Rheumatoid arthritis and the risk of dementia: A systematic review and meta-analysis
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.173623
Source of Support: None, Conflict of Interest: None
Background: The association between chronic inflammation and dementia has been identified in several epidemiologic studies. However, the data on rheumatoid arthritis (RA), one of the most common chronic inflammatory disorders, remains unclear.
Keywords: Alzheimer's disease; dementia, epidemiology; meta-analysis; rheumatoid arthritis
Dementia is a common disorder in older adults characterized by a decline in the cognitive function of at least one domain. A prevalence study in 2000 suggested that at least 4.5 million older adults in the United States suffered from this ailment. Alzheimer's disease (AD) is the most common subtype of dementia, followed by vascular dementia. Less common subtypes include dementia with Lewy bodies, frontotemporal dementia, and Parkinson's disease with dementia. Several cardiovascular risk factors, including diabetes mellitus, hypertension, dyslipidemia, smoking, and obesity, have been linked to an increased risk of cognitive impairment and dementia.,
The association between chronic inflammation and dementia has been identified in several epidemiologic studies., An autopsy study of patients with AD has found neuroinflammatory response in the cerebral cortex of patients with early-stage AD. This evidence suggests that chronic inflammation could be another risk factor for developing dementia.
Rheumatoid arthritis (RA) is one of the most common chronic systemic inflammatory disorders, with an estimated incidence rate of 5–50 per 100,000 person-years. Synovial joint is the primary site of involvement, though extra-articular diseases and cardiovascular complications from chronic inflammation are also well recognized.,
Patients with RA might be at an increased risk of developing dementia. This study aimed to comprehensively review and summarize all available literature on the association between these two conditions.
Two investigators (P.U. and K.W.) independently searched for the published studies indexed in MEDLINE and EMBASE database from inception to August 2015 using the search strategy that comprised terms for RA and dementia as described in supplementary material 1. References of selected retrieved articles were also manually reviewed. The inclusion criteria were as follows: (1) Cross-sectional, case–control, or cohort study comparing the risk of dementia in subjects with and without RA; (2) odds ratio (OR), relative risk (RR), hazard ratio (HR), or standardized incidence ratio (SIR) with 95% confidence intervals (CI) were provided, or sufficient raw data to calculate those ratios and the corresponding 95% CI were provided; and, (3) participants without RA were used as controls in the cohort study and cross-sectional study, while participants without dementia were used as controls in the case–control study.
Each investigator noted earlier independently determined the study eligibility. Quality appraisal was also performed by the two investigators using Newcastle–Ottawa quality assessment scale,, which assessed the study in three domains, including (1) the selection of the participants for each group, (2) the comparability between the groups, and (3) the ascertainment of the exposure (for case–control study) or the outcome of interest (for cohort and cross-sectional studies). The senior investigator (C.T.) oversaw this systematic review process and served as the deciding vote when the first two investigators had different decisions.
A standardized data collection form was used to extract the following information:First author's last name, title of the study, year of publication, year when the study was conducted, study design, country/state where the study was conducted, method used to identify and diagnose RA and dementia, average duration of follow-up (for cohort study), baseline demographic data of each group, confounders that were adjusted, and adjusted effect estimates with the corresponding 95% CIs. All investigators independently performed this data extraction, and the extracted data were cross-checked. Any discrepancy was evaluated by referring back to the original study.
Review Manager 5.3 software from the Cochrane Collaboration (London, United Kingdom) was used to calculate the pooled estimate effect. Point estimates and standard errors were extracted from each study and were combined by the generic inverse variance method of DerSimonian and Laird. As we planned to combine data from different study designs, a high between-study variance was expected. Therefore, the random-effect model rather than a fixed-effect model was used. As the outcome of interest in this study was relatively infrequent, (odds ratio) OR of case–control study and cross-sectional study was used as an estimate for relative risk (RR) to calculate the pooled estimate. Statistical heterogeneity was assessed by Cochran's Q test and I 2 index. A value of I 2 index of 0%–25% represents insignificant heterogeneity, >25% but ≤50% represents low heterogeneity, >50% but ≤75% represents moderate heterogeneity, and >75% represents high heterogeneity. Comprehensive Meta Analysis version 2.2 software (New Jersey, USA) was used to create funnel plot and to calculate Egger's linear regression test to assess the possibility of publication bias.
Our search strategy yielded 1,844 potentially relevant articles (1,573 articles from EMBASE and 271 articles from MEDLINE). After exclusion of 261 duplicated articles, 1,583 articles underwent title and abstract review. Of them, 1,422 articles were excluded in the first round of review, as they were case reports, editorials, review articles, correspondences, basic science studies, or interventional studies; 161 articles underwent a second round of abstract review, and 141 articles were excluded at this stage, as they were not conducted in patients with RA/dementia or did not report the association of interest, leaving 20 articles for a full-length article review. Nine articles were excluded because they were descriptive studies without a control group, while six case–control studies were excluded because they did not report the exposure of interest (RA). Three cohort studies and two cross-sectional studies met our eligibility criteria and were included in our data analyses.,,,, [Figure 1] outlines our literature search and review process. The characteristics as well as the quality assessment of the included cohort and cross-sectional studies are illustrated in Table 1] and [Table 2], respectively.
The pooled risk ratio of dementia in patients with RA versus the control subjects was 1.61 (95% CI, 1.10–2.37). The statistical heterogeneity was high, with an I 2 index of 91%. Subgroup analysis according to the study design revealed an increased pooled risk ratio in both subgroups even though the statistical significance was not reached in the cohort study subgroup (RR, 1.34; 95% CI, 0.93–1.94). The statistical heterogeneity was also markedly reduced when the point estimates were pooled according to the study design (I 2 index of 57% for cohort studies and 0% for cross-sectional studies). Forest plots of the overall analysis and subgroup analysis are illustrated in [Figure 2].
Evaluation for publication bias
Visualization of the funnel plot [Figure 3] did not provide suggestive evidence for publication bias, as the graph was symmetric. Moreover, there was no evidence of publication bias detected by Egger's regression test (P = 0.65).
Our study is the first study to provide a comprehensive evaluation on the association between RA and dementia. We found a statistically significant increase in the risk of dementia among patients with RA, with an overall 1.61-fold increased risk compared with non-RA controls. The relative risks from individual studies ranged from 1.15 to 2.77.
The pathophysiologic mechanisms underlying this increased risk are not well characterized. One possible explanation is related to the effect of chronic inflammation from RA on the brain, as several studies have suggested a significant role of inflammation in the pathogenesis of dementia. For example, increased proinflammatory cytokines have been shown to stimulate microglial activation and neuronal damage in vitro, and animal studies have demonstrated high levels of inflammatory cytokines in brains with neurodegeneration., Moreover, the association between dementia and elevated inflammatory markers has been found in several population-based epidemiologic studies.,
Accelerated atherosclerosis from RA is another possible underlying mechanism, as atherosclerotic cardiovascular diseases are well-established risk factors for dementia., The role of inflammation in the pathogenesis of atherosclerosis is well recognized. Several in vivo and in vitro studies have demonstrated that endothelial cell injury from inflammatory cytokines, activated inflammatory cells, and oxidative stress promoted the progression of atherosclerosis.,, Furthermore, epidemiologic studies have shown a higher incidence of cardiovascular disease among patients with autoimmune inflammatory disorders.,,,,
Nonetheless, we acknowledge that there were some limitations in this study, and thus, the results should be interpreted with caution. First, most of the included studies were medical registry-based studies, which would raise a concern over coding inaccuracy and misclassification. The study by Husni et al., used a pharmaceutical database to verify the diagnosis of RA and dementia, which raised a serious concern of incompleteness of case identification because milder cases of RA and dementia, who did not require pharmacological interventions, would not be detected. To address this concern, we conducted a sensitivity analysis by excluding this study from the meta-analysis. The new pooled RR slightly decreased to 1.38 but remained statistically significant (95% CI, 1.00-1.93). Second, the statistical heterogeneity was high in this study. We believe that the difference in study design was the main source of heterogeneity, as subgroup analysis according to the study design revealed markedly reduced I 2 statistics for both subgroups. Third, this is a pooled analysis of observational studies, which can only demonstrate an association but not causality. Therefore, we cannot conclude that RA itself or some unknown confounders were accountable for the elevated risk.
In conclusion, our meta-analysis demonstrated a significantly increased risk of dementia among patients with RA, with 61% excess risk. Further studies are required to better understand the underlying pathogenesis and to develop optimal strategy to manage this increased risk.
Financial support and sponsorship
Conflicts of interest
We do not have any financial or non-financial potential conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2]