Left atrial function in patients with COVID-19 and its association with incident atrial fibrillation/flutter.

Myocardial injury in Coronavirus disease 2019 (COVID-19) has been associated with adverse outcomes; however, associations between myocardial injury and arrhythmias, such as atrial fibrillation/flutter (AF), are not well established in this population. , Recent advances in two-dimensional echocardiography (2DE), including speckle-based strain, enable the quantification of left atrial (LA) strain (LAS), a measure of atrial deformation that has previously been shown to be predictive of AF and cardiovascular events in stable outpatients. , We aimed to compare echocardiographic measures of LA function between hospitalized COVID-19 patients and COVID-19-negative controls to test the hypothesis that COVID-19 patients have reduced LA function as reflected by abnormal LAS and LA emptying fraction (LAEF). We then tested the hypothesis that among COVID-19 patients, LA dysfunction and cardiac biomarker elevation are associated with incident AF.

This study was approved by the Johns Hopkins Institutional Review Board. From March 25, 2020, to June 20, 2020, we retrospectively studied hospitalized adults who underwent clinically indicated 2DE with adequate image quality in accordance with the American Society of Echocardiography guidelines. , Our cohort included 80 patients with COVID-19 and 34 controls without COVID-19, selected by frequency matching from patients admitted to intensive or intermediate care units with one or more respiratory problems. All patients had 2DE during admission and were followed to discharge or death. Patients were excluded for any history of atrial arrhythmia. The 2DEs were blindly analyzed offline for LAEF and reservoir (peak longitudinal) LAS using a vendor-independent strain application (TomTec Imaging Systems, Munich, Germany). Demographic, clinical, and biomarker data, obtained within 72 hours of 2DE, were taken from the electronic medical record. Atrial fibrillation/flutter was diagnosed by inpatient telemetry. Comparisons were made between COVID-19 patients and controls and between COVID-19 patients with and without AF. The Wilcoxon rank-sum test was used for continuous variables, and the χ2 test was used for categorical variables. Logistic regression was performed to investigate associations between AF and clinical variables.

Patient demographics are listed in Tables 1 and 2 . COVID-19 patients had lower LAS (28.2% [22.9%-34.1%] vs 32.6% [27.7%-38.8%], P = .026) and LAEF (55.7% [50.8%-62.6%] vs 64.1% [58.6%-71.9%], P < .001) compared with controls. Traditional cardiovascular risk factors, inflammatory and cardiac biomarkers, and mortality were similar between groups; however, there was a higher incidence of intensive care unit (ICU) admission and lower incidence of acute respiratory distress syndrome (ARDS) in the control group (Table 1).

Table 1

Characteristics of COVID-19 patients versus COVID-19-negative controls

Characteristics	Total cohort (N = 114)	COVID-19+ (n = 80)	Controls (n = 34)	P value
Age, years, median	61 [51-71]	61 [51-70]	61 [53-72]	.77
Gender, female	47 (41)	32 (40)	15 (44)	.68
Race
White	23 (20)	13 (16)	10 (29)	.11
African American	59 (52)	39 (49)	20 (59)	.33
Hispanic	15 (13)	14 (18)	1 (3)	.035
Other	18 (16)	14 (16)	4 (12)	.44
BMI, kg/m2	29.2 [25.6-34.5]	29.4 [26.4-34.9]	25.9 [24.1-34.1]	.09
Comorbidities
Hypertension	80 (70)	56 (70)	24 (71)	.95
Diabetes mellitus	41 (36)	33 (41)	8 (24)	.07
Hyperlipidemia	49 (43)	39 (49)	10 (29)	.06
Congestive heart failure	14 (12)	11 (14)	3 (9)	.46
Coronary artery disease	13 (11)	10 (13)	3 (9)	.57
Clinical variables∗
Troponin I, ng/mL	0.03 [0.03-0.10]	0.03 [0.03-0.10] n = 74	0.03 [0.03-0.09] n = 26	.72
NT-proBNP, pg/mL	393 [120-1844]	337 [111-1,495] n = 66	1,134 [220-2,116] n = 17	.12
C-reactive protein, mg/dL	12 [3.8-18.0]	12 [3.9-18.4] n = 61	5.7 [2.9-11.6] n = 10	.29
Ferritin, ng/mL	786 [410-1,689]	915 [509-1,689] n = 57	347 [197-2,234] n = 12	.12
D-dimer, mg/L FEU	2.3 [0.9-7.8]	2.1 [0.8-8.1] n = 79	3.8 [2.2-7.6] n = 9	.14
Clinical events
ICU admission	89 (78)	58 (73)	31 (91)	.03
Shock	57 (50)	44 (55)	13 (38)	.10
VTE	30 (26)	23 (29)	7 (21)	.37
ARDS	51 (45)	49 (61)	2 (6)	<.001
Death	16 (14)	11 (14)	5 (15)	.89
Time from admission to 2DE, days	3 [1-8]	4 [2-9]	3 [1-7]	.50
Echo parameters
LVEF, %	62.5 [55.0-67.5]	62.5 [53.8-63.8]	62.5 [62.5-67.5]	.011
LAVI, mL/m2	21.6 [17.6-29.5]	20.6 [16.3-28.7]	27.8 [20.1-33.0]	.003
LAEF, %	58.8 [50.8-64.8]	55.7 [50.8-62.6]	64.1 [58.6-71.9]	<.001
LA reservoir strain, %	29.4 [23.6-35.7]	28.2 [22.9-34.1]	32.6 [27.7-38.8]	.026

Data are presented as median [interquartile range] or n (%). GLS, global longitudinal strain; LVEF , left ventricular ejection fraction; VTE, venous thromboembolism.

Johns Hopkins Hospital laboratory reference ranges: troponin I <0.04 ng/mL; NT-ProBNP 0-125 pg/mL; C-reactive protein <0.05 mg/dL; ferritin 13-150 ng/mL; D-dimer <0.49 mg/L fibrinogen-equivalent units.

Table 2

Characteristics of COVID-19 patients who did and did not develop AF

	COVID-19 + No-AF (n = 56)	COVID-19+ AF (n = 24)	P Value
Age, years	60 [48-68]	66 [60-75]	.017
Gender, female	21 (38)	11 (46)	.49
Race
White	4 (7)	9 (38)	.001
African American	28 (50)	11 (46)	.73
Hispanic	11 (20)	3 (13)	.44
Other	13 (23)	1 (4)	.040
BMI, kg/m2	29.2 [26.4-34.9]	30.1 [26.3-36.8]	.79
Comorbidities
Hypertension	39 (70)	17 (71)	.92
Diabetes mellitus	23 (41)	10 (42)	.96
Hyperlipidemia	25 (45)	14 (58)	.26
Congestive heart failure	9 (16)	2 (8)	.36
Coronary artery disease	7 (13)	3 (13)	>.99
Clinical variables∗
Troponin I, ng/mL	0.03 [0.03-0.05]	0.07 [0.03-0.17]	.011
NT-proBNP, pg/mL	231 [97-846] n = 47	946 [388-3,997] n = 19	<.001
C-reactive protein, mg/dL	13 [2.5-18.0] n = 42	11.8 [4.3-20.0] n = 19	.66
Ferritin, ng/mL	945 [529-1,860] n = 40	758 [506-1,077] n = 24	.41
D-dimer, mg/L FEU	1.8 [0.7-7.6]	3.1 [1.1-9.7]	.41
Clinical events
ICU admission	35 (63)	23 (96)	.002
Shock	26 (46)	18 (75)	.019
VTE	17 (30)	6 (25)	.63
ARDS	30 (54)	19 (79)	.03
Death	6 (11)	5 (21)	.23
Time from admission to 2DE, days	4 [2-7]	5 [2-16]	.29
Echo parameters:
LVEF, %	62.5 [55.0-67.5]	57.5 [47.3-62.5]	.044
LV GLS, absolute value, %	16.9 [14.4-19.5] n = 50	16.7 [15.0-17.4] n = 13	.56
LAVI, mL/m2	20.1 [16.0-26.6]	21.6 [17.4-28.9]	.30
LAEF, %	57.3 [47.5-62.6]	54.6 [51.2-61.7]	.70
LA reservoir strain, %	30.4 [26.1-35.8]	22.3 [20.6-27.8]	<.001

Data are presented as median [interquartile range] or n (%). GLS, global longitudinal strain; LVEF , left ventricular ejection fraction; VTE, venous thromboembolism.

Johns Hopkins Hospital laboratory reference ranges: troponin I <0.04 ng/mL; NT-ProBNP 0-125 pg/mL; C-reactive protein <0.05 mg/dL; rerritin 13-150 ng/mL; D-dimer <0.49 mg/L fibrinogen-equivalent units.

COVID-19 patients who developed AF (n = 24; 30%) had higher troponin I, N-terminal pro-brain natriuretic peptide (NT-proBNP), ICU admission, ARDS, and shock; they were overall older and more often Caucasian compared with COVID-19 patients without AF (Table 2). Despite similar LA volume index (LAVI) and LAEF, LAS was significantly lower in the AF versus non-AF group of COVID-19 patients (22.3% [20.6%-27.8%] vs 30.4% [26.1%-35.8%], P < .001; Figure 1 ). On univariable logistic regression, LAS, ICU admission, and ARDS were associated with AF, while troponin and NT-proBNP levels were not. These findings persisted on multivariable regression adjusted for age, sex, and body mass index (BMI; Supplemental Table 1).

Figure 1

Example of reduced peak longitudinal/reservoir LAS in a COVID-19 patient who developed AF during admission. Average LAS here is 20% (normal, >38%).

These findings show that hospitalized COVID-19 patients have reduced LA function compared with COVID-19-negative controls with similar degrees of critical illness, and this dysfunction is more pronounced in COVID-19 patients who develop AF. Importantly, we report an independent association between LAS and AF among COVID-19 patients, even after adjustment for confounders.

Myocardial injury via serum and echocardiographic findings is associated with AF in the present COVID-19 population studied. Associations between AF and inflammatory markers such as C-reactive protein in COVID-19 have been reported. We found an insignificant trend toward higher inflammatory markers in COVID-19 patients compared with controls that was not associated with AF. The observation of more incident AF in patients with higher troponin and NT-proBNP levels, which are associated with worse outcomes in COVID-19, supports hypotheses involving COVID-19-related myocardial injury beyond that of generalized critical illness.

The LAVI was lower in COVID-19 patients compared with controls and similar in COVID-19 patients with and without AF, suggesting that LA dysfunction developed acutely rather than in the setting of chronic remodeling. Both LAVI and LAS have been shown to predict AF and cardiovascular outcomes in the outpatient setting. The association of reduced LAS with AF in COVID-19 suggests that LAS may have greater utility than LAVI in identifying atrial injury in this population. Furthermore, reduced LAS may represent a higher-risk COVID-19 phenotype that warrants closer monitoring for cardiac complications, including AF.

Limitations to our study include the modest sample size, cross-sectional design, and dependence on 2DE image quality, which allowed for LAS measurement in the two-chamber or four-chamber apical view. Since it is not routinely measured, baseline LAS remains unknown, and causality cannot be inferred based on cross-sectional design. ARDS was less frequent in the control group; however, when this was adjusted for, LAS remained significantly associated with AF (P < .001). Despite excluding patients with a history of AF, prior undiagnosed paroxysmal AF remains a potential confounder. Lastly, the findings here may only apply to COVID-19 inpatients.

Systemic inflammation in COVID-19 may contribute to an atrial myopathy that leads to increased risk of atrial arrhythmias. Evaluation of LA mechanics by 2DE provides useful data for risk stratification. Further studies are needed to confirm these findings in larger populations and define underlying mechanisms.