NAFLD and AF are two pathologic conditions that are highly prevalent in Western countries and that share multiple cardiometabolic risk factors. Presently, the published research on the association between AF and NAFLD (or liver function tests) is sparse. In a large retrospective cohort study, it has been reported that the prevalence of ALT elevations (i.e. defined as serum ALT >40 U/L), as surrogate markers of NAFLD, among a routine clinical care population with AF was high (i.e. 27.6%), although the incidence of new persistent and significant ALT elevations was uncommon 
. More interestingly, the Framingham Heart Study investigators have recently shown that moderately elevated serum ALT or AST levels (>40 U/L for either marker) were independently associated with an increased incidence of AF over a 8-year follow-up period in a community-based cohort of 3,744 adults, who were free of clinical heart failure at baseline 
To our knowledge, this is the first prospective study to examine the role of NAFLD as detected by ultrasonography (which is a more accurate measure of liver fat than serum transaminase levels) in predicting development of incident AF in patients with type 2 diabetes, who were clinically free from AF at baseline. The major finding of our study was that NAFLD was significantly associated with an increased risk of incident AF during a follow-up period of 10 years. Notably, and more importantly, this association was independent of numerous clinical risk factors for AF.
In accordance with previously published reports, we found that older age, LVH and longer PR interval on ECG (i.e. a measure of left atrial size) were strong predictors of incident AF 
. It is well known that LVH causes LV dysfunction and left atrial enlargement, which may lead to fibrosis and electrical remodelling of the atrium, providing a pathophysiological substrate for subsequent development of AF 
. Recently, the Framingham Heart Study investigators published a clinical risk score for development of AF in 10 years that incorporated the presence of age, sex, BMI, systolic BP, hypertension treatment, longer PR interval and history of heart failure 
. Similarly, the Atherosclerosis Risk in Communities study showed that a 10-year clinical risk score incorporating age, race, smoking, systolic BP, hypertension treatment, electrocardiographic LVH, electrocardiographic left atrial enlargement, diabetes, CHD and heart failure was predictive for development of AF in a multi-ethnic, community-based cohort of individuals 
Although there are few data on cardiac function among patients with NAFLD, preliminary evidence indicates that there is a strong relationship between NAFLD and early LV diastolic dysfunction in both non-diabetic and type 2 diabetic individuals 
. It is likely that LV diastolic dysfunction plays a role in AF pathogenesis either by increasing pressure that can affect stretch receptors in pulmonary veins triggers and other areas of the atria or by inducing direct structural changes in atrial myocardium 
. Interestingly, two large population-based studies have also shown that moderately elevated serum GGT levels, as surrogate markers of NAFLD, are independently associated with an increased risk of incident heart failure 
. Collectively, as reported above, our findings confirm and extend to patients with type 2 diabetes, using liver ultrasound for diagnosing NAFLD, the recent results reported by Sinner et al.
demonstrating that NAFLD (as detected by serum transaminase levels) is an independent predictor of new-onset AF in the adult general population.
The underlying mechanisms responsible for the association between NAFLD and increased risk of incident AF require further study. Speculatively, they could include some of the following. Firstly, the association between NAFLD and incident AF is simply a consequence of the shared risk factors and comorbid conditions. However, it is important to underline that in our study NAFLD was associated with an increased risk of incident AF, independently of age, sex, hypertension, electrocardiographic LVH and other clinical risk factors included in the 10-year Framingham Heart Study-derived AF risk score. The odds ratio was not attenuated after adjustment for these potential confounders, thus suggesting that the increased risk of incident AF associated with NAFLD, cannot be fully explained by these shared AF risk factors. Again, the increased risk of AF associated with NAFLD also remained, even after excluding participants with a documented history of previous CHD and heart failure. Secondly, it could be postulated that NAFLD is a marker of ectopic fat accumulation in other tissues, including both the myocardium and pericardium. Rijzewijk et al
and Ng et al
showed that the intra-myocardial fat content, as detected by proton magnetic resonance spectroscopy, was greater in patients with type 2 diabetes than in nondiabetic controls, and was associated with LV diastolic dysfunction. Interestingly, in the study by Rijzewijk et al
there was also a significant, positive association between intra-myocardial and intra-hepatic fat content. Recently, it has been also reported that increased pericardial fat volume was associated with both increased left atrial dimensions 
and increased prevalence of AF 
, independently of multiple established risk factors. Moreover, Shin et al
. reported that total and inter-atrial epicardial adipose tissues were larger in AF patients than in matched controls and were independently associated with left atrial remodeling among patients with AF 
. Preliminary experimental evidence suggests that adipocytes from epicardial or retro-sternal adipose tissues could directly modulate the electrophysiological properties and ion currents, causing higher arrhythmogenesis, in isolated rabbit left atrial myocytes 
. Thirdly, because in our study NAFLD was associated with increased AF incidence, independently of multiple potential confounders, it is also possible to speculate that NAFLD is not only associated with the risk of AF as the consequence of the shared risk factors but that NAFLD per se
might partly contribute to the development and persistence of AF. This process might occur through the systemic release of pathogenic mediators from the steatotic and inflamed liver, including C-reactive protein, interleukin-6, tumor necrosis factor-alpha, plasminogen activator inhibitor-1 and other inflammatory cytokines. Importantly, several studies have shown that these pathogenic mediators are remarkably higher in patients with NAFLD than in those without 
, and may play a role in the development and persistence of AF, possibly by inducing structural and/or electrical remodeling of the atria 
. These pathways may represent a novel pathogenic mechanism by which structural changes resulting from chronic inflammation can perpetuate AF. These findings require further testing and confirmation in larger clinical trials. Nevertheless, these pathways might provide a potential target for pharmacological interruption or reversal of atrial structural remodeling 
Our study has some important limitations. First, our cohort comprised of type 2 diabetic patients of European extraction, so that the results cannot be generalized directly to other ethnic groups. Second, there were a relatively small number of clinical events during the follow-up and, therefore, the results should be interpreted with some caution. Third, the diagnosis of NAFLD was based on ultrasonography that is relatively insensitive to the presence of smaller amounts of hepatic steatosis (<33% liver fat infiltration) and that cannot distinguish NASH from other forms of NAFLD (although, that said, the overall sensitivity and specificity of ultrasonography for detecting moderate and severe hepatic steatosis are ~85% and ~95% respectively, when compared to liver biopsy as a gold-standard) 
. Although some non-differential misclassification of NAFLD on the basis of ultrasonography is likely (i.e., some of the control patients with diabetes could have mild hepatic steatosis and undetected NAFLD, despite normal serum liver enzymes and a negative ultrasonography examination); this limitation would serve to attenuate the magnitude of our effect measures towards the null. Thus, we reason that our results can probably be considered a conservative estimate of the relationship between NAFLD and increased AF incidence. Since hepatic ultrasonography was assessed at baseline only, we could not investigate the relationship of changes (development or resolution) in hepatic steatosis over time to incident AF risk. Fourth, the diagnosis of LVH was based on widely accepted ECG criteria (that have a very high specificity but a relatively low sensitivity when compared with echocardiographic findings) 
. Unfortunately, no echocardiographic measurements were available in this study. However, our data have been also adjusted for systolic BP and hypertension treatment, which are likely to capture almost all patients with LVH not detected by classical ECG voltage criteria. In addition, it is important to recognise that the additional incorporation of echocardiographic measurements only slightly improved the predictive ability of the 10-year Framingham Heart Study-derived risk score for the development of AF 
. Finally, we cannot exclude residual confounding.
Notwithstanding these limitations, our study has important strengths, including its prospective design, the long duration of follow-up (10 years), the relatively large number of participants of both sexes, the diagnosis of hepatic steatosis by ultrasonography (which was performed in all patients by a single experienced radiologist), the complete nature of the dataset, and the ability to adjust for baseline AF risk factors included in the 10-year Framingham risk prediction model 
In conclusion, our study is the first to demonstrate that ultrasound-diagnosed NAFLD is closely associated with an increased incidence of AF in patients with type 2 diabetes, independently of important clinical risk factors for AF. Further studies are needed to confirm this finding in other populations, to elucidate the responsible mechanisms for this association, and to explore whether pharmacological interventions aimed at improving NAFLD effectively reduce the incidence of AF in patients with type 2 diabetes. In the interim, from the perspective of clinical practice, it is important that specialists and practicing clinicians be aware of the link between NAFLD and AF, especially because of the high and growing prevalence of these two pathologies.