Left atrial function in elite athletes: A meta‐analysis of two‐dimensional speckle tracking echocardiographic studies

We sought to investigate left atrial (LA) volume, function, and strain in elite athletes by a meta‐analysis including echocardiographic studies that provided volumetric and strain analysis of LA phasic function.


| INTRODUCTION
Elite athletes are exposed to intensive training that is associated with hemodynamic adaptation and cardiac remodeling. Most of studies have been focused on left ventricular (LV) changes induced by training. However, an important part of cardiac adjustment to increased cardiac output during effort is played by left atrial (LA) remodeling.
There is a consensus regarding LA dilatation in athletes and recent large meta-analysis confirmed that LA linear dimensions, as well as LA volume, was significantly enlarged in athletes. 1,2 It is more complicated when debate comes to the point of LA function. Namely, LA function determines diastole, which has the key role in athletes. 3 Nevertheless, the assessment of LA function is not an easy task because it does not represent evaluation of only one parameter, but the whole set of parameters (total, passive, and active emptying fractions) that resemble LA reservoir, conduit, and active function, respectively. 4 LA reservoir function shows the ability of the LA to store pulmonary venous return during left ventricular contraction and isovolumetric relaxation. 4 Conduit LA function represent the ability to passively transfer blood into the LV. The LA active function shows the LA contraction during the last diastolic phase that enables up to 30% of LV stroke volume in athletes. 5 There is lack of agreement regarding the importance of each of these functions in the athletes. In some of the studies, these LA functions were decreased, 6,7 whereas in some others there was no significant difference between athletes and controls. 2 The introduction of strain in echocardiography significantly changed our perspective and improved our knowledge about cardiac mechanics and provided better and more detailed insight into LA mechanics in athletes. 8,9 The strain evaluation provided even more information about each phase of cardiac diastole and therefore LA functions. However, the results obtained by strain-and volume-derived analysis of LA phase function were not always concurrent.
The importance of LA remodeling in athletes is relevant because it is related with overall and cardiovascular mortality in global population, 10,11 and also because of significantly increased risk of atrial fibrillation. 12,13 Considering the fact that regular physical activity is considered as beneficial, it is important to make distinction between this kind of physical activity and extensive effort in elite athletes and to emphasize the fact that elite athletes are under the increased risk of arrhythmias and particularly atrial fibrillation during middle-age. 14 LA remodeling has central role in the later.
This systematic review and meta-analysis provided comprehensive information on the LA strain in the athletes and summarized current information on this interesting topic.

| METHODS
The present study was performed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. 15 Medical literature was reviewed to identify all articles investigating LA size and function as assessed by echocardiography in athletes. The OVID-MEDLINE, PubMed, and Cochrane CENTRAL databases were searched for English-language articles without time restriction up to February 2018 through focused, high-sensitive search strategies.
Studies were identified by crossing the following search terms: "athletes," "left atrial size," "left atrial volume," "atrial function," "atrial strain," "atrial strain rate," "echocardiography," "2D speckle echocardiography." References from relevant studies were screened for supplementary articles. Any observational (either cross-sectional or longitudinal) study comparing LA size and function, as assessed by standard and 2D speckle echocardiography, in athletes and in sedentary controls was included in the meta-analysis, without any restriction about the types of physical training Inclusion criteria were: (a) full articles published in English in peerreviewed journals; (b) studies reporting data on LA strain (SD, SE or confidence interval) by echocardiography in elite/professional athletes; (c) minimum set of data including age, gender, body surface area (BSA), or body mass index (BMI).
Titles and abstracts were screened independently by two authors (CC, EG) who discarded irrelevant studies to the topic. Case reports, reviews, editorials, letters were excluded from qualitative analyses, but screened for potential additional references. Two authors (CC and EG) independently assessed retrieved abstracts and full text of these studies in order to determine eligibility according to inclusion criteria.
A third reviewer (CS) solved disagreements on study judgments. Data extraction were performed by one reviewer (CC) and independently verified by another reviewer (CS).
The first literature search identified a total of 159 papers. After the initial screening of titles and abstracts, studies were excluded and were reviewed; of these, nine studies fulfilled the inclusion criteria and contained sufficient details to be included in the final review 6,7,16-22 ( Figure S1, Supporting Information).

| Statistical analysis
The primary aim of the meta-analysis was to compare LA size (as assessed by LAV indexed to body surface area) and function, as assessed by 2D speckle tracking echocardiography, both expressed as continuous variables, in competitive high-trained athletes compared to sedentary controls.
To this purpose, a pooled analysis of the above mentioned continuous variables was performed using fixed or random effects models by Comprehensive Meta-Analysis Version 2, Biostat, Englewood, New Jersey. Standard means difference (SMD) with 95% confidence interval (CI) was calculated to evaluate the statistical difference of variables between athletes and controls. The limit of statistical significance was set at P < 0.05. Demographic and clinical data provided by selected studies are expressed as absolute numbers, percentage, mean ± SD (SD), mean ± SE (SE). Heterogeneity was estimated by using I-square, Q, and tau-square values; random effect model was applied when heterogeneity across studies was high (I 2 > 75). Publica-

| Clinical characteristics of athletes
All cases were highly trained competitive professional athletes from 10 different sports. In order to provide a detailed information on the physical exercise characteristics of the sports included in this analysis each athlete sample was allocated to the following groups based on Mitchell's classification: A1 (low dynamic, low static); A2 (low dynamic, moderate static); A3 (low dynamic, high static); B1 (moderate dynamic, low static); B2 (moderate dynamic, moderate static); B3 (moderate dynamic, high static); C1 (high dynamic, low static); C2 (high dynamic, moderate static); C3 (high dynamic, high static). 23 Table 1

| Clinical characteristics of controls
All subjects belonging to the control group had never been involved in competition or recreational sports and were not engaged in any regular training program. In non-trained individuals (68.3% men) mean age ranged from 23.0 ± 6.0 20 to 45.0 ± 5.0 years 17 (pooled mean 29.0 ± 1.8 years) ( Table 2). Average BSA ranged from 1.63 ± 0.10 m 222 to 2.0 ± 0.30 m 216 (pooled mean 1.83 ± 0.03 m 2 ). Average office systolic BP ranged from 110 ± 13 mm Hg 22 to 129 ± 18 mm Hg, 7 diastolic BP from 71 ± 9 mm Hg 22 to 81 ± 14 mm Hg 7 (pooled mean values being 119 ± 1.4 mm Hg and 76.1 ± 1.0 mm Hg, respectively). Average LVMI varied from 63 ± 22 g/m 220 to 86 ± 10 g/m 219 (pooled mean 75 ± 5.0 g/m 2 ).  Table S1 summarizes technical data regarding the brand and model of ultrasound system, probe frequency, echocardiographic protocols used to measure LAV and software used to measure LA myocardial deformation. Gating for LA strain assessment was done in two possible ways: the first and larger group of authors used a R-R gaiting, 6,7,16,18,20 whereas the second group of investigators used P gaiting. 17 (Table S1). Atrial myocardial deformation was measured offline from 2D echocardiographic images using commercial dedicated software with frame rates between 40 and 90 frames per second. Table S1 summarizes technical data regarding the brand and model of ultrasound system, probe frequency, echocardiographic protocols used to measure LAV, and softwares used to measure LA myocardial deformation.

| Left atrial volume and strain in athletes and controls
In the present meta-analysis, the following 2D speckle echocardiographic parameters were considered: global peak atrial longitudinal strain (PALS), global peak atrial contraction strain (PACS), LA peak longitudinal strain rate during ventricular systole (LASR syst ), LA peak longitudinal strain rate during late diastole (LASR late ). Figure 1 shows the assessment of LA strain and strain rate in four-and twochamber view.
Pooled average global PALS was 37% ± 1.2% (CI: 34.6%-39.5%) in athletes and 38.5% ± 1.5% (CI: 35.3%-41.2%) in controls. Figure 2 reports Most studies documented an increase of LA dimension in athletes. 6 The CHILD study failed to observe any difference in LA reservoir and contractile function determined by volumetric and strain parameters between pre-adolescent athletes and controls at baseline. 29 The same authors, however, documented a reduced reservoir and contractile LA function after 5 months of intensive supervised training. 29 A reduced contractile, but not reservoir function of LA has been also reported. 16 Unchanged LA volume-and strain-derived functions during lifetime training, in front of gradual increment in LA conduit and reservoir vol- Our meta-analysis showed that LA global longitudinal strain tended to be reduced in athletes, a finding which is compatible with a reduced reservoir function, whereas no differences in LA contractile function estimated by strain method were found between athletes and controls. It should be pointed out, however, that LA late diastolic strain rate, a marker of LA contractile function, was lower in athletes.
These differences are probably related to the small sample size and statistical power of studies investigating elite athletes. Furthermore, at least three different parameters may account for LA function according to strain method: longitudinal strain, strain rate, and velocity. In healthy volunteers, these parameters of LA mechanical function have been reported to be unrelated with echocardiographic parameters of LV diastolic function, LA volume and function. 31 To these differences may be related our results on LA atrial contraction strain and late diastolic strain rate in athletes.
Our meta-analysis revealed that athletes had reduced global LA longitudinal strain, which resembles LA reservoir function. The authors who reported these findings did not provide explanations for their results. 19,22 The reduced LA reservoir function is strong predictor of incident cardiovascular events, atrial fibrillation occurrence and recurrence, and worse outcome in patients with heart failure with preserved ejection fraction. [32][33][34] Our present findings showed that athletes had reduced LA pump function. The reduction of LA pump function is associated with increased risk of atrial fibrillation, 35 which could explain higher incidence of atrial arrhythmias in elite athletes later in life. This is an important clinical implication of our results.
Increased LAVI in our meta-analysis suggests that LA cavity reacts to the increased hemodynamic pre-load in elite athletes to adapt the elevated venous return and maintain normal contractile function, efficient emptying function, and stroke volume. However, decreased LA reservoir function together with decreased LA booster pump function in athletes in our meta-analysis should be also discussed in the reflection of methods used for LA phasic function. Even though strain-and volume-derived parameters used for evaluation of LA phasic function should closely correlate, this is not always the case. 36  F I G U R E 3 Forest plot for standard means difference (SMD) global peak atrial contraction strain (PACS) in elite athletes and healthy nonathletic controls; data from seven studies (CI, confidence intervals P = 0.29) Recent study showed that LA pump function, estimated by LA strain, significantly increased immediately after marathon and especially in highly trained athletes. 37 The authors reported significant positive correlation between the increase in LA pump function and maximal oxygen consumption measured 1 week before marathon, 37 which could indicate that the increment in LA pump function is associated with better cardiorespiratory fitness and may significantly improve performance of athletes.
In pathological conditions, such as arterial hypertension LA remodeling considers impairment of LA reservoir and conduit function as a consequence of increased afterload and elevated LV stiffness. 38  Total emptying fraction of LA has been shown to better predict mortality compared to LA volume index in a global population, 10 suggesting that LA reservoir function is a more reliable marker of LA remodeling and LV dysfunction than LA enlargement. In a recent study, LA peak longitudinal strain, as a marker of LA reservoir function, has been shown to be the best predictor of atrial fibrillation in male endurance veteran athletes. 16 Moreover, LA contractile function has been reported to be an independent predictor of life-threatening ventricular arrhythmias, 35  imposes is if echocardiography is accurate enough to provide precise information. The reported inter-and intraobserver variability of LA volumetric and strain parameters were high, 36 which confirms the reliability of these data. There is a significant correlation between LAVI obtained by echocardiography and cardiac magnetic resonance (CMR), which still remains the "gold standard "of cardiac imaging.

| Limitations
Limitations and strengths of our meta-analysis should be underlined.
The cross-sectional nature of all studies included in the meta-analysis does not allow to assess the causal relationship between training and LA remodeling in elite athletes. Available data regarding parameters of volumetric and LA function are scanty, thus, we could not perform a meta-analysis of all LA indices. Different authors used various types of LA strain assessment (R-R and P gating). Predominantly was used R-R-gating, but this might be considered as a limitation. Recently published meta-analysis did not show a significant difference between R-R and P-P gating with regard to the practice of measurement of atrial strain. 39 For this meta-analysis, we used aggregated and not original data, which was additional limitation. However, willingness of authors to share their original study is usually limited and therefore analysis of aggregated data represents common and acceptable method. The intra-and inter-observer variability of LA volumes and strain may be also considered as a limitation, but investigations showed that this kind of variability with modern software was unremarkable. 36 The number of investigators in this field is very limited, which is understandable due to very specific population of elite athletes who are usually referred to small number of medical centers, and therefore several references originated from the same research group. Considering already limited number of included studies, this limitation was not possible to avoid.