Potential preventive role of low-dose methotrexate against incident recorded psychosis: a retrospective cohort study based on electronic health records.
Recent evidence suggests that low-dose methotrexate might have antipsychotic properties. However, it remains unknown whether low-dose methotrexate is associated with a reduced risk of incident recorded psychosis in real-world data, whether these associations extend to other putatively immune-related common psychiatric conditions, or whether similar associations are observed with other disease-modifying anti-rheumatic drugs (DMARDs).
In this retrospective cohort study using electronic health records (TriNetX US Collaborative Network), we identified adults with rheumatoid arthritis (age ≤45 years at treatment initiation; data extracted from 1 January 2000 to 21 December 2025). The 5-year risk of an incident recorded psychosis (primary outcome) and bipolar, depression, or anxiety disorders (secondary outcomes) were compared between low-dose methotrexate and each of 14 comparator drugs used in rheumatoid arthritis - three primary comparators (non-steroidal anti-inflammatory drugs; NSAIDs, naproxen, diclofenac and celecoxib) and 11 secondary and exploratory DMARDs comparators. Cumulative incidences and ratios of restricted mean time lost (rRMTL) are reported. Results were Bonferroni-corrected for multiple comparison.
Comparator cohort sizes ranged from 1161 to 21,445 (mean ages 33.7-37.4 years). For the primary outcome of psychosis, initiation of low-dose methotrexate was associated with lower 5-year incidence of newly recorded psychosis than initiation of naproxen (rRMTL 0.69, 95% CI 0.55-0.87) or diclofenac (rRMTL 0.71, 0.56-0.89); no significant difference was observed versus celecoxib or biologic DMARDs. For the secondary outcomes, low-dose methotrexate was similarly associated with lower 5-year incidence of newly recorded bipolar disorder, depression, and anxiety compared with non-selective NSAIDs naproxen and diclofenac, with the mood and anxiety associations extending to the selective cyclo-oxygenase-2 inhibitor celecoxib.
In a real-world rheumatoid arthritis cohort, initiation of low-dose methotrexate was associated with lower 5-year incidence of newly recorded psychosis, bipolar disorder, depression, and anxiety than initiation of non-selective NSAIDs. Given that extensive trials of broad anti-inflammatories have yielded limited psychiatric benefit, the differential profile of low-dose methotrexate is consistent with a mechanism beyond simple inflammation suppression. We hypothesise potentiation of regulatory T cell-mediated control of systemic inflammation and neuro-glial regulation. The active-comparator observational design cannot establish causation; these findings are hypothesis-generating and confirmatory inference will require interventional studies.
UK Research and Innovation (UKRI) Medical Research Council [grant number UKRI4403] Mental Health Platform. The National Institute for Health and Care Research (NIHR) Oxford Health Biomedical Research.
In this retrospective cohort study using electronic health records (TriNetX US Collaborative Network), we identified adults with rheumatoid arthritis (age ≤45 years at treatment initiation; data extracted from 1 January 2000 to 21 December 2025). The 5-year risk of an incident recorded psychosis (primary outcome) and bipolar, depression, or anxiety disorders (secondary outcomes) were compared between low-dose methotrexate and each of 14 comparator drugs used in rheumatoid arthritis - three primary comparators (non-steroidal anti-inflammatory drugs; NSAIDs, naproxen, diclofenac and celecoxib) and 11 secondary and exploratory DMARDs comparators. Cumulative incidences and ratios of restricted mean time lost (rRMTL) are reported. Results were Bonferroni-corrected for multiple comparison.
Comparator cohort sizes ranged from 1161 to 21,445 (mean ages 33.7-37.4 years). For the primary outcome of psychosis, initiation of low-dose methotrexate was associated with lower 5-year incidence of newly recorded psychosis than initiation of naproxen (rRMTL 0.69, 95% CI 0.55-0.87) or diclofenac (rRMTL 0.71, 0.56-0.89); no significant difference was observed versus celecoxib or biologic DMARDs. For the secondary outcomes, low-dose methotrexate was similarly associated with lower 5-year incidence of newly recorded bipolar disorder, depression, and anxiety compared with non-selective NSAIDs naproxen and diclofenac, with the mood and anxiety associations extending to the selective cyclo-oxygenase-2 inhibitor celecoxib.
In a real-world rheumatoid arthritis cohort, initiation of low-dose methotrexate was associated with lower 5-year incidence of newly recorded psychosis, bipolar disorder, depression, and anxiety than initiation of non-selective NSAIDs. Given that extensive trials of broad anti-inflammatories have yielded limited psychiatric benefit, the differential profile of low-dose methotrexate is consistent with a mechanism beyond simple inflammation suppression. We hypothesise potentiation of regulatory T cell-mediated control of systemic inflammation and neuro-glial regulation. The active-comparator observational design cannot establish causation; these findings are hypothesis-generating and confirmatory inference will require interventional studies.
UK Research and Innovation (UKRI) Medical Research Council [grant number UKRI4403] Mental Health Platform. The National Institute for Health and Care Research (NIHR) Oxford Health Biomedical Research.
Authors
Corsi-Zuelli Corsi-Zuelli, Taquet Taquet, Deakin Deakin, Upthegrove Upthegrove
View on Pubmed