Tocilizumab (TCZ) is the first drug with the ability to suppress interleukin-6 receptor (IL-6R)-dependent inflammatory reactions. It was first approved in 2005 by the Pharmaceutical and Medical Devices Agency (PMDA) for moderate to severe rheumatoid arthritis (RA), followed by approvals from the European Medicines Agency (EMA) in 2009 and the U.S. Food and Drug Administration (FDA) in 2010 for patients with an inadequate response to conventional disease-modifying antirheumatic drugs (DMARDs) (1, 2). Beyond RA, TCZ has since been investigated and approved for several other immune-mediated conditions, including cytokine release syndrome, juvenile idiopathic arthritis (JIA), systemic JIA (sJIA), and giant cell arteritis (GCA) (3).

In real-world clinical practice, patients with conditions such as RA or sJIA frequently present with complex complications and usually require the prolonged use of multiple medications in combination, rather than depending exclusively on TCZ as a single-agent therapy, due to the challenges of effectively managing the condition (4–6). TCZ is frequently used in combination with non-steroidal anti-inflammatory drugs (NSAIDs), glucocorticoids, and conventional or synthetic disease-modifying antirheumatic drugs (DMARDs) (1, 7). Although this combination strategy is essential for achieving disease control, it also raises important clinical concerns regarding potential drug–drug interactions (DDIs) that may reduce the therapeutic effect of the treatment or lead to unexpected adverse outcomes. Therefore, a comprehensive understanding of potential DDIs is necessary to optimize therapeutic safety in patients receiving TCZ-based regimens.

Several studies have reported these DDI risks. For example, a large-scale, multicenter clinical trial comparing TCZ monotherapy with TCZ combined with methotrexate exhibited the highest incidence of adverse events (AEs), with 92.6% (126/136) of patients experiencing at least one AE (8). Additionally, data from phase III and IV clinical trials further indicated that a higher incidence of elevated alanine transaminase and aspartate transaminase was observed in patients receiving TCZ in combination with methotrexate compared to methotrexate alone (9). Furthermore, the combined use of TCZ and high-dose glucocorticoids has been associated with an increased risk of serious infections in pediatric patients (3).

Despite these findings, the detailed data on safety associated with DDIs between TCZ and the drugs usually used clinically in combination are still limited in the real world, owing to its relatively short time on the market and inadequate accumulation of reliable clinical data. Thus, this study aimed to utilize the Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) database to investigate DDIs between TCZ and its frequently co-prescribed medications, including DMARDs, glucocorticoids, and NSAIDs, and provide evidence supporting the safety of TCZ for clinical decision-making. Notably, the FAERS database serves as a critical resource in this study, as it provides a large volume of real-world practical data derived from post-marketing AE reports, addressing the scarcity of real-world evidence in existing TCZ DDI research (10).

This is the first study to systematically evaluate the DDIs between TCZ and its most frequently co-prescribed medications in clinical practice. It provides a new reference for the safety monitoring of TCZ combined with 11 drugs in clinical practice, offering supportive information for clinical applications and contributing to the rationalization and safety improvement of medication use. The data for this study were obtained from the OpenVigil 2.1 platform, which provides streamlined access to the FAERS and offers significant advantages as a global spontaneous reporting database for pharmacovigilance research. These advantages include deriving real-world evidence from worldwide reports that enhances generalizability beyond clinical trial settings (14), providing an accessible analytical interface with preprocessed data that eliminates complex computational requirements (16), and possessing generation capacity for detecting rare adverse events and drug interactions often missed in premarketing studies. However, several limitations inherent to pharmacovigilance databases must be acknowledged, including potential underreporting, reporting biases, incomplete clinical information, and the inability to establish definitive causal relationships (17, 18). Therefore, results from OpenVigil should be considered as starting points for further research rather than as definitive proof of drug safety issues.

Although all four models belong to frequency-based signal detection methods, their underlying principles and statistical emphases differ. For example, the reporting ratio method is directly based on the number of reports, making it relatively intuitive but susceptible to bias. In contrast, the Ω shrinkage measure model reduces the influence of extreme values by introducing a shrinkage estimate, thereby improving stability in cases with small sample sizes. The combination risk ratio model integrates multidimensional risk ratio information and is theoretically more comprehensive, though it may be less stable when data are sparse. The chi-squared statistics model is based on the chi-squared test and is suitable for detecting signals with significant frequency differences; it shows a high degree of synergy with the Ω shrinkage model. Through pairwise consistency analysis (Table 3 and Figures 2a,b), we found that certain model combinations—such as the Ω shrinkage measure model and the chi-squared statistics model—exhibited strong agreement, suggesting a higher degree of synergy in signal identification. Therefore, in this study, we initially included all adverse events detected by at least one model for the analysis of drug interaction relevance. After obtaining the initial results, we focused specifically on those adverse events that were not only associated with drug interactions but also detected by at least two models.

Several AE records identified in this study, such as glossodynia, hand deformity, and rheumatic fever, were relatively rare. The results showed that all four frequency-based statistical models included in this study detected signals for the aforementioned rare AEs, and the corresponding RRR_diff values (relative reporting rate differences) for these AEs were significantly higher than the predefined threshold (RRR_diff >0), indicating a significant positive association with potential DDIs. Although these models are capable of identifying statistically significant associations from spontaneous reporting data, they cannot directly establish causality or elucidate definitive biological mechanisms. In addition, in spontaneous adverse drug reaction reporting systems such as the FAERS, the data originate from unstructured clinical practice reports. Consequently, the occurrence of extremely rare AEs (e.g., rheumatic fever) is more likely attributable to complex disease backgrounds, unrecognized concurrent infections, and ambiguous or erroneous diagnostic coding rather than a direct causal relationship with the drugs administered (19). Therefore, we advocate cautious interpretation of these AE signals and recommend that clinical practice place greater emphasis on common AEs that may have a causal relationship.

Glossodynia was primarily observed in reports involving the co-administration of tocilizumab (TCZ) and glucocorticoids, particularly prednisone. Although glossodynia is not a typical adverse reaction to TCZ or glucocorticoids, previous studies have documented glossodynia as an AE associated with tocilizumab (20). In addition, prolonged use of glucocorticoids may be associated with oral mucosal lesions or sensory disturbances, potentially through mechanisms involving local immunomodulation or neuroinflammatory processes (21). We suggest that, when patients report taste abnormalities,—especially during glucocorticoid therapy—clinicians should comprehensively evaluate their medication history and overall clinical status. Although current evidence remains limited, individual susceptibility or indirect effects of drug interactions cannot be excluded.

Hand deformity was predominantly identified among patients in the group receiving combination therapy of TCZ and disease-modifying antirheumatic drugs (DMARDs) (e.g., methotrexate, hydroxychloroquine, and leflunomide). Hand deformity is typically a late manifestation of rheumatoid arthritis (RA) resulting from disease progression. In our signal analysis, however, it was associated with an increased frequency of AE reports related to combination therapy. This association may reflect inadequate disease control, poor therapeutic response, or failure of the treatment regimen to effectively halt disease progression. It is unlikely to represent a direct drug-induced morphological alteration. Accordingly, we propose interpreting this as a potential monitoring signal for treatment response; when such reports emerge, clinicians should reassess the patient’s disease activity and the adequacy of the treatment regimen.

Rheumatic fever was associated with the concomitant use of TCZ and non-steroidal anti-inflammatory drugs (NSAIDs), particularly diclofenac. Rheumatic fever is an autoimmune inflammatory disorder triggered by infection with Group A β-hemolytic streptococcus (GAS), mediated through molecular mimicry-mediated immune cross-reactivity (22). Its appearance in AE reports is more plausibly attributed to complex underlying disease factors or misreporting. Nevertheless, its detection serves as a reminder for vigilance when NSAIDs are used for pain or inflammation control, as potentially unrecognized and uncontrolled infections (e.g., streptococcal) may precipitate immune-mediated systemic responses. We recommend that, for patients presenting with unexplained systemic inflammatory symptoms, potential infectious causes should be investigated and the indication for NSAID use carefully reassessed.

Notably, a strong positive correlation between anti-cyclic citrullinated peptide antibody (ACPA) positive and DDIs was observed when TCZ was used in combination with methotrexate, hydroxychloroquine, sulfasalazine, leflunomide, and diclofenac. Although anti-CCP is traditionally regarded as a diagnostic biomarker for rheumatoid arthritis (RA), it is categorized as a laboratory abnormality in the FAERS database and primarily classified under the “General Investigations” in SOC. When clinicians associate Anti-CCP positivity with drug therapy in reports (e.g., significant elevation of Anti-CCP titers), data abstractors map this to the corresponding MedDRA Preferred Term, categorizing it as an AE. This documentation reflects clinicians’ focus on treatment-related laboratory parameter changes rather than solely diagnostic utility. While Anti-CCP is a biomarker, its aberrant changes (e.g., treatment-emergent positivity or titer surges) may signal potential risks (e.g., treatment failure or disease progression), aligning with FAERS’ definition of an AE. To the best of our knowledge, there has been no relevant research for mechanism reported to date. ACPA is a specific marker for RA and is not only an indicator for the diagnosis but also a predictor of disease outcome (23). It was reported that higher levels of Anti-CCP antibody positivity were associated with more adverse clinical consequences (24). Therefore, patients with RA receiving the above combination therapy should be closely monitored for cutaneous and mucosal symptoms, with periodic reassessment of serological parameters, including ACPA.

When TCZ is combined with leflunomide in patients aged ≥18 years, attention should be paid to a potential decrease in therapeutic efficacy. A potential mechanism for the decreased treatment efficacy might be related to the formation of anti-TCZ antibodies; these antibodies have been reported as a common cause for lower drug serum concentrations and as poorer response to treatment (25). Previous studies have reported that 3–5% of patients were detected to have anti-TCZ antibodies (26). Moreover, in this study, co-administration of TCZ with hydroxychloroquine, leflunomide, sulfasalazine, or methotrexate showed the strongest DDI signals for increased C-reactive protein levels. The strongest DDI signals may be related to the production of anti-TCZ antibodies. Although treatment with TCZ dramatically reduced CRP concentrations in the plasma and CSF, the effect rapidly diminished in the weeks after treatment was stopped. Notably, a CRP reduction of <20% from baseline to 4 weeks of treatment on TCZ is a poor prognostic marker and should elicit considerations for changing treatment to another agent (28). In addition, fixed-dose might be a better dosing strategy for achieving a more consistent drug exposure compared to weight-based dosing (29).

Our study found that when TCZ is used in combination with ibuprofen or naproxen, joint swelling had the strongest correlation with drug interaction, and the incidence of joint swelling might be higher in males and adults. To the best of our knowledge, this result has not been reported in previous research. In contrast, TCZ inhibits the progression of joint damage by suppressing IL-6 signaling, while NSAIDs alleviate pain and inflammation in surrounding tissues through cyclooxygenase enzyme inhibition. This may be related to the TCZ binds to soluble IL-6R in a dose-dependent manner (30), and in patients with higher body weight or high disease activity, the standard dose (e.g., 8 mg/kg) may be insufficient to fully suppress the IL-6 signaling pathway, which may lead to the production of anti-TCZ antibodies in patients and result in higher disease activities. Therefore, it is essential that close monitoring be continued during dose reduction in the remission phase to ensure patient safety, while clinicians must remain vigilant for potential disease recurrence.

Sex is well known to influence risk, severity, and treatment outcomes of RA. Compared with male patients, female patients tend to exhibit higher disease activity scores, worse prognosis (24), and are less likely to achieve remission. Consistent with these findings, our study found that female patients showed a greater susceptibility to infusion-related reactions and infections when TCZ was combined with dexamethasone, as well as increased abdominal discomfort when co-administered with methylprednisolone. The most common AE associated with TCZ was infections, which may be due to TCZ being a direct inhibitor of the interleukin-6 receptor (IL-6), and inhibition of IL-6 was associated with immunosuppression (31). Nguyen et al. reported three comparable cases of patients with RA treated with TCZ who developed a severe prosthetic Staphylococcus aureus infection (32). Furthermore, C-reactive protein levels decrease rapidly after initiating TCZ treatment. During TCZ administration, both CRP and erythrocyte sedimentation rate (ESR) levels may remain low, potentially leading to infections that might not be recognized (33). It should be stressed that TCZ can suppress acute-phase reactions, such as fever and an increase in CRP, and induce transient decreases in neutrophil counts, thereby obscuring the signs and symptoms associated with infection, possibly leading to a delay in diagnosis (33). Therefore, the impact on CRP, neutrophils, and signs and symptoms of infections should be considered when evaluating a patient for a potential infection.

However, several noteworthy limitations inherent in this study should be acknowledged. First, the OpenVigil FDA platform lacks detailed patient-level information, especially regarding comorbidities and concomitant medication, which limits further assessment of patient disease status and AE severity. Second, the scope of entities eligible to submit AE reports is broad: not only healthcare professionals but also patients, their family members, patient representatives, and pharmaceutical manufacturers can submit AE reports. This may lead to duplicate or incomplete reporting. Third, in clinical practice, patients may concurrently use three or more suspect drugs. Finally, since some of the adverse events (AEs) included in this study lacked patient age and sex information in the database, and the database only allows filtering reports for a specific age range by setting minimum age and maximum age thresholds (in years), precise individual patient age could not be obtained. Thus, to some extent, it limited the subgroup analyses from comprehensively capturing all potential DDI-related AEs and may have introduced a certain degree of signal omission bias. However, the FAERS database can only extract AE information associated with the combination of two drugs, making it impossible to investigate interactions among three or more drugs. Despite the inherent limitations, analyses of real-world data, such as those collected in the FAERS database, may provide valuable reference information and complement findings from randomized controlled trials (34). Clinicians and pharmacists should remain vigilant for potential DDI-related AEs.

Healthcare providers should remain vigilant for potential DDI-related AEs when TCZ is used in combination with DMARDs, glucocorticoids, or NSAIDs. Particular attention should be given to signals of decreased treatment efficacy, which may be associated with the formation of anti-TCZ antibodies, as well as to musculoskeletal, cutaneous, and gastrointestinal events. Monitoring serological parameters (e.g., CRP and anti-CCP), skin/mucosal symptoms, and signs of infection is recommended during combination therapy to support medication safety.