Safety of TNF-α inhibitors therapy in patients with rheumatoid arthritis: an umbrella review

Summary

Background

Rheumatoid arthritis (RA) is a chronic systemic autoimmune disorder that is characterized by progressive synovial inflammation and leads to joint destruction, functional decline, and reduced life expectancy. Although tumor necrosis factor-alpha inhibitors (TNFi) have revolutionized RA management by preventing structural damage, their safety profiles remain controversial. Current safety assessments remain fragmented across isolated meta-analyses, failing to address competing risks or incorporate recent pharmacovigilance data. This study compared the adverse effects of anti-TNF therapy in RA and provided a stratified safety profile for optimal treatment selection.

Methods

The research team searched four databases, namely PubMed, Web of Science, EMBASE, and the Cochrane Library, for relevant articles published from January 2000 to March 2025. The review adhered rigorously to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines. For each meta-analysis, we recalculated effect sizes as either odds ratio (OR), relative risk (RR) or standardized mean differences (SMD), accompanied by their respective 95% confidence intervals (CIs). Methodological quality of included meta-analyses was evaluated with A Measurement Tool to Assess Systematic Reviews 2 (AMSTAR 2), while the certainty of evidence was assessed with the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. To address potential overlap among primary studies across meta-analyses, the investigators utilized the Graphical Representation of Overlap for Overviews (GROOVE) tool, followed by classification of the synthesized evidence into four levels based on predefined criteria. This umbrella review was prospectively registered on PROSPERO (CRD42025638409).

Findings

We synthesized results from 12 meta-analyses. The result indicated that TNF-α inhibitor therapy in RA patients was associated with (a) significantly increased risk of serious infection (OR = 1.63, 95% CI: 1.31, 2.04, p < 0.001, I² = 3%); (b) reduced risks of all cardiovascular events (RR = 0.60, 95% CI: 0.40, 0.89, p = 0.01, I² = 54%), myocardial infarction (RR = 0.75, 95% CI: 0.58, 0.98, p = 0.03, I² = 29%); and insulin resistance (SMD = −0.82, 95% CI: −1.38, −0.25, p < 0.01, I² = 92%). However, no significant associations were observed for malignant tumors (OR = 1.22, 95% CI: 0.83, 1.79, p = 0.30, I² = 0%) or insulin sensitivity (SMD = 1.31, 95% CI: −0.34, 2.96, p < 0.01, I² = 93%). Substantial heterogeneity existed in all main analyses.

Interpretation

TNF-α inhibitors are potentially associated with a higher risk of serious infection, while may have beneficial effects on cardiovascular events, myocardial infarction and insulin resistance but no association was found with other indications. However, the inconsistent methodological across studies reduces statistical power and limits the reliability of these findings, which must be interpreted cautiously.

Funding

This work was supported by National Key R&D Program of China (No. 2023YFB4606705), National Natural Science Foundation of China (No. 82272611, 82472522, 82072506 and 92268115), Hunan Provincial Science Fund for Distinguished Young Scholars (No. 2024JJ2089), Science and Technology Innovation Program of Hunan Province (No. 2023SK2024) and Natural Science Foundation of Hunan Province (No. 2023JJ30949).

Research in context.

Evidence before this study

We searched four databases: PubMed, Web of Science, Cochrane Library and Embase from January 2000 to March 2025. Inclusion criteria were as follows: Clinically diagnosed rheumatoid arthritis (RA), anti-tumor necrosis factor (anti-TNF-α) therapy evaluation, and prespecified outcomes; exclusion criteria comprised: non-primary research, duplicate/short-follow-up studies, lack of controls, or unavailable full texts. We assessed the quality of included studies using standardized tools, with most studies rated as moderate or higher. Previous syntheses examined isolated adverse events, which are inadequate to balance competing risks. To address these shortcomings, our study analyzed randomized controlled trials and focused on comprehensive outcomes to investigate the safety of anti-TNF therapy in RA patients.

Added value of this study

This study adds significant value to existing evidence by accessed the qualities of included meta-analyses (MAs) and integrated and analyzed multiple outcomes including the risk of malignancies, the risk of serious infections and cardiovascular events. The findings combined with existing evidence, suggests that TNF-α inhibitors are potentially associated with a higher risk of serious infection, while they may have beneficial effects on cardiovascular events, myocardial infarction and insulin resistance. Further evidence is needed to clarify the effect of anti-TNF therapy on the risk of malignant tumors and insulin sensitivity in RA patients.

Implications of all the available evidence

Our findings may have implications for clinical practice as they provide more integrated evidence of safety of treatment strategies for patients at risk of exposures. It is noting that the inconsistent methodological across studies reduces statistical power and limits the reliability of these findings, which must be interpreted cautiously.

Introduction

Rheumatoid arthritis (RA) manifests as a chronic systemic autoimmune disorder characterized by synovial inflammation that destroys joints progressively. At diagnosis, patients frequently present with joint deformities and progressive functional impairment,1, 2, 3, 4, 5, 6 diminishing their quality of life, and reducing life expectancy by 3–10 years.^3,7 The disease exhibits an annual incidence of 3 per 10,000 individuals and a prevalence of 1%. Both measures increased with age, peaking between 35 and 50 years.⁸ The severe clinical features of RA heavily burden healthcare systems and incur substantial direct medical costs throughout its progression.^6,9

Pharmacological management of rheumatoid arthritis (RA) involves three main therapeutic classes: nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroids, and disease-modifying antirheumatic drugs (DMARDs). DMARDs are further categorized into conventional synthetic agents (e.g., methotrexate) and biological agents, including TNF-α inhibitors (TNFi) and other biologics.^8,10, 11, 12 As a cornerstone of biological therapy, anti-tumor necrosis factor (anti-TNF) agents specifically target the proinflammatory cytokine TNF-α, effectively reducing synovial inflammation and protecting joint structures to mitigate disease progression.^13,14 Five TNF-α inhibitors—adalimumab, golimumab, infliximab, certolizumab, and etanercept—currently hold clinical approval for rheumatological use.¹⁵

However, TNF-α serves as a pivotal cytokine that orchestrates physiological inflammatory responses and immune regulation in healthy individuals. It initiates inflammatory cascades by binding to TNFR1/2 receptors, activating NF-κB and MAPK pathways to induce expression of secondary cytokines (IL-1β, IL-6) and adhesion molecules (VCAM-1/ICAM-1), facilitating leukocyte recruitment to sites of infection or injury.16, 17, 18 TNF inhibitor therapies for RA inevitably produce side effects including serious adverse events, heightened cancer risk, serious infections (including tuberculosis), heart failure, insulin resistance, and bone marrow suppression.^2,19, 20, 21, 22 Although some meta-analyses (MAs) have assessed the safety of anti-TNF therapy in RA patients, ongoing research advancements contradict earlier studies.^23,24 Moreover, existing syntheses predominantly adopted a compartmentalized approach to examine isolated adverse events. Which inadequately guides clinical decision-making when balancing competing risks. Meanwhile, emerging safety signals remain conspicuously absent form current risk-benefit assessments.

Therefore, a systematic synthesis of current evidence is imperative to reconcile existing discrepancies and propose novel mechanistic hypotheses for future investigation. The umbrella review methodology systematically incorporates large-scale evidence and applies rigorous overlap control, quantitatively tackling and reducing duplicate data inflation. This study aims to investigate the risk of adverse effects of anti-TNF therapy in RA patients by direct and indirect comparisons, providing a stratified safety profile to inform clinical decision-making for optimal treatment selection.

Methods

This umbrella review systematically synthesized evidence from multiple MAs across all relevant outcomes.^25,26 The study design compared individuals receiving anti-TNF therapy (experimental group) with control populations not receiving this treatment. Methodology followed the Cochrane Handbook guidelines for umbrella reviews26, 27, 28 and was prospectively registered on PROSPERO (CRD42025638409). The review adhered to both Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Supplementary material A) and A Measurement Tool to Assess systematic Reviews 2 (AMSTAR 2) Guidelines.^29,30 (Supplementary material B). Two independent reviewers performed all data retrieval, extraction, and evaluation procedures, with a third reviewer resolving any discrepancies through consensus review.³¹

Search strategy and selection criteria

A comprehensive search was conducted across four databases: PubMed, Web of Science, Cochrane Library and Embase, from January 2000 to March 2025. The search strategy utilized a combination of subject terms and free words to conduct the search. The search terms included “Tumor Necrosis Factor-alpha”, “Tumor Necrosis Factor Inhibitors”, “Arthritis, Rheumatoid”, and “Meta-analysis.”, etc. (Supplementary material C).

The initial screening of titles and abstracts was performed by a single reviewer to exclude clearly irrelevant studies. Subsequently, two reviewers independently assessed the full texts of potentially eligible studies. Included MAs met the following predefined criteria: (a) enrollment of participants with clinically diagnosed RA; (b) evaluation of anti-TNF-α therapies; and (c) reporting of prespecified outcomes including malignant tumors, severe infections, cardiovascular events (CVE), myocardial infarction (MI), and measures of insulin resistance/sensitivity, etc. Exclusion criteria comprised: (a) non-primary research publications (commentaries, guidelines, case reports, reviews, conference abstracts, and preprints); (b) duplicate studies with shorter follow-up durations; (c) studies lacking appropriate control groups; and (d) unavailable full-text articles.

Data extraction and quality assessment

Two independent reviewers performed data extraction and quality assessment, with a third reviewer adjudicating any discrepancies. Extracted data included: author names, publication year, sample sizes, and number of original studies included in each meta-analysis. Primary outcomes encompassed the risk of malignancies, the risk of serious infections (defined as infection that causes significant complications requiring antimicrobial therapy or hospitalization), all CVE (including all events like myocardial infarction (MI), congestive heart failure (CHF), and cerebrovascular accident (CVA)), MI, Homeostatic Model Assessment for Insulin resistance (HOMA) and Quantitative Insulin Sensitivity Check Index (QUICKI). Moreover, we documented funding sources of included studies to assess potential sponsorship bias, and collected all available follow-up data relevant to each outcome measure. For each meta-analysis, we recalculated effect sizes as either odds ratio (OR), relative risk (RR) or standardized mean differences (SMD), accompanied by their respective 95% confidence intervals (CIs). Methodological quality was evaluated by AMSTAR 2, with arbitration by a third reviewer to resolve disagreements and establish consensus.^26,32 To assess the quality and reliability of the MAs included in this umbrella review, we used the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) manual evaluation method to assess study design, risk of bias, consistency of results, indirectness, and uncertainty within the MAs. The evidence was then classified into four levels according to the criteria for evidence classification: I (Convincing evidence), II (Highly suggestive evidence), III (Suggestive evidence), IV (Weak evidence), and NS (Not significant). Detailed classification criteria are shown in sTable 1.²⁵

Overlapping discovery and processing

When multiple MAs evaluated identical outcome, overlapping original studies were identified using the Graphical Representation of Overlap for Overviews (GROOVE) tool.³³ This visualization tool divides overlap into four levels: Very high (>15%), High (10 to <15%), Moderate (5 to <10%) and Slight (<5%) (Supplementary material D). To address overlap, this study implemented the following hierarchical approach: (a) Cochrane reviews were prioritized over non-Cochrane reviews. (b) For non-Cochrane reviews with significant overlap (≥10%), the review with the highest AMSTAR 2 score was prioritized. If scores were identical, preference was given to the review with the most RCTs or the most recent publication. (c) For overlaps below 10%, all MAs were incorporated into a meta-meta-analysis.

Statistical analyses

For outcome measures with <10% overlap, meta-meta-analysis was conducted. Effect sizes and 95% confidence intervals (CIs) were pooled using a random-effects model, and results were visualized through forest plots. Heterogeneity was assessed using Cochran's Q and I² statistics, with I² values interpreted as follows: low (I²: <25%), low to moderate (I²: 25–50%), moderate to high (I²: 50–75%), or high (I²: >75%). Standardized mean differences (SMDs) were categorized as: less than 0.20 = small effect; 0.20 to 0.50 = medium effect; and greater than 0.50 = large effect. We assessed the statistical significance of the pooled effect size using p values, with p < 0.05 was considered statistically significant. All MAs were performed using Review Manager (version 5.4; The Nordic Cochrane Centre, Copenhagen, Denmark).

Ethics

Our data were obtained from publicly available published literature and did not involve the collection or processing of personally identifiable information; therefore, no ethical review was required. For the same reason, additional participant informed consent was not necessary.

Role of funding source

The funds obtained in this study were used to cover the costs associated with data collection and analyses, including expenses for data gathering, cleaning, and statistical processing.

Results

Search result

A total of 302 articles was initially retrieved according to the search strategy, 216 of which were excluded for duplicates, and 61 studies were excluded by reading the titles and abstracts. Through reading the full text, 13 more studies were excluded (Supplementary material E). Finally, 12 studies were included. Fig. 1 depicts the literature screening procedure.

Fig. 1.

Identification of studies via databases and registers.

Study characteristics

In this article, the included studies were all published between 2000 and 2025. A total of 12 studies were included in the research, with sample sizes ranging from 297 to 236,525 participants, and follow-up durations spanning 2 months to 5 years. Table 1 presents basic information about all the included studies. All MAs reported outcomes in patients with RA with and without TNFi treatment. Regarding malignancy risk,^34,36 two MAs reported no significant association between TNFi therapy and overall cancer incidence in RA patients, though Moulis et al.³⁶ cautioned that prolonged exposure risks remain uncertain. Three MAs reported the risk of serious infections.^23,24,34 Bongartz et al.²⁴ and Liu et al.²³ reported a significant increase in serious infections but Thompson et al.³⁴ did not find apparent heterogeneity. Two MAs reported the risk of all cardiovascular events (CVEs) and myocardial infarction (MI).^22,35 According to Barnabe et al.,³⁵ anti-TNF therapy is relevant to a reduced risk of all cardiovascular events and MI. Roubille et al.²² also reported positive cardiovascular effects of anti-TNF therapy in RA. One study reported the clinical efficacy of TNF antagonists in peripheral insulin resistance and/or sensitivity quantified. Lim et al.³⁷ reported an alleviation in insulin resistance, but also reported a nonsignificant impact on insulin sensitivity in RA patients.

Table 1.

Basic information of the included studies.

Author	Year	Region	No. of studies included	Sample Size	Anti-TNF–Treated Participants	Controls	Populations	Follow-up Duration	Outcomes
Bongartz, T24	2006	USA	9	5005	INF + MTX/DMARD ADA + MTX/DMARD	Placebo Placebo + MTX	RA patients only	12 weeks	The risk of serious infections and malignancies
Thompson, A. E34	2011	UK	6	3419	INF + MTX ADA + MTX ETA + MTX	MTX	RA patients only	6–12 months	The risk of serious infections and malignancies
Barnabe, C35	2011	Canada	16	108,328	INF ADA ETA	Placebo/DMARD	RA patients only	21 weeks–5 years	All cerebrovascular events, myocardial infarction
Moulis, G36	2012	France	33	23,555	INF/ADA/ETA/GLM/CZP + MTX	Placebo	RA patients only	12–104 weeks	The risk of malignancies
Roubille, C22	2015	Canada	28	236,525	ETA/INF/CZP/GLM mono ETA/INF/CZP/GLM + MTX ETA/INF/CZP/GLM + DMARD	MTX/DMARD	RA patients only	NA	All cerebrovascular events, myocardial infarction
Liu, X. L23	2024	China	21	82,470	TNFi	No TNFi	RA patients only	NA	The risk of serious infections
Lim, W. S37	2024	Singapore	12	297	INF/ADA/ETA mono	NA	RA patients only	8–78 weeks	Insulin resistance and insulin sensitivity

INF: infliximab; ADA: adalimumab; ETA: etanercept; GLM: golimumab; CZP: certolizumab; MTX: methotrexate; DMARD: disease-modifying antirheumatic drug; RA: rheumatoid arthritis; TNFi: Tumor Necrosis Factor inhibitors; NA: not access.

Results of umbrella review

Risk of malignant tumor

Five MAs reported the risk of malignant tumors,^{21,24,34,36,38} with three studies having a high AMSTAR-2 score,^21,36,38 one study having a moderate AMSTAR-2 score²⁴ and one study having a low AMSTAR-2 score.³⁴ We analyzed the overlap in MAs using the GROOVE tool and described the results in Supplementary material D. Due to the solution in the methodology, two MAs,^34,36 with a total of 39 studies included. The analysis conducted using a random effects model revealed statistically significant variations within the anti-TNF therapy and control groups (OR = 1.22, 95% CI: 0.83, 1.79, p = 0.30, I² = 0%) (Fig. 2).

Fig. 2.

Effect of anti-TNF therapy compared with control therapy on the risk of malignant tumor in patients with rheumatoid arthritis.

Risk of serious infections

There are three MAs^23,24,34 with a total of 34 studies included. The analysis using a random effects model showed statistically significant differences between the anti-TNF therapy and control groups (OR = 1.63, 95% CI: 1.31, 2.04, p < 0.001, I² = 3%) (Fig. 3).

Fig. 3.

Effect of anti-TNF therapy compared with control therapy on the risk of serious infection in patients with rheumatoid arthritis.

Risk of all cardiovascular events

There are two MAs^22,35 with a total of 21 studies included. Between the anti-TNF therapy and control groups, the analysis using a random effects model found statistically significant differences (RR = 0.60, 95% CI: 0.40, 0.89, p = 0.01, I² = 54%) (Fig. 4).

Fig. 4.

Effect of anti-TNF therapy compared with control therapy on the risk of all cardiovascular events in patients with rheumatoid arthritis.

Risk of myocardial infarction

There are two MAs^22,35 with a total of 12 studies included. The analysis using a random effects model showed statistically significant differences between the anti-TNF therapy groups and control groups (RR = 0.75, 95% CI: 0.58, 0.98, p = 0.03, I² = 29%) (Fig. 5).

Fig. 5.

Effect of anti-TNF therapy compared with control therapy on the risk of myocardial infarction in patients with rheumatoid arthritis.

Peripheral insulin resistance and/or sensitivity quantified

Three studies reported anti-TNF-α biologic therapy may improve insulin sensitivity and ameliorate insulin resistance, they all had a moderate AMSTAR-2 score. We analyzed the overlap in MAs using the GROOVE tool and described the results in Supplementary material D. By combining the methodology and the time-bound nature of the included studies, the results of Lim et al.³⁷ are more reliable. In the study, the use of anti-TNF-α biologics led to a significant reduction in insulin resistance by HOMA score (the Homeostasis Model Assessment of Insulin Resistance) (SMD: −0.82, 95% CI: −1.38, −0.25, p < 0.01, I² = 92%) but the impact on the risk of malignant tumor and insulin sensitivity remains uncertain (SMD: 1.31, 95% CI: −0.34, 2.96, p < 0.01, I² = 93%). The results were rated as Moderate and the level of evidence was Ⅳ (Supplementary material F).

Discussion

Our study reveals that TNF-α inhibitor therapy in RA patients was associated with significantly increased risk of serious infection but potentially have beneficial effects on cardiovascular events, myocardial infarction and insulin resistance. However, the therapy shows no significant associations with malignant tumors or insulin sensitivity. The risk of malignant tumor, serious infection, and MI displays no significant heterogeneity, confirming the reliability of these results.

This study compared its finding with existing literature. Nishanthi et al.¹⁰ noted a slight rise in serious infections and a decreased cardiovascular risk, yet found no increased cancer risk. According to Geiler et al.,¹³ RA patients exposed to TNF inhibitors face a heightened infection risk, including bacterial infections, respiratory and soft tissue infections. Although this exposure does not elevate overall malignancy risk, it potentially associated with an increased risk of lymphomas and haematological malignancies. Damjanov et al.³⁹ demonstrated that TNF inhibitor treatment reduces the first cardiovascular event by approximately 50%, with no apparent rise in solid malignancy risk. These findings closely align with our observations. However, He et al.² found that 10 anti-TNF therapies neither affect the risk of serious infections nor malignant tumors. They also reported not all TNF inhibitor therapies have a beneficial effect upon reducing the risk of acute cardiovascular and/or cerebrovascular events.

The scarcity of available data inevitably introduces potential heterogeneity into our findings, as showed by the pooled results of all cardiovascular events (I² = 54%). Some studies also report conflicting conclusions. These discrepancies may arise from methodological and clinical variabilities outlined below: Table 1 reveals notable variations in sample sizes and follow-up durations across the included studies. These studies used diverse TNF-α inhibitors with varying mechanisms of action - from monoclonal antibodies to receptor fusion proteins, while most permitted concomitant DMARD use, particularly methotrexate. Such combination therapies potentially weaken the observed treatment effects, potentially explaining both the residual heterogeneity in our pooled estimates and the conflicting conclusions from other studies.^40,41 For instance, our study included larger sample sizes, whereas He et al.² included a boarder literature collection and conducted subgroup analyses stratified by TNFi type and concomitant DMARD use. Furthermore, other potentially influential factors such as TNFi dosage regimens and patient characteristics (particularly comorbidities) remain inadequately assessed due to insufficient reported data. This data scarcity prevented meaningful subgroup analyses of these biologically significant effect modifiers in treatment response. Additionally, Lundh et al.⁴² found that industry-sponsored trials are significantly more likely to adopt study designs with higher risks of bias—particularly in outcome measurement and selective reporting—than independently funded research. Given the inconsistent reporting of conflict-of-interest disclosures across included trials, sponsorship bias potentially contributes to unmeasured heterogeneity. Consequently, although most pooled outcomes exhibited acceptable heterogeneity levels (I² < 30%), the uncertainty surrounding these unmeasured confounders warrants cautious interpretation of the results. Cardiovascular disorders (CVDs) dominate as the most prevalent comorbidity in RA patients. Prior literature identified RA as an independent risk factor for cardiovascular disease, remaining significant after adjusting for traditional risk factors (e.g., hypertension, diabetes). RA itself heightens cardiovascular morbidity and mortality driven by chronic inflammation accelerating endothelial dysfunction and plaque instability.43, 44, 45 Pooled data from Avina-Zubieta et al.⁴⁴ revealed that RA patients faces a 48% excess risk of incident cardiovascular disease (RR = 1.48, 95% CI: 1.36, 1.62) and a 68% elevated myocardial infarction risk (RR 1.68, 95% CI: 1.40, 2.03) compared with general population. Although TNF-α inhibitors substantially reduce this risk—our analysis shows a relative risk reduction—therapeutic intervention may achieve only partial risk normalization rather than absolute protection. This disparity highlights that while TNF inhibition modifies the inflammatory component of cardiovascular risk, adjunctive management of metabolic comorbidities and endothelial dysfunction remains imperative.

Despite the proven effectiveness of TNF inhibitors in rheumatoid arthritis treatment, they are not the preferred option in all clinical scenarios.⁴⁶ Conventional synthetic DMARDs, such as methotrexate, have been used to treat RA since the 1980s and often serves as the first line medication today. However, numerous studies link them with limited efficacy, increased mortality, considerable side-effects and high financial costs.^11,46,47 Aaltonen et al.⁴¹ compared the efficacy and safety of anti-TNF drugs with methotrexate, finding no significant differences in efficacy, yet TNF inhibitors, especially etanercept (ETN), showed a reduced risk of adverse events and emerge as the safest alternative, although methotrexate offered a cost advantage. In recent years, biosimilar have emerged to replace the high-cost TNF inhibitors, sharing a highly similar biomolecular structure but not an identical one.^1,48 The availability of many TNFi biosimilars continues to grow, with expectations that their introduction will lower drug costs. Multiple clinical trials have compared the efficacy and safety of TNFi biosimilars with the original TNFi drugs, revealing that TNFi biosimilars demonstrate efficacy and safety profiles closely resembling those of their reference biologics.49, 50, 51

The review faces several limitations: (a) the limited number of included studies prevents subgroup analyses of gender, individual TNF-α inhibitors or the type of disease, potentially introducing bias. (b) The varying quality of the included MAs raises concerns about bias. (c) This paper addresses very heterogeneous outcomes, each with different underlying pathophysiology and timelines. The grouping of these outcomes together without subgroup stratification (e.g., patient demographic, clinical characteristics, symptom duration and follow-up duration) affects the internal coherence of the review. (d) RA exhibits considerable heterogeneity in treatment response, with therapeutic options including TNFi, methotrexate, and targeted synthetic DMARDs (e.g., JAK inhibitors). Insufficient disaggregated data in the included studies restrict our analysis to evaluating composite adverse effects when these therapies combine. Future research should address these limitations to ensure better consistency.

In conclusion, TNF-α inhibitors are potentially associated with a higher risk of serious infection but may reduce cardiovascular events, myocardial infarction and insulin resistance. Further evidence is needed to clarify the effect of anti-TNF therapy on the risk of malignant tumor and insulin sensitivity in RA patients. However, the inconsistent methodological across studies reduces statistical power and limits the reliability of these findings.

Contributors

Concept and design: R. X., P.T.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Y.L.

Statistical analysis: P.T., Y.L., R.X.

Obtained funding: P.T., Y.L., R.X.

Administrative, technical, or material support: Y.L.

P.T., Y.L., R.X. have verified the underlying data.

All authors read and approved the final version of the manuscript.

Data sharing statement

The data that support the findings of this study are available on request from the corresponding author, upon reasonable request and with the provision of a data sharing agreement.

Declaration of interests

The authors declare that they have no competing interests.