key: cord-0958280-g9qlo3xh
authors: Toraih, Eman A.; Elshazli, Rami M.; Hussein, Mohammad H.; Elgaml, Abdelaziz; Amin, Mohamed Nasreldien; El‐Mowafy, Mohammed; El‐Mesery, Mohamed; Ellythy, Assem; Duchesne, Juan; Killackey, Mary T.; Ferdinand, Keith C.; Kandil, Emad; Fawzy, Manal S.
title: Association of cardiac biomarkers and comorbidities with increased mortality, severity, and cardiac injury in COVID‐19 patients: A meta‐regression and Decision tree analysis
date: 2020-06-12
journal: J Med Virol
DOI: 10.1002/jmv.26166
sha: c01946bba84d2088f8a8aa37dfd69aa250aba522
doc_id: 958280
cord_uid: g9qlo3xh

BACKGROUND: Coronavirus disease‐2019 (COVID‐19) has a deleterious effect on several systems, including the cardiovascular system. We aim to systematically explore the association of COVID‐19 severity and mortality rate with the history of cardiovascular diseases and/or other comorbidities and cardiac injury laboratory markers. METHODS: The standardized mean difference (SMD) or odds ratio (OR) and 95% confidence intervals (CI) were applied to estimate pooled results from the 56 studies. The prognostic performance of cardiac markers for predicting adverse outcomes and to select the best cutoff threshold was estimated by ROC curve analysis. Decision tree analysis by combining cardiac markers with demographic and clinical features was applied to predict mortality and severity in COVID‐19 patients. RESULTS: A meta‐analysis of 17,794 patients showed patients with high cardiac troponin I (OR=5.22, 95%CI=3.73‐7.31, p<0.001) and AST levels (OR=3.64, 95%CI=2.84‐4.66, p<0.001) were more likely to develop adverse outcomes. High troponin I >13.75 ng/L combined with either advanced age >60 years or elevated AST level >27.72 U/L was the best model to predict poor outcomes. CONCLUSIONS: COVID‐19 severity and mortality are complicated by myocardial injury. Assessment of cardiac injury biomarkers may improve the identification of those patients at the highest risk and potentially lead to improved therapeutic approaches. This article is protected by copyright. All rights reserved.

The first incidence of coronavirus disease 2019 infection is considered the most serious infection worldwide, most of the infected individuals suffer from mild or moderate symptoms that begin in the first week after infection. The most common mild symptoms include fever, fatigue, and cough. However, infected patients may suffer from serious complications that vary in their degrees between different individuals such as dyspnea, severe pneumonia, and organ dysfunction 1 . based on the previous facts, the diagnosis of COVID-19 cannot be based on specific symptom detection and the only specific detection test depends on identification

Using the key terms, a total of 4021 articles were retrieved using the search strategy.

After screening by the abstract and title of 1541 studies, 160 articles were selected for full-text assessment. Of these, 104 were excluded due to lack of enough data, and 56

were included for qualitative analysis. Pairwise comparison meta-analysis was conducted; 29 articles to compare between the severe and non-severe presentation of COVID-19 disease, 7 records to compare between cohorts who developed cardiac injury and those who are not, 6 records to compare between patients who were admitted to the ICU and those admitted to the general hospital ward and 16 studies to compare between survivors and expired patients ( Figure 1A) . The study included a total of 56 studies (52 retrospective and 4 prospective studies) published from January 24, 2020, to May 7, 2020 1, . These included 17,794 COVID-19 patients from China (13 cities) and overseas ( Figure 1B-C) . The main characteristics of eligible studies are demonstrated in Table 1 .

The demographic characteristics of COVID19 patients are shown in Table 2 . The median Accepted Article age of 17,364 COVID-19 patients across 53 studies ranged from 32 to 74 years in patients with a good prognosis and 47 to 77 years in patients with poor outcomes. Pooled estimates revealed significantly higher age in critical/expired cases (SMD = 1.0, 95%CI = 0.72 to 1.31, p <0.001) than non-critical group. The results from 54 articles with a total sample size of 17,702 patients showed that the proportion of males was significantly higher in critical cases (OR = 1.50, 95%CI = 1.36 to 1.69, p <0.001). Evidence of heterogeneity and publication bias were observed for age data (I 2 = 97.1%, p <0.001, Egger's p = 0.041), but not for gender (I 2 = 26.5%, p = 0.041, Egger's p = 0.58).

The laboratory examination of the included studies is demonstrated in Table 2 

Summarizing analysis revealed a 93% increased risk of poor prognosis in cohorts who experienced chest pain or tightness (OR = 1.93, 95%CI = 1.14 to 3.28, p = 0.014). In addition, meta-analysis showed that COVID-19 patients who developed complications were more likely to have adverse outcomes with higher risk of mortality ( 

Furthermore, as depicted in Table 2 patients who received antibiotics (OR = 3.36, 95%CI = 1.66 to 6.77, p = 0.001), glucocorticoids (OR = 3.52, 95%CI = 2.51 to 4.93, p <0.001), immunoglobulins (OR = 3.41, 95%CI = 1.90 to 6.14, p <0.001), and hydroxychloroquine (OR = 6.67, 95%CI = 2.0 to 22.2, p = 0.002) had higher risk for poor prognosis.

However, noteworthy, there was significant heterogeneity between studies (I2 = 67.9% to 84.6%), and only two studies had reported hydroxychloroquine.

Table S1 summarizes pooled estimates for seven cardiac biomarkers, eight comorbidities, and nine secondary complications in COVID-19 patients with severe presentation compared to non-severe cohorts, who developed secondary cardiac injury versus not, ICU admitted patients versus general ward patients and survived versus expired. The Forest plot for the pooled analyses is presented in Figures S1-S11. Funnel plots for assessment of publication bias are depicted in Figure S12 . Meta-regression to assess the impact of study characteristics as sample size, the city of the study, and timing of publications as moderators for the study effect size of each pairwise comparison is Accepted Article demonstrated in Table S2 .

To assess the impact of study characteristics as sample size, the city of the study, and timing of publications as moderators for the study effect size, meta-regression was performed. Results of studies comparing critical/expired patients with non-critical cases Table 3) .

Receiver Operating Characteristics (ROC) curves were first employed to analyze the prognostic performance of cardiac markers for predicting adverse outcomes and to select the best cutoff threshold with high sensitivity and specificity. The highest area under the curves (AUC) were for myoglobin (AUC = 0.91 ± 0.07, p = 0.002) and high-sensitive cTnI (AUC = 0.89 ± 0.04, p <0.001) at the cutoff values of 72 ng/mL and 13.75 ng/L, respectively, followed by NT-proBNP (AUC = 0.86 ± 0.06, p <0.001) and AST (AUC = 0.84 ± 0.04, p <0.001). Combining cardiac markers with demographic and clinical features, decision tree analysis was used to predict mortality and severity in COVID-19 patients. Age, cTnI, and AST levels were able to classify patients into high and low-risk patients (Figure 2A-B) . High troponin I over 13.75 ng/L combined with either advanced age over 60 years or elevated AST level over 27.72 U/L were the best model to predict poor outcomes (classification accuracy = 81.03%, precision = 74.1%, recall = 86.0%, and This article is protected by copyright. All rights reserved.

diagnostic odds ratio = 20.8). After conversion of SMD to OR, meta-analysis showed that patients with high cTnI (OR = 5.22, 95%CI = 3.73 to 7.31, p <0.001) and AST levels (OR = 3.64, 95%CI = 2.84-4.66, p <0.001) were more likely to develop adverse outcomes for COVID-19 disease.

Our meta-analysis has several important aspects. We include a robust sample size with broad, global geographic reach. Utilizing a two-arms meta-analysis for 56 articles and 17794 COVID-19 subjects, our findings reveal the association of COVID-19 mortality with high levels of cardiac biomarkers. We amplify previous smaller meta-analyses and the single site or regional studies. Furthermore, as of May 8, 2020, we enclosed a larger number of studies and patients, and involved more cardiac biomarkers, demographics, and clinical data than prior studies, demonstrating multiple predictors of cardiac injury, poor prognosis, severity, ICU admission, and mortality. Additionally, for prognostic risk assessment, we employed decision tree model analysis for both serum biomarkers and the clinical data and performed ROC curves analyses. Although our analysis included 169 hospitals located in 11 countries in Asia and Europe, it is largely retrospective.

Meta-regression analyses indicated the pooled results were independent to study characteristics and decision tree analysis revealed that cTnI, AST, and potentially other serum biomarkers could be predictors of risk. One significant limitation, inherent in the use of meta-analyses to guide further clinical practice is the heterogeneity across studies, including differences in study methods. This article is protected by copyright. All rights reserved. This article is protected by copyright. All rights reserved.

The exact pathway by which elevated biomarkers leads to death with COVID-19 with systemic inflammatory activity may include myocarditis, thrombosis, and additionally unstable coronary atherosclerotic plaque rupture. Hence, beyond the predominant pulmonary complications, severity, and mortality sources include viral myocarditis, cytokine-driven myocardial damage, microangiopathy, and acute coronary syndromes 75 . Therefore, biomarkers may identify a heightened inflammatory response, including endothelial dysfunction and microvascular damage.

There are several limitations to our analysis and review. The actual cause of mortality may be obscured by unmeasured or unknown confounders, underestimated by analysis of multivariable regression. Understanding CVD-associated mortality must integrate biomarker data with cardiac imaging and physiologic and structural abnormalities.

Additionally, the percentage of patients with sepsis has been under-reported in our report and cardiac injury may correlate with the prevalence of shock with severe COVID-19. 76 Another limitation of these data is the lack of a determination of timing and estimated glomerular filtration rate (eGFR) as factors. Although cardiac biomarkers may reflect myocardial injury, inflammation, and remodeling, interpretation of biomarkers in chronic kidney disease (CKD) can be complicated by decreased urinary clearance and/or overall CKD-associated chronic inflammation. The prognostic power of future biomarker analyses for COVID-19 mortality should be trended over time and account for the degree of renal dysfunction. 77 Finally, in consideration of the immense COVID-19 global mortality, over 360,000 deaths 78 , with over 100,000 deaths in the U.S. alone 79 at the time of manuscript submission, despite our relatively large sample size, our data will require This article is protected by copyright. All rights reserved.

ongoing supplementation, to overcome inherent statistical bias and confirming our results.

In conclusion, COVID-19 severity and mortality are compounded by vascular and myocardial injury. Elevated cardiac injury biomarkers may improve the identification of those patients at the highest risk and potentially lead to improved therapeutic approaches.

None.

All authors declare no conflict of interest. 

Potential Effects of Coronaviruses on the Cardiovascular System: A Review

Software Review: Borenstein

Random-effects model for meta-analysis of clinical trials: an update

Quantifying publication bias in meta-analysis

Clinical features, laboratory characteristics, and outcomes of patients hospitalized with coronavirus disease 2019 (COVID-19): Early report from the United States

Analysis of myocardial injury in patients with COVID-19 and association between concomitant cardiovascular diseases and severity of COVID-19

Clinical and immunological features of severe and moderate coronavirus disease 2019

Suspected myocardial injury in patients with COVID-19: Evidence from front-line clinical observation in Wuhan, China

Surviving Sepsis Campaign: Guidelines on the Management of Critically Ill Adults with Coronavirus Disease 2019 (COVID-19)

Age-dependence of mortality from novel coronavirus disease (COVID-19) in highly exposed populations: New York transit workers and residents and Diamond Princess passengers. medRxiv

Preliminary Estimate of Excess Mortality During the COVID-19 Outbreak