Risk of dementia in patients with atrial fibrillation: Short versus long follow‐up. A systematic review and meta‐analysis

Abstract Background No previous meta‐analyses have compared the risk of dementia, due to an underlying atrial fibrillation (AF), in the short‐term versus the long‐term period. Aim To perform an update meta‐analysis of studies examining the association between AF and dementia and the relative impact of follow‐up period. Methods Data were obtained searching MEDLINE and Scopus for all investigations published between 1 January 2000 and March 1, 2021 reporting the risk of dementia in AF patients. The following MeSH terms were used for the search: “Atrial Fibrillation” AND “Dementia” OR “Alzheimer’s disease”. From each study, the adjusted hazard ratio (aHR) with the related 95% confidence interval (CI) was pooled using a random effect model. Results The analysis was carried out on 18 studies involving 3.559.349 subjects, of which 902.741 (25.3%) developed dementia during follow‐up. A random effect model revealed an aHR of 1.40 (95% CI: 1.27–1.54, p < 0.0001; I2 = 93.5%) for dementia in subjects with AF. Stratifying the studies according to follow‐up duration, those having a follow‐up ≥10 years showed an aHR for dementia of 1.37 (95% CI: 1.21–1.55, p < 0.0001, I2 = 96.6%), while those with a follow‐up duration <10 years has a slightly higher aHR for dementia (HR: 1.59, 95%CI: 1.51–1.67, p < 0.0001, I2 = 49%). Nine studies showed that the aHR for Alzheimer’s disease (AD) in AF patients was 1.30 (95%CI: 1.12–1.51, p < 0.0001, I2 = 87.6%). Conclusions Evidence suggests that patients with AF have an increased risk of developing dementia and AD. The risk of dementia was slightly higher when the follow‐up was shorter than 10 years.


| INTRODUCTION
Atrial fibrillation (AF) represents the most common cardiac arrhythmia affecting elderly patients, 1,2 affecting about 9% of adults aged 80 years or older. 3 In Europe alone, prevalence of AF in 2010 was around 9 million among individuals older than 55 years and is expected to reach 14 million by 2060. 4 Aging itself exerts significant structural changes on the atrial bundles, characterized by an excessive accumulation of fibrillary collagen in the extracellular space which leads to a progressive agedependent cardiomyocyte loss and concomitant fibrosis replacement. 5 AF is associated with multiple comorbidities, including the development of vascular dementia, but also of the major cause of dementia syndrome, Alzheimer's disease (AD). [6][7][8] This association appears to be multifactorial, and no one model will explain the association completely. Cerebrovascular events, such as stroke/transient ischemic attack, but also subclinical abnormalities, in primis microbleeds and chronic cerebral hypoperfusion (CCH), may reasonably account for this observed relationship. 6,9,10 Consistently, animal studies suggest that long-term AF decreases cardiac output and may precede CCH and the consequent hypoxia. 6 In turn, these adverse events impair the clearance and enhance the accumulation of amyloid-β peptides collection in cerebral vessels, therefore increasing AD risk. 10 A further support on the link between AF and dementia emerged from observational studies showing that, among subjects with either prevalent or incident AF, the treatment with anticoagulant drugs was associated with a decreased risk of cognitive impairment or dementia. 11,12 However, before speculating about the potential treatment strategies able to reduce the risk of dementia due to an underlying AF, a more precise estimation of that risk, especially in those patients having such arrhythmic disease for a long time, remains essential. Moreover, AF is a progressive disease that becomes more difficult to treat with increasing duration and in this regard, aging plays a fundamental role, also the onset on multiple comorbidities which may tigger and maintain AF. 13 Compellingly, no previous analyses have compared the risk of dementia, due to an underlying AF, in the short-term versus the long-term period.
Therefore, in the present manuscript we performed a systematic review and meta-analysis aimed to evaluate the long-term relationship (>10 years) between AF and dementia in population-based studies.

| MATERIALS AND METHODS
The study was performed in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines (Table S1). 14  The selection of studies was independently conducted by 2 authors (MZ, GZ.) in a blinded fashion. Any discrepancies in study selection were resolved consulting a third author (CC). The following MeSH terms were used for the search: "Atrial Fibrillation" AND "Dementia" OR "Alzheimer's disease." Moreover, to ensure comprehensiveness, reference lists of retrieved studies and previous review articles were screened for additional relevant studies. Studies were included if: (1) they provided data regarding the risk of dementia in patients with confirmed AF; (2) the risk of dementia was expressed as adjusted hazard ratio (aHR) with relative 95% confidence (3) they reported information regarding the dementia diagnosis. Conversely, studies reporting the occurrence of AF in patients with mild cognitive impairment (MCI) as well as randomized controlled trials, case reports, review articles, abstracts, editorials/letters, and case series with less than 15 participants were excluded from the analysis. Pre-clinical studies (i.e., in-vitro or animal studies) were also excluded by the final analysis. If a study involved the same population, only the most recent investigation was included (overlapping cohort).

| Outcomes and data extraction
The primary outcome of the meta-analysis was the development of any kind of dementia in AF patients. The secondary outcomes were the comparison of dementia risk after considering the duration of the follow-up (<10 vs. ≥ 10 years) and the correlation between AF and risk of Alzheimer's disease (AD). For all investigations reviewed we extracted the year of publication, gender (males %), follow-up duration, total number of participants and dementia patients and methods ZUIN ET AL.
-1489 used for both AF and dementia diagnosis. Two authors (A.P. and C.B.) revised and extracted the data; in case of discrepancies a third author was consulted (C.C.).

| Quality of studies
The quality of included studies was graded using the Newcastle-Ottawa quality assessment scale (NOS). 15 Specifically, three authors (M.Z., G.Z. and C.C.) performed the quality assessment; in case of discrepancies, a fourth author was consulted and then, the debate was resolved by consensus.

| Statistical analysis
From each study, the adjusted hazard ratio (aHR) with the related 95% confidence interval (CI) was pooled using a random-model while a forest plot was adopted to visually evaluate the results. Statistical heterogeneity between groups was measured using the Higgins I 2 statistic. Specifically, a I 2 = 0 indicated no heterogeneity while we considered low, moderate, and high degrees of heterogeneity based on the values of I2 as <25%, 25%-75% and above 75% respectively.
Moreover, tau-squared (τ 2 ) was also calculated to see the extent of variation among the effects observed in different studies. To evaluate the publication bias, both the visual inspection of the funnel plots and the Egger's test were used. A predefined sensitivity analysis (leaveone-out analysis) was performed removing one study at the time. To further appraise the impact of potential baseline confounders, metaregression analyzed using the length of follow-up of each study and the latitude of the population enrolled, as moderator variables were performed. Analyses were carried out using comprehensive metaanalysis software (CMA), version 3. The HRs were compared by using the software R (R software-version 3.6.3) (http://www.r-project. org/). A p-value <0.05 was considered statistically significant.

| Search results and study characteristics
A total of 1,460 articles were retrieved after excluding duplicates. The initial screening excluded 425 articles because they did not meet the inclusion criteria, leaving 674 articles to assess for eligibility. Subsequently, after evaluation of the full-text articles, 656 were excluded and 18 investigations met the inclusion criteria ( Figure 1). 11,16-31 As shown in Table S2, the number of subjects under oral anti-coagulant therapy was precisely reported in some of the studies considered in the analysis. 16 follow-up. The demographical characteristics as well as the criteria adopted to define AF and to diagnose dementia are shown in Table 1.
The studies included into the meta-analysis resulted of moderate-high quality according to the NOS (Table 2). 15 Table 3. The pooled analysis, using a random effect

| Risk of dementia in patients with atrial fibrillation based on follow-up length
The studies reviewed were further stratified according to the followup duration to determine the risk of AF over time. Specifically, eight studies, based on 3.232.981 subjects, had a follow-up ≥10 years [18,[21][22][23][24][25][29][30]; in these patients, the aHR for dementia was 1.  Figure S2 (Panels A and B, respectively).

| Risk of Alzheimer's disease in patients with atrial fibrillation
Nine investigations performed a sub-analysis on the risk of developing AD in patients with AF. As showed in Figure 5, the aHR was 1.30 (95%CI: 1.12-1.51, p < 0.0001, I 2 = 87.6%). Again, no bias was detected using the Egger's test (t = 0.898, p = 0.398) or by the visual assessment using the funnel plot ( Figure S3).

| DISCUSSION
Our meta-analysis confirms previous evidence showing an increased risk of dementia in patients affected by AF. 20  compared to those with a follow-up length ≥10 years (+37%  [33][34][35][36] ; interestingly, all these conditions have been associated with both AD and VD. 35 The present results highlight that follow-up period plays a pivotal role on the risk of dementia in AF patients. Indeed, recent investigations have elucidated the progressive nature of AF, that becomes more difficult to treat with increasing duration. This aspect is believed to be mainly due to the occurrence of electrical, contractile, and structural remodeling of the atria, which creates a fertile environment for the propagation of AF. [37][38][39] Furthermore, AF generally occurs in the setting of underlying heart disease, such as coronary artery disease, hypertension, valve disease, congestive heart failure, and thyroid dysfunction, which may both trigger and maintain the arrhythmia. Moreover, the occurrence of such comorbidities naturally increases with aging, 40,41 therefore, a longer follow-up may be more adequate to discriminate the impact of a progressive disease on these patients. Despite sensitivity analysis excluded that no single study had an undue impact on the combined HR, meta-regression analysis showed that the length of follow-up duration was a significant source of statistical heterogeneity. Conversely, the latitude of the populations enrolled did not explained the heterogeneity, potentially suggesting that life quality and/or the standard of care did not act as confounding factors. Among the potential strategies suggested for the reduction of dementia onset due to an underlying AF, some investigations have suggested the potential benefices obtained from the administration of oral anticoagulants. 12,42 Unfortunately, these drugs remain under administered in elderly patients with dementia, even in the presence of AF. 43,44 Other findings showed that more invasive treatment, such as catheter ablation, may be another optional treatment to reduce the burden of dementia in these patients, but in most of cases but it remains an invasive procedure with related risk especially in elderly subjects with serval comorbidities and a potential source of silent strokes and cognitive impairment. 23 perspective, dementia and AF seems to be linked by the appearance of micro embolic events as well as microinfarcts, as those observed in AD patients. 24,46 Proinflammatory states are actively implicated in both the genesis and perpetuation of AF, as well as in the promotion of hypercoagulability and thrombus formation, predisposing to stroke. 29 Previous analyses have demonstrated that the timing for the initiation of oral anticoagulation, and the quality of the drug, had a pivotal role in decreasing the risk of dementia in these patients. 47,48 Indeed, patients treated with warfarin with a lower TTR, are at higher risk of dementia. 22 However, some of these issues have been partially overcome with the recent introduction of novel oral anticoagulant agents (NOACS). NOACs, such as dabigatran, rivaroxaban, and apixaban are at least as effective as warfarin in preventing strokes in patients with AF. Moreover, these drugs reduce the risk of intracranial hemorrhage, which represents one of the most important adverse events of oral anticoagulation in elderly with AF. 49 In this regard, a recent meta-analysis comparing NOAC and Warfarin has shown that the former significantly reduced the occurrence of dementia and bleeding events. 18 We cannot assess this aspect in our analysis, since the studies reviewed did not systematically report type, duration, and quality of anticoagulation treatment. Moreover, this aspect was beyond the aim of our study which was to provide an updated evaluation of the risk of developing dementia in AF patients.
In a similar fashion, the therapeutic strategy adopted for the treat-

| CONCLUSIONS
Patients with AF have an increased risk of developing dementia and AD, and the risk is slightly higher in the studies with a follow-up shorter than 10 years. Further analyses are needed to confirm our results and to assess the potential benefit of a more aggressive therapy in those patients with AF and long-life expectances. Indeed, elderly patients generally experience different comorbidities limiting the use of interventional treatments and therefore forcing to a medical treatment, which in several cases is not conclusive or allows some relapse of the arrhythmias with hemodynamic consequences and or complications.
Therefore, it appears useful to adopt a patient's tailored approach also considering the risk of dementia in the long-term period which should promote the resolution of AF and potentially avoid watchful within approaches also if the arrhythmia is well tolerated and without hemodynamic effects.