Article Text
PDF
Abstract
Background The diagnosis of myocardial infarction (MI) in the presence of heart failure (HF) presents a clinical problem. While diagnostic algorithms using high-sensitivity cardiac troponin have been established for suspected MI, their accuracy in patients with HF remains uncertain. This study aims to assess the diagnostic accuracy of high-sensitivity troponin I (TnI) levels in identifying acute MI among patients with HF, focusing on baseline, absolute and relative TnI changes.
Methods Data from 562 individuals admitted to the emergency department with suspected MI were retrospectively analysed. Two-point TnI and baseline brain natriuretic peptide (BNP) test results were available. HF status was determined based on clinical, laboratory and instrumental criteria.
Results Among the 562 patients, 299 (53.2%) were confirmed having MI. Baseline TnI demonstrated predictive capability for MI in the overall population (area under the curve (AUC) 0.63), while TnI relative change exhibited superior performance (AUC 0.83). Baseline TnI accuracy varied significantly by group, notably decreasing in the third group (severe HF) (AUC 0.54) compared with the first and second groups (AUC 0.67 and AUC 0.71, respectively). TnI relative change demonstrated consistent accuracy across all groups, with AUCs of 0.79, 0.79 and 0.89 for the first, second and third groups, respectively, even after adjustment for age, sex and glomerular filtration rate.
Discussion Troponin relative change is a reliable predictor of MI, even in patients with acute HF. Baseline TnI accuracy is influenced by HF severity. It is essential to consider HF status and BNP levels when employing high-sensitivity cardiac troponin testing to rule out suspected MIs.
- Myocardial Infarction
- HEART FAILURE
- Acute Coronary Syndrome
Data availability statement
Data are available upon reasonable request. Available on reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Baseline troponin I (TnI) levels have demonstrated predictive capability for myocardial infarction (MI) in the general population.
WHAT THIS STUDY ADDS
TnI relative change is a reliable predictor of MI in patients with concomitant heart failure (HF), even after adjusting for age, sex and estimated glomerular filtration rate.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The study highlights the influence of HF severity on baseline TnI testing accuracy, emphasising the importance of considering HF status in diagnostic decision-making.
Introduction
The accurate diagnosis of acute myocardial infarction (MI) in the presence of heart failure (HF) may be challenging. HF and acute MI may present with similar symptoms and sometimes as concomitant diseases. Furthermore, in HF, cardiac biomarkers, particularly troponin (Tn), are frequently elevated due to chronic and acute myocardial injury, even in the absence of MI, and this complicates the interpretation of this biomarker for the diagnosis of classic acute MI. Moreover, this issue has become more evident over the last 10 years, along with the growing use of highly sensitive assays for Tn (hs-Tn). Many diagnostic algorithms using hs-Tn testing for the rule-out of patients with suspected MI have been proposed and validated. Still, none have tested the diagnostic accuracy of these algorithms in patients with HF. Furthermore, data from large trials are lacking because many studies on acute coronary syndrome (ACS) have excluded patients with acute HF and vice versa. Therefore, while a large amount of data exists regarding Tn use and prognostic utility in patients with HF, data on the accuracy of Tn assay to diagnose MI in HF are lacking.
To address this gap, we aim to evaluate the accuracy of high-sensitivity TnI (hs-TnI) levels in diagnosing acute MI in patients with clinically manifest HF, focusing on baseline, absolute and relative TnI changes.
Methods
Study design and setting
The Acute coRonary sYndrOmes proSpective regisTry Of Ferrara (ARYOSTO) is a prospective, single-centre study collecting data about baseline characteristics, treatment and outcomes of all patients admitted to the University Hospital of Ferrara with a suspected or confirmed diagnosis of ACS. The ARYOSTO Study started in May 2015 and is still ongoing. The study is registered on ClinicalTrials.gov with the identifier NCT02438085. In December 2018, the emergency department (ED) of the University Hospital of Ferrara adopted a new protocol for patients admitted for suspected ACS, including two hs-TnI dosages (0–3 hours) and brain natriuretic peptide (BNP) assessment. The protocol suggested considering a change greater than 20% of TnI as suggestive of a coronary cause of TnI release. For the present purpose, we considered all the consecutive patients admitted to ED for suspected ACS from December 2018 to December 2019. The main inclusion criteria were: (1) admission to ED for suspected ACS; (2) availability of two hs-TnI values according to the internal rule-out protocol; (3) undergoing a baseline BNP blood test; (4) age greater than 18 years. Exclusion criteria were refusal of the informed consent and diagnosis of ST-segment elevation MI.
Definitions
To determine the final diagnosis for each patient, two independent physicians reviewed the available medical records (including patient history, physical examination, results of laboratory testing, radiological testing, ECG and echocardiography of the patient from the time of ED presentation to the time of discharge). In situations of diagnostic disagreement, cases were reviewed and adjudicated in conjunction with a third cardiologist. MI was diagnosed according to the consensus definition of the Joint European Society of Cardiology/American College of Cardiology Foundation/American Heart Association/World Heart Federation Task Force for the fourth Universal Definition of Myocardial Infarction.1 We defined myocardial injury as cardiac Tn elevation above the 99th percentile upper normal limit (UNL).
MI diagnosis was therefore based on the presence of acute myocardial injury with at least one of the following: (1) clinical presentation of myocardial ischaemia, (2) new ECG changes suggesting ischaemia, (3) pathological Q waves, (4) evidence of newly developed non-viable myocardium or regional wall-motion abnormality by imaging (ie, echocardiography), and (5) evidence of a coronary thrombus by angiography.
Significant Tn change was defined as a positive delta of at least 20% from baseline Tn to the second Tn assessment.
The diagnosis of HF was adjudicated by two expert cardiologists using clinical information (evidence of suggestive signs and symptoms recorded during ED stay), blood test results (ie, BNP) and instrumental data (i.e., echocardiography, chest X-ray), and according to European Society of Cardiology Guidelines. According to this, a group of patients with acute HF was identified. This group was further split according to the median BNP values. Three groups were finally identified: patients without HF (group 1), patients with HF and BNP <397 ng/mL (group 2), and patients with HF and BNP ≥397 ng/mL (group 3).
Data collection and laboratory assays
All clinical, treatment and outcome data were prospectively collected using a dedicated electronic case report form (eCRF). Specialised personnel performed this procedure. The eCRFs were periodically monitored and verified with source data. The internal ED protocol includes anamnestic records, physical examination, intensive monitoring, ECG, oxygen saturation, blood pressure, laboratory tests and chest X-ray. The eCRFs captured all these variables. hs-TnI was measured using the High-Sensitivity Troponin I Assay (Access Immunoassay System, Beckman Coulter, Erlangen, Germany) at ED arrival and 3 hours later. The UNL was 20 ng/mL for males and 12 ng/mL for females. The limit of detection was 2 ng/mL. BNP was measured with the Access Immunoassay System (Beckman Coulter, Erlangen, Germany).
Statistical analysis
Baseline characteristics were summarised as means, SDs, medians, IQRs or percentages. The Jonckheere-Terpstra test was used to determine the p values for trends and test for significant differences in baseline characteristics across groups. The outcome was the diagnostic accuracy of baseline TnI and delta in TnI to predict MI. Patients were categorised by HF status, and the outcome occurrence was analysed accordingly. Discrimination of diagnostic measures was assessed using the area under the receiver operating characteristic (ROC) curves (AUC). Supplementary analyses are provided in the online supplemental appendix and include ROC curves multivariate analysis adjusted for sex, age and estimated glomerular filtration rate (eGFR) (online supplemental table 1) and ROC curves univariate analysis stratified by history of coronary artery disease (CAD) (online supplemental table 2). Assuming an expected AUC of 0.85 for the TnI assay, with an estimated prevalence for MI of 40% in patients admitted to the emergency room for suspected MI, we calculated that at least 170 patients per group were needed to perform the analysis adequately. All the analyses were conducted using STATA V.16.1 (College Station, Texas, USA).
Supplemental material
Results
Baseline characteristics of the study population
The present analysis enrolled 562 patients who were divided into three groups according to the diagnosis of HF. 154 patients did not present with HF at admission (group 1), 203 had HF with a BNP below 397 ng/mL and 205 had HF with a BNP greater than 397 ng/mL. Median values of BNP at baseline were 56.5 (IQR 35–90) pg/mL, 208.0 (IQR 151–293) pg/mL and 938 (IQR 584–1504) pg/mL for each group (p<0.001). The diagnosis of MI was confirmed in 299 (53.2%) cases. Patients in the second and third groups were older and presented with a lower left ventricular ejection fraction (LVEF) and eGFR.
Baseline characteristics of the overall population and according to groups are shown in table 1.
Baseline characteristics of the cohort, overall and according to heart failure (HF) status
Diagnostic accuracy of baseline TnI assessment
Patients with a confirmed diagnosis of MI were equally distributed in the three groups: 82 (53.2%) in the first group, 116 (57.1%) in the second group, 101 (49.3%) in the third group (p=0.39) (table 2). 74% of patients with definite MI underwent coronary artery revascularisation among those without HF, 70% among those with mild HF and 64% among those with more severe HF (p for trend=0.18).
Baseline troponin (Tn) and change Tn levels in the overall cohorts and according to heart failure (HF) status
Median baseline TnI was 49 ng/L (IQR 14–197), 47 ng/L (IQR 18–306) and 57 ng/L (27-324) in each group (p=0.011) (table 2 and figure 1).
Baseline TnI values and change in TnI according to HF status. HF, heart failure; TnI, troponin I.
The AUC for prediction of MI by baseline TnI in the overall population was 0.63 (95% CI 0.58 to 0.68). The AUC was 0.67 (95% CI 0.56 to 0.76) in the first group, 0.71 (95% CI 0.63 to 0.79) in the second group and 0.54 (95% CI 0.46 to 0.72) in the third group (table 3 and figure 2). The test for equality of ROC curves gave a p value of 0.009. According to the Youden index, the best prediction cut-off was 18.5 ng/L, 22.5 ng/L and 19.5 ng/L for each group, respectively (table 3). ROC curve outcomes adjusted for sex, age and eGFR are provided in online supplemental table 1.
ROC analysis and AUC: overall and according to heart failure (HF) status
Accuracy of baseline TnI to predict myocardial infarction according to HF status. BNP, brain natriuretic peptide; HF, heart failure; TnI, troponin; TPR, true positive rate; FPR, false positive rate.
Diagnostic accuracy of TnI absolute change
Median absolute change in TnI was 51 ng/L (IQR 1.5–406.5), 70 ng/L (IQR 3–650) and 26 ng/L (IQR 2–274) in each group (p=0.11) (table 2 and figure 1).
The AUC for prediction of MI by TnI absolute delta change (ng/L) in the overall population was 0.82 (95% CI 0.79 to 0.88). The AUC was 0.81 (95% CI 0.72 to 0.89) in the first group, 0.86 (95% CI 0.79 to 0.93) for the second group and 0.82 (95% CI 0.75 to 0.89) in the third group (table 3 and figure 3A). The test for equality of ROC curves gave a p value of 0.66. The best cut-off for MI prediction in the overall cohort, according to the Youden index, was 29.4 ng/L. The best cut-off was 87.4 ng/L, 49.4 ng/L and 30.9 mg/L for each group (table 3). ROC curve outcomes adjusted for sex, age and eGFR are provided in online supplemental table 1.
Accuracy of TnI absolute change (A) and relative change (B) to predict MI according to HF status. BNP, brain natriuretic peptide; HF, heart failure; MI, myocardial infarction; TnI, troponin I; TPR, true positive rate; FPR, false positive rate.
Diagnostic accuracy of TnI relative change
Median relative change (%) in TnI was 75.3 ng/L (IQR 8.4–348.2), 108.3 ng/L (IQR 9.1–431.0) and 25.9 ng/L (IQR 2.2–165.3) in each group (p=0.001) (table 2 and figure 1).
The AUC for the prediction of MI by TnI relative change (%) in the overall population was 0.83 (95% CI 0.79 to 0.88). The AUC was 0.79 (95% CI 0.69 to 0.88) for the first group, 0.79 (95% CI 0.70 to 0.88) for the second group and 0.89 (95% CI 0.83 to 0.95) for the third group (table 3 and figure 3B). The test for equality of ROC curves gave a p value of 0.08. The best cut-off for MI prediction in the overall cohort, according to the Youden index, was 42.5%. The best cut-off was 75.2%, 43.0% and 21.1% for each group (table 3). ROC curve outcomes adjusted for sex, age and eGFR are provided in online supplemental table 1.
Discussion
HF is a medical condition that is characterised by increased levels of natriuretic peptides due to high wall stress. It is typically accompanied by symptoms of congestion and/or low cardiac output. In many cases, acute HF is caused by an acute MI. Additionally, MI can also trigger worsening HF in patients who already have subclinical or chronic HF. Regardless of the setting, HF can lead to a higher release of Tn, which can make the diagnosis of MI more challenging.2 This study evaluated the clinical utility of baseline, absolute and relative TnI changes in the diagnosis of MI using a well-validated hs-TnI assay. Interpretation of Tn values in patients with HF should be consistent with guidelines for the diagnosis of MI, which require evidence of a rise and fall of the marker above the 99th percentile of a normal reference population, ECG changes or imaging evidence for new loss of functional myocardium in a setting of typical coronary ischaemia.1 Nonetheless, Tn in the setting of HF may be raised for many reasons, and in some studies on acutely decompensated HF, is reported as abnormal in almost all patients.3 4 In the ADHERE trial, on more than 105 000 patients with acute HF, 75% have detectable TnT results. Abnormal Tn levels in HF have already been linked to an increased risk of mortality, independently of other prognostic markers (e.g., natriuretic peptides, LVEF or age). However, an increased Tn value is a common finding also in chronic HF, both in patients with ischaemic heart disease and those with non-ischaemic HF. While in this setting, elevated Tn levels are not directly associated with decompensation, they are linked to a greater risk of death or hospitalisation.5 The determinants of the rise in Tn in HF are many: occurrence of type II MI from increased transmural pressure, small-vessel coronary obstruction, endothelial dysfunction, reduced oxygen delivery secondary to anaemia and hypotension, infiltrative disease (ie, amyloidosis) and toxic exposures.6 7 Despite this, a type I MI must always be considered when a patient with HF has elevated Tn, particularly if other signs suggesting ACS are present. Unfortunately, in patients with chronic abnormalities (ie, history of MI, left ventricular dysfunction), non-invasive diagnostic testing such as ECG or echocardiogram may be less specific for the diagnosis of MI.
In our cohort, diagnosis of acute MI was done in a similar proportion among patients with and without diagnostic criteria of HF (table 2).
Patients with HF and higher BNP levels showed more severe markers of myocardial dysfunction, corroborated by a lower LVEF, higher prevalence of fluid congestion at the chest X-ray and lower eGFR (table 1).
Our analysis confirmed that patients with HF and higher BNP values present with higher baseline TnI levels and that baseline Tn testing was less accurate in identifying patients with MI in the overall population and especially in those with higher degrees of HF (table 3).
On the other hand, absolute and relative changes in TnI were accurate in predicting MI in this group, but a smaller relative change in TnI compared with those with less severe and without HF was observed. A possible explanation could be that in patients with HF, the already increased values of Tn at baseline may mask the change in Tn due to the MI. On this basis, caution should be advised, particularly in judging small relative changes in Tn in those with acute HF.
The findings from the analyses remained consistent even following adjustments for primary confounding factors and determinants of BNP levels, such as age, gender and eGFR. Stratification based on a history of HF and CAD was also undertaken and gave consistent results.
Furthermore, we found that the best cut-off for specificity and sensitivity of Tn change to discriminate MI in patients with more severe HF was lower than those without HF (table 3).
Similarly, a previous report analysing the diagnostic accuracy of a Tn algorithm for non-ST-elevation MI (NSTEMI) in patients with chronic kidney disease, which is another condition that may increase baseline Tn, showed that specificity to detect NSTEMI was reduced in patients with renal failure and suggested the use of different cut-off values to improve the accuracy of NSTEMI identification.8
It should be noted that the cut-off values we have identified are solely intended for scientific speculation and should not be recommended as an alternative to the current rule-in and rule-out protocols, which are based on Tn change cut-offs aimed at maximising sensitivity.
Nonetheless, it is intriguing that the best relative change in TnI cut-off for predicting MI in patients with more severe HF resulted in 21.1% and provided a high sensitivity (1.00) and good specificity (0.75). This would perfectly align with the recommendation to consider a change in Tn of 20% as significantly associated with MI.
The findings of this study suggest that TnI relative change is a reliable and consistent predictor of MI even in patients with acute HF, regardless of their baseline BNP levels. These findings may give the clinicians a more comprehensive understanding of Tn change in patients with HF and highlight the importance of considering HF status and natriuretic peptide levels when using high-sensitivity cardiac Tn testing to rule out patients with suspected MI. Clinicians may need to be cautious when interpreting the diagnostic accuracy of baseline TnI in patients with more severe HF or higher BNP levels, as the accuracy of baseline TnI was notably lower in these patients, and the Tn rise during MI is usually lower compared with patients without HF.
Early and accurate diagnosis of MI in patients with acute HF is crucial, as timely and appropriate treatment can improve patient outcomes and reduce morbidity and mortality.
Therefore, the results of this study may have significant implications for clinical practice and may improve patient care in the management of suspected MI and HF.
Limitations
One of the main limitations of this study is that we cannot delineate the interplay and the sequence of action between HF and MI in our population. Specifically, it may be possible that in some patients, MI was the leading cause or a precipitating trigger of acute HF. On the other hand, it would be possible that some patients had pre-existing HF complicated by the occurrence of MI. While we cannot resolve the complexity of real-world cases with this study, we believe that our data may reassure clinicians about the accuracy of Tn change to diagnose those presenting with concomitant HF.
Furthermore, due to the single-centre nature of the study and the specific diagnostic algorithm applied, these findings are not directly generalisable to all other Tn assays and testing protocols; nonetheless, there is no logical reason to think that also other assays (eg, hs-TnT testing, different delta testing timing) would behave differently in patients with HF. On the other hand, despite this not being a prespecified analysis of the ARYOSTO registry, data were prospectively collected, and due to institutional protocol, there was a standardisation of Tn testing, both in terms of tools and timing.
Conclusion
Hs-TnI delta change (0–3 hours) showed a great ability to discriminate MI independently from concomitant acute HF and outperformed baseline TnI, particularly in patients with more severe HF.
Our analysis showed that TnI relative delta provides a reliable and consistent ability to detect MI even in patients with acute HF, offering consistent diagnostic accuracy across subjects with different degrees of HF.
Data availability statement
Data are available upon reasonable request. Available on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was conducted in accordance with the ethical principles of the Declaration of Helsinki. All patients gave informed written consent for inclusion in the registry.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @matser89, @GianlucaCampo78
Contributors MS, PT, RC, FM and SB—concept and design. MS, BDP, AC and VS—analysis and interpretation of data. MS, PT, GC, MA, FS, DM and LDI—critical writing and revising the intellectual content. MS, BDP, PT, AC, GC, RC, SB, FM, MA, DM, LDI and VS—final approval of the version to be published. MS acts as guarantor.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.