Association between anti-tumor necrosis factor alpha exposure and new-onset autoimmune diseases in inflammatory bowel disease: a nationwide case-control study in Korea

Article information

Intest Res. 2026;.ir.2025.00195

Publication date (electronic) : 2026 February 10

doi : https://doi.org/10.5217/ir.2025.00195

Shin Ju Oh¹, Ji Eun Kim¹, Su Jin Jeong², Chang Kyun Lee^,1, the Big Data Research Group (BDRG) of the Korean Society of Gastroenterology

¹Department of Gastroenterology, Center for Crohn’s and Colitis, Kyung Hee University Hospital, College of Medicine, Kyung Hee University, Seoul, Korea

²Department of Statistics Support, Medical Science Research Institute, Kyung Hee University Medical Center, Seoul, Korea

Correspondence to Chang Kyun Lee, Department of Gastroenterology, Center for Crohn’s and Colitis, Kyung Hee University Hospital, College of Medicine, Kyung Hee University, 23 Kyungheedae-ro, Dongdaemun-gu, Seoul 02447, Korea. E-mail: changkyun.lee@khu.ac.kr

Received 2025 September 2; Revised 2025 November 21; Accepted 2025 December 3.

Abstract

Graphical abstract

INTRODUCTION

Tumor necrosis factor alpha (TNF-α) is a pro-inflammatory cytokine that plays a pivotal role in the inflammatory cascade underlying chronic immune-mediated inflammatory diseases (IMIDs) [1]. Since their introduction in 1991, synthetic anti-TNF-α antibodies have markedly improved clinical outcomes in conditions such as rheumatoid arthritis, ankylosing spondylitis, psoriasis, and inflammatory bowel disease (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC) [2]. Despite the emergence of novel therapies targeting specific cytokines or molecular pathways, anti-TNF-α agents remain a cornerstone of IBD treatment in routine clinical practice [3,4]. However, as highlighted in recent guidelines, the optimal use of these agents is an area of ongoing investigation. Recent real-world evidence suggests that certain subgroups of patients with IBD—such as those with concomitant ankylosing spondylitis—require anti-TNF-α therapy more frequently owing to increased disease burden and severity [5].

While the therapeutic efficacy of anti-TNF-α agents is well established, their use is associated with an increased risk of adverse events, notable infections and malignancies, necessitating ongoing surveillance during treatment [6-8]. More recently, paradoxical immune-mediated adverse events—where anti-TNF-α therapy appears to induce immune-mediated diseases rather than suppress them—have been increasingly recognized [9]. These adverse events include psoriasis, interstitial lung disease (ILD) [10-12], central nervous system (CNS) demyelinating disorder [13,14], lupus and systemic vasculitis [15,16]. Notably, most of these events have been documented in patients with rheumatic arthritis [17,18].

With the growing global prevalence of IBD and expanding use of anti-TNF-α agents, concerns about these paradoxical autoimmune phenomena have grown [19,20]. These events challenge the conventional understanding of TNF-α inhibition by suggesting that blocking one arm of the inflammatory pathway may inadvertently activate alternative immune responses or trigger autoantibody production. Importantly, these autoimmune conditions are conceptually distinct from the classical extraintestinal manifestations (EIMs) of IBD. Unlike EIMs, which are generally linked to underlying IBD activity, these conditions are believed to represent paradoxical, treatment-related autoimmune phenomena associated with TNF-α blockade. Distinguishing between these entities is essential for accurately interpreting the pathophysiological context of anti-TNF-associated autoimmune events. However, the underlying etiopathogenesis remains poorly understood, and population-level data specifically addressing autoimmune disease risk in IBD patients are limited. Understanding whether anti-TNF-α exposure contributes to the development of new-onset autoimmune diseases in patients with IBD is critical for personalized risk stratification and long-term management. Therefore, herein, we conducted a large-scale, nationwide, population-based case-control study to evaluate the association between anti-TNF-α therapy and incident autoimmune disease in patients with IBD.

METHODS

RESULTS

1. Baseline Characteristics and Risk of Autoimmune Disease

In total, 119,739 patients with IBD were included in the study after applying the exclusion criteria. Patients with only a diagnostic code and no prescription for IBD-specific medication (n =488,530) and those with a prior diagnosis of an autoimmune disease before their IBD diagnosis (n=5,935) were excluded from the initial dataset. Among the included patients, 8,586 were subsequently diagnosed with autoimmune diseases, such as psoriasis, ILD, SLE, systemic vasculitis, or CNS disease. These patients comprised the final case group (Fig. 1). The remaining 111,153 patients with IBD who did not develop any autoimmune diseases during the study period served as the control group. Baseline characteristics of the study population are shown in Table 1. The median age of the patients in the case group was 47 years, and 50.9% were men. The median duration from IBD diagnosis to the onset of autoimmune disease in the case group was 4.17 years (interquartile range, 1.55–7.68 years). Of the total patients, 11,294 (10.2%) in the control group and 1,183 (13.8%) in the case group were treated with anti-TNF agents. Demographic and baseline characteristics of the case and control groups were balanced using 1:1 matching. However, even after matching, the case group had higher burdens of comorbidity, more frequent hospitalizations, and greater use of immunosuppressants and anti-TNF-α therapies. In addition, 1,029 cases (12.0%) and 620 (7.2%) controls received combination therapy, which was defined as the simultaneous use of conventional therapy and anti-TNF-α agents for at least 90 days.

Fig. 1.

Study flow for the inclusion of cases and controls. ICD-10, International Classification of Diseases, 10th Revision; IBD, inflammatory bowel disease; UC, ulcerative colitis; CD, Crohn’s disease.

Table 1.

Baseline Characteristics of Cases and Controls in Study Population

Exposure to anti-TNF-α agents among patients with IBD was associated with a significantly increased risk of developing autoimmune diseases. Detailed analysis revealed an OR of 1.52 in the multivariable analysis, which was adjusted for several confounding factors, indicating a 52% higher risk in patients with IBD exposed to anti-TNF-α agents than in those not exposed to anti-TNF-α agents. This association remained consistent across the various statistical models. In the propensity score matched analysis, the risk remained elevated, with an OR of 1.65 (95% CI, 1.44–1.89; P<0.001), reinforcing the strength and reliability of the observed relationship. When individual autoimmune diseases were analyzed separately, certain conditions showed a particularly strong association with anti-TNF-α exposure. Specifically, psoriasis was associated with a 58% increase in risk (OR, 1.58; 95% CI, 1.34–1.86; P<0.001); ILD demonstrated an even higher risk, with an OR of 1.88 (95% CI, 1.33–2.66; P<0.001), indicating nearly double the risk as that for non-exposed individuals. In contrast, other autoimmune conditions, including CNS disease, SLE, and systemic vasculitis, did not show a statistically significant association with anti-TNF-α agent exposure in either the adjusted or matched models (Table 2).

Table 2.

Risk for Autoimmune Disease According to Anti-TNF-α Agent Exposure in Patients with IBD

We explored the dose-dependent association between anti-TNF-α exposure and the development of autoimmune disease by analyzing the cumulative number of anti-TNF-α prescriptions as a continuous variable (Fig. 2). Using a smoothing spline regression model, we observed a non-linear relationship, showing that the estimated risk increased sharply with initial exposure (mean risk, 1.52; 95% CI 1.33–1.71 for <30 prescriptions) and plateaued around 30 prescriptions (mean risk, 1.82; 95% CI, 1.64–2.01 for 30–64 prescriptions). The risk reached its highest point at approximately 64 prescriptions (risk, 1.83; 95% CI, 1.57–2.13) and showed a gradual decline beyond this point (mean risk, 1.69; 95% CI, 1.50–1.87 for >64 prescriptions), although the risk remained elevated compared to that at the lowest exposure levels.

Fig. 2.

Dose-response relationship between anti-tumor necrosis factor alpha (anti-TNF-α) use and autoimmune disease risk.

2. Stratified Risk by Immunosuppressant Exposure and Concomitant Therapy

Table 3 presents subgroup analyses stratified by any history of immunosuppressant use, irrespective of timing relative to anti-TNF-α initiation. In both strata, anti-TNF-α exposure was significantly associated with an increased risk of overall autoimmune disease, although the effect size was modestly attenuated among patients with a history of immunosuppressant use compared with that among non-users. For specific autoimmune conditions, this association was statistically significant only for psoriasis, and the relationship persisted in immunosuppressant users with a slightly reduced magnitude. No significant associations were observed between ILD, CNS disease, lupus, and vasculitis in either subgroup.

Table 3.

Risk of Autoimmune Disease Associated with Anti-TNF-α Agents, Stratified by Immunosuppressant Use

Table 4 presents the sensitivity analysis focusing on concomitant therapy, defined as ≥90 days of overlapping anti-TNF-α and immunosuppressant use. In multivariable logistic regression, patients receiving concomitant therapy demonstrated a significantly elevated risk of overall autoimmune disease compared with those without anti-TNF-α exposure (OR, 1.51; 95% CI, 1.22–1.87; P<0.001), and this association remained robust after propensity score matching (OR, 1.43; 95% CI, 1.04–1.98; P=0.029). Anti-TNF-α monotherapy also showed a trend toward increased risk relative to the no-exposure group but did not meet the conventional significance threshold (OR, 1.19; 95% CI, 0.99–1.42; P=0.050) (data not shown).

Table 4.

Difference in risk of Autoimmune Disease Associated with Anti-TNF-α Agents in Subgroup Analysis

DISCUSSION

This nationwide study explores the complex relationship between anti-TNF-α therapy and new-onset autoimmune diseases in patients with IBD. While anti-TNF-α agents are cornerstone treatments for IBD, providing substantial relief from inflammation and other debilitating symptoms, our findings highlight a significant association of anti-TNF-α therapy with new-onset autoimmune diseases. Patients exposed to anti-TNF-α agents had a 65% higher likelihood of developing an autoimmune disease compared to those with no such exposure, even after adjusting for multiple confounding factors. This risk was notably higher for specific conditions such as psoriasis and ILD. Stratified analyses showed that the increased risk of autoimmune disease with anti-TNF-α therapy was consistent regardless of prior immunosuppressant use and was further elevated with ≥90 days of concomitant use, suggesting a synergistic effect.

The occurrence of new-onset autoimmune disease following anti-TNF-α therapy has been documented across various populations and healthcare settings. According to Pérez-De-Lis et al. [9] from the BIOGEAS Registry, autoimmune phenomena have been increasingly reported with the widespread use of biological agents, particularly TNF-α inhibitors. Although these agents effectively target key immune pathways in diseases such as rheumatic disorders, cancer, and IBD, they also paradoxically induce autoimmune responses. Commonly reported conditions include psoriasis, CNS demyelination, ILD, and lupus. These findings are supported by large population-based studies reporting elevated risks of rheumatoid arthritis, psoriasis, and hidradenitis suppurativa after anti-TNF-α therapy in patients with IBD. A pooled hazard ratio of 1.76 was observed, indicating a 76% increased risk. Furthermore, an active comparator analysis against azathioprine monotherapy reaffirmed the heightened autoimmune risk linked to anti-TNF-α agents [24].

Our dose-response analysis revealed a non-linear association between cumulative anti-TNF-α exposure and the risk of autoimmune disease. The risk increased steeply with early exposure and plateaued after approximately 30 prescriptions, with a modest decline after 64. This pattern suggests that the emergence of paradoxical immune phenomena is more closely linked to early-phase immune dysregulation triggered by TNF-α blockade, rather than simply being a function of the cumulative dose. One possible explanation is that early exposure to anti-TNF-α agents may trigger immune dysregulation or unmask latent autoimmunity in genetically or immunologically susceptible individuals. Conversely, patients who continue long-term therapy without developing autoimmune complications may represent a subgroup with a more stable immune tolerance or less inherent susceptibility. Additionally, clinical selection bias may play a role, as patients who experience adverse immune reactions early in treatment are more likely to discontinue therapy, thereby enriching the long-term exposure group with individuals who are less prone to such effects.

In our study, anti-TNF-α exposure was significantly associated with psoriasis and ILD, with psoriasis showing a 58% increased risk, and ILD showing nearly twice the risk, compared to that in unexposed patients. Prior studies consistently report a link between anti-TNF-α therapy and psoriatic conditions in IBD and other IMIDs, but the risk varies by psoriasis subtype and patient characteristics [25]. A systematic review and metaanalysis by Xie et al. [26] reported an overall incidence of psoriasis and/or psoriasiform lesions at 6.0% (95% CI, 5.0%–7.0%). Bae et al. [27] also indicated a higher risk of psoriasis, particularly in younger male patients, which is somewhat different from the demographic characteristics highlighted by Xie et al. [26] However, the consistent increase in the risk of psoriatic disease across different populations underlines a common pharmacological effect, regardless of regional or demographic variations. Despite the evidence that patients on anti-TNF-α therapy may experience exacerbations of existing pulmonary conditions or develop new-onset ILD [12,28-30], the current incidence is likely underestimated owing to the rarity of disease, the short observation periods in randomized controlled trials, and the diagnostic methods used to identify anti-TNF-α-associated ILD. However, by utilizing a nationwide registry that provides a comprehensive and high-quality dataset representative of the Korean population, we were able to obtain significant evidence of an association.

Our study found no statistically significant association between anti-TNF-α exposure and CNS disease, lupus, or vasculitis. This aligns with data from the British Society for Rheumatology Biologics Registry in Rheumatoid Arthritis, which reported only a marginal, non-significant increase in CNS demyelinating events among anti-TNF-α users—an effect that disappeared in sensitivity analyses [16]. However, given the rarity of these events and diagnostic challenges, as highlighted in studies such as those by Kunchok et al. [13], who observed increased CNS risk in broader autoimmune population, underestimation remains possible in observational settings. These findings underscore the need for larger targeted studies and continued clinical vigilance, particularly for neurological and vascular adverse events, despite the absence of strong statistical signals.

The mechanistic pathways through which anti-TNF-α agents induce autoimmune diseases are not fully understood but are believed to involve the disruption of cytokine networks and immune homeostasis. Several hypotheses have been proposed to explain these mechanistic pathways. These include cytokine remodulation, which results in the overexpression of certain inflammatory molecules such as interferon and the interleukin (IL)-17/IL-23 axis [31]. This overexpression can initiate an inflammatory process, facilitating the activation and proliferation of pathogenic T cells, particularly in individuals genetically predisposed to such reactions. Additionally, the immunogenic nature of most TNF-α inhibitors often leads to the development of anti-drug antibodies and complex immune responses [32,33].

In our study, we evaluated whether the risk of autoimmune disease associated with anti-TNF-α therapy is influenced by the use of other medications, specifically immunosuppressants. When patients were stratified simply by the presence or absence of immunosuppressant use, both groups showed a significant increase in risk with anti-TNF-α therapy. Although the risk was slightly lower in the immunosuppressant group, this difference was not statistically significant, suggesting no clear protective or interactive effects. This finding implies that the small differences observed between the groups may be driven more by the underlying patient characteristics than by the pharmacological interaction of immunosuppressants. To examine this relationship more precisely, we performed an additional analysis categorizing patients into 3 groups: anti- TNF-α monotherapy, no anti-TNF-α exposure, and concomitant therapy. In this analysis, patients receiving concomitant therapy had a higher risk of autoimmune disease than those receiving anti-TNF-α monotherapy. This contrasts with some prior studies reporting a potential protective effect of combination therapy with anti-TNF-α agents and traditional immunomodulators [34]. This apparent discrepancy suggests the importance of treatment timing, cumulative exposure, and potential paradoxical immune reactions. Historical immunosuppressant use may reflect milder disease activity or an independent immunomodulatory effect; however, prolonged concurrent exposure to 2 potent immunosuppressive agents may synergistically enhance immune dysregulation, thereby increasing the risk of autoimmune events. These findings emphasize the need for careful consideration of combination therapy in clinical practice and suggest that the autoimmune risk profile differs depending on the nature and duration of immunosuppressant exposure. Our analysis focused on treatment exposures occurring prior to the onset of autoimmune disease, allowing us to more clearly evaluate the temporality between anti-TNF-α therapy and autoimmune outcomes. Further studies that systematically assess treatment modifications after autoimmune disease onset—such as discontinuation of anti-TNF-α therapy or switching to alternative agents—would strengthen the clinical interpretation of these findings. Incorporating post-outcome treatment trajectories will also be important to clarify how therapeutic decisions evolve after autoimmune complications and how they influence long-term outcomes.

Despite these insights, this study had certain limitations that must be acknowledged. First, despite the use of robust propensity score matching, the observational nature of this study limits our ability to infer definitive causal relationships. Second, this study relied on claims data, which, although comprehensive, may be prone to diagnostic misclassification or incomplete coding. In particular, the diagnoses of autoimmune diseases are based on ICD-10 codes without confirmation by histopathology or specialist assessment, raising the possibility of misclassification. Third, although we adjusted for multiple potential confounders, residual confounding from unmeasured variables, such as family history, genetic susceptibility, and prior immune-related events, cannot be fully ruled out. Additionally, in this study, EIMs commonly associated with IBD, such as arthritis and skin involvement, were intentionally excluded to focus on the risk of other new-onset immune-mediated diseases. While this approach allowed for clearer attribution of autoimmune outcomes to anti-TNF-α exposure, it may underestimate the full spectrum of immune-related adverse events in clinical practice. Furthermore, during the study period, reimbursement for newer biologics and small-molecule agents in Korea began only near the end or after our observation window, preventing meaningful comparison across different biologic classes. This reimbursement context also limited our ability to distinguish biologic-naive from biologic-experienced patients. Because nearly all anti-TNF-α users in our study initiated treatment before other agents became reimbursed or were routinely used in clinical practice, they can be reasonably regarded as biologic-naive at the time of initial exposure. Rare non-reimbursed or trial-based use of other biologics could not be identified in claims data, but such cases are expected to be exceedingly uncommon and unlikely to influence our findings. We acknowledge that some patients may have become biologic-experienced by switching from one anti-TNF-α agent to another, which could theoretically influence risk profiles. Our study did not stratify analyses based on anti-TNF-α switching status. Although our dose-response analysis indicates that the highest risk occurs during the earliest phase of anti-TNF exposure, we were unable to determine whether switching confers additional risk beyond initial exposure. Future studies with detailed longitudinal tracking of biologic sequencing, including anti-TNF-α switching and later transitions to newer biologic agents, will be important to clarify risk differences between biologic-naive and biologic-experienced patients.

In conclusion, our nationwide analysis demonstrated that anti-TNF-α therapy is significantly associated with an increased risk of new-onset autoimmune diseases in patients with IBD, particularly psoriasis and ILD. This risk persists regardless of immunosuppressant use and is further elevated with prolonged combination therapy. These findings highlight the need for individualized risk-benefit assessments and close monitoring of paradoxical immune responses, especially during the early phase of treatment. Future prospective studies incorporating integrative multi-omics approaches are needed to validate these associations and better elucidate the temporal dynamics and individual susceptibility underlying these paradoxical immune responses.

Supplementary Material

Supplementary materials are available at the Intestinal Research website (https://www.irjournal.org).

Article information Continued