Association Between Non-alcoholic Fatty Liver Disease and the Risk of Dementia

A Nationwide Cohort Study

Gi-Ae Kim; Chi Hyuk Oh; Jung Wook Kim; Su Jin Jeong; In-Hwan Oh; Jin San Lee; Key-Chung Park; Jae-Jun Shim


Liver International. 2022;42(5):1027-1036. 

In This Article


This nationwide study aimed to investigate the association between NAFLD and the risk of dementia. To explore the association between the NAFLD status in middle age with dementia development in later life, the study included subjects aged 40–69 years who ended up becoming about 50–79 years at the end of the follow-up. Based on an analysis of data from 4 031 948 subjects registered in the NHIS cohort, with an observation period of 9.5 years, the study found that the risk of dementia among NAFLD subjects was higher than that among non-NAFLD subjects. The results were consistent in multivariable-adjusted analysis, competing risk analysis and propensity score-matched analysis. The association between NAFLD and dementia risk was stronger among females than males and among non-obese NAFLD than obese NAFLD.

In recent years, some suggestions have been made regarding a possible link between NAFLD and cognitive impairment.[12,14] However, data on the association between NAFLD and dementia are scarce. A recent study that investigated the association between NAFLD and incident dementia concluded that there was no clear link between these diseases; however, the prevalence of NAFLD in the cohort of the study was only 3.3%, which was substantially lower than the global prevalence.[33] Besides, the potential competing risk of death was not considered in the analysis.[33] Nevertheless, this study found the association between NAFLD and an increased risk of dementia in a large cohort whose prevalence of NAFLD was 31%, comparable to that in previous reports.[34,35] The annual incidence rate of dementia among the study population was 1.08% which also corresponds to the rate previously reported.[36] The association of NAFLD and a higher risk of dementia was consistent when clinical variables and competing risk of death were considered. Yet, there has been a study showing a conflicting result. Involving 656 biopsy-proven NAFLD patients and 6436 matched controls, the study found no association between NAFLD and dementia even considering competing risk of death.[37] Thus, future studies validating the findings on the relationship between NAFLD and dementia should follow.

Although our study found that NAFLD is associated with an increased risk of developing dementia, it was not possible to determine the mechanism responsible for this association. The possible underlying mechanism may be related to the results of a study which demonstrated that PPARα signalling in the hippocampus is linked to the control of fatty acid metabolism in the liver.[15,20] Studies also indicate that the initiation and progression of Alzheimer's disease is associated with aberrant cholesterol metabolism and inflammation via liver X receptors (LXR), which regulate neuroinflammation and decrease amyloid-B peptide accumulation.[16] Another possible explanation may be that liver steatosis and dementia share some risk factors, including diabetes, obesity and physical inactivity.[13] In addition, the potential role of vascular damage among NAFLD patients in the context of metabolic syndrome can be considered as a possible pathogenetic factor of developing dementia.[38,39] Future studies should follow to elucidate the underlying mechanism behind the association between NAFLD and an increased risk of dementia.

When the association of NAFLD with dementia risk was analysed based on BMI, little difference was observed in the development of dementia by the status of NAFLD among obese patients. Among the non-obese patients, however, patients with NAFLD showed higher risk of dementia compared to those without NAFLD. Although non-obese NAFLD subjects are generally considered to carry a less severe form of liver disease than obese NAFLD subjects, this notion has recently been challenged.[8] Suggestions have been made that non-obese patients with NAFLD have higher mortality rates and increased morbidities.[8] A recent meta-analysis also showed that both non-obese and lean NAFLD patients had substantial long-term liver and non-liver comorbidities.[40] Future studies may shed more light on the relationship between NAFLD and dementia, especially for non-obese patients with NAFLD.

The association between NAFLD and dementia risk was more prominent among females in this study. This appears to be owing to the higher incidence rate of dementia observed among females. As for the male subjects whose death accounted for approximately 75% of all deaths, the incidence rate of dementia was relatively low, possibly because of the fact that they did not survive long enough to develop dementia. A study showed that the lifetime risk of dementia at age 45 was 1 in 5 among females but 1 in 10 among males which may have derived from the higher cardiovascular mortality among males.[41]

The study has several limitations. First, as a study based on observational data, it is not free of potential bias or confounding factors. We employed various approaches to overcome this challenge and included multivariable adjustment, competing risk analysis and propensity score-matching analysis. The use of multiple rigorous strategies helped adjust for the differences found at baseline and the tested outcomes. Second, the study employed HSI to diagnose NAFLD. Since HSI includes BMI and diabetes along with sex in its calculation, other risk factors for dementia rather than NAFLD might have given some influence to our results. The potential of NAFLD patients being classified as non-NAFLD or vice versa was not completely avoided. Yet, for a study involving such a large population, it was impossible to obtain a histological or radiology-based diagnosis of all subjects. Although there was an inevitable chance of misclassification, the prevalence of NAFLD in this study was 31%, which did not deviate significantly from that reported in previous studies. Further, HSI was developed and validated based on a large cohort of Korean individuals and is reproducible;[26] indeed, several previous studies have used it, and it has a relatively fair accuracy of fatty liver discrimination (AUROC 0.81).[25,26] Third, by using the G31 code in defining dementia, it is possible that the study may have included some patients with a mild neurological disorder in the dementia group. As this study was conducted based on claims and health check-up data available in the NHIS database, the study was not able to exclude cases in which patients with dementia were included among patients with G31 code. The study was not able to take into account the subtypes of dementia, such as Alzheimer's disease, Lewy body dementia, vascular dementia or frontotemporal dementia since the database used for this study was limited with the precise classification of dementia. Lastly, the study was not able to include the blood pressure change trajectories in its analysis.

To conclude, this nationwide study found that NAFLD was associated with an increased risk of dementia. The association between NAFLD and dementia risk was prominent among females and non-obese NAFLD subjects. Considering that the prevalence of both NAFLD and dementia is increasing, thereby resulting in substantial clinical and economic burdens worldwide, our findings suggest that it would be advisable to keep in mind the potential of dementia when treating patients with NAFLD. Future studies should investigate the causes of this association and verify the findings in other demographic settings.