*3.7. Risk of Bias Assessment*

Based on quality assessment of diagnostic accuracy studies questionnaire (QUADAS-2), four domains of criteria for risk of bias and three for applicability were analyzed, and the risk of bias was assessed as "low risk", "high risk", or "unclear risk" (Table S1: Supplement 1) [10]. Most studies had a low or moderate risk of bias and clearly defined their objectives and the main outcomes (Table S3, Figure S6, Supplement 8, 9). The QUADAS-2 analysis for bias evaluation showed all domains to have had low risk of bias (≤40%). Additionally, there was no evidence for publication bias, as evaluated by Egger's test for our findings.

#### **4. Discussion**

Despite its failure in about one-third of treated patients, cardiac resynchronization therapy (CRT) still remains the best treatment for symptomatic heart failure patients on full medical therapy, as stated in the European and American guidelines [23,24]. One of the known causes of symptoms in such patients is atrial arrhythmia, which is known to be related to LA cavity enlargement and disturbed function, both of which may improve in CRT responders [16,18]. The regression of atrial arrhythmia with CRT treatment has been reported and interpreted on the basis of reversed LA cavity remodeling [25,26]. Despite these suggestions, the exact contribution of LA function in cardiac reverse remodeling related to CRT remains poorly established [27]. This meta-analysis evaluated the relationship between LA function in patients who received CRT for heart failure.

*Findings:* Our analysis shows that CRT responders had no baseline difference in LA strain compared CRT non-responders, but the LA EF was higher in responders. At follow-up, both LA strain and the LA EF only significantly increased in responders. The increase in LA strain was associated with a fall in LVESV and a rise in the LVEF. Similarly, although with a less significance, the increase in the LA EF correlated with the fall in LVESV and the increase in the LVEF. Finally, the increase in LA strain correlated with the increase in the LA EF.

*Data Interpretation:* LA function is an integral part of cardiac function, and its pump function normally contributes by at least one third to overall LV filling, which increases with age [28]. In heart failure, responders to CRT are mainly those with worse LV dyssynchrony, which itself compromises LA emptying and consequently stroke volume. Additionally, reduced LA emptying results in raised LA pressure and, consequently, myocardial stretch, which then leads to cavity function instability and arrhythmia. Studies have shown that an increase in LA volume is the most accurate predictor of atrial arrhythmia [29,30].

With an optimum response to CRT, LV systolic function improves and the ejection fraction increases, and ESV falls as stroke volume increases. These changes have been shown to have significant hemodynamic effects on overall cardiac performance and symptoms [30]. Furthermore, an improved LV pump function results in better LA emptying as a further contribution to stroke volume. The other side of the benefit from CRT lies at the LA myocardial function level; as has been shown by our results, LA strain increases parallel to the increased cavity emptying fraction, thus providing further evidence for the improvement in LA integral function in the form of myocardial intrinsic function and overall pump performance. It is of interest that these aspects of improvements of LA function did not happen independently of the LV, but they were associated, although modestly, with the fall in LVESV and the increase in ejection fraction [31–33]. Such a relationship is of clinical and academic interest, since the described fall in LVESV, usually referred to as a sign of LV reverse remodeling, seems to happen also in a similar way in the left atrium, with an increase in emptying fraction [20,34]. Thus, although the term reverse remodeling might sound non-specific, our results shed light on some of its ingredients in the setting of LV and LA structure and function changes in response to CRT [15–21]. Finally, our results highlight the fact that the LA is not only a conduit chamber but also a more complex anatomical and function structure, both in and of itself and in its relationship with the LV [35,36]. Our recently published meta-analysis [5] showed a concordant relationship between LA indexed volume and LV volume and function with CRT. The documented improvement of LA myocardial intrinsic function and emptying function is expected to be associated with a fall in cavity pressure with its direct implications on the frequency of atrial arrhythmias known in patients with significant heart failure [11].

*Limitations:* The analysis of LA function and LVESV and/or LVES was based on a small number of studies, so its results should be seen as having modest accuracy until proven in a larger number of studies. The data included in the meta-analysis were collected from the published papers, on whose quality we did not have control; we had to trust the academic merit of the investigators. We were unable to comment on the relevance of our findings in controlling atrial arrhythmia in the analyzed studies because of the limited available data. Likewise, we had hoped to provide evidence for long term benefits from CRT, but, again, such information was not available in the analyzed studies. Finally, we sought to assess the relationship between the individual and combined LA and LV function changes

with CRT against symptoms in more detail, but the available data on LA function parameters that could be used in such analysis were very limited. It would have been of great interest to analyze subgroups of patients according to the concordant/discordant relationships between improvements of LA and LV functions, but, again, such data were not uniformly available in the small studies we analyzed.

*Clinical Implications:* The left atrium is an integral component of the overall cardiac structure and function, and it should be seen more than just a conduit. Based on the anatomical myocardial fiber architecture, the association between the left atrial and left ventricular function changes further strengthen such relationship, particularly in the setting of HF with a reduced EF and increased diastolic pressures. Our findings may assist in explaining the well documented lack of symptomatic improvement with CRT in patients with atrial fibrillation, since significant components of the mechanisms of LA emptying and myocardial contraction that contribute to the overall cavity strain does not exist [37].
