Silent cerebral microvascular disease and the longitudinal risk of cognitive decline in atrial fibrillation Authors: Yusuke Kondo, Keitaro Senoo, Miyo Nakano, Yuki Shiko, Yohei Kawasaki, Masahiro Nakano, Takatsugu Kajiyama, Raita Uchiyma, Masato Yamanouchi, Satoru Kobayashi, Yoko Pearce, Toshihiro Saito, Ryo Takemura, Kusano Kengo, Takashi Kurita, Takashi Uno, Yoshio Kobayashi Published: 2025 ABSTRACT AbstractThe SKAF study aimed to investigate the incidence of asymptomatic cerebrovascular disorders and cognitive decline in patients with nonvalvular atrial fibrillation (NVAF). We enrolled patients with NVAF aged ≥ 60 years and those with sinus rhythm (SR) for comparison. We collected the head magnetic resonance imaging test and Montreal Cognitive Assessment (MoCA) data at baseline and 24 months follow-up. The primary endpoint was the composite of silent ischemic stroke and microbleeding, and the secondary endpoint was cognitive decline, both over the 24 months of follow-up. In total, 313 patients were enrolled. The SR and NVAF groups comprised 99 and 213 patients, respectively (male, 193 [62%]; age, 72.2 ± 6.5 years). The composite of silent ischemic stroke and microbleeding showed no significant difference between the two groups (17.4% vs. 17.4%; p = 1.00). However, the NVAF group exhibited a significantly larger decline in cognition changes in MoCA score than the SR group (− 1.04 vs. 0,000; p = 0.041). The risk of silent cerebral microvascular disease was equivalent between patients with SR and NVAF. Clinically, NVAF affected cognitive decline.Keywords: Atrial fibrillation, Silent stroke, Microbleeding, Cognitive decline, EdoxabanSubject terms: Cardiology, Neurology Introduction Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia worldwide, and its clinical burden will continue to expand as the population ages, with coexisting cardiovascular disease. Over the next four decades, its incidence is expected to grow by 150%1. AF significantly increases the risks of mortality, stroke, heart failure, cognitive impairment, and dementia2–5. Approximately one in five patients with AF exhibits clinically silent brain infarcts on systematic brain magnetic resonance imaging (MRI)4. These silent brain infarcts are associated with cognitive dysfunction similar to a clinical stroke, potentially explaining the link between AF and cognitive decline in individuals without stroke history. Oftentimes, patients with AF present with a high burden of cerebral microbleeding (CMBs) and white matter lesions. Prospective studies are necessary to elucidate brain lesion progression and their effects on cognitive decline. Although newer oral anticoagulants such as rivaroxaban, dabigatran, apixaban, and edoxaban effectively prevent clinical stroke and reduce dementia risk in patients with AF, prospective data on the development of new brain lesions and their association with cognitive decline in patients undergoing oral anticoagulation therapy for AF remain unavailable6. This exploratory study aimed to investigate the incidence of asymptomatic cerebrovascular disorders and cognitive decline in patients with nonvalvular AF (NVAF) over a 2-year follow-up period using predefined, systematic brain MRI. Methods Patient population The Silent Cerebral Microvascular Disease and Longitudinal Risk of Cognitive Decline in AF (SKAF) study is a prospective, investigator-initiated, multicenter, observational study conducted at seven centers in Japan. In this study, we enrolled 200 patients without a history of cerebral infarction or dementia, with a definitive diagnosis of NVAF, and with and without anticoagulant therapy with edoxaban. For comparison, we included 100 patients with sinus rhythm (SR), aged 60 years or older. The exclusion criteria comprised mitral stenosis, artificial valve implantation (mechanical or bioprosthetic), cardiac intervention within 1 month before enrollment, and contraindications to MRI. Furthermore, all patients underwent screening using the mini-mental state examination (MMSE) and the Geriatric Depression Scale-Short Version to rule out dementia and depression7,8. Those with scores ≤ 24 points on the MMSE and ≥ 10 points on the Geriatric Depression Scale-Short Version were excluded. Figure 1 shows a flowchart reporting the numbers and reasons for excluding the screened cohort from the final study population. Study flowchart. Flowchart of the participant selection process MRI, magnetic resonance imaging. The final study population comprised 310 participants: 213 patients with AF (115 paroxysmal and 98 persistent) and 97 controls in SR without AF history. AF was defined as paroxysmal if self-terminating within 7 days and persistent if at least one AF episode lasted longer than 7 days according to the direct patient interviews and/or an exhaustive medical record search. All patients provided written informed consent before enrollment. The study conformed to the latest version of the Declaration of Helsinki and obtained approval from the ethics committees at the Chiba University, Graduate School of Medicine (Approval no. G29011). Baseline evaluation All screened patients underwent extensive clinical assessment, including medical history (targeting AF type and duration, heart disease occurrence, and comorbidities), thromboembolic risk assessment, physical examination, and electrocardiography. The thromboembolic risk assessment comprised the CHADS2 score and the congestive heart failure, hypertension, age ≥ 75 years (doubled), diabetes mellitus, prior stroke or transient ischemic attack (TIA) or thromboembolism (doubled), vascular disease, age 65–74 years, sex (CHA2DS2-VASc) score. Clinical variables At baseline and yearly in-person study visits, trained study personnel collected patient information such as patient demographics, prior medical history, interventional and medical treatments, and risk factors by using standardized case report forms. Weight and height were measured directly, and body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Clinical stroke was defined as a new focal neurological dysfunction with clinical, imaging, or pathological evidence of focal infarction of ischemic, hemorrhagic, or undetermined origin. Supplementary Material details the definitions of the stroke subtypes. A TIA was defined as a new transient focal neurological dysfunction without evidence of focal infarction. Clinical stroke or TIA occurring during the follow-up period was documented on predefined case report forms and independently adjudicated by two stroke neurologists using anonymized source documentation (medical records, including discharge summaries and imaging reports). Discordant adjudications were resolved through consensus. MRI Brain MRI was conducted using a 1.5-T or 3.0-T scanner. The imaging protocol included a sagittal T1-weighted spin–echo sequence (repetition time, 400 ms; echo time, 13 ms; slice thickness, 5 mm; field of view, 230 mm; matrix size, 192 × 256), an axial T2-weighted fluid-attenuated inversion recovery (FLAIR) sequence (repetition time, 8,500 ms; echo time, 112 ms; inversion time, 2,500 ms; slice thickness, 5 mm; field of view, 240 mm; matrix size, 154 × 256), and a diffusion-weighted sequence (repetition time, 3,200 ms; echo time, 99 ms; slice thickness, 5 mm; field of view, 230 mm; matrix size, 128 × 128; bandwidth, 1,502 Hz; gradient strength, 22 mT; diffusion gradient duration, 31 ms; gradient separation, 42 ms in three orthogonal directions; B value, 1,000). We generated an apparent diffusion coefficient map for each diffusion-weighted sequence. All sequences were centered on an axis, the line passing between the anterior and posterior cerebral commissures. According to the anatomic–pathological criteria, areas of silent cerebral infarction (SCI) were defined as focal, sharply demarcated, regularly or irregularly shaped areas that were hyperintense on T2-weighted FLAIR images or isointense on T1-weighted images9. We used T2-weighted FLAIR images to differentiate gliotic ischemic lesions < 3 mm from the perivascular spaces and lacunae (hypointense on T2-weighted FLAIR images)10. To distinguish SCI from subcortical infarction, lesions were classified as SCI if they were asymptomatic, confined to the subcortical white matter or deep gray matter (e.g., putamen, caudate nucleus, or thalamus), ≥ 3 mm in diameter, hyperintense on T2-weighted FLAIR images, and isointense on T1-weighted images. Lesions associated with acute clinical symptoms, those involving large vascular territories, or those extending into the cortical ribbon were excluded. CMBs were assessed using T2-weighted gradient-echo imaging. CMBs were defined as small (2–10 mm in diameter), round or ovoid, well-defined hypointense lesions, reflecting hemosiderin deposits in perivascular spaces. Lesions located in the brain parenchyma and distinct from vascular flow voids, calcifications, or perivascular spaces were considered CMBs, based on established radiological criteria11. Patients with even a single lesion were assigned to the SCI group. We recorded each SCI area and defined small lesions closely grouped together, exclusively localized in the subcortical white matter of the frontal lobe, as a “spotted pattern.” All MRI scans were independently analyzed by two neuroradiologists who were blinded to the clinical data. Conflicts were resolved through consensus with a third expert. Cognitive performance We assessed participants’ cognitive function by using the MMSE and the Montreal Cognitive Assessment (MoCA), exploring five domains: visuospatial/executive functioning, naming, memory, attention, and language7,12. Each domain provides varying valuable information on cognitive function, allowing healthcare professionals to identify potential areas of impairment. Statistical analysis We collected patient characteristics, clinical course, and data from brain MRI and cognitive function tests (MoCA) at baseline and at the 24-month follow-up. This study primarily sought to estimate the incidence rate of SCI in NVAF cases. In this study, based on previous reports, the incidence of SCI over a two-year period in community-dwelling older adults without NVAF was assumed to be approximately 3%. Accordingly, the alternative hypothesis was set as an incidence rate exceeding 3%13. In contrast, since prior studies have reported an incidence rate of approximately 7% in patients with NVAF, this value was used as the threshold for the null hypothesis14. This hypothesis setting was determined by considering both clinical relevance and statistical power. A two-sided significance level of 5% and a power of 80% were used in the analysis. The estimation of the incidence rate was accompanied with the calculation of the exact 95% confidence intervals using the binomial distribution. Additionally, the variables of the NVAF group and the SR group (control) were compared using Fisher’s exact test. After adjusting for confounding factors, we estimated the adjusted risk difference and 95% confidence intervals (CIs) and compared them between patients in each group and between the two patient groups by using the Mantel–Haenszel method. We also used the log-rank test to compare the incidence rates of CMBs between patients in each group and between such groups. Moreover, hazard ratios and 95% CIs were estimated using Cox regression analysis. Similar statistical methods were employed to assess a composite endpoint comprising both asymptomatic SCI and CMBs. For sensitivity analysis, we conducted Cox regression analysis adjusted for patient characteristics to estimate the adjusted hazard ratios and 95% CIs. The secondary analysis aimed to provide supplementary insights into the main analysis results. Multiple adjustments were not performed. The significance level was set at 5% for two-tailed tests, and 95% CIs were calculated. Descriptive statistics was calculated for each group and visit. The null hypothesis of equal means between groups was tested using the t-test. Additionally, the analysis of covariance was conducted to compare changes from baseline to the final time point (24 months) between the NVAF and control groups, with baseline values serving as covariates. To determine the prognostic factors and confounding variables related to the primary outcome measures, we examined the associations between patient characteristics and the primary outcome measures. The cutoff values for each factor were determined according to the existing literature and final analysis. Moreover, the hazard ratios and their 95% CIs were estimated using Cox proportional hazards models to assess effect sizes. We plotted Kaplan–Meier survival curves for each factor and compared group differences using the log-rank test. Optimal models for predicting effects were also explored through model selection. All analyses were conducted using the SAS software (version 9.4; SAS Institute, Cary, NC, USA), with p < 0.05 considered statistically significant. Results The study was initiated in October 2016 with a planned 2-year recruitment period. Figure 1 presents a flowchart outlining the study. As of January 31, 2018, we screened 326 patients and enrolled 313. Among these enrolled patients, 97 had SR (control group), and 213 had NVAF (NVAF group) (male, 193 [91%]; age, 72.2 ± 6.5 years). Of the 195 eligible patients with NVAF, 156 (80%) received anticoagulant therapy with edoxaban. Table 1 lists the baseline characteristics. Anticoagulant therapy was not prescribed for any indication other than NVAF in either the NVAF group or the control group. Four patients with cerebrovascular disorders were included at baseline. All of these conditions were non-stroke cerebrovascular diseases, including internal carotid artery stenosis, cerebral arteriovenous malformations, and cerebral aneurysms. Table 1Comparison of baseline characteristics between the nonvalvular atrial fibrillation and sinus rhythm (control) groups.NVAF, nonvalvular atrial fibrillation; AF, atrial fibrillation; CHADS2: congestive heart failure, hypertension, age > 75 years, diabetes (all 1 point each), and previous stroke (2 points); CHA2DS2-VASc: congestive heart failure, hypertension, age ≥ 75 (doubled), diabetes, stroke (doubled), and vascular disease; AAD, antiarrhythmic drug; COPD, chronic obstructive pulmonary disease; PPI, proton pump inhibitor; PPAR, peroxisome proliferator–activated receptor. Comparison of baseline characteristics between the nonvalvular atrial fibrillation and sinus rhythm (control) groups. NVAF, nonvalvular atrial fibrillation; AF, atrial fibrillation; CHADS2: congestive heart failure, hypertension, age > 75 years, diabetes (all 1 point each), and previous stroke (2 points); CHA2DS2-VASc: congestive heart failure, hypertension, age ≥ 75 (doubled), diabetes, stroke (doubled), and vascular disease; AAD, antiarrhythmic drug; COPD, chronic obstructive pulmonary disease; PPI, proton pump inhibitor; PPAR, peroxisome proliferator–activated receptor. The primary composite endpoints showed no significant difference between the two groups (17.4% vs. 17.4%; p = 1.00). Throughout the 24-month follow-up period, the incidence of silent ischemic stroke was 10.8% and 9.3% in the NVAF and control groups (p = 0.083), respectively, whereas that of CMBs was 8.1% and 9.2% in the NVAF and control groups (p = 1.00), respectively (Table 2). Additionally, potential risk factors, including age, sex, BMI, hypertension, diabetes mellitus, chronic kidney disease, smoking, and alcohol consumption habits, for cerebrovascular disorders were not significant (Table 3). Table 2Comparison of the incidence of cerebrovascular disorders between the NVAF and control groups.NVAF, nonvalvular atrial fibrillation. Comparison of the incidence of cerebrovascular disorders between the NVAF and control groups. NVAF, nonvalvular atrial fibrillation. Table 3Potential risk factors for cerebrovascular disorders.HR, hazard ratio; CI, confidence interval; BMI, body mass index; HT, hypertension; DM, diabetes mellitus; CKD, chronic kidney disease. Potential risk factors for cerebrovascular disorders. HR, hazard ratio; CI, confidence interval; BMI, body mass index; HT, hypertension; DM, diabetes mellitus; CKD, chronic kidney disease. Table 4 displays the MMSE and MoCA scores at baseline and at the 24-month follow-up, and the change in cognitive function during the follow-up period, as assessed by the MoCA score, was significantly greater in the NVAF group compared to the control group (− 1.04, 95% CI: − 2.04 to − 0.04, p = 0.041), indicating that NVAF is associated with cognitive decline. In the analysis of covariance, adjustments were made for potential confounders, including hypertension, diabetes mellitus, dyslipidemia, obesity, smoking, alcohol consumption, educational attainment, cardiovascular disease, and the use of anticoagulant or antiplatelet agents. Furthermore, although this was a post-hoc analysis, after adjusting for confounding factors, the decline tended to be greater in patients with persistent AF compared to those with paroxysmal AF. However, no statistically significant difference was observed between the two groups (− 0.50, 95% CI: − 1.35 to 0.35, p = 0.250). Table 4Comparison of the MMSE and MoCA scores between the NVAF and control groups at baseline (a) and 24 months of follow-up (b).NVAF, nonvalvular atrial fibrillation; MMSE, mini-mental state examination; MoCA, Montreal Cognitive Assessment. Comparison of the MMSE and MoCA scores between the NVAF and control groups at baseline (a) and 24 months of follow-up (b). NVAF, nonvalvular atrial fibrillation; MMSE, mini-mental state examination; MoCA, Montreal Cognitive Assessment. The incidence of major adverse events over the 24-month follow-up was 5.0% in the control group and 19.1% in the NVAF group (p < 0.001). Ischemic stroke and hemorrhagic stroke occurred in four (1.9%) and two (0.93%) patients in the NVAF group, respectively, while none in the control group (Table 5). Table 5Comparison of major adverse events between the NVAF and control groups over the 24-month follow-up period.NVAF, nonvalvular atrial fibrillation; PCI; percutaneous coronary intervention, ACS; acute coronary syndrome. Comparison of major adverse events between the NVAF and control groups over the 24-month follow-up period. NVAF, nonvalvular atrial fibrillation; PCI; percutaneous coronary intervention, ACS; acute coronary syndrome. Discussion This study obtained the following findings in patients aged 60 years without cerebral infarction or dementia history in Japan: (i) both patients with SR and NVAF exhibited a comparable risk of silent cerebral microvascular disease, and (ii) NVAF had a significant influence on cognitive decline. Our study benefited from the inclusion of a control group and the consistent use of a single anticoagulant, edoxaban, for all patients in the NVAF group. Edoxaban is administered once daily, which is advantageous for maintaining medication adherence. This approach also helped reduce variability and potential confounding associated with differences in the pharmacokinetics and pharmacodynamics of different anticoagulants. Additionally, we conducted a detailed characterization of patients with NVAF without a history of cerebral infarction or dementia, which involved regular brain MRI scans and cognitive assessments at baseline and at the 2-year follow-up. Notably, our focus extended beyond SCI to include CMBs. Swiss atrial fibrillation study Similar to our study, the Swiss AF study conducted MRI and cognitive function assessments in patients with NVAF before and after a 2-year follow-up period15. In the Swiss-AF study, 5.5% of patients with AF developed new brain infarcts over a 2-year follow-up. Notably, 85% of these infarcts occurred in patients who were on oral anticoagulation therapy, including both vitamin K antagonists such as warfarin and direct oral anticoagulants, and were clinically silent. In this SKAF study, only 4 (1.9%) of 215 patients with NVAF experienced symptomatic brain infarction. Taken together, the results from the Swiss-AF and SKAF studies suggest that new brain infarcts, regardless of whether they are clinically apparent or silent, are associated with cognitive decline, underscoring the potential contribution of even asymptomatic cerebral events to neurocognitive deterioration in patients with AF. Therefore, anticoagulation therapy alone may not fully prevent progressive brain damage in all patients with AF. This study raises two interesting points. First, among the patients receiving anticoagulant therapy, 33.7% used vitamin K antagonists. Therefore, determining the time spent in the therapeutic range of this drug would be interesting. Additionally, 19.2% of the patients had a history of stroke or TIA during enrollment. In multivariate models, a history of stroke or TIA has been identified as a potential predictor of new silent brain infarcts. Hence, the insight into whether the presence of prior cerebral damage influences the correlation between silent brain infarction and cognitive impairment remains unclear. In patients with AF, cerebrovascular events may have occurred before the anticoagulant therapy was initiated, potentially compromising brain protective mechanisms or causing irreversible damage. Atrial fibrillation increases the risk of stroke-related dementia Various mechanisms have been proposed to explain the association between AF and cognitive dysfunction, with ischemic brain injury being the best understood. Given that AF causes ischemic stroke, the risk of developing cognitive impairment could be increased. According to a meta-analysis, the risk for cognitive impairment is more than double (hazard ratio, 2.4) in patients with AF who have experienced a stroke16. Clinically insignificant or asymptomatic embolic events can also contribute to cognitive decline. The odds of SCI associated with future stroke and cognitive impairment can be doubled by AF14. The use, timing, and effectiveness of anticoagulation therapy are crucial for preventing brain injury and cognitive decline. In a population-based study in Sweden, the incidence rate of cognitive impairment among anticoagulant users was lower than that among nonusers (1.14 vs. 1.78 per 100 patient-years at risk, p < 0.001)17. Furthermore, patients who received anticoagulants at least 80% of the time experienced greater benefits. From a timing perspective, the risk of developing cognitive impairment was lowest in patients who initiated anticoagulation within 1 year of diagnosis. Therefore, early initiation of stroke prevention therapy is important in determining cognitive decline. Atrial fibrillation increases the risk of dementia regardless of stroke The correlation between NVAF and dementia extends beyond vascular dementia, with recent evidence suggesting an increased risk of Alzheimer’s disease and senile dementia. In a previous study involving over 37,000 patients from the Intermountain Healthcare registry, with 27% having AF and 4.1% having dementia, the odds ratios for AF-associated dementia were 1.73, 1.39, 1.06, and 1.44 for vascular dementia, senile dementia, Alzheimer’s disease, and nonspecific dementia, respectively, diagnosed via ICD codes18. While vascular dementia had the highest odds ratio, other forms of dementia were clearly associated with AF. Another study of over 550,000 participants from the Korea National Health Insurance Service confirmed a significant increase in dementia risk with incident AF (hazard ratio, 1.52; 95% CI, 1.43–1.63). Even after adjusting for stroke, the hazard ratio remained significant at 1.27 (95% CI, 1.18–1.37)16. Numerous observational reviews and meta-analyses have agreed that AF independently increases the risk of dementia, irrespective of the manifestation of cerebrovascular events1,18–22. In this SKAF study, while NVAF was associated with cognitive decline, it was not related to microvascular brain lesions. Thus, the association between NVAF and dementia may be attributed to Alzheimer’s disease or senile dementia. Newly proposed mechanism linking atrial fibrillation to cognitive dysfunction NVAF causes blood dynamic abnormalities, such as loss of atrial contractility and asynchronous atrioventricular activity, leading to reduced cardiac output. This outcome can result in transient or chronic cerebral hypoperfusion, potentially causing a stroke and cognitive decline. Studies using MRI in Iceland demonstrated that persistent AF decreases cerebral blood flow and estimates whole-brain perfusion23). This finding may help elucidate the mechanisms underlying cognitive decline. Other potential mechanisms for cognitive decline in AF cases are cerebral microvasculature dysfunction and autonomic dysregulation. Brain blood vessels are highly sensitive to changes in arterial carbon dioxide (CO2) levels, affecting the blood flow24. Autonomic dysregulation, commonly observed in patients with brain injuries such as stroke, also occurs in patients with AF, with reduced cerebrovascular reactivity to CO2 and elevated biomarkers of oxidative stress, inflammation, and endothelial dysfunction25,26. Imaging findings, such as white matter changes and CMBs observed on brain MRI, have been associated with cognitive decline, stroke, and vascular dementia4,13. However, data on whether controlling the rate in patients with AF can prevent cognitive decline remain inconclusive. Although the rates of asymptomatic cerebrovascular events were similar between the NVAF and control groups, cognitive decline as measured by the MoCA was observed only in the NVAF group, suggesting a possible association that warrants further investigation. In the present study, while the NVAF group showed a significant decline in MoCA scores, no significant change was observed in MMSE scores. This discrepancy may reflect the differing sensitivities of the two assessments. The MMSE primarily evaluates basic cognitive functions such as memory, orientation, and language, whereas the MoCA includes more complex and demanding tasks assessing executive function, attention, and visuospatial ability, and is considered more sensitive to subtle cognitive changes, particularly in the early stages of impairment12,27. Previous studies have reported that AF may be associated with gradual declines in higher-order cognitive function even in the absence of overt stroke28. Furthermore, population-based brain MRI studies have demonstrated a high prevalence of subclinical cerebrovascular abnormalities such as silent infarcts and CMBs, which may be linked to mild cognitive impairment29. Therefore, the decline in MoCA scores observed in this study may reflect the cumulative effects of silent cerebrovascular injury or AF-related cerebral remodeling, even in the absence of clinically apparent stroke. These findings underscore the importance of using sensitive screening tools such as the MoCA to detect early cognitive decline in patients with NVAF. To elucidate the optimal management and prevention strategies for the sequelae of AF that can lead to cognitive decline and dementia, we need further research. Improved treatments for AF have reduced mortality caused by cardiovascular diseases. Trends from the Framingham Heart Study suggest that most of the AF-related deaths are not attributable to cardiovascular disease, regardless of age at diagnosis30. As AF therapies further advance, the distribution of causes of death is changing, with death from other causes, such as cognitive dysfunction and dementia, being more likely to occur. Our study has several limitations. First, its relatively small sample size and observational design preclude the establishment of causal relationships. First, its relatively small sample size, inclusion of a control group that did not consist of completely healthy individuals, and observational design preclude the establishment of causal relationships. Second, the generalizability of our findings may be limited, particularly to patient populations who are unable to undergo brain MRI due to implanted cardiac devices or competing medical risks. Third, we did not collect information on the presence of atherosclerosis in the arteries supplying the brain, which may have influenced the outcomes. Additionally, the number of detected brain infarcts was limited, reducing the statistical power to assess predictors and perform subgroup analyses, such as evaluating the association between infarct location and cognitive decline. Moreover, adjustment for randomly missing observations suggests a potential underestimation of the true incidence of brain lesions in individuals with AF. Furthermore, brain MRIs were performed using either 1.5 T or 3.0 T scanners. Given that the sensitivity to detect microinfarcts may differ between these two modalities, this technical variability could have affected the identification and quantification of cerebral microvascular lesions. Therapeutic interventions for heart failure and other cardiac conditions may also affect cognitive outcomes. Although we considered adjusting for these variables, the number of relevant events was insufficient for meaningful statistical analysis, and they were therefore excluded from the final models. Finally, our study did not examine the effects of edoxaban. Data regarding AF duration at the time of anticoagulation initiation, the impact of rhythm-control strategies, and the timing of stroke occurrence relative to cognitive decline were not available. These unmeasured factors could have influenced the interpretation of our findings. The SKAF registry indicated an equivalent risk of silent cerebral microvascular disease in patients with SR and those with NVAF. Clinically, NVAF affects cognitive decline. Disclosures Dr. Kondo received lecture fees from Daiichi Sankyo, Bayer, Abbott Medical Japan, Biotronik Japan, Boston Scientific, Japan Lifeline, Medtronic Japan and research funds from Daiichi Sankyo and Boston Scientific. Dr. Senoo belongs to the endowed department from Biotronik Japan and Japan Lifeline. Drs. Nakano Mi and Nakano Ma belong to the endowed department from Abbott Medical Japan, Biotronik Japan and Fukuda Denshi. Dr. Nakano Ma received lecture fees from Medtronic. Dr. Kajiyama belongs to the endowed department from Medtronic Japan. Dr. Kusano received lecture fees and research grants from Medtronic Japan and Biotronik Japan. Dr. Kurita received lecture fees from Medtronic Japan, Abbott Japan, Japan Lifeline, and Biotronik Japan. Dr. Uno received lecture fees from Elekta Japan, Siemens, and Zeria and educational grant from GE Healthcare Pharma, Guerbet Japan, PDRadiopharma, and Fuji Pharma. Dr. Kobayashi received lecture fees from Abbott Medical Japan, Bayer Japan, Bristol-Myers Squibb, Boehringer Ingelheim, Daiichi Sankyo, and scholarship funds from Takeda Pharmaceutical, Abbott Medical Japan, Terumo, Otsuka Pharmaceutical, Boehringer Ingelheim, Astellas, Daiichi Sankyo, Win International, Japan Lifeline, and Nipro. Acknowledgements We thank all hospital staff who assisted in the data collection. Author contributions Yu.K., K.S., K.K. and K.T conceived and designed the study, Mi.N., Ma.N., T.K, T.U., M.Y., S.K. and Y.P. conducted data analysis, and Yu.K., K.S. and T.S. drafted the manuscript. Y.S. and Yoh.K. provided critical input in the statistical re-analysis of our data. All authors contributed to the interpretation of the data and reviewed the final version of the manuscript. Yos.K. oversight throughout the project. Funding This work was funded by the Daiichi Sankyo Co., Ltd. Data availability The data underlying this article will be shared on reasonable request to the corresponding author. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data underlying this article will be shared on reasonable request to the corresponding author.