Article Text
PDF
Abstract
Objectives To compare the incidence of cardiovascular (CV) events in rheumatoid arthritis (RA) treated with janus kinase inhibitors (JAKi), tumour necrosis factor inhibitors (TNFi), or other biological disease-modifying antirheumatic drugs (bDMARDs), in clinical practice, and to contextualise these findings by comparing to the Swedish RA population and general population at large.
Methods Patients with RA initiating JAKi, TNFi and non-TNFi bDMARDs were identified in the Swedish Rheumatology Quality Register between 2016 and 2021. Through linkages to national registers, a cohort of patients with RA, general population comparators, as well as covariates and incident major acute CV event (MACE, including myocardial infarction, stroke and fatal CV events) were identified until 2022. Crude and age-sex standardised rates were calculated and HRs estimated from multivariable Cox regression models using TNFi as reference.
Results We identified 13 492 patients with RA initiating a JAKi, non-TNFi bDMARD or TNFi treatment. Among 3037 JAKi-initiators, 59 MACE events were observed. The age-sex standardised rates for MACE were similar in the JAKi (0.88 per 100 person years) and TNFi (0.91) cohorts. Fully adjusted models showed no increased rate of MACE with JAKi (HR=0.71, 95% CI 0.51 to 0.99), or non-TNFi bDMARD (HR=0.98; 95% CI 0.78 to 1.23) in comparison to TNFi. We found no evidence that this HR changed over time since treatment initiation. In a CV-enriched subset, we observed higher rates but similar HRs.
Conclusions As used in present clinical practice in Sweden, we found no evidence that CV risk is higher with JAKis than TNFis in RA.
- Rheumatoid Arthritis
- Biological Therapy
- Cardiovascular Diseases
- Epidemiology
Data availability statement
No data are available.
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
Based on the results of the large safety trial (ORAL Surveillance) of the janus kinase inhibitor tofacitinib, regulatory agencies issued warnings regarding the cardiovascular safety profile of the class of janus kinase inhibitors. Data on the cardiovascular safety profile of janus kinase inhibitors as used in clinical practice are scarce.
WHAT THIS STUDY ADDS
The risk of cardiovascular events including myocardial infarction, stroke and cardiovascular deaths is not higher during treatment with janus kinase inhibitors than with biological disease-modifying antirheumatic drugs.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
While reassuring, reason(s) for the seemingly discrepant results compared with the ORAL surveillance trial, and cardiovascular risks with longer term use, should be investigated.
Introduction
Patients with rheumatoid arthritis (RA) are at increased risk of cardiovascular (CV) diseases such as major acute cardiovascular events (MACE) including myocardial infarction (MI). This increase likely represents the net effect of an increased prevalence of CV risk factors (such as smoking) and comorbidities in RA, of short-term and long-term effects of RA-inflammation on the vasculature, and direct and indirect effects of RA treatments.1–6
Although the absolute excess CV risk in RA has declined over years, we and others have recently shown that a risk increase persists among patients with RA initiating treatment with biological disease-modifying anti rheumatic drug (bDMARDs). This was observed even though there is little to suggest that individual bDMARDs would have markedly different effects on CV risk, and even if certain bDMARDs may by themselves affect CV-related biomarkers such as lipid levels.2 7 8
For the most recently approved class of DMARDs, janus kinase inhibitors (JAKis, a targeted synthetic (ts)DMARD), phase 3 clinical trials for tofacitinib and baricitinib have been promising in terms of short-term efficacy and safety in comparison to tumour necrosis factor inhibitor (TNFi).9 10 However, the U.S. Food and Drug Administration (FDA) required a prospective trial comparing tofacitinib to the most common class of RA bDMARD, TNFis, in order to address safety concerns, as the drug development programme indicated increased serum lipid levels (as well as other potential risks). The resulting ORAL surveillance trial was a randomised, open-label, non-inferiority, postauthorisation trial in a CV risk-enriched population that reported patients with RA on JAKi to be at 30% increased risk of MACE compared with those treated with TNFi.11 Since this finding, the FDA has issued warnings for increased risk for CV events (among others) for the entire class of JAKis (https://www.fda.gov/drugs/drug-safety-and-availability/fda-requires-warnings-about-increased-risk-serious-heart-related-events-cancer-blood-clots-and-death).12 The European Medicines Agency has also initiated a safety review for JAKis used in inflammatory disorders (https://www.ema.europa.eu/en/news/ema-starts-safety-review-janus-kinase-inhibitors-inflammatory-disorders).13
With JAKis representing an (effective-wise) much-needed option for the treatment of RA in clinical practice, with CV disease being a major comorbid condition in RA, and with additional indications for JAKis being investigated, this warning is of major clinical concern. It is, however, not clear if the results of this multinational, multicentre non-inferiority trial in a CV-enriched population is applicable to patients with RA in clinical practice. Further, it is also unclear whether the increased risk observed with JAKi versus TNFi in the trial was due to a decreased CV risk with TNFi (proposed to be the case in comparison to non-bDMARDs14), a true increase with JAKi, or both.
We therefore aimed (1) to assess the incidence of CV events in patients with RA treated with JAKi versus TNFi in clinical practice, (2) to contextualise these risks by estimating rates and relative rates in patients with RA treated with non-TNFi bDMARDs, in the entire RA population irrespective of treatment, and in the general population.
Methods
Study design
We conducted a nationwide register-based, active comparator, new-user cohort study that made use of prospectively recorded and linked data from clinical and other Swedish registers.
Setting
In Sweden, healthcare and drug prescriptions are publicly funded with an annual threshold of payment for the individual of €230. The vast majority of patients with RA are treated by rheumatologists at public rheumatology clinics. Use of DMARDs and other drugs approved by the national drug reimbursement agency is at the discretion of individual physicians with guidelines issued by the Swedish Society for Rheumatology.
Data sources
Longitudinal information on patients with RA and their treatments was obtained from the Swedish Rheumatology Quality register (SRQ) as part of a register linkage (Anti-Rheumatic Therapies in Sweden).15 This linkage contains data from the National Patient Register (NPR), the Prescribed Drug Register, the Cause of Death Register, the Swedish Population Register, the Swedish National Cancer Register and Longitudinal integrated database for health insurance and labour market studies (LISA). Online supplemental table S1 describes these sources in further detail.
Supplemental material
Study population, study period and exposures
We identified all individuals aged 18 years or older with RA recorded in the SRQ. Within this population, we identified individuals who initiated treatment with a (1) TNFi (adalimumab, certolizumab pegol, etanercept, golimumab, infliximab), (2) any non-TNFi bDMARD approved for RA (rituximab, abatacept, tocilizumab, sarilumab), or a (3) JAKi (baricitinib, tofacitinib, upadacitinib, filgotinib) any time during our study period (1 January 2016 through 31 December 2021). The JAKi cohort was additionally categorised into the separate JAKi treatments, excluding filgotinib due to too few initiations during the study period (n=7).) Note that treatment initiations for JAKis were only available from 2017 when tofacitinib was introduced to the Swedish market. One individual could contribute to more than one treatment cohort, and with more than one treatment initiation (of different molecules, each representing its own date of treatment start) to each study cohort but only the first initiation of each treatment molecule during the study period was included. For instance, a patient initiating treatment with etanercept who then switched to adalimumab and subsequently to baricitinib contributed both TNFi treatments to the TNFi cohort and the baricitinib treatment to the JAKi cohort. Treatments that were ongoing (vs new starts) at the beginning of the treatment study period were not included. Treatments before the study period did not disqualify entry into the treatment cohorts during the study period, and were included when counting the number of previous b/tsDMARDs. Using validated algorithms, we further identified the entire RA population in Sweden (the ‘all RA cohort’) defined by at least two separate visits listing RA at a rheumatology or internal medicine clinic before or during the study period.16 Through the Swedish Population Register, we identified five general population comparator subjects to every patient with RA contributing to the b/tsDMARD treatment cohorts, individually matched on year of birth, sex and region of domicile.
Outcome
The NPR and Cause of Death register were used to identify the outcomes of interest: MACE, acute MI including unstable angina, stroke, and fatal CV event, using International Classification of Diseases version 10 (ICD-10) codes. Previous validations using medical charts have indicated a positive predictive value for acute MI and stroke ICD-codes of above 90% in patients with and without RA.17–19 Individuals with a history of the outcome in question before start of follow-up were not excluded. Full outcome definitions are available in online supplemental table S2.
Follow-up
For the b/tsDMARD treatment cohorts, start of follow-up was defined as the registered date of treatment start in SRQ. For the all RA cohort, follow-up started at the time-point of the second ICD-10 code registration for RA during the study period. For the matched general population cohort, follow-up started at the time-point of the first recorded b/tsDMARD treatment initiation in the corresponding index-individual with RA. In the main analysis, we used an on-drug approach. Follow-up thus ended at the earliest of a first recorded CV event of interest, or censoring at treatment discontinuation+60 days (for treatment cohorts), emigration from Sweden, death, end of follow-up on 31 December 2022.
Covariates
Information on age, sex, comorbidities (history of: cancer, diabetes, heart failure, ischaemic heart disease, hospitalised infections, lung disease, kidney failure, stroke, surgery, and venous thrombotic events; days in hospital during the previous year) at start of follow-up was obtained from the National Cancer Register and NPR. Information on education and civil status at start of follow-up was obtained from LISA. Disease duration, seropositivity, C reactive protein (CRP), DAS28CRP (disease activity score on 28 joints), smoking status and the number of previous b/tsDMARDs was obtained from the SRQ at the start of follow-up. Use of oral prednisolone, methotrexate, other conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) and prednisolone, plus dispensations of other treatments (beta-blockers, anticoagulants, ACE inhibitors, lipid-lowering drugs, aspirin, diuretics) at start of follow-up were obtained from the Prescribed Drug Register. Full definitions can be found in online supplemental table S3.
Statistical methods
Descriptive statistics of covariates at treatment start were summarised. Events, crude and age-sex-standardised (to the distribution in the TNFi cohort) rates were calculated. HRs comparing cohorts to the TNFi cohort were estimated from Cox proportional hazards models using time-since-treatment initiation as the timescale, and robust SEs were estimated. Models were adjusted for age, sex and number of previous biologics (HR1); additionally, for comorbidities, medication use, civil status and education (HR2); further for RA disease-related factors, smoking, csDMARDs and prednisolone use (HR3). The latter two of these models did not include the general population cohort since some of the covariates were not available/relevant for this cohort. Missing data were essentially zero for all variables except RA-related covariates obtained from rheumatology visits recorded in the SRQ; see online supplemental table S4 for the proportion of missing information. HR3 therefore included missing categories for these variables. Models using a complete case and multiple imputation approach were performed as sensitivity analyses. Note that for the all RA population, only rates were calculated; no comparative analysis was performed since this cohort included patients from the treatment-defined RA cohorts.
Similarly adjusted Cox regression models to estimate the HR separately for tofacitinib and for baricitinib (vs TNFi) were also performed (upadacitinib was not included due to too few exposed/outcomes as a result of later market entry). Analyses were also performed separately by the number of previous b/tsDMARDs (0, 1–2, ≥3), and by time since treatment initiation by including an interaction-term between exposure and the time since start of follow-up (≤1, 1–2, ≥2 years). Flexible parametric survival models were also fitted to allow for non-proportional hazards of the effect of cohort (ie, allow the HR to change across time), and present these smoothly across time since treatment initiation.20 Wald tests were performed to assess statistical significance of non-proportionality.
Comparative analyses were only performed where the number events in each cohort were ≥5. SAS and Stata V.16.1 were used to perform analyses.
Sensitivity analyses
To facilitate comparison of the results to the ORAL surveillance trial, we selected a CV-enriched population (see online supplemental table S5 for definition based on age 50+ years and presence of at least one CV risk factors) and ran our main analyses also in this population. We further performed a sensitivity analyses restricting to patients with a history of one TNFi who started either a second TNFi, a first non-TNFi or a first JAKi, to assess the impact of our study design.
Results
We identified 13 492 patients with RA contributing to the JAKi, non-TNFi bDMARD or TNFi treatment cohorts: the JAKi cohort comprised 3037 treatment initiations (470 tofacitinib, 1973 baricitinib and 587 upadacitinib), the non-TNFi bDMARD 5396, and the TNFi 11 307 treatment initiations. Sixty-two thousand one hundred and two general population comparators were followed. Characteristics at start of follow-up are presented by treatment cohort in table 1. Patients initiating a TNFi tended to have had RA for fewer years, less severe disease and fewer previous b/tsDMARDs than those in the non-TNFi bDMARD and JAKi cohorts. The proportion of JAKi patients with a history of any CV comorbidity was slightly higher than in the TNFi cohort, but all proportions remained at 5% or below.
Baseline characteristics of individuals diagnosed with rheumatoid arthritis in Sweden and initiating treatment with a JAKi, non-TNFi bDMARD or TNFi, and general population comparators 2016–2021
Main analysis
The median (IQR, maximum) follow-up time in the JAKi, non-TNFi bDMARD and TNFi cohorts was 1.62 (2.54, 5.9), 1.60 (2.72, 7.0) and 1.71 (2.74, 7.0) years, respectively. In the JAKi cohort, 59 MACE events were recorded resulting in a crude rate of 0.98 per 100 person years. Of these, 9 occurred in patients treated with tofacitinib and 45 in patients treated with baricitinib. The age/sex-standardised rates of MACE per 100 person years were 0.88 in the combined JAKi cohort, 1.15 in the non-TNFi bDMARD cohort, 0.91 in the TNFi cohort, 1.38 in the all RA cohort, and 0.72 in the general population cohort.
Figure 1 presents the Kaplan-Meier failure function for MACE. Table 2 displays events, crude and standardised rates, and HRs estimated from Cox proportional hazards models for all CV outcomes. Borderline statistically significant decreased rates of MACE and of acute MI were found for JAKi versus TNFi (adjusted HR 0.71, 95% CI 0.51 to 0.99, and 0.58, 95% CI 0.36 to 0.94 for these two outcomes, respectively). No other statistically significant increased or decreased rates of CV outcomes were found in any of the b/tsDMARD cohorts verus TNFi, with little variation in HRs across models, although point estimates were numerically lower for JAKi versus TNFi for all CV outcomes (see online supplemental figure S1). Complete case and multiple imputation using chained equations analyses gave similar results (online supplemental table S6).
Events, person years at risk, crude and standardised rates, and HRs estimated from Cox proportional hazards models in patients with rheumatoid arthritis (RA) treated with JAKis, non-TNFi bDMARDs and TNFis for cardiovascular (CV) disease outcomes
Kaplan-Meier failure function for time-to-major acute cardiovascular event in patients with rheumatoid arthritis treated with janus kinase inhibitor (JAKi), non-tumour necrosis factor inhibitor (TNFi) biological disease-modifying anti rheumatic drug (bDMARD) and TNFi, and general population comparators in Sweden. Note the reduced y-axis range.
Analyses by time since treatment initiation and by number of previous biologics
Figure 2 presents fully adjusted HRs for MACE from the main analysis, by time since treatment initiation, and number of previous b/tsDMARDs. Figure 3 presents the HRs for MACE when considering time since treatment start continuously, from proportional hazards models and when allowing the effect of treatment cohort to vary over time. Although point estimates varied across time since treatment initiation (when allowed to), we found no statistical evidence of a difference in HR across time, indicating that the decreased risk of CV outcomes with JAKi was consistent across time since treatment initiation (see also online supplemental table S7). Further, when the HR was allowed to vary continuously across time since treatment initiation (figure 3), despite increasing across time, the HR remained below 1 during the follow-up. Results from analyses performed separately on strata defined by the number of previous b/tsDMARDs indicated that there was a decreased risk of MACE for those on JAKis and non-TNFi bDMARDs (vs TNFi) with more than three previous b/tsDMARDs (HR3 for JAKi vs TNFi was 0.40, 95% CI 0.21 to 0.76); see also online supplemental table S8 and S9. Sensitivity analyses assessing the effect of our study design, restricting to treatment initiations among individuals with a history of one prior TNFi use, showed results almost identical to those of the main analysis; see online supplemental figure S2.
HRs comparing janus kinase inhibitors (JAKis) and non-tumour necrosis factor inhibitor (TNFi) biological disease-modifying antirheumatic drug (bDMARDs) to TNFis for rates of major acute cardiovascular (CV) event in patients with rheumatoid arthritis in Sweden. Estimated from Cox proportional hazards models adjusting for age, sex, number of previous biological therapies, disease history, concomitant therapies, disease-related factors, smoking and socioeconomic factors. Main analyses, subgroup analyses restricted to a CV-enriched population, estimates by previous number of b/tsDMARDs, and estimates where the HR could vary over time-since treatment initiation.
HRs comparing janus kinase inhibitors (JAKis) and non-tumour necrosis factor inhibitor (TNFi) biological disease-modifying antirheumatic drug (bDMARDs) to TNFis for rates of major acute cardiovascular event in patients with rheumatoid arthritis in Sweden. Estimated from flexible parametric survival models allowing for non-proportional hazards and adjusted for age, sex, number of previous biological therapies, disease history, concomitant therapies, disease-related factors, smoking and socioeconomic factors.
Analyses in a CV-enriched population
Online supplemental table S10 provides descriptive statistics of the CV-enriched population. These patients had longer disease duration, and a larger proportion with comorbidities and medication use. Rates of CV events were higher than in the main analysis and comparative analyses resulted in point estimates for JAKi versus TNFi similar to those in the main analysis (eg, MACE HR3=0.62, 95% CI 0.41 to 0.93, table 3 and figure 1). Finally, post-hoc analyses restricted to those patients without history of ischaemic heart disease, heart failure or stroke (as defined in table 1) showed similar results (JAKi HR3=0.65, 95% CI 0.45 to 0.93; non-TNFi bDMARD HR3=1.02, 95% CI 0.80 to 1.32) versus TNFi.
Events, person years at risk, crude and standardised rates, and HRs estimated from Cox proportional hazards models in patients with rheumatoid arthritis (RA) treated with JAKis, non-TNFi bDMARDs and TNFis for cardiovascular (CV) disease outcomes: CV-enriched population
Discussion
In this observational, population-based, new-user, active comparator cohort study of patients with RA initiating JAKi, non-TNFi bDMARDs or TNFi in clinical practice, we did not observe any increased risk of MACE, MI, stroke or fatal CV events for patients with RA treated with JAKi in comparison to TNFi. Age-and-sex standardised rates of MACE were similar in the TNFi and JAKi cohorts, both approximately two-thirds of that presented for patients with RA in Sweden. We saw an increased incidence of CV events overall in a CV-enriched population but similar relative effect estimates for JAKi versus TNFi as in our main analysis.
In the open-label ORAL surveillance trial, some 4500 patients with RA were randomised 1:1:1 to tofacitinib versus a TNFi (etanercept/adalimumab). During an average of 3.45 years of follow-up (on drug, which in itself may be unexpectedly long in light of typical drug survival data from b/tsDMARDs in RA), 3.4% (tofacitinib arms combined) versus 2.5% (TNFi) developed MACE (HR=1.3, 95% CI 0.9 to 1.9).11 For several reasons, comparison between this multinational multicentre clinical trial and our nationwide population-based and practice-based observational study is not without challenges. First, we performed analyses in a CV-enriched population by age and CV risk factors, and not additionally selected based on MTX cotreatment or disease activity due to too few observations and events. Second, the ORAL surveillance trial consisted of tofacitinib only (moreover in two dosages, of which only the 5 mg two times a day was used in Sweden), whereas in our study, around 75% of all person-time and outcomes in the JAKi cohort pertained to baricitinib exposure. We note, however, that the age at entry (around 60 years) and incidences of MACE in the three arms of the ORAL surveillance trial (0.7–1.1 per 100 person years) were strikingly similar to those of the main analysis in our study (around 60 years, and 0.9–1.2 per 100 person years, respectively), but only around half those of our CV enriched population (around 1.5 per 100 person years), suggesting structural differences between the two populations. On the other hand, in our study, our version did not appreciably affect the HRs. In our study, we noted a higher crude incidence of MACE in the JAKi (vs TNFi) during the second year after treatment initiation, but numerically similar incidences otherwise. Importantly, our findings by time since treatment initiation show no statistically significant difference between rates of MACE in the JAKi versus TNFi across available follow-up. For MACE, the follow-up category 2+ years since treatment initiation had a HR with an upper 95% CI of 1.11, excluding clinically meaningful risk increases in this space. Although HRs by time since trial start were not presented in the ORAL surveillance report, visual inspection of the KM curves for MACE suggests a separation of the incidence curves already at 6 months.11 The reason for the low(-er) incidence of MACE with JAKi than TNFi in our study is unclear. Besides a true effect, it may be related to residual channelling bias, potentially in turn the result of a particular (here: beyond that for TNFi or non-TNFi bDMARDs) concern regarding safety risks with this more recently introduced class of DMARDs for which CV safety concerns (here: mainly for thromboembolic events) were communicated during our study period. Using similar types of data for bDMARDs, we have previously investigated and quantified the impact of such bias, which together with the short average time-on-drug may also be a reason for the seemingly low levels of increased (fatal) CV risks for all b/tsDMARD cohorts versus the general population cohort in our study21; bluntly put: patients selected to start treatment should be ill enough to need it yet expected to remain alive during the coming months. Therefore, in any practice-based drug-study in RA, the immediate mortality in treatment initiating cohorts is expected to be low compared with untreated patients, or even to the general population. Our results are in alignment with existing meta-analyses of MACE and CV outcomes (other than venous thromboembolism) in the shorter, pivotal clinical trials of JAKi, although based on small numbers, which did not suggest any increase in CV event risk with tofacitinib, nor with baricitinib.22 23
Our result of no increased rate of MACE in JAKi versus TNFi is in alignment with other observational studies of the same nature; in a Korean study (HR=0.64, 95% CI 0.33 to 1.22),24 for JAKis versus adalimumab in French study (HR=1.0, 95% CI 0.7 to 1.5)25 and a multicentre cohort study considering CV outcomes for JAKi versus TNFi (pooled incidence rate ratio=0.80, 95% CI 0.48 to 1.35).26 Note that the former two of these studies had maximum follow-up times of approximately 6.75 and 4.5 years, in contrast to our maximum follow-up of 6 years for JAKi; the latter study contained three cohorts with median follow-up of 0.91, 1.32 and 1.02 years for JAKis in comparison to our 1.62 years. Our results are also in alignment with a study by our group, with more limited follow-up, comparing rates of MACE in baricitinib users versus etanercept users (HR=0.83, 95% CI 0.49 to 1.42).27 STAR-RA, a US study that combined several claims databases, also found no evidence of any increased risk for CV endpoints in patients taking tofacitinib (baricitinib was not included) versus TNFi (pooled HR=1.01, 95% CI 0.83 to 1.23), nor in a CV risk-enriched subset (HR=1.24, 95% CI 0.90 to 1.69).28 Similar rates of MACE were found for tofacitinib and bDMARD initiators in a comparative safety study using the Corevitas RA registry in the USA.29
Our study has some limitations. As mentioned, our follow-up time was reflective of the relatively recent introduction of JAKis and our results thus mainly reflect the risk of CV events with JAKi versus TNFi during the first 3–4 years. Extended follow-up will be necessary for the assessment of longer term risks. Although using register-based data sources has many advantages, one disadvantage was the lack of information on potentially important confounders, such as body mass index. It is also possible that residual confounding by indication exists, despite our best attempts of adjustment for a wealth of disease-related and comorbidity-related variables, and stratification by the previous number of biologics. Similarly, despite adjustment for CV history in our analysis, it is possible that residual confounding, or differential CV surveillance and intervention, may remain. By design, our main analysis allowed multiple treatment episodes by class, which might have disproportionally enriched the TNFi treatment cohort with patients failing several drugs of the same mode of action. Importantly, however, the results of our sensitivity analysis conditioning on a history of one TNFi (effectively comparing TNFi, non-TNFi and JAKi as second b/tsDMARD treatment) came very close to the main analysis. We assessed variables at baseline (ie, covariates were updated at every b/tsDMARD treatment start) but did not consider any time-varying effects of these covariates during follow-up. Although we were able to capture a subpopulation with a higher CV risk, low number of observations and events prevented us from selecting a high-CV risk population additionally based on as the very same disease activity requirements as those used in the ORAL surveillance trial (swollen joint count>6, tender joint count>6 and CRP>0.3 mg/dL). Thus, the current results do not necessarily disagree with the ORAL surveillance trial, but pertain to the same patient segment. Although we aimed to study the different JAKis separately, limited number of outcomes in the (smaller) tofacitinib group prevented any comparative analysis for this JAKi versus TNFi. Similarly, low numbers of upadacitinib and filgotinib exposure prevented corresponding assessments of their CV safety. Additionally, adjusting for many confounding factors where the number of events are low, even when not presenting results for events less than 5, may not be entirely justified and should be interpreted in light of potential few events. However, we believe that collectively these adjusted results demonstrate that there is not one large, or several minor, confounders that could be measured. Finally, this study focused on MACE, MI, stroke and fatal CV events; the results should not be generalised to other CV outcomes.
Our study also has a number of strengths. Our setting permitted us to perform a nationwide and population-based study, thus maximising generalisability, and to assess the occurrence of outcome independently of exposure through prospective register linkages,17 which further allowed us to accommodate effects of potential confounders and effect modifiers. Importantly, and in contrast to the ORAL surveillance study, we were able to contextualise the relative risk estimate comparing JAKi to TNFi, and could demonstrate that the CV risk with non-TNFi bDMARDs, and for the entire RA population as a whole, were close to those of the JAKi and TNFi cohorts. Such contextualisation is necessary when comparing two treatments and to the best of our knowledge has not yet been presented. We were also able to present the components of MACE separately, although power did limit our comparative analyses for these separate outcomes.
To conclude, our observational population-based Swedish cohort study suggests that, as used in clinical practice against RA, JAKis are not associated with higher CV risks than bDMARDs.
Data availability statement
No data are available.
Ethics statements
Patient consent for publication
Ethics approval
The study was approved by the Swedish Ethical Review Authority (2021-00096).
Acknowledgments
The ARTIS Study Group conducts scientific analyses using data from the Swedish Biologics Register ARTIS, run by the Swedish Society for Rheumatology. The following were members of the ARTIS Study Group during study completion: Gerd-Marie Alenius (Department of Public Health and Clinical Medicine/Rheumatology, Umeå University), Eva Baecklund (Department of Medical Sciences, Uppsala University), Katerina Chatzidionysiou (Rheumatology, Theme Inflammation and Ageing, Karolinska University Hospital), Nils Feltelius (Swedish Medical Products Agency, Department of Public Health and Caring Sciences, Uppsala University), Helena Forsblad-d’Elia (Department of Rheumatology and Inflammation Research, Sahlgrenska Academy, University of Gothenburg), Alf Kastbom (Department of Biomedical and Clinical Sciences, Linköping University), Lars Klareskog (Department of Medicine Solna, Karolinska Institutet), Ann Knight (Department of Medical Sciences, Uppsala University), Elisabet Lindqvist (Department of Clinical Sciences Lund, Lund University, Skåne University Hospital), Ulf Lindström (Department of Rheumatology and Inflammation Research, Sahlgrenska Academy, University of Gothenburg), Carl Turesson (Rheumatology, Department of Clinical Sciences Malmö, Lund University, Skåne University Hospital), Christopher Sjöwall (Department of Biomedical and Clinical Sciences, Linköping University), and Johan Askling (Department of Medicine Solna, Karolinska Institutet and Rheumatology, Theme Inflammation and Ageing, Karolinska University Hospital).
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
Contributors All coauthors participated in the design of the study. HB conducted the statistical analyses. All authors contributed to interpretation of the results. HB and JA contributed to the drafting of the manuscript. All authors contributed to the critical revision of the manuscript for important intellectual content. HB is responsible for the overall content as guarantor.
Funding The underlying linkages and some of the salary costs for the ARTIS b/tsDMARD safety monitoring programme were covered through agreements between Karolinska Institutet and the following companies: AbbVie, BMS, Eli Lilly, Pfizer, Samsung Bioepis, and Sanofi. The research environment was also supported through grants from the Swedish Research Council, the Swedish Heart-Lung foundation, the Swedish Cancer Society, Vinnova, NordForsk, and agreements between Region Stockholm and Karolinska Institutet (ALF).
Competing interests Karolinska Institutet, with JA as principal investigator, has or has had research agreements with AbbVie, Astra-Zeneca, BMS, Eli Lilly, Galapagos, MSD, Pfizer, Roche, Samsung Bioepis, Sanofi, and UCB, mainly in the context of safety monitoring of biologics via ARTIS/Swedish Biologics Register.
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.