Atrial Function as an Independent Predictor of Postoperative Atrial Fibrillation in Patients Undergoing Aortic Valve Surgery for Severe Aortic Stenosis


Postoperative atrial fibrillation (POAF) is a common, clinically relevant, but hardly predictable complication after surgical aortic valve replacement. The aim of this study was to test the role of preoperative left atrial longitudinal strain as a predictor of POAF in clinical practice.


Sixty patients scheduled for aortic valve replacement for severe isolated aortic stenosis, in stable sinus rhythm, were prospectively enrolled and underwent full clinical, biochemical, and transthoracic echocardiographic assessment on the day before surgery. Left atrial strain–derived peak atrial longitudinal strain (PALS) and peak atrial contraction strain (PACS) were obtained. The occurrence of POAF was evaluated during the hospital stay after the intervention.


POAF was present in 26 of 60 patients (43.3%). Among all clinical variables examined, age showed a significant correlation with POAF ( P = .04), while no significant differences were noted regarding preoperative symptoms, cardiovascular risk factors, medications, and biochemical data. As for the echocardiographic parameters, only PALS and PACS showed strong, significant correlations with the occurrence of arrhythmia ( P < .0001 on univariate analysis), with areas under the curve of 0.87 ± 0.04 (95% CI, 0.76–0.94) for PALS and 0.85 ± 0.05 (95% CI, 0.73–0.93) for PACS. In two comprehensive multivariate models, PALS and PACS remained significant predictors of POAF (odds ratio, 0.73 [95% CI, 0.61–0.88; P = .0008] and 0.72 [95% CI, 0.59–0.87; P = .0007]). No significant interaction was detected between PALS or PACS and other clinical and echocardiographic variables, including age, E/E′ ratio, and left atrial enlargement.


PALS and PACS indexes are routinely feasible and useful to predict POAF in patients with severe isolated aortic stenosis undergoing surgical aortic valve replacement.


  • POAF is common after AVR for AS.

  • Preoperative assessment of longitudinal atrial strain is feasible and reproducible.

  • Reduced systolic and late diastolic atrial strain are strong predictors of POAF.

  • The predictive role of atrial strain is independent of atrial enlargement.

  • Atrial strain largely exceeds age and other clinical and echocardiographic parameters in predicting POAF.

Atrial fibrillation (AF) is a common arrhythmia worldwide, causing substantial morbidity and mortality. It is now well documented that an atrial abnormality (frequently inflammation or fibrosis) caused by underlying cardiac or noncardiac disease might act as a substrate for the development of the arrhythmia, which is, however, extremely difficult to predict.

The onset of AF can be precipitated by various conditions, and cardiothoracic surgery is recognized as a potent arrhythmic trigger. Indeed, postoperative AF (POAF) is the most common complication after cardiac surgery, occurring in up to 50% of cases. The pathogenesis of POAF is complex and multifactorial, and the clinical consequences are relevant: a prolonged intensive care unit (ICU) stay, an increased risk for perioperative stroke, and myocardial infarction.

Patients who undergo surgical aortic valve replacement (AVR) for severe aortic stenosis (AS) are at consistent risk for developing POAF, given the augmented pressure imposed on the left atrial (LA) chamber by ventricular stiffness and hypertrophy and the increased age and comorbidities of candidates for surgery in recent years.

Two-dimensional (2D) speckle-tracking echocardiography (STE) is emerging as a powerful diagnostic tool to evaluate LA mechanics, and it might be useful in this particular context, revealing subtle atrial dysfunction before surgery. According to recent studies, a reduction in LA longitudinal strain is associated with atrial tissue remodeling and fibrosis ; it has been correlated with adverse cardiac outcomes and occurrence of paroxysmal AF. Moreover, a limited number of studies, conducted with particular software, have demonstrated an association between the occurrence of POAF following cardiac surgery and reduced preoperative atrial strain parameters, mainly addressing atrial longitudinal deformation during systole. However, whether this approach is reliable and able to provide a refined characterization of the atrial function in clinical practice is still unknown.

The aim of this study was to investigate whether evaluation of LA strain at different phases of the cardiac cycle may be useful to predict POAF in patients undergoing AVR for severe isolated AS, routinely referred for echocardiographic evaluation before surgery. A secondary purpose of the study was to analyze the potential relationships of LA strain with atrial volumes and ventricular filling parameters detected by standard 2D echocardiography, as well as with clinical and biochemical data.


Eligibility Criteria

Between January 2015 and November 2015, we prospectively enrolled patients with severe isolated AS, scheduled for AVR at the Department of Cardiac Surgery, University of Verona. Aortic valve stenosis severity was defined according to current ESC Guidelines as indexed aortic valve area < 0.6 cm 2 /m 2 . The patients were evaluated the day before surgery and had to be in stable sinus rhythm as documented by 12-lead electrocardiography; other exclusion criteria were significant coronary artery disease, presence of other more than moderate valvular disease and valvular prosthesis, history of congenital cardiac abnormalities or cardiac tumors, emergency surgery, severe chronic obstructive pulmonary disease, inability to provide informed consent, and left atrial profile not entirely detectable from the apical four-chamber view during routine preoperative echocardiography. Clinical data including age, gender, coronary artery disease, New York Heart Association functional class, history of angina and syncope were collected for each patient. The presence of preoperative paroxysmal AF was determined according to previous documentation of AF on electrocardiography or a report of paroxysmal AF in the medical history. It was also evaluated from Holter electrocardiograms obtained before admission in patients with syncope. The following cardiovascular risk factors were noted: hypertension (previous diagnosis of hypertension or peripheral arterial pressure > 140/90 mm Hg on at least two different measurements at our evaluation), hyperlipidemia (low-density lipoprotein cholesterol > 100 mg/dL or use of lipid-lowering drugs), smoking (current smoking or any smoking habit in the past 10 years), and diabetes (previous diagnosis of diabetes mellitus or glycated hemoglobin ≥ 6.5%). Preoperative use of medications was documented. Peripheral arterial pressure was measured with a mercury sphygmomanometer at the upper arm in a recumbent position at the time of echocardiography, and venous blood samples were drawn the morning before surgery and postoperatively in the ICU; in particular, serum potassium relative to the first sample collected within 2 hours after arrival in the ICU was taken into account. Measurements of the levels of serum electrolytes, lipids, creatinine, lipid profile, hemoglobin, and other biochemical blood attributes were obtained using standard laboratory procedures. We analyzed P-wave duration and the P-R interval from preoperative electrocardiograms. The total intervention time (from sternal incision to the exit of the patient from the operating room) was taken into account. All subjects provided written informed consent in accordance with the research protocol approved by the local institutional review board (University Hospital of Verona ethics committee).

Standard Echocardiography

All patients underwent complete transthoracic echocardiography the day before surgery using an HD15 or iE33 ultrasound system (Philips, Best, The Netherlands), equipped with an S5 transducer (1–5 MHz). Left ventricular (LV) end-diastolic and end-systolic diameters, wall thickness, and aortic root diameters were measured using 2D echocardiography from the parasternal long-axis view, and LV mass was calculated, as recommended by American Society of Echocardiography and European Association of Cardiovascular Imaging guidelines. LV ejection fraction was calculated using Simpson’s biplane method. LA volumes were measured using the area-length method from the apical four-chamber view and subsequently indexed to body surface area. Doppler tissue imaging–derived systolic and diastolic velocities (S′, E′, and A′) were measured from the lateral and septal edge of the mitral annulus, and the mean E/E′ ratio was calculated. A comprehensive evaluation of LV diastolic function was performed as recommended in the latest American Society of Echocardiography guidelines, taking into account mitral inflow Doppler measurements, tricuspid regurgitation velocity, Doppler tissue imaging indexes, and LA volume index. Right ventricular function was estimated using tricuspid annular plane systolic excursion and the maximal lateral tricuspid annular velocity measured by Doppler tissue imaging. Systolic pulmonary arterial pressure was determined from the tricuspid regurgitation jet velocity using a modified Bernoulli equation, and this value was added to an estimate of right atrial pressure by means of the diameter and collapsibility of the inferior vena cava (3 mm Hg if the inferior vena cava diameter was ≤21 mm, collapsing >50% with inspiration; 15 mm Hg if it was >21 mm, collapsing <50% with inspiration; and 8 mm Hg in the intermediate situations).

Aortic valve area was evaluated using the continuity equation and indexed to body surface area. Because a quantitative method was not possible in all patients for technical reasons, we performed a multiparametric evaluation of mitral regurgitation (MR) and aortic regurgitation (AR) as recommended, and a semiquantitative classification was attempted: absence of regurgitation was defined as grade 0, and valvular regurgitation was graded as mild (grade 1), mild to moderate (grade 2), moderate (grade 3), or severe (grade 4).

Two-Dimensional STE

For the speckle-tracking echocardiographic analysis of LV and LA function, 2D grayscale images were acquired in the standard apical four-, three-, and two-chamber views at a frame rate of ≥40 frames/sec. Three consecutive heart cycles were recorded during a quiet breath hold. The analysis of recordings was performed offline by an experienced echocardiographer, using dedicated semiautomated acoustic tracking software (QLAB 9; Philips). To calculate LV global longitudinal strain (GLS), three points (two at the opposite edges of the mitral annulus and one at the apex) were positioned at the LV endocardium’s inner border in each of the three apical views, and a region of interest was automatically defined between the endocardial and epicardial borders; after manual adjustment in case of suboptimal tracing, GLS was automatically calculated from strain in the three apical views.

To calculate LA longitudinal strain, the LA endocardial border was semiautomatically traced in a four-chamber view with optimal visualization of the atrium during the cardiac cycle; after marking three points at the endocardial inner border (two at the lateral and septal edges of the mitral annulus and one at the center of the atrial roof), a region of interest was delineated, composed of six segments. QRS onset was used as a reference point; after the segmental tracking quality analysis and the eventual manual adjustment of the region of interest, the software generated longitudinal strain curves for each segment and an average deformation curve. Peak atrial longitudinal strain (PALS), which corresponds to the end of the reservoir phase, was calculated by averaging the peak values observed in all LA segments (global PALS); peak atrial contraction strain (PACS) was obtained from the average deformation curve as the positive peak at the onset of the p wave, corresponding to the end of the conduit phase in late diastole (global PACS). To assess the reproducibility of LA strain, 15 patients were randomly selected, and speckle-tracking analysis was repeated on the same beats 1 month later by the same operator and by a second echocardiographer.

POAF Detection

Continuous telemetry monitoring was available postoperatively for every patient during ICU stay and to the end of the third postoperative day during the subsequent hospital stay. Electrocardiographic telemetry was then continued in patients who had arrhythmic events or life-threatening complications, and it was discontinued in clinically stable patients. For the latter, arrhythmic episodes were then noticed by detection of vital parameters and documented on electrocardiography. POAF was defined as any episode of sustained AF (>30 sec) registered by telemetry and/or 12-lead electrocardiography. Total intervention time, and the main postoperative complications (respiratory complications with necessity of prolonged invasive ventilation, infections, necessity for blood transfusions after the first postoperative day, stroke, advanced atrioventricular block) were also determined according to hospital notes and discharge letters.

Statistical Analysis

The Kolmogorov-Smirnov test was performed to assess if data were normally distributed. Data are expressed as mean ± SD. A P value < .05 was considered to indicate statistical significance. Differences between group means were evaluated using Student’s t tests (continuous variables) or χ 2 analyses (categorical variables). Among the different clinical and echocardiographic data analyzed, univariate logistic regression analyses were performed to detect significant predictors of POAF. Two multivariate logistic regression models were also created including PALS or PACS and other variables with significant association with POAF at univariate analysis. Box plots were elaborated to show the relationship between global PALS and PACS and the occurrence of POAF in the whole study population and in patients with normal and enlarged left atria, respectively. Subgroups analyses were performed to test the accuracy of PALS and PACS in predicting POAF. Subgroups of patients were identified on the basis of clinical and echocardiographic parameters; the probability of interaction with POAF was tested for each of the variables taken into account. Receiver operating characteristic curves were generated to assess the overall performance of PALS and PACS to predict POAF and to detect the cutoff values of the two indexes with the best predictive accuracy for POAF. Bland-Altman analysis was performed to assess inter- and intraobserver agreement regarding measures of LA strain, and intraclass correlation coefficients were calculated. Coefficients of variation were calculated using the logarithmic method. Analyses were performed using StatView release 5.0 (SAS Institute, Cary, NC) and SPSS release 17.0 (SPSS, Chicago, IL).


Seventy-one patients with severe isolated AS were evaluated by our echocardiography team during the enrollment period. Sixty patients (84%) fulfilled the eligibility criteria and formed the study population. Among the other patients, three refused participation in the study, three had severe chronic obstructive pulmonary disease, and five had permanent or persistent AF. The majority of the patients enrolled received bioprostheses (58 of 60 [97%]), while two of them were implanted with mechanical valves. POAF was then documented in 26 of 60 patients (43.3%); Table 1 shows the correlation between demographic and clinical parameters and POAF. Age was slightly higher in the POAF group (73.6 ± 7.6 vs 69.3 ± 8.1 years, P = .04); body mass index (BMI) was slightly lower in the POAF group, and this difference showed weak statistical significance (28.8 ± 4.8 vs 26.8 ± 22.8 kg/m 2 , P = .04 by t test, P = .05 by logistic regression). No significant differences resulted from the analysis of other variables, including arterial pressure, preoperative symptoms, and risk factors, history of paroxysmal AF, medications, and biochemical data. Five of 60 patients (5%) had first-degree atrioventricular block at the time of enrollment; no significant differences in P-R interval duration were noted between patients without and those with POAF (mean P-R interval, 161 ± 37 vs 165 ± 18 msec; P = .6). Similarly, there were no significant differences in P-wave duration (99 ± 6 and 92 ± 10 msec in the non-POAF and POAF groups, respectively, P = .1). No significant correlation was found between total intervention time and POAF or between serum potassium pre- and postoperative levels and the occurrence of arrhythmia. POAF occurred on the first postoperative day in the majority of cases (61% of the total). Arrhythmia appeared on the second postoperative day in seven patients (27% of cases), on the third postoperative day in two patients, and in one patient on the fifth postoperative day.

Table 1

Clinical variables in patients with or without POAF

Variable Non-POAF group ( n = 34) POAF group ( n = 26) P Odds ratio 95% CI P
Clinical data
Age (y) 69.3 ± 8.1 73.6 ± 7.6 .04 1.07 1.00–1.16 .04
Men 50.0 50.0 .99 1.00 0.37–2.77 .99
BSA (m 2 ) 1.9 ± 0.2 1.8 ± 0.2 .20 0.18 0.01–2.99 .20
BMI (kg/m 2 ) 28.8 ± 4.8 26.8 ± 2.8 .04 0.85 0.72–0.99 .05
Systolic AP (mm Hg) 135 ± 13 133 ± 16 .70 1.00 0.95–1.03 .70
Heart rate (beats/min) 70.4 ± 9.3 68.9 ± 11.7 .60 0.99 0.94–1.04 .60
NYHA class III or IV 35.3 37.5 .90 1.1 0.37–326 .90
Angina 24.2 16.0 .40 0.59 0.16–2.26 .40
Syncope 18.2 20.0 .90 1.12 0.30–4.21 .90
Previous paroxysmal atrial fibrillation 6.0 11.5 .40 2.08 0.32–13.51 .40
Hypertension 78.8 87.9 .40 1.79 0.41–7.82 .40
Diabetes 18.2 21.7 .80 1.25 0.33–4.72 .70
Smoking 45.4 21.7 .10 0.33 0.10–1.11 .10
Biochemical data
Total cholesterol (mg/dL) 181 ± 45 185 ± 37 .70 1.00 0.99–1.01 .70
HDL (mg/dL) 54 ± 13 59 ± 16 .20 1.03 0.98–1.07 .20
LDL (mg/dL) 101 ± 42 107 ± 39 .60 1.00 0.99–1.02 .60
TG (mg/dL) 127 ± 74 105 ± 49 .20 0.99 0.98–1.01 .20
Glucose (mg/dL) 105 ± 28 107 ± 30 .80 1.00 0.98–1.02 .80
Albumin (mg/dL) 39.2 ± 3.1 38.3 ± 2.5 .30 0.90 0.74–1.08 .20
Creatinine (μmol/L) 85.2 ± 33.6 85.4 ± 29.6 .90 1.00 0.98–1.02 .90
Hemoglobin (g/dL) 13.4 ± 1.5 13.9 ± 1.7 .20 1.25 0.87–1.80 .20
Preoperative K + (mEq/L) 3.5 ± 0.5 3.6 ± 0.6 .60 1.60 0.46–5.62 .50
Postoperative K + (mEq/L) 3.8 ± 0.8 3.6 ± 0.7 .40 1.15 0.51–2.62 .70
Preoperative medications
ACE inhibitors/ARBs 53.1 50.0 .80 0.88 0.29–2.62 .80
Statins 50.0 40.9 .50 0.69 0.23–2.07 .50
Oral antidiabetics 12.1 18.2 .50 1.61 0.35–7.26 .50
Insulin 6 4.5 .80 0.74 0.06–8.67 .70
Intervention time (min) 213 ± 41 210 ± 61 .80 0.99 0.98–1.01 .90

ACE , Angiotensin-converting enzyme; AP , arterial pressure; ARB , angiotensin receptor blocker; BSA , body surface area; HDL , high-density lipoprotein; LDL , low-density lipoprotein; NYHA , New York Heart Association; TG , triglyceride.

Data are expressed as mean ± SD or as percentages.

Student’s t test or χ 2 test.

Logistic regression.

Drug treatment with amiodarone was given in almost all cases to treat POAF (88%); spontaneous cardioversion to sinus rhythm was observed in three patients. Notably, none of the patients received amiodarone or antiarrhythmic drugs preoperatively. Restoration of sinus rhythm occurred in <24 hours in the majority of patients (16 of 26 [61%]), and only three of 26 (11%) were finally discharged in AF. Pharmacologic or spontaneous cardioversion was achieved between 24 and 48 hours after surgery in three patients (11%) and between 48 and 72 hours in two other patients (8%). One patient had frequent episodes over the whole hospital stay; another, who had septic shock in the postoperative course, had persistent AF for 20 days and was successfully treated with electrical cardioversion.

Complications were noted in nine patients (35%) in the POAF group and in six patients (18%) in the non-POAF group ( P = .10). The following postoperative main complications occurred: two strokes (both in the POAF group), three cases of advanced atrioventricular block (all in the non-POAF group) requiring permanent pacemaker implantation, three cases (two in the POAF group, one in the non-POAF group) of persistent anemia requiring blood transfusion after the first postoperative day, two cases of severe sepsis (both in the POAF group), and six instances of respiratory complications (four in the POAF group, two in the non-POAF group; infections, pleural effusions) that required prolonged invasive ventilation. No significant difference in the occurrence of any of the complications listed was noted between the two groups, but they tended to be more frequent in the POAF group (except for atrioventricular block). Patients with POAF had a significantly longer median ICU stay (1.3 ± 0.7 vs 2.5 ± 3.0 days, P = .02). The difference in the total length of hospital stay was, however, not significant (8.4 ± 3.0 and 9.6 ± 3.3 days for the POAF and non-POAF groups, respectively, P = .2).

Table 2 shows all echocardiographic parameters. Notably, LA strain analysis was feasible in all enrolled patients, and among all the analyzed variables, only global PALS and PACS were significantly associated with POAF ( P < .0001 by t test). Mean atrial volume was larger in patients with POAF, but the difference was not statistically significant; no significant differences were noted among LV systolic and diastolic functional parameters (including LV GLS), aortic valve area, or mean gradients. Classifying diastolic function, we found that three of 60 patients (5%) had no (grade 0) diastolic dysfunction, 25 of 60 patients (42%) had grade I dysfunction, 28 of 60 (47%) had grade II dysfunction, and two of 60 (3%) had grade III dysfunction; diastolic function was indeterminate in two patients. No significant association was noted between the grades of diastolic dysfunction and the occurrence of POAF ( P = .09 by logistic regression). There were no significant associations between the occurrence of POAF and data relative to AR and to other valve diseases. In particular, 17 of 60 patients (28%) had grade 2 (mild to moderate) and two of 60 (3%) had grade 3 (moderate) AR, while the majority of patients (41 of 60 [68%]) had no or mild AR.

Table 2

Echocardiographic variables in patients with or without POAF

Variable Non-POAF group ( n = 34) POAF group ( n = 26) P Odds ratio 95% CI P
2D M-mode parameters
LVEDV (mL) 123 ± 33 133 ± 26 .20 1.01 0.99–1.03 .20
LVEF (%) 60 ± 9 58 ± 10 .60 0.98 0.93–1.04 .60
LV mass index (g/m 2 ) 149 ± 42 161 ± 45 .30 1.00 0.99–1.02 .30
LA volume 71.2 ± 20.3 78.6 ± 28.3 .20 1.01 0.99–1.04 .20
LA volume index (mL/m 2 ) 37.2 ± 9.8 42.5 ± 14.3 .10 1.04 0.99–1.09 .10
TAPSE (mm) 22.9 ± 5.2 24.4 ± 4.4 .30 1.07 0.94–1.21 .30
Bicuspid AV 17.6 3.8 .09 0.19 0.02–1.66 .10
Stroke volume index (mL/m 2 ) 40.7 ± 7.6 44.6 ± 11.4 .20 1.05 0.99–1.11 .10
AV peak gradient (mm Hg) 82.5 ± 19.9 79.1 ± 19.7 .50 0.99 0.96–1.02 .50
AV mean gradient (mm Hg) 52.2 ± 14.0 49.4 ± 12.8 .40 0.98 0.95–1.02 .40
AVA (cm 2 ) 0.72 ± 0.18 0.75 ± 0.20 .60 1.9 0.12–31.4 .60
AVA index (cm 2 /m 2 ) 0.38 ± 0.08 0.41 ± 0.11 .20 32.2 0.13–79.8 .20
MR (0–4) 0.8 ± 0.5 1.0 ± 0.5 .20 1.83 0.68–4.98 .20
AR (0–4) 1.1 ± 0.8 1.2 ± 0.9 .60 1.17 0.62–2.21 .60
E (cm/sec) 85 ± 30 84 ± 22 .90 1.00 0.98–1.02 .90
E/A ratio 0.8 ± 0.4 1.0 ± 0.7 .20 1.89 0.60–5.69 .20
DT (msec) 262 ± 57 248 ± 73 .40 0.99 0.98–1.01 .40
sPAP (mm Hg) 31.1 ± 7.8 32.9 ± 9.1 .50 1.03 0.95–1.11 .50
Pulmonary vein S/D peak velocity ratio 1.3 ± 0.4 1.00 ± 0.5 .10 0.16 0.01–1.97 .10
Tissue Doppler
Mean S, mitral annulus (cm/sec) 6.3 ± 1.3 6.4 ± 1.3 .80 1.06 0.74–1.52 .70
Mean E, mitral annulus (cm/sec) 5.8 ± 2.0 6.3 ± 1.7 .40 1.11 0.84–1.47 .50
Mean A, mitral annulus (cm/sec) 8.5 ± 1.7 7.6 ± 2.4 .10 0.83 0.64–1.08 .20
Mean E/E′ ratio 14.8 ± 4.6 14.0 ± 5.5 .20 0.96 0.86–1.06 .40
S, right ventricle (cm/sec) 12.2 ± 2.5 11.8 ± 2.8 .70 0.99 0.81–1.20 .90
E, right ventricle (cm/sec) 9.4 ± 2.8 10.0 ± 2.7 .40 1.07 0.89–1.30 .50
A, right ventricle (cm/sec) 14.2 ± 2.9 15.7 ± 4.0 .20 1.09 0.94–1.27 .30
LV GLS (%) −14.8 ± 2.8 −14.6 ± 3.4 .80 0.97 0.88–1.14 .70
LA global PALS (%) 27.1 ± 6.7 18.1 ± 5.3 <.0001 0.75 0.64–0.87 .0002
LA global PACS (%) 14.3 ± 4.7 8.4 ± 3.1 <.0001 0.70 0.58–0.84 .0001

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Apr 15, 2018 | Posted by in CARDIOLOGY | Comments Off on Atrial Function as an Independent Predictor of Postoperative Atrial Fibrillation in Patients Undergoing Aortic Valve Surgery for Severe Aortic Stenosis

Full access? Get Clinical Tree

Get Clinical Tree app for offline access