Predicting new onset atrial fibrillation post acute myocardial infarction: Echocardiographic assessment of left atrial size

Atrial fibrillation (AF) commonly occurs following acute myocardial infarction (AMI). Left atrial (LA) size has been reported to predict new onset AF in this cohort, however, the optimal metric of left atrial size for risk stratification following AMI is unknown.

diastolic dysfunction, and valvular dysfunction. [4][5][6] This milieu can lead to an interdependent cycle of AF, myocardial ischemia and heart failure. 7 AF can complicate the post-AMI period by driving supply/demand ischemia, embolic events, and heart failure. AF also impacts AMI pharmacotherapy, often necessitating the addition of anticoagulation to antiplatelet regimens, with attendant increased bleeding risks. 8 Multiple studies have associated AF with worse short-and long-term outcomes following AMI. 1,9,10 However, an independent relationship has not universally been demonstrated. 11 Left atrial (LA) size is a well-recognized predictor of incident AF in a wide range of cohorts, 12 including following AMI. 7,13,14 Multiple metrics of echocardiographic quantification of LA size have been described, including LA area and LA volume. Contemporary guidelines currently recommend the use of maximal LA volume (LAVmax), with indexation to body surface area (LAVImax). 15 However, increasing evidence suggests that indexed minimal left atrial volume (LAVImin) is a superior correlate of left ventricular filling pressure, and has a greater predictive ability for mortality post AMI over LAVImax. 16 We sought to evaluate the incidence of new onset AF in a cohort of patients with incident AMI, and investigate the predictive utility of LA size metrics and diastolic dysfunction for new onset AF.

Study design
This study utilized the AMI cohort established by Prasad et al., consist-ing of consecutive patients with first-ever AMI (including STEMI and NSTEMI) undergoing invasive coronary angiography with concurrent transthoracic echocardiography at a tertiary referral centre between January 2013 and December 2014. 16,17 All patients received guideline directed medical therapy for AMI, unless contraindicated.
Exclusion criteria included previous AMI, indeterminate diastolic function or confounders of diastology (e.g., acute AF, paced rhythm, and significant mitral valve disease), significant hemodynamic instability (shock, acute pulmonary oedema, requirement for mechanical ventilation, intra-aortic balloon pump, or ventricular tachycardia) or insufficient image quality. 16 Previously known diagnosis of AF at presentation was incorporated as an additional exclusion criterion for the present study.

Echocardiographic analysis
Comprehensive echocardiography was typically performed within

Atrial fibrillation data collection
The presence of current or previous AF episodes was determined from review of electronic medical records. Subsequent AF episodes post index AMI presentation were identified by reviewing the statewide electronic medical records system for recorded AF diagnoses at all available clinical encounters, including emergency presentations, hospital admission, or outpatient clinic review.

Statistical analysis
The IBM SPSS Statistics package (v26) was used for all statistical analyses. Univariate analyses were performed using Pearson's chi squared analysis for categorical data and Student's T test for continuous data. Multivariable analysis with binary logistic regression was used to identify independent predictors of new onset AF post AMI. Three multivariable models were created to assess the predictive ability of echocardiographic metrics of LA size (LA area, LAVImax, LAVImin). STEMI/NSTEMI presentation and revascularization with PCI were excluded from multivariate analysis due to collinearity with coronary artery bypass grafting (CABG). Right atrial area was also excluded due to collinearity with LA size.

Ethics
Ethical approval for the study was received from the institutional human research ethics committee. Note: Data are expressed as n (%) for binary data and mean ± standard deviation for continuous data. Abbreviations: CABG, coronary artery bypass grafting; IHD, ischemic heart disease; IVS, interventricular septum; LVEDVI, indexed left ventricular end diastolic volume; NSTEMI, non-ST elevation myocardial infarction; PCI, percutaneous coronary intervention; RVSP, right ventricular systolic pressure; STEMI, ST elevation myocardial infarction.

Study cohort demographics and AF diagnoses
There were 718 patients screened for inclusion in the parent study, with 433 in the current study cohort after applying exclusion criteria. The mean age was 61.1 years, and 74.6% were male (see Table 1).
Thirty one (7.2%) were found to have new AF diagnoses post AMI at a median follow up of 3.8 years (IQR 3.3-4.2 years). The median time to AF diagnosis was 28 days (IQR 0-287 days).  In the three alternate multivariable models for the prediction of new onset AF post AMI (see Table 2), the only metric of LA size found to be independently predictive for new onset AF in its model was LAV-Imin (OR 1.096, p < .001). Age (p = .009) and CABG (p < .001) were also independently associated with new onset AF in this model. Neither LA area nor LAVImax was independently predictive of new onset AF in their models (p > .05). In the LAVImax-based model, age (p = .001) and CABG (p = .006) were independent predictors of new onset AF. In the LA area-based model, only age was an independent predictor (p = .001).

Echocardiographic risk factors for new onset AF
When the multivariable models were repeated in the subgroup of 376 patients who did not undergo CABG, LAVImin remained the only LA size metric independently associated with subsequent AF diagnosis (p = .001). LA area and LAVImax were not significantly associated in their respective models (p > .05). Age remained independently predictive in all three models.

Summary of key findings
LAVImin was an independent predictor of new onset AF post AMI in the present study. Furthermore, LAVImin outperformed echocardiographic assessment of diastolic function and alternative metrics of left atrial size for the prediction of new onset AF post AMI, including the contemporary standard of LAVImax.

Left atrial size and atrial fibrillation
LA size reflects the chronicity and severity of LV diastolic dysfunction, and is associated with elevated LV filling pressures. 18 Atrial stretch can facilitate initiation and maintenance of AF, and LA enlargement often reflects structural and electrophysiological atrial remodeling. 19,20 As such, it is unsurprising that LA size predicts the onset of AF in the general population, 21 as well as its recurrence post cardioversion 22

Left atrial volume: LAVImax versus LAVImin
Biplane assessment of LA volume provides a comprehensive assessment of LA size, has been shown to outperform LA diameter in predicting cardiac outcomes, 26,27 and is a powerful prognostic marker across a wide range of cardiac cohorts, 24 including prediction of incident AF in the general population, 26 and survival post AMI. 28 While LA volume varies significantly throughout the cardiac cycle, it is conventionally assessed by maximal LA volume (LAVmax) measured at end ventricular systole with indexation to body surface area (LAVImax), in keeping with ASE/EACVI guidelines. 24,29 In post-AMI cohorts, some studies have suggested that maximal LA volume predicts new onset AF, 13,14 although this finding is not universal, 30 particularly on multivariable analysis. 31 These inconsistencies may be in part attributable to differences in study cohorts, imaging protocols, and covariates recorded and corrected for in multivariable models. However, some authors have also highlighted the limitations of LAVmax as a metric of LA size, in that LA volumes at end-ventricular systole are dependent upon the degree of atrial stretch mediated by LV longitudinal systolic function, and therefore may be less reflective of the relationship between LA size and LV diastolic function. 32 In contrast, recent evidence suggests that LAVmin is a superior correlate of LV filling pressure than LAVmax, 33

Other echocardiographic risk factors for atrial fibrillation post myocardial infarction
Beyond LA size, previous studies have reported other echocardiographic predictors of new onset AF post AMI to include LVEF, mitral regurgitation (MR) and Doppler metrics of diastolic dysfunction. 7,40,45 These parameters are hypothesized to drive AF through atrial pressure loading, with subsequent atrial remodeling in the longer term. 25 The impact of MR on AF risk could not be assessed in our study, as significant MR confounds assessment of diastolic function, and as such these patients were excluded from the parent cohort. While both diastolic dysfunction and reduced LVEF suggested a trend toward increased AF risk in the present study, neither was statistically significant on univariate analyses. Furthermore, the present cohort displayed relatively preserved LV systolic function (mean LVEF > 50%), which is likely attributable in part to the recruitment of incident AMIs only, rather than patients with recurrent ischemia and/or chronic ischemic cardiomyopathy. Furthermore, a degree of LV dysfunction seen on early echocardiographic assessment may represent reversible acute myocardial stunning, which may have a less significant impact on long term AF risk.

Clinical risk factors for atrial fibrillation post myocardial infarction
Previous studies have suggested that clinical and biochemical risk factors for new onset AF post AMI include age, female gender, obesity and revascularization with CABG. 14,39,41,44,[46][47][48][49] We similarly found an independent association with both age and CABG, which supports a model of new onset AF post AMI as outlined in Figure 2. A lower incidence of AF was seen in the STEMI cohort and patients undergoing revascularization via PCI. This is likely due in part to an inverse relationship with revascularization via CABG, as patients with STEMI typically undergo emergent PCI. Furthermore, NSTEMI cohorts often have greater burden of cardiovascular comorbidities, 50 which may independently drive AF risk.

Limitations and considerations
While inter-operator variability in echocardiographic measurements may have been present, all cardiac sonographers followed identical echocardiography protocols, and a previously published Bland-Altman analysis of measurements from two separate sonographers on 20 randomly selected cases from the dataset suggested satisfactory intraand inter-observer reproducibility for LAVImax and LAVImin. 16 Timing of echocardiography relative to revascularization differed between AMI types, with NSTEMIs frequently undergoing echocardiography prior to angiography, while patients with STEMI typi-cally underwent emergent revascularization at presentation with an echocardiogram performed later in the admission.
Although the present study focused on echocardiographic assessment of LA size, the role of LA function and LA strain in risk stratification would certainly be of interest for future studies.

CONCLUSIONS
LAVImin is an independent predictor of new onset AF post AMI, and appears to outperform both grade of diastolic dysfunction, and alternative metrics of LA size (LA area and LAVImax) for risk stratification. Further studies are warranted to more clearly define the echocardiographic predictors of AF post AMI, and evaluate whether LAVImin holds similar advantages over LAVImax in other cohorts.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
The data underlying this study are available upon reasonable request to the authors.