Safety of Janus kinase inhibitors compared to biologic therapy in patients with inflammatory bowel disease
Article information
Abstract
Background/Aims
There is a paucity of data regarding the comparative safety of Janus kinase inhibitors (JAKi) compared to biologic therapy in inflammatory bowel disease (IBD).
Methods
A retrospective cohort study was conducted using the U.S. Collaborative Network to assess the risk of adverse events in patients with IBD on JAKi compared to biologic therapies (tumor necrosis factor inhibitor, ustekinumab or vedolizumab). Outcomes assessed included infections, major adverse cardiovascular events (MACE), malignancy, venous thromboembolism (VTE), gastrointestinal perforations and laboratory abnormalities within 2 years. 1:1 propensity score matching (PSM) was performed.
Results
There were 3,186 patients in the JAKi cohort (mean age 44.3 ± 17.1, 46.5% male sex, 85.7% ulcerative colitis, 72% tofacitinib) and 60,462 patients in the biologic cohort. After PSM, there was no difference in the risk of bacterial (adjusted odds ratio [aOR], 1.08; 95% confidence interval [CI], 0.82–1.43), viral (aOR, 1.06; 95% CI, 0.84–1.32), or fungal infections (aOR, 1.10; 95% CI, 0.84–1.44), MACE (aOR, 0.72; 95% CI, 0.47–1.10), malignancy (aOR, 0.98; 95% CI, 0.71–1.35), and VTE (aOR, 0.95; 95% CI, 0.67–1.34). This was consistent in subgroup analyses based on IBD type, JAKi type and age group. There was a higher risk of low-density lipoprotein ≥ 190 mg/dL (aOR, 1.78; 95% CI, 1.03–3.05) and grade 3 or higher lymphopenia (aOR, 2.28; 95% CI, 1.60–3.26) in the JAKi cohort. There was no difference in the 4-year risk of MACE or malignancy in the tofacitinib cohort compared with the biologic cohort.
Conclusions
Our real-world study did not show an increased risk of adverse events except severely elevated LDL and lymphopenia in patients on JAKi compared to biologic therapy.
INTRODUCTION
Janus kinase inhibitors (JAKi) have rapidly emerged as an important therapeutic class in the management of inflammatory bowel disease (IBD), which includes ulcerative colitis (UC) and Crohn’s disease (CD) [1,2]. Despite their considerable potential, there are several safety concerns related to JAKi. Early studies identified an increased incidence of infections, particularly herpes zoster, and laboratory abnormalities such as elevated low-density lipoprotein (LDL), total cholesterol, and lymphopenia [3,4]. Furthermore, select trials raised concerns about cardiovascular and thromboembolic risks. The Oral Rheumatoid Arthritis Trial (ORAL) Surveillance study in patients with rheumatoid arthritis (RA) who were over the age of 50 with at least one risk factor for cardiovascular events reported a higher incidence of major adverse cardiovascular events (MACE) in participants receiving tofacitinib relative to those on tumor necrosis factor inhibitor (TNFi) [5]. Although the absolute differences were modest, this triggered cautionary regulatory updates for JAKi, especially in older patients with additional risk factors [5,6]. Another worrisome signal emerged from an integrated safety analysis in RA, which suggested potential increases in malignancy and venous thromboembolism (VTE) [7,8]. However, it is unclear whether these risks apply uniformly to IBD.
Multiple observational and post-marketing studies in the UC population have since examined these potential risks. These results indicate that, although the risk of viral infections (e.g., rates of herpes zoster) is increased, the incidence of MACE in short-term follow-up remains comparable to that observed with biologic therapy [9,10]. Yet these data are frequently limited by small sample sizes, short observation periods, or lack of direct head-to-head comparison groups [4,11]. Balancing the efficacy profile of JAKi against emerging safety signals highlights the need for larger, more comprehensive studies that directly compare JAKi to standard biologic therapy across a broad IBD population.
Against this background, the current study was conducted to examine the risk of adverse events (AEs) in patients with IBD receiving JAKi compared to those on biologic therapy. Specifically, the primary aim of the study was to assess the 2-year risk of AEs in patients with IBD on JAKi compared to patients on biologic therapy. The AEs evaluated included infections, cardiovascular and thromboembolic complications, malignancies, gastrointestinal perforation, all-cause mortality, and significant laboratory abnormalities.
METHODS
1. Database
Most health care organizations within the U.S. Collaborative Network of the TriNetX database (TriNetX LLC, Cambridge, CA, USA) are large academic medical institutions which contain inpatient and outpatient facilities. The data represents the entire patient population of the organization. The de-identification process is determined and done at a network-level and attested through formal determination by a qualified expert as defined in the HIPAA Privacy Rule. TriNetX obfuscates patient counts < 10 to ensure patient anonymity. Clinical variables are derived directly from electronic health records (EHRs) of included health care organizations as well as retrieved through a built-in natural language processing system that extracts variables from clinical documents. Robust quality assurance is achieved at the time of extraction from EHRs before inclusion in the database, in a systematic and standardized format. The process also includes data cleaning which rejects patient records that do not meet the TriNetX quality standards. The database does not include claims data or data collected from randomized clinical trials. The database includes patient data regarding demographics, diagnosis, procedures, laboratory values and medications. The interface only provides aggregate counts and statistical summaries to protect patient health information and ensures that the data remain de-identified at all levels of data retrieval and dissemination. Hence, institutional review board approval is not required.
2. Study Participants and Cohorts
A retrospective cohort study was conducted using the U.S. Collaborative Network in TriNetX, a multi-institutional database. TriNetX is a global federated research network which provides real-time access to de-identified EHRs of more than 120 million patients within 70 health care organizations in the United States. Patients in the JAKi cohort included adults ≥18 years old who had at least one International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) code in their EHRs for UC (K51*) or CD (K50*) plus a RxNorm code for tofacitinib (1357536) or upadacitinib (2196092) between June 1, 2018 and January 22, 2023. The start date of the study was chosen as tofacitinib was first approved for UC in the United States in May 2018 [12]. The end date of the cohort was chosen to allow for a 2-year follow-up in all patients. In prior administrative or claims-based studies, IBD case definitions incorporating at least one ICD-10-CM diagnosis code and a concordant IBD-specific prescription achieved positive predictive value ≥80% and specificity ≥85% [13]. TriNetX is an established data source for IBD research and has supported several previously published studies [14-16]. The biologic cohort (control) included patients with UC or CD who had a RxNorm code for infliximab, adalimumab, vedolizumab or ustekinumab during the same time frame. These case definitions to identify the IBD cohort within the database have been validated and utilized in previously published studies. To ensure patients initiated a JAKi or biologic therapy, those who had a RxNorm code for an alternative advanced therapy or a Current Procedural Terminology (CPT) code for IBD-related surgery within 30 days were excluded from the cohorts. Additionally, patients who had RxNorm codes for a JAKi within 2 years of the index biologic therapy prescription were also excluded from the control cohort to ensure patients were not exposed to a JAKi during follow-up.
3. Study Aims and Outcomes
The primary aim of the study was to assess the 2-year risk of AEs between the JAKi and the biologic cohort. The risk was also evaluated for each biologic therapy that included TNFi, ustekinumab, and vedolizumab. AEs were categorized by infection, MACE, malignancy, VTE, gastrointestinal perforation, acne, laboratory abnormalities, and any-cause mortality. Outcome ascertainment was 30 days after the index prescription of the medication. Infectious AEs were further divided by bacterial, viral, fungal, other non-specific infections, sepsis, or septic shock. Non-specific infections included acute upper or lower respiratory infections, urinary tract infections, otitis externa or media, skin and subcutaneous infections, myositis, osteomyelitis, and septic arthritis. This was independently evaluated to capture patients who developed infections but did not have a causative infectious agent identified, which is common in the outpatient setting. The infectious organisms and outcomes, as well as other non-infectious complications included, were based on a study by Hoisnard et al. [17] who reported AEs of JAKi from a World Health Organization international pharmacovigilance database. Outcomes were identified by ICD-10-CM codes or Logical Observation Identifiers Names and Codes (LOINC) for positive laboratory tests for the organism of interest. MACE included de novo acute myocardial infarction, angina pectoris, cerebral infarction, and acute heart failure. VTE included de novo deep vein thrombosis and/or pulmonary embolism. Malignancy included de novo solid organ and/or hematologic malignancy. We only included de novo cases given the challenge of differentiating new AEs in patients with a history of the same from use of the same ICD-10-CM codes which could have predated the initiation of JAKi or biologic therapy. All ICD-10-CM, RxNorm, and CPT codes can be found in the Supplementary Table 1. Laboratory abnormalities included very high levels of LDL, defined by values ≥190 mg/dL; moderate-to-severe neutropenia, defined by values <1,000 cells/μL; severe lymphopenia defined by values <500 cells/μL; and moderate-severe elevations in liver enzymes defined by values ≥100 U/L for aspartate and/or alanine aminotransferase. These values were chosen to identify patients with clinically significant abnormalities [18-21]. Patients who did not have these laboratory values during follow-up were excluded from this analysis. AEs were also compared to each biologic class. We also performed subgroup analyses based on IBD type, type of JAKi, and age groups (18–49 and ≥50 years old). Censoring events included end of study period, loss to follow-up and death.
The secondary aim of the study was to assess the 4-year risk of AEs between the tofacitinib cohort and the biologic cohort. The time frame was adjusted to only include patients who were initiated on these therapies before January 1, 2022. Given upadacitinib was not approved for UC until March 2022, only patients on tofacitinib were included. We also required patients to have a prescription of tofacitinib or biologic therapy after 2 years to ensure on-going long-term use given the mean duration of treatment was approximately 3 years in the ORAL Surveillance study.
4. Statistical Analysis
All statistical analyses were conducted using the TriNetX software using the browser-based real-time analytics feature, TriNetX Live (TriNetX LLC). Baseline characteristics of cohorts were described using means, standard deviations, and proportions. One-to-one (1:1) propensity score matching (PSM) was performed to balance the covariates between the cohorts. Covariates were adjusted based on the outcome of interest. The following covariates were included: age, race, IBD type, other autoimmune diseases, hypertension, disorders of lipid metabolism, diabetes mellitus, chronic lower respiratory diseases, ischemic heart disease, heart failure, cerebral infarction, transient ischemic attack, thrombophilia, family history cardiovascular disease or stroke, atrial fibrillation/flutter, nicotine dependence, chronic kidney disease, primary sclerosing cholangitis, cirrhosis, alcohol related disorders, human immunodeficiency virus disease, history of malignancy, mood disorders, immunodeficiency disorders, neutropenia, malnutrition, recent mean laboratory values (hemoglobin, C-reactive protein, albumin, calprotectin), body mass index, aspirin use, statin use, insulin use and previous corticosteroid use. We also included previous oral steroid use, hospitalization requiring intravenous steroid use, as well as laboratory values described as a proxy for disease severity and activity. The TriNetX platform utilizes input matrices of the user-identified covariates to conduct logistic regression analysis to obtain propensity scores for all individual subjects. The propensity scores generated are used to match patients using greedy nearest-neighbor algorithms with a caliper width of 0.1 pooled standard deviations. TriNetX randomizes the order of rows to eliminate bias resulting from nearest-neighbor algorithms. Standardized mean difference after PSM indicates the success of matching covariate between the 2 cohorts. A standardized mean difference <0.1 indicates that the difference between the cohorts for the covariate is small. All propensity-matched tables can be found in the Supplementary Tables 2-5. After PSM, the risk of each outcome was calculated and expressed as adjusted odds ratios (aOR) with 95% confidence intervals (CIs). All analyses were intention to treat analysis.
RESULTS
1. Characteristics of the Study Population
There were 3,186 patients in the JAKi cohort. The mean age of the cohort was 44.3 ± 17.1 years, 46.5% had male sex and 78% had White. UC accounted for 85.7% of cases, and 72% were on tofacitinib. The mean duration of IBD prior to initiation of JAKi was 41.2 months. After PSM, 76.4% were on tofacitinib and 23.6% on upadacitinib. There were 60,462 patients in the biologic cohort. The mean age of the cohort was 39.4 ± 17.9 years, 45.6% had male sex, and 73.6% had White. CD comprised 62.8% of cases. In the biologic cohort, 72.6% were receiving infliximab and/or adalimumab, 24.9% were on vedolizumab and 30.6% were on ustekinumab. A total of 13.8% of patients were on 2 or more biologic therapies with a different mechanism of action. After PSM, 69.5% were on infliximab and/or adalimumab, 19.7% were on ustekinumab, and 28.2% were on vedolizumab. The mean duration of IBD prior to initiation of biologic therapy was 23 months. All demographic parameters, co-morbid diseases, and laboratory parameters before and after PSM can be found in the Supplementary Table 2-5. The mean follow-up before PSM was 670.4 ± 165 days and 679.6 ± 156 days and after matching was 670.2 ± 165 days and 685.8 ± 145 days in the JAKi and biologic cohort respectively. The median follow-up was 730 days in both cohorts before and after matching.
2. Risk of Infectious Complications
After PSM, there was no difference in the risk of bacterial (aOR, 1.08; 95% CI, 0.82–1.43), viral (aOR, 1.06; 95% CI, 0.84–1.32) or fungal infections (aOR, 1.10; 95% CI, 0.84–1.44) between the JAKi and biologic cohort (Table 1). There was a lower risk of non-specific infections in the JAKi cohort (aOR, 0.74; 95% CI, 0.61–0.90) compared to the biologic cohort. There was also no difference based on the specific causative agent, including herpes zoster (aOR, 1.36; 95% CI, 0.94–1.98). There was no difference in the risk of infectious complications after comparing to each biologic class (Table 2). Subgroup analysis based on IBD type, type of JAKi, and age groups did not show a higher risk for infections in the JAKi cohort compared to the biologic cohort.
Two-Year Risk of Adverse Events in Patients with IBD in the JAKi and Biologic Cohorts after Propensity Score Matching
Two-Year Risk of Adverse Events in Patients with IBD in the JAKi Compared to Each Biologic Class in the Biologic Cohort after Propensity Score Matching
3. Risk of Non-Infectious Complications
Within 2 years, 38 (1.27%) and 52 (1.75%) patients developed MACE in the JAKi and the biologic cohort, respectively. In total, 76 (2.62%) and 78 (2.67%) patients developed de novo malignancy in the JAKi and the biologic cohort, respectively. After PSM, there was no difference in the risk of MACE (aOR, 0.72; 95% CI, 0.47–1.10) and malignancy (aOR, 0.98; 95% CI, 0.71–1.35) between the 2 cohorts. There was a lower risk of acute heart failure in the JAKi cohort (aOR, 0.58; 95% CI, 0.33–0.99) compared to the biologic cohort. When comparing to each biological class, JAKi cohort had a lower risk of MACE (aOR, 0.64; 95% CI, 0.44–0.94) compared to the TNFi cohort. There was no difference in the risk of MACE and malignancy based on subgroup analysis (Tables 3-5). There was no difference in the risk of VTE between the 2 cohorts (aOR, 0.95; 95% CI, 0.67–1.34). Fewer than 10 patients developed gastrointestinal perforations in the 2 cohorts. In terms of risk of acne, we did not observe an increased risk (aOR, 1.28; 95% CI, 0.97–1.69) in the JAKi cohort compared to the biologic cohort. However, when compared to each individual biologic agent, we observed an increased risk in the JAKi cohort compared to TNFi (aOR, 1.62; 95% CI, 1.20–2.19), vedolizumab (aOR, 1.76; 95% CI, 1.29–2.41), and ustekinumab (aOR, 1.88; 95% CI, 1.36–2.58).
Two-Year Risk of Adverse Events in the JAKi and Biologic Cohort Based on IBD Type after Propensity Score Matching
Two-Year Risk of Adverse Events in the JAKi and Biologic Cohort Based on Type of JAKi after Propensity Score Matching
Two-Year Risk of Adverse Events in the JAKi and Biologic Cohort Based on Age Groups after Propensity Score Matching
4. Risk of Laboratory Complications
After PSM, the JAKi cohort had a higher risk of de novo LDL ≥190 mg/dL (aOR, 1.78; 95% CI, 1.03–3.05) and grade 3 or higher lymphopenia (aOR, 2.28; 95% CI, 1.60–3.26) compared to the biologic cohort. There was no difference in the risk of elevated transaminases (aOR, 0.84; 95% CI, 0.60–1.16) and moderate-severe neutropenia (aOR, 0.91; 95% CI, 0.43–1.95) between the 2 cohorts (Fig. 1).
Risk of adverse events in patients with inflammatory bowel disease in the Janus kinase inhibitor and biologic cohort within 2 years. MACE, major adverse cardiovascular events; VTE, venous thromboembolism; PE, pulmonary embolism; LDL, low-density lipoprotein.
5. Four-Year Risk of AEs
After PSM, there was no difference in the risk of bacterial, viral, or fungal infections between the tofacitinib cohort and the biologic cohort (Table 6). When specifically evaluating the risk of herpes zoster, we found a higher risk in the tofacitinib cohort (aOR, 2.23; 95% CI, 1.08–4.59) compared to the biologic cohort. There were 20 (3.18%) and 21 (3.35%) patients who developed MACE in the tofacitinib cohort and biologic cohort, respectively. There were 29 (5.01%) and 38 (6.66%) patients who developed malignancy in the tofacitinib cohort and biologic cohort, respectively. There was no difference in the risk of MACE (aOR, 0.94; 95% CI, 0.50–1.76) and malignancy (aOR, 0.74; 95% CI, 0.45–1.21) between the 2 cohorts. There was also no difference in the risk of VTE (aOR, 1.05; 95% CI, 0.52–2.10) between the 2 cohorts.
Four-Year Risk of Adverse Events in Patients with Inflammatory Bowel Disease in the Tofacitinib and Biologic Cohort after Propensity Score Matching
DISCUSSION
Our study, utilizing real-world data from a US-based multicenter EHR database, is one of the few direct head-to-head comparisons of adverse event rates in a large IBD population receiving JAKi compared to biologic therapy. Our results demonstrate that while certain events, such as elevated LDL levels and moderate-severe lymphopenia, were more common in the JAKi cohort, the overall rates of infections, malignancy, MACE, and thromboembolic complications were similar between the 2 cohorts. Interestingly, the risk of herpes zoster and VTE was not increased; there was a higher proportion of patients who developed VTE in the JAKi cohort who were ≥50 years old and herpes zoster in the overall JAKi cohort, but this did not reach statistical significance. Given limited real-world data on the long-term safety of JAKi, our study also demonstrated a comparable safety profile compared to biologic therapy. Notably, there was no difference in the risk of MACE and malignancy, and the rates were comparable to those observed in the ORAL Surveillance study [5]. Our results were consistent based on type of IBD and JAKi and in different age groups. Given the need for more long-term safety data, we also assessed the 4-year risk of AEs in a cohort of patients on tofacitinib with >2-year exposure to tofacitinib and only observed an increased risk of herpes zoster. In the primary analysis, the risk of AEs was also compared to each biologic agent with a different mechanism of action; however, additional subgroup analysis was limited for the non-TNFi biologic agents due to smaller sample size and low event rates making reporting of data challenging as the database obfuscates patient counts less than 10.
Our results align with prior clinical trials and observational studies in several respects, particularly with regards to infections and lipid abnormalities. In a large administrative claims-based study comparing new users of JAKi with TNF antagonists between 2016 and 2023, Ahuja et al. [22] reported a higher risk of overall infections with JAKi, but no increased risk of serious infections requiring hospitalization, VTE, or MACE. Our results complement and extend these observations using a multicenter EHR-based dataset and a broader comparator framework that included TNFi, vedolizumab, and ustekinumab. After PSM, we did not observe an increased risk of bacterial, viral, or fungal infections in the overall JAKi versus biologic analysis and similarly did not identify a higher risk of these infection categories in the JAKi versus TNFi comparison. We also found no difference in VTE risk and no overall increase in MACE, with lower MACE odds observed in the JAKi cohort compared with TNFi. The difference between our neutral infection signal and the higher overall infection risk reported by Ahuja et al. [22] may relate to variations in infection definitions, coding granularity, and capture of outpatient events in claims versus EHR data, as well as differences in cohort composition and residual confounding. Collectively, these findings support a broadly comparable real-world safety profile of JAKi relative to biologic therapies for major adverse outcomes, while reinforcing the importance of continued pharmacovigilance and risk stratification. Also, several studies have shown that tofacitinib-treated patients had a higher susceptibility towards to viral infections (especially herpes zoster), with incidence rates ranging from 1% to 5% over 12 to 24 months [1,3,23,24]. In our study, herpes zoster occurred at a higher rate in the JAKi cohort with an increased risk compared to biologic therapy with longer use. While our analysis was adjusted for history of vaccination, it is possible that there were other factors that influenced the development of herpes zoster in the biologic cohort. We did observe an increased risk of herpes zoster with more long-term follow-up in the JAKi cohort. We also evaluated short-term risk (<6 months) and observed an increased risk of herpes zoster in the JAKi cohort (0.71% vs. 0.31%: aOR, 2.30; 95% CI, 1.09–4.86). This supports the notion that JAK blockade may compromise immune surveillance for latent viruses [3]. Recent analyses have revealed that JAKi increases LDL by an average of 11 mg/dL and high-density lipoprotein (HDL) by 8 mg/dL, typically stabilizing by week 12 [25,26]. This pattern echoes findings from clinical trials, where lipid shifts reflect redistribution rather than enhanced synthesis [26-28]. Although elevated LDL can theoretically raise cardiovascular risk, the concurrent rise in HDL may offset such concerns, and no direct link to higher MACE incidence has been established [27]. In our IBD cohort, we similarly observed greater odds of severely elevated LDL (≥ 190 mg/dL) among the JAKi cohort, aligning with known on-target metabolic effects, but we did not observe an increased risk of MACE [4,10,27,29].
In IBD-focused analyses, multiple studies found no significant difference in MACE incidence between JAKi and biologic therapy, with cumulative MACE rates of 1%–4% over follow-up periods of 6 months to 3 years [30-32]. For instance, Kochar et al. [31] observed that tofacitinib had a cardiovascular event rate similar to TNFi in patients with IBD (hazard ratio, 2.50; 95% CI, 0.37–6.18). Qapaja et al. [30] reported no increased risk of MACE events (myocardial infarction, stroke, and all-cause mortality) with either oral small molecules or biologic therapy from 30 days to 3 years post-prescription. Deepak et al. [32] reported MACE rates well below 1% in a cohort of 260 patients with UC on tofacitinib; however, the median follow-up was 6 months. In contrast, the ORAL Surveillance study indicated a higher incidence of MACE with tofacitinib—3.4% versus 2.5% for TNFi—over nearly 4 years, yielding a hazard ratio of 1.33 (95% CI, 0.91–1.94) [5]. Although these findings led to concerns for older patients with RA and multiple cardiovascular risk factors, event rates were low. No large IBD cohort with more than 2 to 3 years of follow-up has investigated whether MACE rates remain unchanged beyond this interval. Our study in a small cohort of patients >50 years old did not demonstrate an increased risk of MACE within 4 years. Further observational or prospective investigations are necessary to determine whether cardiovascular risk evolves with prolonged JAKi use in elderly patients with IBD who have additional risk factors for MACE.
As with prior studies, tofacitinib’s action on inhibition of JAK2 raised concerns regarding cytopenia, particularly neutropenia and anemia [33]. However, phase 2 trials in CD and UC reported no clinically severe events, while RA extension data indicated mostly mild or moderate declines [1,34,35]. Although 60.2% of patients on 5 mg twice daily and 30.6% on 10 mg twice daily experienced moderate-to-severe lymphopenia, these decreases did not correlate with serious infections [35]. It is important to note that none of these trials differentiated anemia secondary to drug-induced cytopenia from active IBD, leaving potential for confounding by disease activity. Yet, our finding of significantly higher lymphopenia rates parallels earlier evidence, underscoring a possible on-target effect without a definitive infection risk increase.
Our study has several notable strengths. This study included the largest sample size of patients with IBD on JAKi and provided real-world data comparing its safety profile to biologic therapy. Our study reported the risk of several serious AEs with key subgroup analyses to increase the robustness of our findings. We also reported data on the long-term safety of JAKi, specifically tofacitinib, which adds to the growing body of evidence regarding its safety in comparison to biologic therapy. We had a diverse cohort from HCOs across the United States, allowing for generalizability when interpreting the results. We utilized PSM to reduce the impact of confounding variables despite the retrospective nature of the study.
Our study also has several limitations that warrant consideration. The retrospective design of the study and reliance on ICD-10-CM codes could result in underreporting or misclassification of AEs, especially infectious complications, if mild or sub-clinical. We attempted to mitigate this by including positive laboratory test for the organism of interest. Our primary aim was to evaluate the 2-year risk of AEs; however, certain complications—particularly malignancy and some cardiovascular outcomes—may require long-term follow-up to ascertain the risk adequately. We did assess 4-year risk of AEs; however, this was limited to tofacitinib, and robust subgroup analyses could not be performed due to the smaller sample size and low event rates. The relatively small proportion of patients with CD in our cohort reflects more recent approval of upadacitinib and limits our ability to detect subgroup differences in that population if present. While we accounted for numerous confounders in our propensity score model, unmeasured variables—such as baseline vaccination status, endoscopic disease activity scores, or specific lifestyle factors—could influence AE rates. Additionally, it was not possible to determine if patients were on combination therapy of TNFi with a thiopurine or methotrexate which can affect the results. Cumulative corticosteroid dose can be associated with adverse outcomes; however, due to database constraints, we were unable to obtain data regarding this. There is also a possibility of indication bias as JAKi may have been preferentially prescribed in more refractory and severe cases of IBD; however, we would have observed an increased risk of several AEs related to higher disease activity in this cohort. It is also possible that patients could have met the outcome of interest outside of the participating HCOs or left the HCO. However, most HCOs in TriNetX include large academic centers. Additionally, on further evaluation of the cohorts, 89% and 91% in the JAKi and biologic cohort respectively, did not have an end of patient record during the 2-year follow-up. However, the magnitude of missed events and subsequent bias is difficult to determine and there is no prior data regarding this to our best knowledge. Finally, we were unable to perform subgroup analysis based on dosing of JAKi due to substantial missing data and a small sample size in patients with available data.
In conclusion, our study suggests that JAKi use was not associated with an increased risk of infections, MACE, VTE, and malignancy compared to biologic therapy in a real-world IBD cohort. Long-term safety data with regards to tofacitinib were also reassuring. These findings should reassure clinicians who incorporate JAKi into the management of moderate-to-severe IBD, yet they also highlight the importance of vigilant monitoring, especially with respect to vaccination, cardiovascular risk factors, and laboratory abnormalities. In the interim, a personalized, risk-based approach—balancing oral convenience and immunologic potency against the potential for infections, thrombosis, and metabolic abnormalities—remains the optimal strategy for guiding therapeutic decisions in clinical practice.
Notes
Funding Source
The authors received no financial support for the research, authorship, and/or publication of this article.
Conflict of Interest
Kochhar GS has served on advisory boards for CoreVetas Research, Eli Lilly, and GIE Medical; has acted as a consultant for Boston Scientific Endoscopy, Olympus Endoscopy, and Pentax Endoscopy; has received speaker fees from Eli Lilly; and holds stock options in DigBi Health. Dulai PS has served on advisory boards for and/or has received research support and/or speaker or consultation fees from AbbVie, Abivax, Adiso, Bristol Myers Squibb, Geneoscopy, GlaxoSmithKline, Janssen, Eli Lilly, Pfizer, Roivant, and Takeda. The remaining authors declare no conflicts of interest.
Data Availability Statement
The data underlying this article will be made available upon reasonable request to the corresponding author.
Author Contributions
Conceptualization: Kochhar GS, Khataniar H, Dulai PS, Desai A. Data curation: Kochhar GS, Khataniar H, Dulai PS, Desai A. Methodology: Kochhar GS, Dulai PS. Supervision: Khataniar H. Writing–original draft: Kochhar GS, Khataniar H, Dulai PS, Desai A. Writing–review & editing: Kochhar GS, Dulai PS. Approval of final manuscript: all authors.
Supplementary Material
Supplementary materials are available at the Intestinal Research website (https://www.irjournal.org).
Supplementary Table 1.
ICD-10-CM Codes Utilized to Identify Adverse Events in the JAKi and Biologic Cohorts
Supplementary Table 2.
Baseline Characteristics of Patients in the JAKi Cohort and Biologic Cohort before and after Propensity-Score Matching to Evaluate the Risk of Infectious Complications
Supplementary Table 3.
Baseline Characteristics of Patients in the JAKi Cohort and Biologic Cohort before and after Propensity-Score Matching to Evaluate the Risk of MACE, Malignancy and Other Non-Infectious Complications
Supplementary Table 4.
Baseline Characteristics of Patients in the JAKi Cohort and Biologic Cohort before and after Propensity-Score Matching to Evaluate the Risk of Laboratory Abnormalities
Supplementary Table 5.
Baseline Characteristics of Patients in the Tofacitinib Cohort and Biologic Cohort before and after Propensity-Score Matching