Having explored the pathophysiological mechanisms through which DMARDs may mitigate cognitive decline, it is essential to assess how these theoretical benefits translate into clinical outcomes.

Both csDMARDs and bDMARDs have shown the potential to reduce systemic inflammation, offering protection against neuroinflammation and cognitive impairment. For instance, a population-based case-control study from Taiwan examined the impact of DMARDs on dementia risk in 20,707 patients with RA compared with 62,121 matched individuals without RA. The study found that only 1.2% of patients with RA developed dementia, compared with 2.0% of participants who did not have RA. Notably, DMARD users exhibited a significantly lower incidence of dementia (0.7%) than non-users (2.3%) (adjusted hazard ratio (HR): 0.63, 95% CI 0.55–0.72). A dose-response relationship was observed, with the highest cumulative doses providing the greatest protection, particularly in women and patients under 65 years. However, all DMARDs were grouped together, potentially obscuring differences between individual drugs [45].

To fully understand whether these treatments impact cognitive functions, we must further examine studies investigating the effects of DMARD therapies on cognition in patients with RA and related conditions.

To provide a comprehensive overview, Table 2 summarizes the key studies investigating the impact of DMARD therapies on dementia risk in patients with RA.

Table 2. Summary of studies on DMARD exposure and cognitive function outcomes in patients with rheumatoid arthritis7.1 Effect of Conventional Synthetic DMARDs and Biologic DMARDs on Cognitive Function

A population-based cohort study examined dementia risk factors in 886 patients with RA with a mean age of 65 years [46]. The cohort was followed for a median of 8.5 years, during this period, 103 individuals developed dementia, resulting in a cumulative incidence of 7.0% at 10 years and 16.1% at 20 years. Key risk factors included older age at RA onset, which increased dementia risk by 1.14 times per year, as well as the presence of rheumatoid nodules, hypertension, and large joint swelling, all of which doubled dementia likelihood. Cardiovascular comorbidities such as ischemic stroke and heart failure, along with psychological factors such as anxiety and depression, were also significantly associated with increased dementia risk. In terms of treatment, 47.5% of patients used methotrexate, 40.1% used hydroxychloroquine, and 6.3% used bDMARDs. However, no significant associations were found between csDMARDs (methotrexate: HR 1.17, 95% CI 0.72–1.88; hydroxychloroquine: HR 0.72, 95% CI 0.46–1.13; or other DMARDs: HR 1.19, 95% CI 0.68–2.07), bDMARDs (HR 1.18, 95% CI 0.52–2.67), glucocorticoids (HR 1.48, 95% CI 0.86–2.56), or NSAIDs (HR 1.07, 95% CI 0.50–2.28), and dementia risk [46]. The low rate of bDMARD use limited definitive conclusions regarding the impact on cognitive outcomes.

A subsequent retrospective cohort study focused on the association between specific RA treatments and dementia risk in 8925 matched patients with RA treated with TNFα-i versus methotrexate, utilizing data from two large US insurance databases [47]. This study suggests that there is no significant difference in the risk of dementia or AD between TNFα-i and methotrexate (Optum database: HR 0.69, 95% CI 0.45–1.05; Medicare database: HR 1.14, 95% CI 0.66–1.96), despite limitations such as a short follow-up and reliance on insurance claims.

Only one observational study, the RESIST study, has specifically investigated whether patients with RA with mild cognitive impairment treated with csDMARDs or bDMARDs experience slower cognitive decline [48]. This study enrolled 251 participants aged 55 and older, with 157 receiving csDMARDs and 94 receiving TNFα-i (etanercept, infliximab, golimumab, adalimumab, or certolizumab), and followed them for 18 months using the Free and Cued Selective Reminding Test with Immediate Recall (FCSRT-IR) and the MoCA. The results indicated no significant changes in FCSRT-IR scores in either group, with mean differences of 0.7 (95% CI = – 0.4, 1.8) for csDMARDs and 0.7 (95% CI = – 0.7, 2.2) for TNFα-i. Similarly, MoCA scores showed no significant differences, with mean differences of 0.5 (95% CI = – 0.2, 1.1) for csDMARDs and 0.6 (95% CI = – 0.1, 1.3) for TNFα-i, suggesting protective effects on cognitive functions from both drug classes. A significant proportion of participants was also on methotrexate (76.7% in the csDMARD group and 60.3% in the TNFα-i group), which may confound results owing to its potential protective effects against dementia. Baseline differences, including longer RA duration and higher FCSRT-IR scores in the TNFα-i group, may have influenced outcomes.

7.2 Effects of Conventional Synthetic DMARDs on Cognitive Function

Conventional synthetic DMARDs represent the cornerstone of RA management and are considered first-line therapy [49, 50]; this class includes agents such as methotrexate, sulfasalazine, leflunomide, hydroxychloroquine, and azathioprine. While primarily used for their antiinflammatory and immunosuppressive properties, their role in modifying the risk of neurodegenerative diseases, including dementia, has become a growing area of research interest. Several observational studies have reported conflicting results on the association between csDMARDs and dementia risk.

Some studies have suggested that long-term use of csDMARDs might increase the risk of vascular dementia owing to the potential prothrombotic effects and impact on cerebrovascular function, which could exacerbate cerebrovascular pathology in predisposed individuals. Chou et al. reported 63% higher odds of dementia in patients with RA treated with csDMARDs compared with those not using csDMARDs, with an even higher risk specifically for vascular dementia (odds ratio (OR) 1.78, 95% CI 1.21–2.61) [31]. Conversely, the use of csDMARD was not associated with AD; in addition, other population-based studies reported that low-dose csDMARDs, such as methotrexate, may confer a protective effect. In a retrospective matched case-control study involving patients over 50 years of age with RA (486 dementia cases and 641 controls), it was found that prior methotrexate use was associated with a lower risk of dementia (OR 0.71, 95% CI 0.52–0.98); moreover, methotrexate use for more than 4 years was associated with the lowest dementia risk (OR 0.37, 95% CI 0.17–0.79) [51]. In the study of Zhou et al. (2020), the use of methotrexate use was found to be associated with a lower risk of AD and all-cause dementia (OR 0.58, 95% CI 0.56–0.60) compared with a control group receiving NSAIDs or steroids but not DMARDs, in patients diagnosed with RA [52]. In a large retrospective analysis (n = 82,828), Huang et al. found that the RA cohort had a lower risk of all-cause dementia (HR 0.63, 95% CI 0.55–0.72) compared with individuals who did not have RA, with an even more pronounced reduction observed in the subgroup of DMARD users (including both csDMARDs and bDMARDs) (HR 0.48, 95% CI 0.39–0.58) [45]. In this regard, the role of bDMARDs and tsDMARDs, either in combination with or as alternatives to csDMARDs, remains unclear owing to conflicting evidence across different studies [47, 53, 54].

The potential role of csDMARDs to modify the course of established dementia has been explored less extensively than their impact on dementia risk. Most available evidence is derived from small observational studies or post hoc analyses in patients with RA with mild cognitive impairment or early stage dementia [10]. Some studies have suggested that methotrexate is associated with mood changes and confusion and may accelerate cognitive decline in patients with pre-existing dementia [15, 55, 56], potentially through mechanisms related to folate metabolism disruption [57]. However, this effect may be counterbalanced by the positive impact of reduced systemic inflammation on cognitive function, and there is currently no definitive evidence supporting this association [15, 55, 56]. Overall, the evidence suggests that the relationship between csDMARDs and dementia risk is complex and likely influenced by factors such as drug type, dosage, treatment duration, adjunctive therapies, and the presence of comorbid conditions.

7.3 Effects of Biologic DMARDs on Cognitive Function

Although prior research has suggested an increased risk of dementia, particularly vascular dementia, with csDMARD use, and no clear difference between classes of DMARDs, the evidence on their cognitive impact remains inconsistent. Several studies have investigated whether bDMARDs, particularly TNFα-i, can reduce the risk of dementia or cognitive decline in patients with RA. Findings have been inconsistent; some research shows potential cognitive benefits, while other studies report no significant associations. In the following sections, key findings from clinical studies assessing these cognitive outcomes will be reviewed.

Several investigations have demonstrated significant positive effects, particularly with TNFα-i, in reducing dementia risk and preserving cognitive function in patients with RA. These studies differ in terms of data collection methods, patient populations, and specific cognitive outcomes, but together, they offer evidence that bDMARDs may provide protective cognitive benefits. One significant difference in these studies, compared with previously mentioned studies showing comparable effects among all DMARDs, is their scale and the use of national databases to investigate dementia risk in large populations of patients with RA. This broader approach facilitates a more comprehensive dataset; however, it may obscure the specific effects of individual treatments.

Two studies specifically examined the impact of TNFα-i on dementia risk in patients with RA, each with a different population and approach, but both suggested a potential trend toward reduced dementia risk with TNFα-i use. A nested case-control analysis, which included 13,474 TNFα-i users and 3774 csDMARD users from the British Society for Rheumatology Biologics Register, reported a lower incidence of dementia in the TNFα-i group (0.31 versus 0.85 per 1000 person-years), although it was not statistically significant [58]. A retrospective analysis of over 21,000 patients with RA from Northern Ireland found a non-significant 16% lower risk of dementia in TNFα-i users [54]. Both studies faced limitations, considering the relatively small number of dementia cases, which reduced statistical power.

Other studies including larger cohorts reported reductions in dementia risk with TNFα-i use, with etanercept consistently showing the strongest effect, while other TNFα-i such as adalimumab have shown more variable results. To explore these effects more specifically in patients with RA, two key studies focused on the impact of TNFα-i therapy on dementia risk, offering valuable insights. A nested case-control design using data from over 8.5 million adults in a US insurance database examined the relationship between RA treatments and AD risk [59]. This study focused on patients with RA treated with DMARDs, including methotrexate, prednisone, and bDMARDs such as TNFα-i (including adalimumab, infliximab, and etanercept) and rituximab. Among the population, 9253 individuals (0.11%) were diagnosed with AD. The prevalence of AD was significantly higher in patients with RA (0.79%) compared with individuals who did not have RA (0.11%) (p < 0.0001). Anti-TNF agents were linked to a lower risk of AD (OR 0.45 95% CI 0.23–0.90). Etanercept showed the strongest protective effect (odds ratio (OR) 0.30, 95% CI 0.08–0.89), while adalimumab and infliximab did not show significant effects.

A large-scale retrospective case-control analysis reviewed electronic health records from over 56 million US patients to investigate the effects of TNFα-i on dementia risk across various inflammatory conditions, including RA and psoriasis [52]. Of this population, 28.22% were seniors over 65 years old. The study found that patients with RA had a significantly higher risk of dementia compared with those without RA. TNF-blocking agents, particularly etanercept (OR 0.34, 95% CI 0.25–0.47) and adalimumab (OR 0.28, 95% CI 0.19–0.39), significantly reduced AD risk. Infliximab also offered some protection (OR 0.52, 95% CI 0.39–0.69), although to a lesser extent. These agents similarly decreased the risk of broader dementia outcomes, with etanercept (adjusted OR 0.30, 95% CI 0.26–0.35) and adalimumab (adjusted OR 0.35, 95% CI 0.30–0.41) showing notable effects. Methotrexate was also associated with lower risks of dementia (adjusted OR 0.64, 95% CI 0.61–0.68), and combining it with TNFα-i, particularly etanercept (adjusted OR 0.53) and adalimumab (adjusted OR 0.61), provided even greater protection. Both studies had similar limitations regarding claims data and clinical codes for AD diagnosis, which could introduce misclassification errors. Neither study provided detailed data on treatment duration or adherence, which may affect long-term cognitive outcomes. The relatively young average age of participants in these studies, along with the short follow-up periods, also limited the generalizability of the findings, particularly in older patients with RA, where dementia risk is typically higher [52, 59].

A large-scale, retrospective cohort study involving US veterans extended the follow-up to 20 years, offering a comprehensive view of how TNFα-i treatment impacts dementia and AD risk [60]. This study examined TNFα-i both as a class and individually (including adalimumab, etanercept, certolizumab, golimumab, and infliximab), allowing for a deeper understanding of their specific long-term protective effects. In this population, the prevalence of AD was 0.60%, and 1.79% were diagnosed with dementia The incidence of dementia was 5% in veterans receiving TNFα-i treatment compared with 6% in those without. TNFα-i use was linked to a 36% reduction in dementia risk and a 43% reduction in AD risk. Adalimumab showed the strongest protective effect (HR 0.57, 95% CI 0.41–0.80), while etanercept also demonstrated a significant reduction in dementia risk (HR 0.72, 95% CI 0.52–0.98). The study indicated that TNFα-i use was consistently associated with a lower risk of dementia over follow-up periods of 5, 10, 15, and 20 years (HR 0.67–0.72, p < 0.05). Among patients aged 65 and older, TNFα-i use was associated with a significantly lower risk of dementia (HR 0.67, 95% CI 0.48–0.95) and AD (HR 0.57, 95% CI 0.39–0.83). Despite these findings, the study had limitations, including a predominantly male population and reliance on International Classification of Diseases (ICD) codes for dementia diagnosis, as well as not accounting for non-TNFα-i biologics, which were used less frequently in the cohort.

7.4 Recent Insights into Targeted and Biologic DMARDs on Cognitive Function

Recent studies have built upon earlier research by evaluating the effects of tsDMARDs and bDMARDs on dementia risk in patients with RA. By employing larger cohorts and more advanced methodologies, these studies provide a deeper understanding of how different bDMARD classes, including tsDMARDs, influence cognitive outcomes, broadening the perspective on their potential role in mitigating dementia risk.

A large-scale cohort study examined the risk of AD and related dementias (ADRD) in patients with RA treated with tsDMARDs, including TNFα-i [61]. Using a new-user, active comparator design, the study analyzed data from 22,569 matched pairs of older patients with RA starting treatment with tofacitinib, tocilizumab, or TNFα-i compared with those on abatacept. This design aimed to isolate the specific effects of other targeted therapies on ADRD risk. The participants had an average age of 72 years and were predominantly women, with most in the TNFα-i group being DMARD-naive. The incidence of ADRD ranged from 14 to 18 per 1000 person-years, dropping to 2–4 per 1000 person-years under stricter definitions. No significant differences in ADRD risk were found between the treatment groups, with HR ranging from 0.50 to 1.29. Although TNFα-i showed a potential reduction in ADRD risk in patients with preexisting cardiovascular disease (HR 0.76), this result was inconsistent and not statistically significant. Secondary outcomes related to AD also showed no significant differences. Strengths of the study included its large sample size and multiple analytic approaches, addressing biases in observational research. However, limitations included a relatively short follow-up period, which may have hindered the detection of long-term cognitive effects, and reliance on Medicare claims data, raising concerns about ADRD diagnosis misclassification. The use of abatacept as a comparator may have further reduced the ability to detect differences between treatments.

Building on previous findings about tsDMARDs and dementia risk, a large-scale study investigated the effects of bDMARDs and tsDMARDs on cognitive function in patients with RA [53]. Utilizing national claims data, the research assessed dementia risk in a substantial cohort compared with csDMARDs. The study analyzed 233,271 DMARD initiations among 141,326 patients with RA, with an average age of 67 years and a median follow-up of 3.1 years, during which 3794 dementia cases were recorded. The csDMARD group had 1711 cases (20.1 per 1000 person-years), while the b/tsDMARD group recorded 2083 cases (17.0 per 1000 person-years) After adjusting for confounding factors, b/tsDMARD use was associated with a 19% reduction in dementia risk (HR 0.81, 95% CI 0.76–0.87) [53]. Subgroup analyses showed similar reductions for patients using TNFα-i-bDMARDs, non-TNFα-i-bDMARDs, and tsDMARDs compared with csDMARDs. Moreover, b/tsDMARD users, regardless of prior methotrexate use, showed consistent reductions in dementia risk (HR 0.84) compared with csDMARD users. No significant differences in risk reduction were found on the basis of cardiovascular risk factors or age groups. While the large sample size and statistical adjustments strengthened the findings, limitations included reliance on claims data, which may have introduced misclassification biases, and a relatively short follow-up period of 3.1 years, which could have limited long-term risk detection. In addition, the study did not differentiate between dementia subtypes.