Janus Kinase inhibitors associated with reduced anterior uveitis risk in autoimmune diseases

Purpose

To evaluate the impact of Janus kinase inhibitors (JAKi) compared to anti-tumor necrosis factor-α antibodies (anti-TNF) on anterior uveitis (AU) risk in patients with autoimmune diseases.

Methods

This retrospective cohort study analyzed deidentified global electronic medical records from December 2011 to December 2023, with 2-year follow-up. Adults with ankylosing spondylitis, psoriasis, psoriatic arthritis, or inflammatory bowel disease newly prescribed JAKi or anti-TNF were included. After 1:1 propensity score matching, 3,332 matched pairs were analyzed. AU incidence was assessed using Kaplan-Meier analysis.

Results

JAKi was associated with lower AU risk compared to total anti-TNF (HR: 0.43, 95% CI: 0.20–0.92) and non-monoclonal anti-TNF (HR: 0.28, 95% CI: 0.11–0.68). In ankylosing spondylitis, JAKi showed lower risk versus all anti-TNF biologics: total (HR: 0.30, 95% CI: 0.10–0.91), non-monoclonal (HR: 0.33, 95% CI: 0.11–0.99), and monoclonal anti-TNF (HR: 0.26, 95% CI: 0.09–0.78).

Conclusions

JAKi use is associated with reduced AU incidence compared to anti-TNF therapies, particularly in ankylosing spondylitis, supporting JAKi as a potential treatment option for autoimmune-related uveitis.

Key Messages

What is known

• Anterior uveitis (AU) is a significant ocular complication in autoimmune diseases, particularly ankylosing spondylitis.

What is new

• JAKi treatment is associated with significantly lower AU risk compared to anti-TNF therapies (HR: 0.43, 95% CI: 0.20-0.92).

• The protective effect is most pronounced in ankylosing spondylitis patients across all anti-TNF categories.

• JAKi appears to offer superior protection against AU compared specifically to non-monoclonal anti-TNF (HR: 0.28, 95% CI: 0.11-0.68).

Introduction

Among non-infectious uveitis (NIU), acute anterior uveitis (AU) is the most prevalent, constituting approximately 81% of all cases. AU frequently serves as the initial manifestation of spondyloarthropathies (SpA), including ankylosing spondylitis, psoriatic arthritis, inflammatory bowel diseases such as Crohn’s disease and ulcerative colitis, and psoriasis [1–4]. The recurrent nature of AU and its potential complications present critical challenges in prevention, early diagnosis, and effective treatment [2].

Current therapies for SpA-related diseases, including AU, face substantial limitations. Even with monoclonal anti-TNF, which is considered first-line biologic therapy for SpA-related uveitis by international guidelines [4], frequent relapses, and complications persist, emphasizing the need for alternative therapeutic strategies [3, 5].

JAKi represents a promising class of targeted immunomodulatory agents for managing NIU, by inhibiting the JAK-STAT signaling pathway, which plays a central role in immune-mediated inflammation [6]. Preclinical studies in experimental autoimmune uveitis (EAU) models have demonstrated the efficacy of JAKi, particularly JAK1 inhibitor agents such as upadacitinib, in reducing inflammatory markers and preventing relapses [7, 8]. Early clinical evidence further supports their efficacy in managing refractory non-infectious inflammatory ocular diseases, including AU, with notable reductions in relapse rates [9].

Recent clinical trials have explored several novel therapeutic agents for NIU, including JAKi. However, these Phase 2/3 trials have primarily targeted non-anterior uveitis, despite AU being the most common form of autoimmune-associated ocular inflammation [10–12]. Current research is limited by significant methodological challenges, including small sample sizes, short follow-up durations, and heterogeneous assessment protocols [9, 13].

To address these gaps, this study aims to evaluate the efficacy of JAKi in reducing the risk of AU in patients with underlying autoimmune diseases. By directly comparing JAKi with established anti-TNF agents, this study aims to generate robust clinical evidence to guide future therapeutic strategies for refractory AU in the context of systemic autoimmune diseases.

Methods

The data used in this study was collected on January 13–14, 2025 from the TriNetX “Research” Network, which provided access to electronic medical records (diagnoses, procedures, medications, laboratory values, genomic information) from approximately 135,645,991 patients across 103 healthcare organizations. All data was obtained from TriNetX, a global federated network of electronic medical records encompassing approximately 150 million patients from over 200 healthcare organizations. All data underwent de-identification in compliance with Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule standards. This study received ethical approval from the Institutional Review Board (IRB) of Taichung Veterans General Hospital (IRB NO: CE24430C).

Study design

This retrospective cohort study examined adults aged 18 years or older identified via International Classification of Diseases, 10th Revision (ICD-10) codes for ankylosing spondylitis, psoriasis, psoriatic arthritis, or inflammatory bowel disease (including ulcerative colitis and Crohn’s disease) who received JAKi or anti-TNF. To reduce enrollment bias, we excluded patients with pre-existing NIU within 12 months prior to the index date. Notably, baseline use of anti-uveitis medications (e.g., steroids, csDMARDs) was not an exclusion criterion but was strictly balanced using 1:1 propensity score matching. The study population was divided into two cohorts: patients receiving JAKi (tofacitinib or upadacitinib) and those receiving anti-TNF (adalimumab, infliximab, golimumab, etanercept, or certolizumab pegol). The selection of tofacitinib and upadacitinib was based on their U.S. Food and Drug Administration (FDA) approval for treating ankylosing spondylitis, psoriatic arthritis, ulcerative colitis, and Crohn’s disease. Anti-TNF were further divided into two groups (monoclonal and non-monoclonal anti-TNF) and compared to the JAKi group. Monoclonal anti-TNF (adalimumab, infliximab, and golimumab) have an established therapeutic role in NIU treatment, supported by preclinical and clinical studies [13], particularly highlighted by adalimumab’s FDA approval for uveitis treatment [14]. The patient enrollment commenced on December 20, 2011, coinciding with the first FDA approval of tofacitinib, and concluded at the end of 2023.

Patients who initiated biologic therapy with either a JAK inhibitor or an anti-TNF were included, with the index date defined as the initial prescription. The primary outcome of interest was the occurrence of AU, identified via ICD-10 codes. Study participants were followed for 24 months from the index date until either the occurrence of AU or the end of the study period. To maintain treatment specificity and minimize confounding, we applied the following exclusion criteria: (1) diagnosis of non-infectious uveitis (NIU) within 12 months prior to the index date, to reduce confounding from preexisting ocular inflammation; (2) combination therapy (concomitant JAKi and anti-TNF); (3) initiation of other biologics (anti-IL-17, IL-23, IL-12/23, IL-6) after the index date; (4) crossover between JAKi and anti-TNF; and (5) lack of documented continuation of JAKi or anti-TNF therapy between 6 months and 2 years after the index date. The TriNetX platform captures longitudinal medication data using standardized RxNorm codes, enabling the reliable detection of medication changes. In this study, patients were identified and excluded if any non-index therapeutic class codes were detected following the index date, ensuring that risk estimates reflect clinical outcomes associated with single-class exposure.

Statistical analysis

A 1:1 propensity score matching (PSM) methodology was implemented to minimize baseline differences between the JAKi and Anti-TNF groups. Balance between matched groups was evaluated using standardized mean differences (SMD), with an SMD < 0.1 indicating negligible variation. Matching variables included demographic factors (age, sex, race) smoking status, clinical comorbidities (hypertension, diabetes mellitus, ischemic heart disease, vitamin D deficiency), disease activity markers (ESR, CRP levels), and prior medication use (glucocorticoids, csDMARDs). Variables included in the propensity score matching were selected based on their established or potential association with uveitis risk. Specifically, vitamin D deficiency and nicotine dependence were included as covariates because they have been identified as significant modifiable risk factors for non-infectious uveitis in several epidemiological studies [15, 16]. Notably, a previous diagnosis of uveitis was not included as a matching covariate because it served as an exclusion criterion, ensuring the analysis cohort was free of recent NIU at baseline. These covariates were collected within 6 months before the index date. We conducted stratified analyses to evaluate treatment efficacy across different medication classes and disease subtypes. Anti-TNF were categorized into three groups: monoclonal anti-TNF, non-monoclonal anti-TNF, and the combined total anti-TNF cohort. JAKi were stratified into tofacitinib (a non-selective JAKi) and upadacitinib (a selective JAK1 inhibitor) subgroups. Further stratification was performed based on underlying autoimmune conditions, including ankylosing spondylitis, psoriasis, psoriatic arthritis, ulcerative colitis, and Crohn’s disease. All comparative analyses maintained a 2-year follow-up period to ensure consistent assessment of outcomes.

Data extraction and statistical analyses were conducted in January and February of 2025 using the built-in functions of the TriNetX platform, which facilitate patient selection, data analysis, and result generation. To comply with HIPAA regulations, TriNetX displays event counts ≤ 10 as obscured values in the output interface; however, the platform’s statistical engine performs all computations using exact, individual-level data within a secure environment, and all reported HRs and confidence intervals are derived from the platform. Cox regression models were employed to estimate hazard ratios (HRs) and corresponding confidence intervals (CIs). Kaplan-Meier survival analyses were utilized to evaluate survival probability, with censoring applied for patients lost to follow-up. Patient demographics and diagnoses were determined using ICD-10 codes as detailed in Table 3 in Appendix. Our study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology statements for cohort studies (Table 4 in Appendix).

Table 3.

ICD-10 and other codes used in the analysis

Population
Ankylosing spondylitis	ICD-10-CM M45
Psoriasis	ICD-10-CM L40
Psoriatic Arthritis	ICD-10-CM L40.5
Ulcerative colitis	ICD-10-CM K51
Crohn’s disease	ICD-10-CM K50
Intervention and control
JAKi	Tofacitinib (RxNorm 1357536)Upadacitinib (RxNorm 2196092)
Anti-TNF	Adalimumab (RxNorm 327361)Infliximab (RxNorm 191831)Golimumab (RxNorm 819300)Etanercept (RxNorm 214555)Certolizumab pegol (RxNorm 709271)
Other JAKi	Baricitinib (RxNorm 2047232)Abrocitinib (RxNorm 2591476)
Monoclonal anti-TNF	Adalimumab (RxNorm 327361)Infliximab (RxNorm 191831)Golimumab (RxNorm 819300)
Non-monoclonal anti-TNF	Etanercept (RxNorm 214555)Certolizumab pegol (RxNorm 709271)
anti IL-17 Ab	Secukinumab (RxNorm 1599788)Ixekizumab (RxNorm 1745099)Brodalumab (RxNorm 1872751)Bimekizumab (RxNorm 2668041)
anti IL-23 Ab	Risankizumab (RxNorm 2166040)Guselkumab (RxNorm 1928588)Tildrakizumab (RxNorm 2053436)
IL-12/IL-23 p40 Inhibitors	Ustekinumab (RxNorm 847083)
anti IL-6 Ab	Tocilizumab (RxNorm 612865)Sarilumab (RxNorm 1923319)
Outcome
Non-infectious uveitis	ICD-10-CM D86.83, H20.9, H20.04x, H20.1x, H20.02x, H20.00, H20.01x, H30.2x, H35.06x, H20.82x, H30.81x, H44.11x, H44.13x, H30.01x, H30.02x, H30.03x, H30.04x, H30.14x, H30.89x, H30.9x, H31.22, H31.8
Anterior uveitis	ICD-10-CM H20.9, H20.04x, H20.1x, H20.02x, H20.00, H20.01x
Propensity score matching covariates
Comorbidities	Nicotine dependence (ICD-10-CM F17)Vitamin D deficiency (ICD-10-CM E55.9)Hypertension (ICD-10-CM I10)Hyperlipidemia (ICD-10-CM E78.5)Diabetes mellitus (ICD-10-CM E08-E13)Ischemic heart disease (ICD-10-CM I25)Cerebrovascular diseases (ICD-10-CM I60-I69)
Medication	Glucocorticoids (VA HS051)Azathioprine (RxNorm 1256)Methotrexate (RxNorm 6851)Mycophenolate mofetil (RxNorm 68149)Cyclosporine (RxNorm 3008)

Abbreviations: ICD-10-CM, International Classification of Diseases, Tenth Revision, Clinical Modification; JAKi, janus kinase inhibitors; anti-TNF, anti-tumor necrosis factor-α antibodies; anti IL-17 Ab, anti-interleukin 17 antibodies; anti IL-23 Ab, anti-interleukin 23 antibodies; IL-12/IL-23 p40 Inhibitors, interleukin 12/interleukin 23 p40 inhibitors; anti IL-6 Ab, anti-interleukin 6 antibodies

Table 4.

STROBE Statement—Checklist of items that should be included in reports of cohort studies

	Item No	Recommendation	Page No
Title and abstract	1	(a) Indicate the study’s design with a commonly used term in the title or the abstract	1, 3
		(b) Provide in the abstract an informative and balanced summary of what was done and what was found	3
Introduction
Background/rationale	2	Explain the scientific background and rationale for the investigation being reported	5
Objectives	3	State specific objectives, including any prespecified hypotheses	5
Methods
Study design	4	Present key elements of study design early in the paper	6
Setting	5	Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data collection	6-7
Participants	6	(a) Give the eligibility criteria, and the sources and methods of selection of participants. Describe methods of follow-up	6-7
		(b) For matched studies, give matching criteria and number of exposed and unexposed	6-7
Variables	7	Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria, if applicable	6-7
Data sources/ measurement	8*	For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe comparability of assessment methods if there is more than one group	6-7
Bias	9	Describe any efforts to address potential sources of bias	6-7
Study size	10	Explain how the study size was arrived at	6-7
Quantitative variables	11	Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why	7
Statistical methods	12	(a) Describe all statistical methods, including those used to control for confounding	7
		(b) Describe any methods used to examine subgroups and interactions	6-7
		(c) Explain how missing data were addressed	7
		(d) If applicable, explain how loss to follow-up was addressed	7
		(e) Describe any sensitivity analyses	N/A
Results
Participants	13*	(a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed	8
		(b) Give reasons for non-participation at each stage	8
		(c) Consider use of a flow diagram	8
Descriptive data	14*	(a) Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential confounders	8-10
		(b) Indicate number of participants with missing data for each variable of interest	8-10
		(c) Summarise follow-up time (eg, average and total amount)	8
Outcome data	15*	Report numbers of outcome events or summary measures over time	8-10
Main results	16	(a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included	8-10
		(b) Report category boundaries when continuous variables were categorized	8
		(c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period	N/A
Other analyses	17	Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyses	9-10
Discussion
Key results	18	Summarise key results with reference to study objectives	11
Limitations	19	Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and magnitude of any potential bias	12
Interpretation	20	Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from similar studies, and other relevant evidence	11-12
Generalisability	21	Discuss the generalisability (external validity) of the study results	11-12
Other information
Funding	22	Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based	13

*Give information separately for exposed and unexposed groups.Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at http://www.strobe-statement.org.

Results

At the time of data extraction, the TriNetX database comprised 135,645,991 individuals, and a total of 1,434,915 patients were included in our final analysis. The study flowchart is illustrated in Fig. 2 in Appendix. The baseline demographic characteristics, comorbidities, and laboratory measurements between JAKi and Anti-TNF groups were well-balanced after PSM (Table 1). Baseline characteristics—including demographics, laboratory parameters, medication use, and comorbid conditions—demonstrated excellent balance between the JAKi and Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF cohorts, with all SMD value < 0.1. The mean follow-up duration (SD) was 480.2 (261.4) days for the JAKi group and 611.8 (213.6) days for the Anti-TNF group, with similar patterns observed in the subgroup comparisons.

Fig. 2.

Study flowchart. Figure 2 Patient selection flowchart from the TriNetX Research database. Abbreviations: yrs, years; mths, months; JAKi, janus kinase inhibitors; Anti-TNF, anti-tumor necrosis factor-α antibodies; Anti IL-17 Ab, anti-interleukin 17 antibodies; Anti IL-23 Ab, anti-interleukin 23 antibodies; Anti IL-12/IL-23 Ab, anti-interleukin 12/interleukin 23 p40 antibodies; Anti IL-6 Ab, anti-interleukin 6 antibodies; NIU, non-infectious uveitis; PSM, propensity score matching

Table 1.

Demographics of study population before and after propensity score matching

	Pre-matching			Post-matching
	JAKi group (n = 3,333)	Anti-TNF group (n = 47,775)	SMD	JAKi group (n = 3,332)	Anti-TNF group (n = 3,332)	SMD
Age, mean (SD), yr	49.5 (15.0)	45.1 (15.8)	0.288	49.5 (15.0)	48.8 (15.6)	0.043
Sex, n (%)
Female	1,896 (56.9)	23,010 (48.2)	0.175	1,895 (56.9)	1,862 (55.9)	0.020
Races*, n (%)
White	2,465 (74.0)	34,653 (72.5)	0.032	2,464 (73.9)	2,449 (73.5)	0.010
Black	185 (5.6)	3,391 (7.1)	0.064	185 (5.6)	204 (6.1)	0.024
Asian	139 (4.2)	1,672 (3.5)	0.035	139 (4.2)	148 (4.4)	0.013
Laboratory, mean (SD)
Erythrocyte sedimentation rate (mm/h)	19.9 (20.0)	17.5 (19.1)	0.122	19.9(20.0)	18.1(18.5)	0.097
C reactive protein (mg/L)	10.3 (21.0)	10.2 (24.4)	0.003	10.3(21.0)	10.0(24.5)	0.014
Diagnosis, n (%)
Ankylosing spondylitis	148 (4.4)	4,173 (8.7)	0.174	148 (4.4)	141 (4.2)	0.010
Psoriasis	799 (24.0)	11,305 (23.7)	0.007	799 (24.0)	794 (23.8)	0.004
Psoriatic arthritis	662 (19.9)	7,985 (16.7)	0.082	662 (19.9)	676 (20.3)	0.010
Ulcerative colitis	1,079 (32.4)	7,499 (15.7)	0.398	1,078 (32.4)	1,145 (34.4)	0.043
Crohn’s disease	395 (11.9)	15,772 (33.0)	0.524	395 (11.9)	374 (11.2)	0.020
Comorbidities, n (%)
Nicotine dependence	110 (3.3)	1,719 (3.6)	0.016	110 (3.3)	115 (3.5)	0.008
Vitamin D deficiency	347 (10.4)	4,103 (8.6)	0.062	346 (10.4)	276 (8.3)	0.072
Hypertension	588 (17.6)	6,520 (13.6)	0.110	587 (17.6)	524 (15.7)	0.051
Hyperlipidemia	344 (10.3)	3,135 (6.6)	0.136	343 (10.3)	310 (9.3)	0.033
Diabetes mellitus	284 (8.5)	3,069 (6.4)	0.080	284 (8.5)	283 (8.5)	0.001
IHD	115 (3.5)	1,125 (2.4)	0.065	115 (3.5)	103 (3.1)	0.020
CVD	41 (1.2)	435 (0.9)	0.031	41 (1.2)	39 (1.2)	0.006
Medication, n (%)
Glucocorticoids	1,327 (39.8)	17,397 (36.4)	0.070	1,326 (39.8)	1,378 (41.4)	0.032
Azathioprine	30 (0.9)	2,743 (5.7)	0.273	30 (0.9)	35 (1.1)	0.015
Methotrexate	327 (9.3)	4,302 (8.5)	0.024	308 (9.2)	310 (9.3)	0.002
Mycophenolate mofetil	10 (0.3)	68 (0.1)	0.034	10 (0.3)	10 (0.3)	< 0.001
Cyclosporine	41 (1.2)	309 (0.6)	0.061	41 (1.2)	41 (1.2)	< 0.001

*The racial and ethnic classifications (e.g., White, Black and African American, Asian) are directly taken from the original terminology provided in the TriNetx database. Specifically, “Black and African American” has been shortened to “Black” due to space constraints

Abbreviations: JAKi, janus kinase inhibitors; anti-TNF, anti-tumor necrosis factor-α antibodies; SMD, standardized mean difference; SD, standard deviation; yr, year; IHD, ischemic heart disease; CVD, cerebrovascular diseases

After a follow-up duration of 2 years, JAKi demonstrated significantly lower risk of AU compared to the non-monoclonal anti-TNF group (HR: 0.28, 95% CI: 0.11–0.68) and total anti-TNF group (HR: 0.43, 95% CI: 0.20–0.92). However, no significant differences were observed between JAKi and monoclonal anti-TNF groups for AU risk. The results are presented in Table 2.

Table 2.

Comparison of Anterior Uveitis Risk Among Different JAKi and Anti-TNF Groups

Case Group	Compared to Monoclonal anti-TNF				Compared to Non-monoclonal anti-TNF				Compared to Anti-TNF
	N	Event, n*	HR (95% CI)		N	Event, n*	HR (95% CI)		N	Event, n*	HR (95% CI)
		(Case / Control)				(Case / Control)				(Case / Control)
JAKi	3,332	≦ 10 / 23	0.47 (0.22, 1.02)		2,367	≦ 10 / 26	0.28 (0.11, 0.68)		3,332	≦ 10 / 25	0.43 (0.20, 0.92)
Tofacitinib	2,016	≦ 10 / 16	0.50 (0.22, 1.17)		1,496	≦ 10 / 19	0.27 (0.10, 0.73)		2,129	≦ 10 / 15	0.54 (0.23, 1.28)
Upadacitinib	1,553	≦ 10 / 11	0.13 (0.02, 1.02)		1,144	≦ 10 / 16	0.09 (0.01, 0.65)		1,542	≦ 10 / ≦10	0.13 (0.02, 0.99)

*To protect patient privacy, TriNetX obscures any event count ≦ 10. While these counts are displayed as “≦10” in the table, the Hazard Ratios (HR) and associated statistical metrics were calculated by the platform. Abbreviations: N, number of patients in each matched cohort group; JAKi, janus kinase inhibitors; anti-TNF, anti-tumor necrosis factor-α antibodies; HR, hazard ratio; CI, confidence interval; SD, standard deviation; AU, anterior uveitis

Kaplan-Meier analysis demonstrated significantly higher 2-year AU-free survival in JAKi users versus anti-TNF users (99.62% vs. 99.14%, log-rank p = 0.026, Fig. 1a). Stratified analyses revealed more pronounced differences when comparing JAKi with non-monoclonal anti-TNF agents (Fig. 1b and 99.67% versus 98.74%, log-rank p = 0.003), while the comparison with monoclonal anti-TNF agents did not reach statistical significance (Fig. 1c and 99.62% versus 99.20%, log-rank p = 0.051).

Fig. 1.

Kaplan-Meier survival analysis of AU risk between JAKi and anti-TNF biologics. Figure 1 a, Comparison of survival without anterior uveitis between JAKi and Anti-TNF. b, Comparison of survival without AU between JAKi and Non-monoclonal Anti-TNF. c, Comparison of survival without AU between JAKi and Monoclonal Anti-TNF. The Kaplan–Meier survival plot illustrates the two-year survival probability for patients prescribed with JAKi and Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF. A zoomed-in view is provided in the lower right corner of each figure. The dark blue lines represent patients prescribed with JAKi, whereas the green dotted line represents those on Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF. Figures are produced by R (version 4.4.0).Abbreviations: JAKi, janus kinase inhibitors; anti-TNF, anti-tumor necrosis factor-α antibodies

In stratified analyses comparing individual JAK inhibitors with non-monoclonal anti-TNF therapy (Table 2), both tofacitinib (HR: 0.27, 95% CI: 0.10–0.73) and upadacitinib (HR: 0.09, 95% CI: 0.01–0.65) were significantly associated with a lower incidence of AU. When compared with the total anti-TNF group, only upadacitinib showed a significant association with AU risk (HR: 0.13, 95% CI: 0.02–0.99). Neither tofacitinib nor upadacitinib demonstrated significant differences in AU risk when compared specifically with monoclonal anti-TNF therapy.

Stratified analyses of different systemic autoimmune diseases showed that the effectiveness of JAKi in preventing AU retained in patients with ankylosing spondylitis when compared to total anti-TNF (HR: 0.30, 95% CI: 0.10–0.91), non-monoclonal anti-TNF (HR: 0.33, 95% CI: 0.11–0.99), and monoclonal anti-TNF (HR: 0.26, 95% CI: 0.09–0.78) (Fig. 3 in Appendix). The outcomes for each stratification analysis are illustrated in Fig. 3 in Appendix. In the Kaplan-Meier survival analysis of the ankylosing spondylitis subgroup, the 2-year AU-free survival probability was significantly higher with JAKi compared to total anti-TNF (98.74% versus 95.62%, log-rank p = 0.023), non-monoclonal anti-TNF (98.66% versus 95.64%, log-rank p = 0.037), and monoclonal anti-TNF agents (98.74% versus 95.18%, log-rank p = 0.010) (Fig. 4 in Appendix).

Fig. 3.

AU risk of JAKi and anti-TNF biologics by disease entities. Fig. 3 Associations between JAKi and Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF on AU across different diseases. The forest plot depicts the comparative associations of JAKi and Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF with the risk of AU across different diseases. A lower hazard ratio indicates a reduced risk of the outcome with JAKi users compared to Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF users. Figures are produced by R (version 4.4.0).Abbreviations: JAKi, janus kinase inhibitors; anti-TNF, anti-tumor necrosis factor-α antibodies; HR, hazard ratio; CI, confidence interval; AU, anterior uveitis

Fig. 4.

Kaplan-Meier plot of AU-free survival in ankylosing spondylitis. Fig. 4 a, Comparison of survival without anterior uveitis between JAKi and Anti-TNF in ankylosing spondylitis subgroup. b, Comparison of survival without AU between JAKi and Non-monoclonal Anti-TNF in ankylosing spondylitis subgroup. c, Comparison of survival without AU between JAKi and Monoclonal Anti-TNF in ankylosing spondylitis subgroup. The Kaplan–Meier survival plot illustrates the two-year survival probability for ankylosing spondylitis patients prescribed with JAKi and Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF. A zoomed-in view is provided in the lower right corner of each figure. The dark blue lines represent patients prescribed with JAKi, whereas the green dotted line represents those on Anti-TNF/non-monoclonal anti-TNF/monoclonal anti-TNF. Figures are produced by R (version 4.4.0). Abbreviations: JAKi, janus kinase inhibitors; anti-TNF, anti-tumor necrosis factor-α antibodies; AS, ankylosing spondylitis

Discussion

Our study identifies a significant association between JAKi use and reduced incidence of AU, compared to anti-TNF. This association was especially notable in patients with ankylosing spondylitis, suggesting a potential therapeutic role for JAKi in managing autoimmune-related uveitis. These findings are consistent with mechanistic studies implicating JAK-STAT signaling in uveitis pathogenesis.

The JAK/STAT pathway is a key regulator of autoimmune and inflammatory processes, including those involved in uveitis. It mediates cytokine signaling that drives the differentiation of pathogenic Th1 and Th17 cells, primarily through STAT1 and STAT3 activation [17–20]. Experimental autoimmune uveitis (EAU) models have demonstrated that JAK inhibition effectively reduces inflammation by suppressing Th17 cell proliferation, IL-23/IL-17 A signaling, and inflammatory cell infiltration into ocular tissues [21]. JAKi also modulate Th1-driven pathways, with tofacitinib showing a distinct mechanism by reducing IFN-γ production [22, 23].

Research has demonstrated significantly elevated JAK-1 and JAK-3 expression in CD3 + T cells and CD14 + monocytes from ankylosing spondylitis patients. Importantly, JAK-1 expression levels correlate with inflammatory markers (ESR, CRP) and clinical manifestations including peripheral arthritis, enthesitis, and importantly, uveitis [24]. Additional evidence has identified a JAK2 locus haplotype associated with ankylosing spondylitis pathogenesis [25]. Collectively, these molecular and clinical findings establish the JAK-STAT signaling axis as a promising therapeutic target in uveitis [8, 26, 27].

Our findings demonstrate that JAKi, regardless of their selectivity profile, showed comparable efficacy in reducing AU risk among patients with autoimmune diseases. This suggests that the lower AU risk may be a class effect of JAK inhibition rather than dependent on inhibitor selectivity. We evaluated two JAKi with distinct targeting profiles: tofacitinib, a non-selective JAKi predominantly targeting JAK3 and JAK1, with partial activity on JAK2; and upadacitinib, a selective JAK1 inhibitor with higher specificity [27, 28]. By inhibiting JAK1/JAK3 activity, these agents disrupt downstream signaling cascades responsible for pro-inflammatory cytokine production, particularly IL-17, which plays a central role in uveitis pathogenesis. The discrepancy between individual and combined JAKi analyses may be attributed to reduced statistical power in the smaller sample size in stratified analyses.

Previous evidence, including a landmark registry-based study by Lim et al., has demonstrated that monoclonal anti-TNF antibodies are more effective than etanercept in preventing uveitis recurrence [29]. Several meta-analyses of randomized controlled trials (RCTs) have extended these findings by showing that monoclonal anti-TNF agents significantly reduce new-onset acute uveitis (AU) in axial spondyloarthritis (axSpA), with a pooled incidence-rate ratio of approximately 0.32 compared to placebo [13, 30]. Our findings align with this trend, providing further real-world context for treatment selection. Meanwhile, prospective cohort studies and observational research have consistently indicated that JAKi are effective in refractory cases of NIU and AU. The same network meta-analysis also demonstrated that JAKi offer efficacy comparable to monoclonal anti-TNF and anti-IL-17 agents, particularly in refractory AU cases [13]. However, these studies also highlight key research gaps, including small sample sizes, limited long-term follow-up, and inconsistent subgroup analyses [9, 13]. Recent clinical trials further contextualize the role of JAKi in NIU treatment [10–12], suggesting that newer JAKi, particularly those with combined JAK/TYK inhibition, may enhance treatment efficacy for non-anterior uveitis.

Unlike clinical trials such as NEPTUNE, HUMBOLDT, and VISUAL that primarily focus on non-anterior uveitis [10–12], our study sheds light on AU, adding a broader clinical context for treatment comparisons. We evaluated FDA-approved JAKi (tofacitinib and upadacitinib) that currently lack dedicated NIU trials, highlighting their potential therapeutic role in AU, the most prevalent NIU subtype.

Clinical applicability and patient-centered considerations

The convenience of oral JAKi may improve adherence compared to injectable biologics. Their broad immunomodulatory effects may also provide advantages in complex cases involving systemic autoimmune diseases with multiple inflammatory manifestations. However, differences in safety profiles, particularly infection risks and long-term safety concerns, require further investigation to inform clinical decision-making. Clinicians should consider comorbidities, preferences for administration route, and prior treatment responses when selecting between JAKi and anti-TNF for patients at risk of AU.

Strengths and limitations

This study leverages a large, diverse dataset to evaluate real-world treatment patterns and outcomes, providing valuable insights into the comparative efficacy of JAKi and anti-TNF with comprehensive assessment of glucocorticoid, conventional DMARD-sparing effects along with evaluation of comorbidities through PSM. However, several limitations must be acknowledged. First, due to the retrospective observational design, residual confounding and selection bias cannot be entirely excluded. Although we implemented a 12-month look-back period to exclude recent NIU cases, potential channeling bias (confounding by indication) cannot be entirely ruled out. Clinicians may preferentially prescribe monoclonal anti-TNF to patients with a remote history of uveitis (beyond the exclusion window) given the drug’s established efficacy for ocular inflammation. Consequently, the anti-TNF cohort might have possessed a higher underlying predisposition to uveitis compared to the JAKi cohort. This residual confounding could theoretically inflate the observed uveitis risk in the anti-TNF group, suggesting that our findings may be conservative estimates of the comparative safety/efficacy. Furthermore, the strict exclusion of patients with an NIU diagnosis within 12 months prior to the index date warrants consideration. While this washout period was necessary to capture incident cases and reduce baseline confounding, it inherently selects for a population with a lower baseline ocular risk. Therefore, our reported incidence rates likely underestimate the overall burden of uveitis in broader clinical practice, which includes patients with chronic or frequently relapsing disease. Consequently, our findings are strictly generalizable to patients initiating therapy during a period of ocular quiescence, and may not apply to those with active uveitis at baseline. Second, reliance on electronic medical records may result in disease misclassification and limit diagnostic accuracy. Third, the small sample size in individual JAKi analyses, coupled with low event counts for AU outcomes, constrains the generalizability and robustness of our conclusions for this specific outcome. These low event counts resulted in reduced statistical power and wide confidence intervals, indicating a high degree of statistical uncertainty. Consequently, these comparative estimates should be interpreted with caution, as the study may be underpowered to detect potential differences between treatment groups regarding AU risk. Future studies should prioritize prospective designs, longer follow-up durations, and larger, ethnically diverse cohorts to validate these associations and further explore the long-term safety and efficacy of JAKi in AU.

Conclusion

This study provides evidence supporting JAKi as a potential treatment option for autoimmune-related uveitis. Our findings suggest an association between JAKi use and a reduced incidence of AU, particularly in patients with ankylosing spondylitis. Future large prospective studies are needed to further evaluate this potential benefit and to determine the optimal role of JAKi in the management of patients at risk for autoimmune-related anterior uveitis.

Abbreviations

anti-IL-12/IL-23

anti-interleukin 12/interleukin 23 p40 antibodies

anti-IL-17

anti-interleukin 17 antibodies

anti-IL-23

anti-interleukin 23 antibodies

anti-IL-6 Ab

anti-interleukin 6 antibodies

anti-TNF

anti-tumor necrosis factor-α antibodies

AU

anterior uveitis

CD

cluster of differentiation

CI

confidence intervals

CRP

c reactive protein

csDMARDs

conventional synthetic disease-modifying antirheumatic drugs

DMARD

disease-modifying antirheumatic drugs

EAU

experimental autoimmune uveitis

ESR

erythrocyte sedimentation rate

FDA

U.S. Food and Drug Administration

HIPAA

Health Insurance Portability and Accountability Act

HR

hazard ratios

ICD-10

International Classification of Diseases, 10th Revision

IRB

Institutional Review Board

JAK

Janus kinase

JAKi

Janus kinase inhibitors

NIU

non-infectious uveitis

SD

standard deviation

SMD

standardized mean differences

STAT

signal transducer and activator of transcription

Author Contribution

Conceptualization, Methodology: Shiao-Ling Wu; Yi-Ming Chen; Chien-Chih Chou Writing: Shiao-Ling Wu; Yi-Ming Chen; Chien-Chih Chou; Ssu-Yu Pan Funding, Supervision: Yi-Ming Chen; Chien-Chih Chou Analyzing, visualization: Shiao-Ling Wu; Jun-Fu Lin; Chien-Hsiang Weng; Yun Hsia; Mey-Fann Lee Resources: Shiao-Ling Wu; Yu-Wen Chu; Wei-Ting Ho; De-Xian Li; Hui-Ju Lin; I-Jong Wang. All authors contributed to the study conception and design. The first draft of the manuscript was written by Shiao-Ling Wu and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

Open access funding provided by Taichung Veterans General Hospital. This study was supported by Taichung Veterans General Hospital (grant number TCVGH-1136902B). The funding organization played no role in this study. The authors have no relevant financial or non-financial interests to disclose.

Data Availability

All data and materials supporting the published claims in this study are available from the corresponding author upon request and comply with current ophthalmological field standards. The datasets generated during this study, statistical analysis code, and methodological details are accessible to ensure reproducibility of the research findings.

Declarations

Ethics approval

This study received ethical approval from the Institutional Review Board of Taichung Veterans General Hospital (IRB NO: CE24430C)

Consent to participate / publish

Not applicable. Patients and the public were not directly involved in the design, recruitment, or conduct of this study, as it was based on retrospective analysis of de-identified electronic health records obtained from the TriNetX global research network.

Disclosure of potential conflicts of interest

The authors declared there is no conflict of financial or non-financial interests.