT) assays, are available. Their value in critically ill patients exhibiting no acute coronary syndrome is still not firmly defined.

The aim of our study was to evaluate a protocol-guided approach to detect MI in the critically ill (serial ECGs, hsTnT measurements, and echocardiography) in comparison to a retrospective cohort with only clinically driven detection of MI.

2. Methods

3. Results

4. Prospective group analysis

4.4 Primary outcomes

Patients with MI had higher in-hospital and one-year mortality (Table 4). There was no difference in ICU mortality between patients with MI and with elevated hsTnT only. Hospital mortality and one-year mortality were higher in patients with hsTnT levels above 52 ng/L. Maximal hsTnT levels, but not admission levels, were associated with ICU mortality (ROC curve: AUC 0.803; p = 0.003; sensitivity: 77.8%; specificity: 75.6%, cut off TnT value: 157.50 ng/L) (Fig. 1). Odds ratios of hsTnT $>$52 ng/L, hsTnT $>$70 ng/L and MI to predict outcome are presented in Table 5.

Fig. 1.ROC curve of maximal high sensitive troponin T (hsTnT) predicting ICU-mortaliy.

Table 5.Odds ratios of Myocardial ischemia and elevated high-sensitive troponin (hsTnT) to predict inotropes use, decreased left ventricular ejection fraction (LVEF), regional wall motion abnormalities (RWMA) and mortality.

	Odds ratio (95% CI)	Statistics (p)
hsTnT $>$52 ng/L
Inotropes use	5.33 (1.45 to 19.56)	0.016
LVEF $<$30%	14.76 (0.84 to 260.20)	0.064
RWMA	5.02 (1.07 to 23.60)	0.041
ICU mortality	5.23 (0.625 to 43.725)	0.120
Hospital mortality	3.00 (1.010 to 8.908)	0.048
1-year mortality	6.00 (2.051 to 17.554)	0.001
hsTnT $>$70 ng/L
Inotropes use	3.35 (1.184 to 9.472)	0.022
LVEF $<$30%	9.65 (1.172 to 76.620)	0.035
RWMA	3.87 (1.032 to 14.533)	0.045
ICU mortality	8.38 (1.004 to 69.924)	0.050
Hospital mortality	3.05 (1.133 to 8.230)	0.027
1-year mortality	5.43 (2.1099 to13.971)	0.005
Myocardial ischemia
Inotropes use	1.53 (0.565 to 4.1572)	0.400
LVEF $<$30%	4.06 (1.048 to 15.748)	0.043
RWMA	7.71 (2.414 to 24.645)	0.001
ICU mortality	3.06 (0.740 to 12.643)	0.120
Hospital mortality	3.33 (1.190 to 9.329)	0.020
1-year mortality	4.66 (1.647 to 13.222)	0.004

5. Discussion

Our study confirmed that elevated hsTnT levels are a very common finding in the critically ill. Maximal hsTnT levels during ICU stay can predict ICU, in-hospital and even one-year mortality. Elevated hsTnT levels in critically ill patients can predict the use of inotropes, echocardiographic abnormalities such as lower LVEF and RWMA. Standard diagnostic procedures and laboratory findings are not very effective in diagnosing MI in the critically ill, as we confirmed with our protocol-driven approach. Detected MI additionally stratifies patients with elevated hsTnT to higher hospital and one-year mortality.

In our study, all patients had detectable hsTnT levels during their ICU stay, and $>$95% of them had at least one elevated hsTnT level. To our knowledge, this is the highest reported percentage of patients with elevated cTn levels in the ICU – reflecting a new era of highly sensitive troponin assays [11, 12]. Almost a quarter of patients with elevated hsTnT levels also had other signs of MI (chest pain, ECG abnormalities). All patients with MI had hsTnT levels higher than 100 ng/L.

Diagnosing MI and interpreting elevated cTn levels in critically ill patients is challenging for various reasons, and applying the recent definition of myocardial infarction has its limitations [1, 2]. Communication with critically ill patients is hampered and ischemic chest pain is often masked by analgesic and sedative drugs [4, 6, 11]. Some MI-associated signs, such as hypotension, arrhythmia or pulmonary oedema, are non-specific and could be related to other diseases, such as sepsis or respiratory failure. Continuous ECG monitoring is not as sensitive as frequent 12-lead ECG recordings [13]. Moreover, ECG changes are frequently difficult to interpret, particularly in patients with a bundle branch block, electrostimulation or electrolyte disturbances [1]. The reliability of interpretation depends not only on the physician’s knowledge and experience but also on the clinical information provided [14]. Our study confirmed that the most common ECG change suggesting MI in critically ill patients is ST depression [15, 16].

The rates of elevated cTn observed in previous studies were influenced by different study protocols and troponin assays used. In a meta-analysis of 20 studies, Lim et al. [17] found elevated cTn levels in 43% of 3278 patients. Ostermann et al., [4] using a protocol-based approach with hsTnT, found that 84% of 144 patients had at least one elevated troponin level and 41% met the criteria for MI. Study results were blinded to ICU staff and only 20% of definite MIs were recognised by the clinical team. In a similar study, although with a third-generation Tn assay, Lim et al. [6] found that MI was present in 35.9% of 103 patients, 14.6% had elevated cTn levels only, and 49.5% had no cTn elevation.

Our results confirm that elevated hsTnT and MI are associated with higher in-hospital and even 1-year mortality. The relationship between cTn and mortality is not new [4, 5, 18], although, prior results are not all in agreement. Myocardial injury and MI have been associated with ICU and perioperative mortality [19] regardless of the aetiology of cTn release.

With more widespread use of high sensitivity assays, elevated cTn levels are becoming more difficult to interpret [11, 12]. Current cTn thresholds are valid for non-critically ill patients. The universal upper reference limit lacks specificity in the ICU and the optimal troponin increase still remains to be determined and evaluated in order to avoid overestimation of ischemia [8, 18]. Troponin levels should always be interpreted in the clinical context. Differentiation between myocardial injury and ischemia can be difficult, especially in the critically ill. As mentioned, almost three-quarters of patients in our study had elevated hsTnT levels but no other signs of ischemia, reflecting a high number of patients with myocardial injury due to various pathophysiological processes in critical illness. Docherty et al. [18] demonstrated a rise and fall pattern of cTn levels in all critically ill patients (although with known cardiovascular risk factors) suggesting an “acute hit” mechanism of injury. Critical illness is characterised by increased sympathetic tone, which may be attributable to increased platelet aggregation and microvascular obstruction or even acute plaque rupture [20]. Peter Ammann et al. [21] proved that more than 70% of critically ill patients with elevated cTn levels did not have obstructive coronary disease. Critical illness could represent a stress test where non-critical coronary obstruction – previously undiagnosed, might become important [22, 23]. Many pathologies can cause an imbalance between myocardial oxygen supply and demand, which leads to ischemia [24]. This is also sometimes referred to as demand ischemia, which can be also caused by anaemia, tachycardia and suboptimal diastolic ventricular filling, hypotension or sepsis (inappropriate oxygen consumption) and many others [25, 26].

ECG changes could be non-specific in this setting [27]. We found that patients with MI had higher maximal hsTnT levels than those without clinical signs (ECG changes, RWMA, decreased LVEF, etc.). Also, inotropes were more commonly used in patients with MI, possibly suggesting more profound myocardial dysfunction. The use of vasopressors and mechanical ventilation was not, in contrast to studies by Lim and Ostermann, significantly associated with MI. In our study, previously diagnosed ischaemic heart disease was not significantly associated with MI, however, it was associated with elevated troponin levels. This may imply that myocardial ischemia in the critically ill is truly more often due to supply-demand imbalance or possible toxic effects from mediators than acute plaque rupture or other coronary occlusive events. Another possible explanation is that many patients simply did not have any diagnostics of ischemic heart disease made before ICU admission. Apart from supply-demand ischemia, both Ammann and Ostermann et al. [21, 28] showed that there is a significant association between elevated cTn and markers of systemic inflammation: IL-6, its soluble receptor and TNF-$\alpha$. These substances could mediate myocardial depression, and they, especially TNF-$\alpha$, have been implicated in increased cardiomyocyte permeability. This could explain cTn release without irreversible myocardial injury or necrosis, possibly due to transient loss of membrane integrity and release of cTn from the cytosolic pool [29, 30, 31]. Such cTn release without cell necrosis has been shown in some in vitro studies [32].

In previous studies, troponin elevations have also been described in a heterogeneous group of patients, medical and surgical, with pulmonary embolism, heart contusion, intracerebral pathology, myocarditis or renal failure. Causes for the troponin elevations in these situations are multifactorial and can be sometimes difficult to interpret [11, 25, 26, 33]. Elevated troponin levels without other signs of ischemia are myocardial injuries, which can be acute or chronic, depending on the troponin dynamic [1].

Our study was novel in its emphasis on trans-thoracic echocardiographic imaging use in diagnosing myocardial ischemia. We showed an inverse correlation between hsTnT levels and LVEF. Similar conclusions were made by Ammann et al. [21] and ver Elst et al. [34] who also showed an association between troponin elevation and evidence of LV dysfunction on transoesophageal echocardiography in septic patients. Given that echocardiographic abnormalities, whether global LV dysfunction or RWMA, are associated with increased hsTnT, an echocardiographic exam should be a standard procedure in all these patients. Even though echocardiography is difficult to perform in some patients (obese, COPD, mechanical ventilation), it remains the quickest and simplest technique to assess cardiac function in the critically ill [35].

We found that the incidence of MI was higher and ICU mortality was significantly lower in the prospective group and that maximal hsTnT in our prospective arm predicted ICU mortality. Our study was not designed (sample size was too small) to make definitive conclusions on whether systematic detection and treatment of MI could influence ICU mortality. Previous studies of Lim and Ostermann had an intriguing conclusion that there is no difference in mortality between clinically recognised and unrecognised myocardial ischemic events [4, 6]. In contrast to previous studies, our results were available to the medical team all the time, but specific therapeutic interventions upon the diagnosis of MI were applied only in half of all patients – they received antiplatelet or anti-ischemic treatment and two also underwent angiography. It remains unclear whether traditional antiplatelet, anti-ischemic or interventional treatments are appropriate or beneficial in treating MI in the ICU [3]. To date, there have been no randomised trials in this field and there are currently no guidelines regarding the management of MI in the ICU. Retrospective analyses by Poe et al. [36] showed that patients with high cTn levels had lower 30-day mortality if treated with beta-blockers or aspirin than those who were not. However, the potential for iatrogenic harm from treatment presents a challenge for the physician to decide on a plan of action; recent surgery or coagulation disorders may limit the use of antiplatelets, and cardiovascular instability may preclude the use of beta-blockers and nitrates [2, 3]. Moreover, in the setting of demand ischemia, it may be more appropriate to try to control physiologic variables – tachycardia, hypotension, anaemia. This could even prevent MI and improve outcome. Also, as emphasised in the Universal Definition, early cooperation with cardiologists is of great importance [2]. Our results are additionally supported by studies of post-mortem examinations on ICU patients showing that MI in the ICU is the most frequently missed diagnosis likely to impact outcome [37].

6. Study limitations

Our study had some limitations. Firstly, we conducted a single-centre study. Although our study sample was powered to detect mortality, it was not big enough for other subgroup analyses. While we had several exclusion criteria and recorded known patient comorbidities, we have not recorded smoking history, the presence of dyslipidaemia or family history for cardiovascular diseases, which could be important. We have also not recorded the use of cardioprotective medication. Although we followed our patients for up to one year, autopsy was not performed in all patients. Therefore, we do not know the exact cause of death. And, as stated in discussion, the major limitation of our study is comparison of a prospective protocol-driven approach to detect MI with retrospective analysis of a clinically driven approach.

7. Conclusions

In conclusion, high hsTnT and MI in critically ill patients are common and underdiagnosed. A methodical approach with serial troponin measurements, 12-lead (not only 2 or 3 lead) ECG and echocardiography is more successful in detecting MI compared to clinically driven decision making. Maximal hsTnT during ICU stay predicts ICU mortality. High prognostic value of elevated hsTnT and MI to predict hospital and 1-year mortality was also confirmed. Further research is necessary to better understand the various aetiologies of troponin rise in the ICU patients, to define appropriate diagnostic criteria for MI and injury in order to identify these patients and to develop an optimal treatment plan for them.

Author contributions

GK was responsible for data collection, basic statistics and wrote the manuscript; MM was responsible for data collection, echocardiography revisions, reviewing the manuscript; MG was responsible for data collection, reviewing the manuscript; GV and MP were responsible for the conception of the study, evaluation of data, reviewing the manuscript.

Ethics approval and consent to participate

Approval for this study was obtained from the National Medical Ethics Committee (Affiliation: Ministry of Health of the Republic of Slovenia; approval Number: 0120-317/2017/3) and IRB of General and Teaching Hospital Celje.

Acknowledgment

Authors thank all the medical and nursing staff of both intensive care units for cooperation and support.

Funding

This research was founded by General and Teaching Hospital Celje, there was no additional external funding.

Conflict of interest

The authors declare no conflict of interest. Matej Podbregar is serving as one of the Editorial Board members of this journal. We declare that Matej Podbregar had no involvement in the peer review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to LA.