Atrial Fibrillation Is Highly Prevalent Yet Undertreated in Patients With Biopsy-proven Nonalcoholic Steatohepatitis

Maureen Whitsett; Jane Wilcox; Amy Yang; Lihui Zhao; Mary Rinella; Lisa B. VanWagner


Liver International. 2019;39(5):933-940. 

In This Article


In our large urban tertiary care centre, the prevalence of AF among patients with NASH with mean age of 54 years was 4.6% and among patients <65 years old was 4.0%. In contrast, among patients <65 years old without NASH evaluated at our institution, the prevalence of AF was only 2.7%. The prevalence of AF in US adults <65 years old is estimated to be 2% though it exceeds 9% for patients older than 65 years.[19] Patients with NASH and AF had increased healthcare utilization manifested by higher rates of hospitalization and longer length of stay compared to the NASH cohort without AF. Importantly, only 62.5% of patients with biopsy-proven NASH and AF, all with a documented elevated stroke risk, were receiving guideline-directed therapy for AF aimed at reducing thromboembolic stroke risk. The literature discussing prescriber non-adherence to anticoagulation guidelines in AF is sparse and ranges from 20% to 60%, and the rate of adherence observed in our study is slightly higher yet still indicates undertreatment.[23–26]

Our findings are consistent with other studies, demonstrating an increased prevalence, as well as incidence, of AF in patients with NAFLD.[10,11,27] There is growing epidemiologic evidence which establishes NAFLD as a risk factor for the development of AF, and studies cite a prevalence of AF between 10%-15% in patients with NAFLD.[10,11,27] Coupled with the fact that NASH prevalence is increasing worldwide, it is reasonable to expect that the prevalence and potentially incidence of AF among patients with NASH will also increase in the future. Ongoing clinical and population-based studies are needed to fully document the burden of concurrent AF and NAFLD.

Adherence studies of anticoagulation therapy for AF alone reveal that despite clear guidelines, patient undertreatment and guideline non-adherence remain significant issues.[28–30] This is evident in our cohort, as two patients were inappropriately prescribed antiplatelet therapy instead of anticoagulation therapy. Importantly, no studies, with the exception of the SPAF (Stroke Prevention in Atrial Fibrillation)-1 trial, show benefit for 325 mg of aspirin alone in preventing stroke among patients with AF.[31] As demonstrated in a high-quality meta-analysis including 29 randomized control trials testing antithrombotic therapies of >12 weeks' duration among 28 044 patients, antiplatelets are less effective than anticoagulation for stroke risk reduction in nonvalvular AF and yet they have similar rates of overall bleeding.[32] Understanding the potentially magnified cardiovascular and thromboembolic risk among patients with NASH and concomitant AF, treatment with careful attention to disease-specific guidelines can improve outcomes by preventing thromboembolic events such as stroke.[28,33,34]

Patients with NASH may be undertreated for primary prevention of stroke due to a concern for increased bleeding risk, particularly in cirrhosis. The ability to assess the balance between increased risk of bleeding and hypercoagulability in cirrhosis is complex. Many patients with AF in the setting of cirrhosis are hypercoagulable. However, the risk of bleeding complications while receiving anticoagulation therapy can dampen enthusiasm for the initiation of anticoagulation.[35] Additionally, there are no clear guidelines on how a history of bleeding or further decompensation should impact treatment.[36] Few studies have examined the safety of anticoagulation in cirrhosis specifically for the indication of AF, and those that did demonstrated some benefit in stroke prevention in well-compensated cirrhosis with AF but increased bleeding risk in those with hepatic decompensation.[35,37,38] In our study, 50% (n = 5) of patients with NASH/NASH cirrhosis and concomitant AF were not receiving anticoagulation when it was medically indicated. Only two of these patients had a documented reason, that is, bleeding events, for not receiving therapy. Of the seven patients with AF and NASH cirrhosis in our cohort, there were three patients with a documented history of gastric or esophageal varices receiving anticoagulation therapy, but these patients had well-compensated cirrhosis and low MELD scores.

The pathophysiologic mechanisms delineating how patients with NAFLD may have a propensity to develop AF are not well understood as outlined in a recent excellent review by Anstee et al.[39] Shared risk factors, such as obesity, hypertension, and diabetes, make establishing a direct causative link between the two quite challenging.[11] Studies suggest that the insulin resistance, visceral obesity and the inflammatory environment found in NAFLD may underlie the development of AF.[11,40] Additionally, patients with NAFLD have alterations in cardiac structure which may predispose to AF development.[41] A number of studies have reported increased left atrial volume and size, left ventricular wall thickness and mass, and impaired diastolic function among persons with NAFLD, which are known risk factors for the development of AF.[41–43] In a recent large population-based study of over 3000 black and white middle-aged adults with CT-diagnosed NAFLD, NAFLD was independently associated with subclinical myocardial remodelling and dysfunction independent of established heart failure risk factors.[41] Thus, it is plausible that early myocardial remodelling in the setting of NAFLD state coupled with the progressive inflammatory milieu of NASH may potentially alter cardiac metabolism and function, and thus predispose to the development of AF. The hypothesis that systemic factors rather than direct hepatic processes may underlie the association is supported by the epidemiologic observation that gamma-glutamyl transpeptidase (GGT), a marker of both hepatocyte injury and systemic oxidative stress, is independently associated with incident AF, whereas aminotransferases, as markers of hepatocyte injury alone, have not been independently associated with incident AF.[44]

Fibrosis itself may play a significant role in the development of cardiac remodelling and AF in NASH. A recent cross-sectional analysis from the population-based Multi-Ethnic Study of Atherosclerosis (MESA) demonstrated that early subclinical liver stiffness, a marker of possible fibrosis, as assessed by T1-mapping via magnetic resonance imaging is highly associated with prevalent AF independent of traditional AF risk factors.[45] T1-mapping-derived indices of liver stiffness were also correlated with cardiac structure and functional parameters including left ventricular hypertrophy, left atrial dysfunction, and myocardial fibrosis, findings that might partly explain the mechanisms for the association of liver fibrosis with AF. In addition, analysis from the Oulu Project Elucidating the Risk of Atherosclerosis (OPERA) study demonstrated that increasing liver stiffness, as measured by transient elastography, is associated with increased AF prevalence.[46] A significant portion of our cohort had advanced fibrosis by biopsy and thus the high prevalence of AF among NASH observed in our study may be mediated by the presence of fibrosis. Nevertheless, the mechanisms of the association of liver fibrosis with AF and heart failure are yet to be elucidated.

Strengths and Limitations

Our study has several strengths. First, our findings support recent publications which define AF as highly prevalent in the NAFLD population by extending this finding to the NASH subpopulation, who arguably is at higher risk for both liver- and cardiovascular-related complications related to AF. It is to date the only study to assess anticoagulation use in patients with concurrent AF and NASH, highlighting the need for improved understanding of barriers to adherence to treatment guidelines and the safety of anticoagulation treatment in this particular patient population. There are several limitations of our study that warrant mention. The main limitation is the small size of the biopsy-proven NASH and AF group. While our biopsy-proven patient population is small, prevalence estimates were similar when ICD-9 codes were used to define NAFLD/NASH. In addition, pathology slides from all patients were not available to re-review and so we do not have NAFLD activity scores (NAS) available on all patients, nor was the degree of fibrosis quantified with each pathology read. The small sample size may underestimate the association between AF and NASH as there likely was a significant proportion of patients with likely NASH who did not receive a liver biopsy for diagnosis. Additionally, we relied upon ICD-9 coding to capture patients with a diagnosis of AF, but we may have underestimated the true prevalence of AF in this patient population as some patients may have paroxysmal AF which was undiagnosed. In addition, the increased number of hospital visits among patients with NASH and AF may have led to increased diagnosis of AF due to increased encounters with the healthcare system. A key reference population for AF prevalence was obtained from published data in the general US population and may not be well-matched to our cohort, as our patients had a high comorbidity burden and sought care in a tertiary centre. However, we attempted to address this limitation by including reference data for patients without NASH and who also had AF at our institution. We were not able, however, to perform a matched propensity analysis for healthcare utilization in the AF alone vs NASH and AF cohorts. Finally, we are unable to link adherence to guideline-based anticoagulation management in NASH and AF to hard clinical outcomes, in particular stroke and CVD mortality, due to the small sample size. However, reporting of our single centre experience provides a baseline metric for which we, and others, can use to target quality improvement measures to assess outcomes associated with guideline-based therapy and ultimately improve care and outcomes in the high-CVD risk NASH population.