What's new?

The study presents results based on a large database of consecutive patients with COVID‑19 treated in a single hospital according to the same standard. It shows that patients with pre‑existing heart failure and COVID‑19 had a particularly high risk of in‑hospital death. However, abnormal values of high‑sensitivity cardiac troponin T and N‑terminal pro–B‑type natriuretic peptide (NT‑proBNP) measured on admission allowed us to identify patients with a higher risk of death and prolonged hospitalization, regardless of whether they had heart failure or not. Thus, the study postulates that the markers of cardiac injury should be routinely measured on admission in all patients with COVID‑19, not only in those with a history of cardiovascular disease. The results add to the growing evidence that simple‑to‑obtain cardiac biomarkers, such as cardiac troponin and NT‑proBNP, may help in early risk stratification of patients hospitalized for SARS‑CoV‑2 infection.

Introduction

Although COVID‑19 is a viral disease in which the lungs are the primary and most severely affected organ, in some cases, it can also involve other organs, especially the heart and peripheral vessels. After the early infection stage, a severe hyperinflammation may occur, from which systemic complications are likely to result. The severity of COVID‑19 depends on many factors, the most important being age and pre‑existing comorbidities.^1-4 SARS‑CoV‑2 can also affect the cardiovascular system in a variety of ways. In patients without a prior history of cardiovascular disease, it is likely that the extent of systemic inflammatory response is of vital importance. In individuals with a history of pre‑existing cardiovascular disease or comorbidities, these conditions become apparent or more complicated during COVID‑19, both as a consequence of inflammation and due to the side effects of medication or secondary hospital‑acquired infections or complications.^5-7 Most patients infected with SARS‑CoV‑2 report to the hospital after the viremia phase, when they develop more severe symptoms of the disease: fever, weakness, dyspnea, and hypoxemia. This means that they are at least in the pulmonary phase of the disease, which can be confirmed by lung imaging. At this early stage of hospitalization, it is extremely important to identify the patients who are at an increased risk of death.

For many years, cardiac biomarkers are have been used in clinical practice to identify and assess the progression of cardiovascular diseases. Among these biomarkers, high‑sensitivity cardiac troponin (hs‑cTn) and N‑terminal pro–B‑type natriuretic peptide (NT‑proBNP) are the most frequently used. They are important in various cardiovascular conditions, such as chronic heart failure (CHF), ischemic heart diseases, acute coronary syndromes, or pulmonary embolism. Hs‑cTn, as a marker of cardiomyocyte necrosis, and NT‑proBNP, as a marker of cardiac biomechanical stress, seem to be useful for early stratification of patients hospitalized for COVID‑19—both are associated with damage of the heart, and therefore, their concentration during COVID‑19 indicates involvement of the cardiovascular system.^5,8-10

Although several studies have shown that elevated levels of hs‑cTn or NT‑proBNP are associated with a higher risk of complications and death in patients with COVID‑19,^11-14 some aspects remain to be clarified, especially with respect to individuals with concomitant heart failure. For example, it is unclear whether troponin and / or NT‑proBNP are more useful in terms of risk stratification and mortality prediction than classic inflammatory parameters, such as leukocyte count, C‑reactive protein (CRP), or interleukin 6 (IL‑6). Thus, the aim of this study was to investigate the association between high‑sensitivity troponin T (hs‑cTnT) and NT‑proBNP values on admission to the hospital and in‑hospital mortality in a large cohort of patients with COVID‑19, with particular emphasis on individuals with a prior history of CHF.

Patients and methods

We analyzed medical records of all consecutive patients who were admitted between March 6, 2020 and October 15, 2020 to the University Hospital in Kraków, Poland, which was temporarily converted into an Infectious Disease Hospital designated for the treatment of patients with COVID‑19. The patients were diagnosed with COVID‑19 according to the World Health Organization and Polish guidelines using reverse transcription–polymerase chain reaction (rhino‑oropharyngeal swab positive for SARS‑CoV‑2 RNA).¹⁵ The protocol used for COVID‑19 treatment was in accordance with recommendations of the Polish Association of Epidemiologists and Infectiologists.^16,17 The patient data were obtained from the electronic Hospital Information System. Cardiovascular risk factors and diseases were identified based on the medical history and / or prior treatment, and were defined according to the current European Society of Cardiology guidelines.¹⁸ Information on the demographics, symptoms, laboratory tests (hs‑cTnT measured with the enzyme‑linked immuno‑culture assay [ELICA] using a Cobas Pro device [Roche Diagnostics GmbH, Mannheim, Germany]; NT‑proBNP measured with ELICA using a Cobas Pro device [Roche Diagnostics]; high‑sensitivity CRP (hs‑CRP) measured by nephelometry using a Siemens BN II device [Siemens Healthineers, Erlangen, Germany]; interleukin 6 (IL‑6) measured with ELICA using a Cobas Pro device [Roche Diagnostics]; D‑dimer measured by coagulometry using a Siemens Atellica COAG 360 device [Siemens Healthineers]), treatment options, and clinical outcomes (pneumonia, length of hospital stay, survival at discharge or death) was collected from all participants. Some data were presented in our previous paper, in which we also analyzed the association between cardiovascular drugs and mortality rates.¹⁹ Only cardiac and inflammatory biomarkers on admission (day 1 or 2) were considered. The diagnosis of pneumonia was confirmed using imaging studies such as chest X‑ray or high‑resolution computed tomography (CT) performed soon after admission. The data of survivors and nonsurvivors were compared, and were additionally analyzed according to the presence of previously diagnosed CHF. The study was approved by the Jagiellonian University Ethics Committee (no. 1072.6120.278.2020) and conforms to the ethical principles outlined in the Declaration of Helsinki.

Results

Receiver operating characteristic and logistic regression analyses

In the ROC curve analysis, hs‑cTnT values of 142 ng/l or greater measured on admission had an 80% sensitivity and a 70% specificity for predicting in‑hospital death (AUC = 0.8) (Figure 1). For NT‑proBNP, the cutoff value of 969 pg/ml had a sensitivity and specificity of 79% and 67%, respectively, for predicting in‑hospital mortality (AUC = 0.8) (Figure 2).

Logistic regression analysis showed that in COVID‑19 patients without CHF, each 10‑ng/l increase in the hs‑cTnT concentration was independently associated with a higher risk of death (odds ratio [OR], 1.01; 95% CI, 1.00–1.07; P = 0.009), based on the model of variables described in the Statistical analysis section (detailed results are presented in Supplementary material, Table S1).

Similarly, in patients without CHF, logistic regression analysis of the NT‑proBNP concentration on admission showed that for every 100‑pg/ml increase there was a greater risk of death (OR, 1.02; 95% CI, 1.01–1.03; P <⁠0.001) (Supplementary material, Table S1).

Discussion

The study demonstrated that patients infected with SARS‑CoV‑2 who had elevated serum concentration of hs‑cTnT (>142 ng/l) and / or NT‑proBNP (>969 pg/ml) on admission had a markedly higher risk for in‑hospital death than those with lower values of these parameters. Since hs‑cTn and NT‑proBNP levels were assessed on admission in about 60% and 41% of all patients, respectively, it can be assumed that the doctors ordering these tests were guided by the assessment of the general health status of their patients upon admission. This is evidenced by the fact that most comorbidities, and also pneumonia, were observed more frequently in the patients who had the hs‑cTnT or NT‑proBNP levels measured. Additionally, the high percentage of patients with hs‑cTnT and NT‑proBNP levels evaluated on admission among individuals with CHF (87% and 73%, respectively) supports this hypothesis. Due to the fact that the frequency of myocardial injury (as defined by elevated troponin levels) among hospitalized patients with COVID‑19 increases with greater severity of the disease,^20-22 it may be concluded that the frequency of testing for cardiac markers in the present study should be higher. However, in other studies, the percentage of patients in whom these markers were assessed was similar or even lower. For example, Cunningham et al²³ performed a large analysis of over 48 000 patients with COVID‑19 and found that hs‑cTn and NT‑proBNP were measured on admission in 34% and 25% of individuals, respectively.

The prevalence of elevated cardiac troponin levels in individuals with COVID‑19 varies widely, ranging from 20% in cohorts of hospitalized patients to more than 40% in critically ill patients.^24-26 Among our COVID‑19 cases, cardiac injury, defined by hs‑cTnT levels greater than 99th percentile URL on admission, was present in 41.5% of patients. However, considering the entire study group, regardless of the hs‑cTnT levels being determined on admission, the patients with cardiac injury accounted for 25% of the total cohort (432 out of 1729). These values are therefore similar to the results presented by other authors mentioned above.^24-26

In the present study, higher NT‑proBNP values were observed in patients whose hs‑cTnT levels were measured on admission. The reverse relationship was also observed, that is, higher hs‑cTnT values were found in patients with NT‑proBNP measurements performed on admission. This means that higher values of one of the cardiac markers were accompanied by higher values of the other one. Moreover, higher values of these cardiac biomarkers were seen in more severely ill patients. This finding is consistent with recent observations that myocardial injury in COVID‑19 appears to be predominantly mediated by the severity of critical illness rather than direct injury of the myocardium from viral particles.²⁷ Finally, in our study, the length of hospital stay was longer by 4 days in those with hs‑cTnT or NT‑proBNP levels measured on admission than in those in whom these markers were not assessed. Since the exact mechanisms leading to myocardial injury in COVID‑19 are still not well understood (whether cardiomyocyte damage dominates, as reflected by troponin leak, or whether heart overload is crucial, as reflected by natriuretic peptide increase), it seems justified to determine both biomarkers on admission to the hospital. The relationship between the concentration of cardiac troponins on admission and mortality in mild‑to‑severe and critical COVID‑19 patients was described by many authors.^23,28-31 Meta‑analyses published in 2020 concluded that an increased level of high‑sensitivity troponin I (hs‑cTnI) or hs‑cTnT is strictly associated with the most severe forms of COVID‑19.^32,33 Our observations are in line with these reports.

Another important observation from the present study is a significantly higher in‑hospital mortality rate among COVID‑19 patients with a history of CHF than in those without CHF. Other researchers also reported high mortality (up to 45%) in patients with CHF during hospitalization for COVID‑19.³⁴ In our previous publication we found that, apart from older age and male sex, only diabetes mellitus and HF were independent predictors of in‑hospital death from any cause.¹⁹ In the present study, the patients with HF had higher hs‑cTnT and NT‑proBNP values on admission than the remaining patients. Undoubtedly, this result was influenced by the higher age of patients with CHF and the burden of comorbidities. Unexpectedly, we did not find differences in the frequency of pneumonia, length of hospital stay, or hs‑CRP and IL‑6 values between patients with or without CHF who had had hs‑cTnT or NT‑proBNP levels measured on admission. This fact is not easy to understand. It may be hypothesized that CHF is such a serious disease that its presence could predispose to a more frequent occurrence of viral pneumonia and a higher concentration of inflammatory markers than in individuals without CHF. Unfortunately, in our database we did not collect data on the New York Heart Association functional class of the patients with CHF, which could have also elucidated their health status and better estimate the “at entry risk.” On the other hand, it should not be forgotten that before admission to the hospital, the patients with CHF had already received the usual treatment for this disease, such as ACEIs or ARBs, mineralocorticoid receptor antagonists, β-blockers, and diuretics. As shown in our previous analysis of COVID‑19 patients, the use of these drugs during hospitalization was associated with a better prognosis, regardless of sex, age, and comorbidities.¹⁹ Nonetheless, there are also other observations about advantages and safety of ACEIs or ARBs used in the treatment of HF in COVID‑19 cases.³⁵ A retrospective analysis recently performed in Spain showed that patients with a history of HF hospitalized for COVID‑19 who continued to receive ACEI/ARB treatment had lower rates of in‑hospital mortality and other adverse outcomes as compared with those in whom the therapy was stopped.³⁶ Also of note, cardiac injury is not always associated with unfavorable respiratory outcomes. For instance, in a meta‑analysis published in 2021 by Bansal at al,³⁷ cardiac injury in patients with COVID‑19 was associated with a higher risk of mortality (risk ratio [RR], 7.79), admission to the intensive care unit (RR, 4.06), mechanical ventilation (RR, 5.53), and coagulopathy (RR, 3.86), but was not associated with an increased risk of acute respiratory distress syndrome. It seems that the relationship between developing pneumonia and elevation of cardiac biomarkers during the SARS‑CoV‑2 infection needs further investigation, especially in patients with a prior history of CHF.

The importance of NT‑proBNP levels on admission in patients with CHF complicated by COVID‑19 was also explored before, although not as often as that of cardiac troponin. A single‑center study by Belarte‑Tornero et al³⁸ involved 129 patients (43 CHF and 86 non‑CHF) with COVID‑19. All‑cause mortality was almost 2 times higher in the CHF patients than in those without CHF. A history of CHF was independently associated with 30‑day mortality (hazard ratio, 2.3); however, the patients with CHF and normal hs‑cTnT levels (<⁠14 ng/l) had an excellent prognosis. In that study, an NT‑proBNP level greater than 2598 pg/ml on admission was associated with a higher 30‑day mortality in the patients with CHF.³⁸ Our study included slightly more individuals with CHF (179 cases); however, the cutoff point for increased mortality in patients with COVID‑19 in the ROC curve analysis was lower, namely, 969 pg/ml. In a recently published meta‑analysis with meta‑regression, involving about 18 800 COVID‑19 cases, Zinnellu et al³⁹ confirmed that BNP/NT‑proBNP concentrations were significantly higher in the patients with a more severe disease or nonsurvivor status than in those with low disease severity or survivor status during follow‑up.

In our hospital laboratory, the cutoff point for normal hs‑cTnT value (>99th percentile URL) was 14 ng/l, and the cutoff point for normal NT‑proBNP was 125 pg/ml. In the analyzed cohort of COVID‑19 cases without a history of CHF, even a slightly elevated level of hs‑cTnT above the reference range was associated with significantly higher mortality. As for NT‑proBNP, only an increase by about 100% above the reference range was associated with a significant increase in the number of deaths in patients without a prior history of CHF. Based on this observation, blood test results showing a slight elevation in hs‑cTnT and / or a relatively mild increase in NT‑proBNP concentrations on admission of a patient with SARS CoV‑2 infection should warrant more stringent monitoring because the risk of in‑hospital death may be greater. Other authors reported similar findings.^5,11,22 In a multicenter, retrospective study of approximately 3200 patients with COVID‑19, Qin et al⁴⁰ suggested that to evaluate the prognosis of patients with COVID‑19 on admission, the cutoff threshold of abnormality for hs‑cTnI, creatine kinase (CK)-myocardial band, NT‑proBNP, and CK should be lower than the currently recommended laboratory range. In their dataset, cutoff values of these cardiac markers for worse prognosis were found to be much lower (ca 50%) than the upper reference limits for the general population.⁴⁰

In patients with a SARS‑CoV‑2 infection, several mechanisms could be involved in myocardial injury and the subsequent elevation of cardiac troponin and / or NT‑proBNP levels. These include acute viral infections that may cause myocarditis or pericarditis, hypoxia during the course of the disease, and endothelial dysfunction which may lead to ischemia of the heart and, in some cases, acute myocardial infarction type 2. Bacterial sepsis, cardiac adrenergic activation, and pulmonary embolism could also lead to troponin level elevation.^21,41 Also, comorbidities may increase hs‑cTn values.^2,42 In our study, hypertension, diabetes, coronary artery disease, atrial fibrillation, chronic obstructive pulmonary disease, and HF occurred more frequently in the group of patients who had hs‑cTnT levels measured on admission. Our results are in line with previous reports suggesting that the measurement of hs‑cTn and NT‑proBNP levels may improve risk stratification in patients with COVID‑19 in addition to the assessment of inflammatory or thrombotic markers.^43-45

We must acknowledge some limitations of this study. First, it was limited by its retrospective design, inducing possible bias in data collection and result interpretation (the retrospective nature of the study did not allow us to draw conclusions about the cause‑effect relationship). Moreover, our results were obtained based on cases treated during the first 2 waves of the pandemic, with a similar protocol of COVID‑19 in‑hospital treatment. Therefore, the generalizability of our findings to asymptomatic or symptomatic COVID‑19 patients who were not hospitalized may be limited. However, we presented a large, comprehensive dataset of consecutive patients and checked the data carefully. Second, hs‑cTn and NT‑proBNP levels were not systematically monitored during the hospital stay. Although the patients were treated in a single hospital, the assessment of cardiac troponin and NT‑proBNP levels on admission was not obligatory. Third, data on concomitant diseases, including a history of HF, were obtained based on the available medical records and an interview with the patient or their relatives. Finally, echocardiography was performed on admission in a small subset of the study patients (<⁠3%), which made it impossible to perform a reliable analysis of the heart function in relation to hs‑cTn or NT‑proBNP concentrations.

In conclusion, hs‑cTnT or NT‑proBNP predicted in‑hospital mortality in all patients with COVID‑19 included in our study. The patients with a history of CHF with concomitant COVID‑19 who had elevated cardiac troponin or NT‑proBNP levels at the time of admission to the hospital had the worst prognosis. Elevated values of hs‑cTnT and NT‑proBNP also enabled us to identify the patients with a higher risk of prolonged hospitalization time. For this reason, both cardiac troponin and NT‑proBNP should be routinely measured on admission in all patients with COVID‑19, not only in those with a history of cardiovascular disease. The results of the study support the growing evidence that simple‑to‑obtain cardiac biomarkers may help in early risk stratification of patients hospitalized for SARS‑CoV‑2 infection.