Article Text
PDF
Abstract
Objectives To compare the safety of Janus kinase inhibitors (JAKi) with that of tumour necrosis factor inhibitors (TNFi) and determine drug persistence among patients with rheumatoid arthritis (RA) and spondyloarthritis (SpA).
Methods We analysed data from patients included in BIOBADASER 3.0 and treated with JAKi or TNFi from 2015 to 2023 and estimated the incidence rate ratio (IRR) of adverse events and persistence.
Results A total of 6826 patients were included. Of these, 52% had RA, 25% psoriatic arthritis and 23% axial SpA. Treatment was with TNFi in 86%. The mean duration of treatment was 2.2±2.0 years with TNFi versus 1.8±1.5 with JAKi. JAKis were prescribed in older patients with longer term disease, greater comorbidity and later treatment lines and more frequently as monotherapy. The IRR of all infections and gastrointestinal events was higher among patients with RA treated with JAKi. Drug persistence at 1, 2 and 3 years was 69%, 55% and 45% for TNFi and 68%, 54% and 45% for JAKi. Multivariate regression models showed a lower probability of discontinuation for JAKi (HR=0.85; 95% CI 0.78–0.92) and concomitant conventional synthetic disease-modifying antirheumatic drugs (HR=0.90; 95% CI 0.84–0.96). The risk of discontinuation increased with glucocorticoids, comorbidities, greater disease activity and later treatment lines.
Conclusions Infections, herpes zoster and gastrointestinal adverse events in patients with RA tended to be more frequent with JAKi. However, prognosis was poor in patients receiving JAKi. Persistence was similar for TNFi and JAKi, although factors associated with discontinuation differed by diagnostic group.
- Antirheumatic Agents
- Arthritis, Psoriatic
- Arthritis, Rheumatoid
- Spondylitis, Ankylosing
Data availability statement
The data are stored according to research regulations at the Unidad de Investigación de la Fundación Española de Reumatología. The data contained in the database are available on reasonable request to members of the Spanish Society of Rheumatology.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The safety and persistence of Janus kinase inhibitors (JAKi) and tumour necrosis factor inhibitors (TNFi) for the treatment of rheumatoid arthritis (RA), psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA) has been demonstrated in clinical trials. However, real-world data are needed.
WHAT THIS STUDY ADDS
Treatment with JAKi is associated with a slight increase in the frequency of all infections, herpes zoster and gastrointestinal events. This increase is greater in patients with RA. However, adverse events were mild and had no effect on mortality.
Persistence of JAKi was similar to that of TNFi in RA and axSpA, and slightly better for TNFi in PsA.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
In a real-world setting, the use of JAKi for treatment of RA, PsA or axSpA is an alternative with an acceptable balance between efficacy and safety. Rheumatologists should consider a personalised ‘treat-to-target’ strategy depending on the disease, patient profile and preferences, cost for the health system and availability.
Introduction
Janus kinase inhibitors (JAKi) are oral targeted synthetic disease-modifying antirheumatic drugs (tsDMARDs) used frequently as second-line therapy in rheumatoid arthritis (RA) and spondyloarthritis (SpA).1 With more than 20 years of use in clinical practice, biological DMARDs (bDMARDs) have a robust efficacy and safety profile.2–4 Tumour necrosis factor inhibitors (TNFi) are recommended as the first-line biologic for RA, psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA), either in monotherapy or in combination with methotrexate after failure of conventional synthetic DMARDs (csDMARDs).5–7
JAKis were initially designed for treatment of RA and has now been approved for SpA. They have been marketed since 2012 in the USA and have been evaluated in multiple randomised clinical trials. However, safety in daily clinical practice is still under evaluation, especially after reports of potentially associated cardiovascular events and cancer.8 9 Data from the ORAL Surveillance study prompted the US Food and Drug Agency and the European Medicines Agency to issue an initial warning concerning the safety of tofacitinib and a subsequent warning about JAKi as a group.10 11
The four JAKis marketed in Spain (as of October 2023) are tofacitinib, baricitinib, upadacitinib and filgotinib,1–4 all of which are prescribed and reimbursed by the National Health System.12 Considering the increasing number of currently available options, an estimation of real-world safety and relative effectiveness could help rheumatologists to choose the appropriate therapy.
Real-world registries provide a unique opportunity to compare the safety and efficacy of JAKi and treatments with longer experience of use, such as TNFi, in daily clinical practice. In this study, we used data from the Spanish registry BIOBADASER 3.0 to evaluate and compare the safety profile of JAKi with that of TNFi and determine drug persistence in patients with RA and SpA.
Methods
Study design
BIOBADASER 3.0 (Spanish registry on adverse events (AE) of advanced therapies in rheumatic diseases; https://biobadaser.ser.es/) is an observational, nationwide prospective registry of patients with chronic inflammatory rheumatic diseases starting treatment with a bDMARD (original or biosimilar) or tsDMARD. BIOBADASER 3.0 has been continuously collecting patient data from routine clinical practice since it was first established in 2000.13 The main objectives of the registry are to assess the long-term safety of patients undergoing treatment with bDMARDs/tsDMARDs and to determine effectiveness.14 The registry was established by the Spanish Agency of Medicines and Medical Devices and the Spanish Foundation of Rheumatology in 2000. Phase III was initiated in 2015. The full protocol of BIOBADASER 3.0 is available online (https://biobadaser.ser.es/protocolo.aspx). The study was reported in accordance with the guidelines of Strengthening the Reporting of Observational Studies in Epidemiology and their extension for longitudinal observational drug studies in rheumatology.15 16
Participants
We included adult patients (≥18 years) diagnosed with RA and SpA (PsA and axSpA) according to the criteria of the treating rheumatologist and initiating TNFi or JAKi between 2015 and 2023.
Time points
Baseline was defined as the initiation date for each course of treatment since January 2015. The course of treatment ran from the initiation date to the earliest discontinuation, the end of participation in the registry (patient withdrawal, death or loss to follow-up) or the end of the study period (October 2023).
Treatment groups
The treatment groups were patients with RA, patients with PsA and patients with axSpA treated with TNFi or JAKi. For patients who switched between TNFi and JAKi, each treatment course contributed to the corresponding treatment group; therefore, a patient could be counted in both groups.
Outcomes of interest
AEs during follow-up are collected routinely in BIOBADASER 3.0 (date of occurrence, type and classification according to the Medical Dictionary for Regulatory Activities (version 26.1),17 severity, outcome and concomitant treatment at the time of the AE). Safety was measured as the incidence of AEs during the risk window for each treatment (ie, duration of exposure plus a lag period).
Drug persistence was defined as the drug retention rate at 1, 2 and 3 years after initiation of therapy.
Variables
The data collected were as follows: (1) patient data, namely sex, date of birth, diagnosis, date of diagnosis and comorbidities (Charlson Comorbidity Index)18; (2) data on treatment, including type of TNFi (originals and biosimilars) and tsDMARD, dates of initiation and discontinuation and reason for discontinuation. Sex, age, disease duration, Charlson Comorbidity Index, line of treatment, concomitant medications and disease activity were considered confounding variables for adjustment. Disease activity was categorised as low, medium and high according to the thresholds defined for each disease activity index.19–24
Statistical analysis
Patient characteristics at baseline (initiation of treatment) were described using percentages and mean and SD according to the type and distribution of the variables. We estimated the incidence rate of AEs per 1000 person-years and 95% CI by group. We used the ‘ever taken drug’ risk attribution model for AEs with long latency periods (malignancy and mortality), attributing all remaining AEs while the patient is actively receiving therapy plus a lag period beyond discontinuation. The risk window lag period was twice the drug half life and an additional 30–180 days depending on the type of event (eg, 30 days for respiratory disorders, 90 days for infections and 180 days for cardiovascular disorders).25 Comparisons of the incidence rate between groups were expressed as the incidence rate ratio (IRR), which was obtained by Poisson regression and adjusted for the confounding variables.
Drug persistence was calculated as the number of patients remaining on treatment divided by the number of patients at risk of discontinuation using Kaplan-Meier survival curves (excluding censored patients) and considering any of the reasons for discontinuation (eg, loss of efficacy, AEs) as a completion event. Log-rank tests were used to compare the survival curves of JAKi and TNFi across the different diagnoses. Cox regression (proportional hazards) was used to analyse the factors associated with differences in drug retention rates between JAKi and TNFi. All analyses were performed using Stata V.13.1 (StataCorp, College Station, Texas, USA).
Results
We included 6826 patients, 3513 (51.5%) with RA and 3313 (48.5%) with SpA, including 1742 (25.5%) with PsA and 1571 (23.0%) with axSpA. A total of 5899 (86.4%) patients received 7661 courses of treatment with TNFi, 1642 (24.0%) patients received 1974 courses of treatment with JAKi and 715 (10.5%) received both treatments. Demographic and disease-related variables are shown in table 1. Patients with RA were older, mostly female and had a higher comorbidity index. They received concomitant csDMARDs and glucocorticoids more frequently than patients with SpA.
Main clinical characteristics, by disease and group of treatment
Patients treated with JAKi were older at initiation of treatment (56.3 vs 52.1 years), had a longer disease duration (11.7 vs 9.1 years) and presented a higher comorbidity index. The prescription pattern was also different, with JAKi being more frequently prescribed in later treatment lines in both diagnostic groups and JAKi monotherapy more common in patients with RA. The use of glucocorticoids was more frequent in combination with JAKi than with TNFi in patients with SpA.
The most frequently prescribed JAKi was baricitinib (37.7%), followed by tofacitinib (30.0%), upadacitinib (26.4%) and filgotinib (5.9%). As some of these therapies have no indication in SpA, differences by diagnostic group were recorded. Mean baseline disease activity indices at initiation of therapy were similar in both groups. Patients discontinued TNFi and JAKi for similar reasons, namely inefficacy (56.7% and 60.1%, respectively) and AEs (21% and 26.6%, respectively).
Safety profile
The cumulative duration of exposure for RA was 6432.3 patient-years in the case of TNFi and 3165.1 patient-years in the case of JAKi. The equivalent values for SpA were 10302.9 and 375.8 patient-years, respectively.
The incidence rate of AEs per 1000 patient-years and the crude and adjusted IRR comparing JAKi with TNFi are shown in table 2. The IRR of fatal AEs and cancer was similar between treatments, with no clinical or statistically significant differences in the crude or the adjusted analysis.
Incidence rate per 1000 person-years of main adverse events
The risk of all infections, serious infections and herpes zoster was onefold to twofold higher among patients with RA treated with JAKi in the crude analyses. In the adjusted IRR analyses, the IRR was statistically significant only for all infections and herpes zoster. In patients with SpA, this difference was only statistically significant for all infections in the crude analysis, but not in the adjusted. The incidence of tuberculosis was similar in patients with RA treated with TNFi and JAKi. No cases of tuberculosis were reported among patients with SpA treated with JAKi. The most common infections were upper and lower respiratory tract infections, followed by SARS-CoV-2 infection.
The IRR of cardiac AEs was higher in patients with RA treated with JAKi in the crude but not in the adjusted analyses. No differences were found in the IRR of thrombotic/vascular AEs for patients with RA or SpA. The main presentations for cardiovascular AEs were atrial fibrillation, tachyarrhythmia, chest pain and myocardial infarction.
The IRR of pulmonary and gastrointestinal AEs was higher in patients with RA for JAKi in the crude analysis; however, this difference was not confirmed in the adjusted IRR. The main presentations of pulmonary AE were cold, dyspnoea and cough. There were five cases of pulmonary embolism in patients with RA treated with TNFi, six cases in patients with RA treated with JAKi and four in patients with SpA treated with TNFi. The gastrointestinal AEs observed were diarrhoea, abdominal and oropharyngeal pain, oral ulcers and nausea. No gastrointestinal perforations were reported among patients with SpA treated with JAKi. Additional data for AEs are shown in the online supplemental material.
Supplemental material
The most common causes of death were serious infections, cardiovascular events and cancer.
Drug persistence and associated factors
Drug persistence at 1, 2 and 3 years was 69.2%, 55.3% and 45.2% for TNFi and 68.3%, 54.0% and 44.9% for JAKi, respectively, in the study population as a whole (table 3 and figure 1). When stratified by diagnostic group, persistence of TNFi and JAKi was similar in RA (p=0.86) and axSpA (p=0.30). Significant differences in favour of TNFi were observed in the PsA group (p=0.016) (figure 2). However, follow-up time was limited for patients with PsA and axSpA treated with JAKi, as approval for these indications was more recent (ie, December 2019).12
Kaplan-Meier drug persistence curves by treatment, all patients. JAKi, Janus kinase inhibitor; TNFi, tumour necrosis factor inhibitor.
Kaplan-Meier drug persistence curves by treatment, stratified by disease. JAKi, Janus kinase inhibitor; TNFi, tumour necrosis factor inhibitor.
Drug persistence by group of treatment
Hazard ratios (HRs) with their 95% CI were estimated in both the bivariate and multivariate analyses to identify the variables associated with drug persistence (table 4). In the model that included all diseases, treatment with JAKi (HR=0.85; 95% CI 0.78–0.92), having axSpA (0.84; 0.76–0.95) and concomitant use of csDMARDs (0.90; 0.84–0.96) were associated with lower probability of discontinuation. On the other hand, the variables associated with a greater probability of discontinuation were higher comorbidity index (1.07; 1.03–1.11), concomitant treatment with glucocorticoids (1.13; 1.05–1.21), female sex (1.16; 1.08–1.24), the use of a second line of treatment (1.30; 1.20–1.40) or ≥3 lines (1.53; 1.41–1.65), and medium baseline disease activity (1.19; 1.10–1.28) or high baseline disease activity (1.34; 1.23–1.46). When stratified by disease, the models differed slightly from the general model, especially in patients with PsA and axSpA. Unlike RA, in the SpA models, JAKi, glucocorticoids and the comorbidity index were no longer associated with a lower probability of discontinuation, and female sex was associated with a greater risk of discontinuation, as were ≥3 lines of treatment and baseline disease activity.
Cox proportional hazards regression models for drug discontinuation
Discussion
The real-world safety of JAKi has been compared with that of TNFi in a limited number of studies. This analysis of BIOBADASER 3.0 included 6826 patients treated with TNFi and JAKi in daily clinical practice between 2015 and 2023. We provide real-world evidence of a similar safety profile for JAKi and TNFi.
A higher incidence of infection including herpes zoster, all infections and gastrointestinal events was recorded for patients with RA treated with JAKi; in patients with SpA, incidence was only high for all infections in the crude, but not in the adjusted analysis. However, patients initiating JAKi were older and biologic experienced and had a longer disease duration and higher comorbidity, as reported in other registries.26–29 In the JAK-pot study, a collaboration of European registries of patients with RA treated with bDMARDs and tsDMARDs including data from BIOBADASER, the probability of stopping treatment owing to AE was higher for JAKi than for TNFi.28
Our safety results confirm findings from a pooled safety analysis based on data from randomised clinical trials in patients with RA for tofacitinib,30 baricitinib,31 upadacitinib32 and filgotinib.33 Safety data for SpA are restricted to tofacitinib and upadacitinib, although the results are similar.34 35
Most real-world evidence is for tofacitinib36 because of its earlier market release. Our data are consistent with this finding based on the doses marketed in the summary of product characteristics and authorised by regulatory agencies.10–12
Patients with RA, PsA and axSpA have a higher cardiovascular risk, with the result that AEs such as major cardiovascular and thromboembolic events are expected.37 ,38 However, a higher incidence rate for cardiac AEs was only observed in patients with RA treated with JAKi in the crude analyses. The most common presentations of cardiovascular AEs were atrial fibrillation, tachyarrhythmia, chest pain and myocardial infarction, as reported elsewhere.38 Compared with patients with SpA, patients with RA treated with JAKi were older, had a longer disease duration and a higher comorbidity index, received glucocorticoids more frequently and had higher levels of disease activity. Although we tried to reduce the effect of these variables in the adjusted analysis, there is still residual confounding that may contribute to the observed results. In addition, since 2019, treatment with tofacitinib for RA and SpA has been limited to 5 mg two times per day and to patients under 50 years of age, while in older patients, it can be used if no other options are available.9–12 34 35 No differences were found in the IRR of thrombotic/vascular AEs for patients with RA or SpA.
Global drug persistence rates at 1, 2 and 3 years were good. In patients with RA and axSpA, persistence was similar for TNFi and JAKi; in PsA it was slightly better for TNFi. Data from the DANBIO and DERMIO registries showed similar persistence rates. In patients with RA, the highest drug retention rates were for baricitinib, adalimumab and etanercept, with differences according to the line of treatment. In patients with PsA, persistence was better for TNFi and poorer for tofacitinib. In axSpA, the highest retention rates were for TNFi and the poorest were for tofacitinib.27 Other real-world studies on persistence of TNFi and JAKi in these three diagnoses are scarce and with limited follow-up, although they report results similar to ours.26–29 36 39–41
In our models, the variables associated with drug persistence were different for each diagnostic group. JAKis were associated with a lower probability of discontinuation in the general model and the RA-specific model. In patients with SpA, female sex increased the risk of discontinuation. These results confirm previous observations, although the data published for PsA and axSpA are scarce and limited to 1 year.40 42
In patients with RA, glucocorticoids were associated with a higher probability of discontinuation. This association may be confounded by prescription bias, as glucocorticoids are prescribed to patients with more severe disease. Further research is necessary to clarify the relationship between glucocorticoids and drug persistence.
The strengths of our study include the large group of patients with different diagnoses treated in routine clinical practice. In addition, BIOBADASER is a consolidated registry that has followed up patients from throughout Spain for more than 20 years and in which data quality is monitored.
The main limitation of our study is the complexity involved in measuring variables associated with prescription bias in observational studies. An additional limitation is last resort bias, which is difficult to identify in secondary-use data. For this reason, we also included the line of treatment as an adjustment variable in the Cox regression model, after which the difference between JAKi and TNFi persisted. Similarly, some clinical and important variables associated with various risks (mortality, infection, gastrointestinal perforations and cardiovascular conditions, including thromboembolic risk) are difficult to measure and were not evaluated. Examples include a personal history of previous atherosclerotic disease, thromboembolic events, severe infections, colon diverticulitis, concomitant non-steroidal anti-inflammatory drugs,43 opioids,44 contraceptives with oestrogens and other drugs associated with these risks. For this reason, despite the attempts to model AE risks between treatment groups through adjusted IRR, there might still be residual confounding. In addition, JAKis were recently approved for treatment of axSpA, and the number of patients and long-term data decreased because of censored observations.
Conclusions
We recorded significant differences between patients with RA, PsA and axSpA treated with JAKi and TNFi in a real-world setting. In patients with RA, treatment with JAKi was associated with a slight increase in the frequency of AEs, particularly infections, herpes zoster and gastrointestinal events. However, these AEs were mild and did not increase mortality. Drug persistence was similar for JAKi and TNFi in patients with RA and axSpA, and only slightly higher for TNFi in patients with PsA. Treatment with JAKi and concomitant use of csDMARDs were associated with a lower risk of discontinuation, although this differed by diagnostic group. The variables increasing the probability of discontinuation were comorbidities, later lines of treatment, glucocorticoids and baseline disease activity, with differences by disease. In patients with PsA, a greater probability of discontinuation was associated with female sex, later lines of treatment and disease activity.
Data availability statement
The data are stored according to research regulations at the Unidad de Investigación de la Fundación Española de Reumatología. The data contained in the database are available on reasonable request to members of the Spanish Society of Rheumatology.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by the Hospital Clinic of Barcelona Ethics Committee (approval code: FER-ADA-2015-01). All procedures and materials complied with the International Conference on Harmonization Guidelines for Good Clinical Practice and the principles of the Declaration of Helsinki and with Spanish regulations on data protection and research. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors acknowledge the invaluable contributions of all the participating rheumatologists and their patients and the Spanish Foundation of Rheumatology for providing support to BIOBADASER and medical writing/editorial assistance during the preparation of the manuscript.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Josef S Smolen
Presented at A preliminary version of this work was presented at the EULAR 2021 Meeting (Hernández-Cruz B, Sánchez-Piedra C, Freire-González M, et al on behalf of BIOBADASER. OP0123 Safety profile of JAK-inhibitors versus TNF-inhibitors in real world clinical practice: data from a multicenter register. Ann Rheum Dis 2021;80:68–69).
Contributors All authors made substantial contributions to the design of the work and acquisition of data and their interpretation. LO-V, FS-A and CS-P performed the statistical analysis. BH-C, LO-V and IC wrote the first draft. All other authors participated in the final draft and reviewed it critically. BH-C, LO-V and IC accept full responsibility for the finished work and/or the conduct of the study, had access to the data and controlled the decision to publish. BH-C is the guarantor.
Funding BIOBADASER is partly funded by the Spanish Agency of Medicines and Medical Devices (AEMPS), AbbVie, Bristol-Myers Squibb, Celltrion, Eli Lilly, Galápagos-Gilead, Janssen, MSD, Novartis, Pfizer, Regeneron, Samsung Bioepis and UCB.
Disclaimer The pharmaceutical companies contribute equally through a partnership agreement with the Spanish Society of Rheumatology and have no role in the analysis, interpretation and presentation of data.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.