Cardiac troponin I predicts clinical outcome of patients with cancer at emergency department

Abstract Background The prognostic ability of cardiac troponin I (TnI) has been demonstrated in general populations and among cardiovascular disease patients, but it has not been evaluated in cancer patients. Hypothesis This study assumes to have the prognostic ability of cardiac troponin in cancer patients visiting the emergency department. Methods Cancer patients visiting the emergency department were enrolled in this retrospective cohort study. Patients with previously known coronary artery disease or clinically indicated coronary angiography were not included. The maximal value from Siemens ADVIA Centaur troponin I Ultra assay within 24 hours was assessed. The primary endpoint was 180‐day all‐cause death, including cardiovascular and noncardiovascular death. Results A total of 9135 cancer patients (mean age: 63 years, male gender: 60%) were enrolled. Lowest (0.006 ng/mL), assay‐specific <99th % (0.007‐0.039 ng/mL), below median ≥ 99th % (0.040‐0.129 ng/mL), and above median ≥ 99th % (≥0.130 ng/mL) TnI were found in 4487 (49.1%), 3158 (34.6%), 852 (9.3%), and 638 (7.0%) patients, respectively. There was 3192 (34.9%) all‐cause deaths including 137 (1.5%) cardiovascular and 3047 (33.4%) noncardiovascular deaths in the 180‐day follow‐up period. The risks of all‐cause, cardiovascular, and noncardiovascular death increased across higher TnI strata (hazard ratio [HR] = 1.3‐2.9; 2.1‐9.3; 1.3‐1.8; P < .001, all). These findings were consistent within clinical subgroups including solid and hematologic cancers. Conclusions Cancer patients visiting the emergency department with elevated troponin I were at increased risk of 180‐day death. Cancer patients with elevated TnI may need additional evaluation or careful follow‐up even without cardiovascular disease diagnosis.


| INTRODUCTION
Cancer patients are increasingly visiting emergency departments. 1 The emergency department is often the first place patients turn to when they have a complication or unexpected worsening of their condition. Cancer patients often require management of conditions caused by the cancer itself as well as the comorbidities and treatment effects. Emergency physicians frequently need to quickly evaluate patients' clinical needs, stratify individual risk, determine the most appropriate treatment, and decide whether to admit or provide outpatient care, all with limited time and resources. Case complexity can increase when patients present to the emergency department with little to no medical history information. A simple affordable riskpredicting biomarker could be helpful in such situations.
Cardiac troponin is a highly sensitive and specific biomarker for myocardial infarction, and may be elevated in patients without clinically overt myocardial infarction. 2 Cardiac troponin is frequently elevated in nonischemic cardiac conditions such as arrhythmias, heart failure, pericarditis, and also in noncardiac conditions such as pulmonary embolism, stroke, chronic kidney disease, sepsis or systemic inflammatory response syndrome (SIRS), critical illness, and drugcausing cardiotoxicity. 3 Although the pathophysiology of cardiac troponin elevation in nonischemic cardiac or extracardiac conditions are not well elucidated, elevated cardiac troponin showed prognostic implications not only in acute coronary syndrome but also in various clinical settings regarding patients with nondiagnosed chest pain and patients using statins for primary prevention. [4][5][6][7][8] Therefore, we reasoned that troponin may predict clinical outcomes in patients with cancer. We investigated the ability of cardiac troponin to predict mid-term (180 days) death risk in cancer patients who visited the emergency department of a tertiary hospital.

| Patients
We performed a retrospective single-center cohort study of patients

| Data sources and definitions
All data were retrieved from the Clinical Data Warehouse Center at Samsung Medical Center (Darwin-C). In cases of mortality outside the hospital, the specific cause of death was retrieved from the government operated organization Statistics of Korea. The timing of all clinical events, arrival or discharge time, death, and laboratory tests were defined using the timestamp in patients' electrical medical records.
The study index date was used as the visit time to the emergency department.
The Siemens ADVIA Centaur XP analyzer (Munich, Germany) was used for cardiac troponin I measurement. It has a lowest cardiac troponin I (TnI) analytic sensitivity of 0.006 ng/mL, an upper refence limit at the 99th % of 0.040 ng/mL, and a coefficient of variation <10% at 0.030 ng/mL. The maximum level of TnI within 24 hours after emergency department visit was grouped into four categories using the lowest (0.006 ng/mL), <99th % (0.007-0.039 ng/mL), less than median ≥ 99th % (0.040-0.129 ng/mL), and median ≥ 99th % or higher (≥0.130 ng/mL). In instances of multiple TnI measurements, the highest value measured within 24 hours was selected. The first mea-  R version 3.6 (R Foundation for Statistical Computing, Vienna, Austria) was used for analyses. Hazard ratios (HR) for compared outcomes are reported with a 95% confidence interval (CI). Statistical significance was defined by two-tailed P < .05.

| Study population
A total of 37 747 patients who visited the emergency department and underwent TnI testing within 24 hours were screened for potential cardiovascular disease. Most TnI results were available within 2 hours (median: 1.1, interquartile range: 0.8-1.9 hours). After excluding 28 344 patients without cancer and 268 patients who underwent cardiopulmonary resuscitation, received mechanical circulatory support, had a prior history of coronary artery disease diagnosis or revascularization, or heart transplantation, or underwent urgent coronary angiography within 48 hours of visiting the emergency department, a total of 9135 patients were enrolled in this study ( Figure 1).
A set of multivariate Cox regression models showed that elevated TnI levels were independently associated with cardiac and noncardiac death as well as all-cause death, not only in the basic model but also in the models that included clinical comorbidities, electrocardiographic change, and laboratory results (HR = 1.19-5.63; P < .05 for all, except TnI = 0.007-0.039 ng/mL for cardiac death; Figure 3).

| DISCUSSION
In this retrospective cohort study, elevated TnI was independently associated with higher risk for both cardiac and noncardiac death in cancer patients visiting the emergency department who had not had a previous diagnosis of specific cardiac disease. Higher TnI levels were associated with increasing death risk. These results remained after sensitivity analyses using various multivariate Cox regression models.
To the best of our knowledge, our study is the first to enroll a large Our study showed associations but not causal relationships among elevated cardiac troponin I and clinical outcomes. Interestingly, the risk for noncardiac death was much higher than for cardiac death.
This suggests that the risk of death might be driven mostly by patient comorbidities rather than by complications of myocardial ischemia or necrosis ( Table 2).
Cardiac troponin level has predictive value but it was not compared with other parameters or predicting algorithms because of a lack of standardized guidelines to assess patients visiting the Cardiac troponin is an organ-specific, not disease-specific, biomarker. Although noncanonical expression of cardiac troponin in human cancer cells has been reported, 13,14 it is widely accepted that cardiac troponin is highly specific and sensitive to myocardial injury.
However, elevation of cardiac troponin in cases of nonischemic cardiac disease or extracardiac disease is not uncommon. 15 Elevation of cardiac troponin in cancer patients can be caused by myocardial ischemia secondary to reduced oxygen supply or increased oxygen demand, or by direct cardiac damage secondary to chemotherapy or infiltrative cardiac diseases. 16

| Limitations
Although we enrolled any person who visited the emergency department to minimize selection bias, our study is not free from the inherent limitations of a retrospective single-center study. Troponin testing was performed at the physician's discretion. There were no pre-specified indications for troponin testing. The source data were administrative claims, which lack codes for specific conditions. We did not include patients who underwent coronary angiography during their hospital stay and the cause of TnI elevation was not investigated. The type, biology, and treatment setting for each patient's cancer were not investigated. The severity of chemotherapy-or radiation-induced cardiotoxicity was not assessed.

| Medical knowledge competency
Cancer patients with elevated TnI in the emergency department had increased risk of 180-day death.

| Patient care competency
It is important for emergency physicians, cardiologists, and oncologists to know that cancer patients with higher TnI levels have increased risk of death, which may help them make efficient and effective collaborative decisions about their healthcare strategy regarding aggressive, palliative, or hospice care given the available medical resources in crowded emergency departments.
F I G U R E 3 Panels A -C show Cox regression models using demographics and visiting years. Panels D -F show Cox regression models using clinical parameters in addition to parameters used in Panels A -C. Panels G -I show Cox regression models using laboratory results, vital signs, and ECG in addition to parameters used in Panels D -F. TnI levels and associated data are shown in blue color. TnI, cardiac troponin I; ECG, electrocardiography; SBP, systolic blood pressure; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; BMI, body mass index; WBC, white blood cell

| Translational outlook
Cancer patients usually have complex clinical conditions and comorbidities, and additional studies to design complex algorithms that include various predictors including TnI are needed to make better prognoses for life expectancy.

DATA AVAILABILITY
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.