Abstract
Objective To conduct a systematic review of the effectiveness and safety of pharmacological treatments for adult-onset Still disease (AOSD).
Methods Six databases, 2 trial registries, and conference abstracts were searched from January 2012 to February 2023 for studies of pharmacological interventions in people with AOSD. Outcomes were rates of remission and response, discontinuation of concurrent treatments, complications of AOSD, and treatment-related adverse events. Risk of bias was assessed with the Cochrane risk of bias tool and the Joanna Briggs Institute tool for case series.
Results Forty-four studies evaluated treatments, including nonsteroidal antiinflammatory drugs (NSAIDs), corticosteroids (CS), conventional synthetic disease-modifying antirheumatic drugs (DMARDs), and biologic DMARDs (bDMARDs). For bDMARDs, tocilizumab (TCZ), anakinra (ANK), and canakinumab (CNK) had the most available data. Although 3 randomized controlled trials did not show statistically significant benefits of bDMARDs, metaanalyses showed high rates of complete remission and CS discontinuation. Complete remission was 80% (95% CI 59-92%, I2 36%), 73% (95% CI 58-84%, I2 66%), and 77% (95% CI 29-97%, I2 82%) and CS discontinuation was 57% (95% CI 29-81%, I2 66%), 47% (95% CI 18-78%, I2 79%), and 34% (95% CI 6-81%, I2 59%), respectively, for TCZ, ANK, and CNK. Studies with a higher proportion of patients previously treated with bDMARDs showed a trend toward lower rates of CS discontinuation (P = 0.05). The analyses had high clinical heterogeneity, largely because treatments were prescribed as different lines of therapy.
Conclusion Evidence supports TCZ, ANK, and CNK therapy for AOSD. However, the magnitude of effect and comparative effectiveness of treatments is uncertain.
- adult-onset Still disease
- biologics
- corticosteroids
- disease-modifying antirheumatic drugs
- metaanalysis
- systematic review
Adult-onset Still disease (AOSD) is a rare systemic inflammatory disease of unknown etiology.1,2 Annual incidence and point prevalence of AOSD have been estimated as 0.16 to 0.4 per 100,000 and 2.7 to 3.9 per 100,000 people, respectively.3-6 The main symptoms are fever, joint pain, and evanescent skin rash that may be accompanied by severe systemic manifestations, including serositis.7,8
The clinical course of AOSD can be categorized into 3 patterns: monocyclic (characterized by a single flare with systemic manifestations and potentially joint involvement), polycyclic (characterized by intermittent recurrences that combine systemic manifestations and joint involvement), and chronic (characterized by persistent, worsening joint involvement with regular systemic flares or unremitting continuous disease, and needing long-term treatment).9,10 It has been estimated that 30% of patients have monocyclic AOSD, 30% have polycyclic AOSD, and 40% develop chronic patterns.11
Treatment for AOSD includes nonsteroidal antiinflammatory drugs (NSAIDs), corticosteroids (CS), conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), and biologic DMARDs (bDMARDs). The 3 most commonly used bDMARDs are anakinra (ANK),12,13 canakinumab (CNK),14,15 and tocilizumab (TCZ),16 although TCZ does not have regulatory approval for AOSD.
There have been a number of reviews of bDMARDs for AOSD, including all interleukin-1 inhibitors (IL-1i)17-20 or specifically ANK21-23 or CNK.24 Reviews of more recent treatments (Janus kinase inhibitors [JAKi]25) have also been conducted, but no systematic review has covered the entire range of possible treatments. In addition, no previous review has included publications of recent analyses of data from the Autoinflammatory Disease Alliance (AIDA) international registry dedicated to Still disease.26-28
Therefore, the objective of this systematic review was to evaluate the available evidence for the effectiveness and safety of all pharmacological treatments for AOSD. As the evidence from randomized controlled trials (RCTs) is limited, the review also includes observational evidence. To improve the applicability and homogeneity of findings, the review focuses on studies published in the last 10 years, when the more common bDMARDs came into use and methods for diagnosis and treatment became more consistent with current practice.
METHODS
This systematic review followed the principles embodied in the Cochrane handbook29 and guidance published by the Centre for Reviews and Dissemination (CRD)30 and is reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) statement.31 The protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42022350576).32
Eligibility criteria. Studies of adults with diagnosed AOSD and treated with pharmacological therapies were eligible for review. Studies of any design (except case reports) assessing any licensed or unlicensed treatments were eligible. Studies published since 2012 (as per the date of the first published bDMARD RCT33) in any language were included. Eligible outcomes were rates of remission and response, discontinuation of concurrent treatments, complications of AOSD, and treatment-related adverse events (AEs). Detailed eligibility criteria are shown in Supplementary Table S1 (available with the online version of this article).
Patient consent. Informed patient consent was not required for this systematic review.
Searches. A MEDLINE (OvidSP) search strategy was designed to identify AOSD studies (see Supplementary Figure S1, available with the online version of this article), and the final strategy was peer reviewed by a second information specialist. The strategy comprised a single concept: AOSD (search lines 1-10).
The strategy was not restricted by language, but animal studies, editorials, and news items were excluded. Reflecting the review’s eligibility criteria, the strategy was restricted to studies published from January 2012 to August 2022 (search line 14). An update search was conducted in February 2023 in the same resources using the same search strategies.
Searches were translated appropriately (Supplementary Figure S1, available with the online version of this article) and were conducted in the databases and information resources shown in Supplementary Table S2. The proceedings of the European Alliance of Associations for Rheumatology (EULAR) 2022 Congress were hand searched.
The reference lists of any relevant systematic reviews published in the last 3 years were checked for any eligible studies that might have been missed by the database searches. For all papers included in the review, retractions, erratum notices, or corrections were also checked for included studies.
Study selection and data extraction. Following deduplication, 2 reviewers (ME and M. Cikalo) independently assessed the titles and abstracts of remaining records for relevance against the eligibility criteria, with disagreements resolved through discussion. Full texts for the selected records were obtained and screened by 2 reviewers (ME and M. Cikalo). One reviewer (J. Patterson) extracted data on a previously piloted data sheet, and a second researcher (MC) checked all the data points. For each outcome, data were collected at all timepoints reported.
Risk of bias assessment. One reviewer assessed the risk of bias (ROB) of each included study, and a second reviewer checked the ROB assessment. The comparative studies were assessed using the Cochrane ROB I tool,34 whereas the case series were assessed using the Joanna Briggs Institute (JBI) case series checklist.35 ROB plots were generated using the Risk-of-Bias Visualization (robvis) tool.36
Definitions. As no standardized clinical or operational definitions of remission in AOSD are available, the following definitions were used for reporting outcomes across studies:
• Complete drug-free remission: The disappearance of all clinical and biological manifestations and withdrawal of all pharmacological treatments without a disease flare.
• Complete intervention-free remission: The disappearance of all clinical and biological manifestations and withdrawal of intervention treatment (the treatment evaluated in the included research study) without a disease flare. Patients could continue concomitant medications.
• Complete remission: The disappearance of all clinical and biological manifestations.
• Clinical response: The disappearance of all clinical manifestations.
• Partial response: A partial improvement in clinical and/or biological manifestations.
If studies used different definitions of remission and response, these definitions were mapped onto those listed above in order to provide consistency for analysis.
Statistical analysis. The extracted data were checked for availability and consistency in outcome definitions to assess the feasibility of metaanalysis of the single-arm data. Where there was consistency in outcomes and sufficient available data, metaanalysis was considered. Due to the large clinical and, in some cases, statistical heterogeneity, results were statistically pooled using random-effects metaanalyses, which generated a pooled estimate of the proportion of individuals experiencing each outcome measure for a given treatment. Robustness of the estimates was assessed through tests for heterogeneity (I2), publication bias (Egger test), outliers, and leave-one-out analysis. Potential sources of heterogeneity were evaluated through sensitivity (removing studies with short follow-up) and subgroup (separating studies where more than half, compared with less than half, of participants had previously been treated with a bDMARD) analyses.
All statistical analyses were conducted using R version 4.2.1.37 Additional R packages used included meta (version 6.0-0),38 dmetar (version 0.0.9000),39 tidyverse (version 1.3.2),40 and metafor (version 3.8-1).41 Results were presented as forest plots, with the significance level set at P ≤ 0.05.
RESULTS
Results of the searches. The original searches identified 3926 records. Following deduplication, 2531 records were assessed for relevance (2506 from database searches and 25 from other sources; see Supplementary Figure S2, available with the online version of this article). The update searches identified 229 additional records that were assessed for relevance.
Following screening of titles and abstracts, 2498 records were excluded and 262 full-text publications were assessed. Of these, 194 were excluded, and 44 studies, reported in 68 publications, were included in the review (Supplementary Figure S2, available with the online version of this article). Study, intervention, and patient characteristics are shown in Supplementary Table S3 and Supplementary Table S4. Summary data for the included studies are shown by intervention in Table 1.
Study, intervention, and patient characteristics summary.
Characteristics of the included studies. Across the 44 included studies,4,26,27,33,42-81 there were 3 RCTs (Canakinumab for Treatment of Adult-Onset Still’s Disease to Achieve Reduction of Arthritic Manifestation [CONSIDER],64 Nordström et al,33 and UMIN00001298760), 1 phase II dosing study,58 and 40 case series. CONSIDER64 and UMIN00001298760 were placebo-controlled trials of CNK and TCZ, respectively. Nordström et al33 assessed ANK against csDMARDs. The phase II dosing study58 assessed 2 different doses of the IL-18i tadekinig alfa.
Studies ranged in size from 2 to 356 patients, although only 3 included > 60 patients. The largest 3 case series evaluated CS and csDMARDs56,61 and ANK.42 The 44 studies were conducted across a range of countries across Europe and Asia, and in the USA.4,26,27,33,42-81
Seventeen26,33,43,44,46,47,49,52,57,59,62,63,66,69,71-73 of the 44 studies reported that included patients had to be refractory to conventional treatment including CS and csDMARDs. In other studies, patients had undergone previous treatment, but failure on all conventional management was not an inclusion criterion. Where reported (27/44 studies), duration of follow-up varied widely, ranging from 1 month63,72 to 15 years.44
When comparing different types of interventions, it was apparent that, for some interventions, studies tended to be in patients who were older (TCZ, CNK) and had an AOSD diagnosis for longer (TCZ, CNK; Table 1). For studies of ANK and TCZ, where reported, a lower proportion of patients had previous bDMARD treatment compared with studies of CNK (Table 1).
Quality of the evidence. ROB assessments are shown in Supplementary Figures S3A,B (available with the online version of this article). Of the 4 controlled trials, 2 were open label.33,58 In UMIN000012987,60 the authors noted that patients receiving placebo had higher disease activity, so the 2 arms were not similar in all respects at baseline. Overall, UMIN000012987,60 Nordström et al,33 and Gabay et al58 were considered to have unclear ROB, and CONSIDER64 to have low ROB.
All 4 controlled trials had small sample sizes. In CONSIDER,64 recruitment was stopped prematurely, with enrollment of 36 out of 68 planned patients; therefore, the study was underpowered. Nordström et al33 was also underpowered, with only 22 patients instead of the required 60. The UMIN000012987 trial60 had its sample size reduced while the trial was already underway. Gabay et al58 had a small sample size and no power calculation was reported.
For the case series, 23 were judged to be at a low ROB4,42-46,48-50,52,53,55-57,59,61-63,65,66,68-70 and 17 at a high ROB.26,27,47,51,54,67,71-81 Reporting was often not clear or complete. Ten of the case series showed high ROB in 1 or 2 fields, most commonly relating to poor or nonexistent reporting of patient baseline characteristics or outcomes.51,69,71,72,74-77,79,81 This was often due to the case series being reported in abstracts rather than full papers, with 7 abstracts47,72,75-78,80 flagged as having a high ROB.
Publication bias was assessed for the subset of studies contributing to each of the metaanalyses. Although funnel plots of some of the analyses were asymmetrical, the Egger test produced P > 0.18, suggesting that the quantitative analyses were at low risk of publication bias.82
Results of the review. Supplementary Table S5 and Supplementary Table S6 (available with the online version of this article) present raw data for efficacy and safety outcomes for each study.
Generally, the 3 RCTs did not report significantly positive findings, but all showed trends to positive effect. For the primary outcome, in the TCZ trial,60 American College of Rheumatology (ACR) 50 response at week 4 was higher for TCZ compared with placebo (62% vs 31%, P = 0.24); in the ANK trial,33 remission at 8 weeks was higher for ANK compared with csDMARDs (58% vs 50%, P value was not significant); and in the CNK trial,64 the proportion of patients with a clinically relevant reduction in articular manifestations (ie, change in Disease Activity Score in 28 joints based on erythrocyte sedimentation rate [DAS28-ESR] > 1.2) was higher for CNK compared with placebo (67% vs 41%, P = 0.18).
Data for the most frequently reported outcomes across the review (complete intervention-free remission, complete remission, CS discontinuation, and discontinuation due to AEs) are summarized across studies in Table 2. Following an assessment of the available data, metaanalysis of 2 key outcomes (complete remission and CS discontinuation) was possible for TCZ, ANK, and CNK. Results of these metaanalyses are presented in Figure 1.
Results summary.
Metaanalyses for (A-C) complete remission or (D-F) corticosteroid discontinuation in adult-onset Still disease treated with tocilizumab, anakinra, or canakinumab, respectively. CONSIDER: Canakinumab for Treatment of Adult-Onset Still’s Disease to Achieve Reduction of Arthritic Manifestation.
Complete intervention-free remission. Data on the proportion of patients experiencing complete intervention-free remission were available for TCZ (6 studies including 97 patients50,55,57,62,71,74), ANK (4 studies including 208 patients42,53,62,71), and CNK (1 study of 11 patients76; Table 2). A large range of results was observed. Nonetheless, between 14% and 73% of patients (across 6 studies4,55,57,62,71,74) achieved complete intervention-free remission with TCZ. The 4 ANK studies42,53,62,71 reported between 2% and 39% of patients with intervention-free remission, whereas the single CNK study76 reported 9%.
Complete remission. Complete remission was reported for 9 interventions (NSAIDs,56 CS,56,61 csDMARDs,33,45,56 TCZ,4,44,50,62,69,71 ANK,33,42,47,52,54,62,71 CNK,27,64,66,68 infliximab [IFX],44 etanercept [ETN]44,47 and JAKi63,72,80,81), where each showed large ranges in outcome across studies (Table 2). For 3 interventions, there was enough data for metaanalysis, which showed that across different lines of bDMARD treatment (effect of line of treatment shown in subgroup analysis section below), the proportion of patients achieving complete remission was 80% (95% CI 59-92%, I2 36%), 73% (95% CI 58-84%, I2 66%), and 77% (95% CI 29-97%, I2 82%) for TCZ, ANK, and CNK, respectively, with large overlapping 95% CIs (Figure 1A-C).
In the sensitivity analysis, removing studies with results < 3 months (or not reporting) follow-up (Supplementary Figure S4, available with the online version of this article) resulted in similar findings, but with higher rates of complete remission and reduced heterogeneity for TCZ and CNK: TCZ at 85% (95% CI 69-94%, I2 0%), ANK at 73% (95% CI 56-86%, I2 72%), and CNK at 84% (95% CI 48-97%, I2 26%).
Discontinuation and dosage of CS. Discontinuation of CS treatment was reported for 5 interventions (csDMARDs,33 TCZ,50,51,59,60,62,65,71 ANK,33,42,52,54,62,71 CNK,27,42,68,73 JAKi72,81; Table 2). Most data were available for bDMARDs and meta-analysis was conducted for 3 of these interventions. Metaanalysis showed that the proportion of patients who discontinued CS across different lines of bDMARD treatment (effect of line of treatment shown in subgroup analysis section below) was 57% (95% CI 29-81%, I2 66%), 47% (95% CI 18-78%, I2 79%), and 34% (95% CI 6-81%, I2 59%) in studies of TCZ, ANK, and CNK, respectively (Figure 1D-F). For this outcome, all studies had > 3 months of follow-up, and no studies were removed for short follow-up in the sensitivity analysis.
A number of studies reported CS dosage, and all showed a reduction from baseline. For TCZ, 10 studies showed a reduction,26,43,49-51,57,59,60,65,69 and for 7 of these, this was significant compared with placebo60 or baseline.49,51,57,59,65,69 For ANK, 4 studies reported reductions from baseline33,42,53,54 and 2 of these reported significant differences.53,54 For CNK, 3 studies showed a reduction,27,42,68 which was reported to be significant in 2 studies.27,68 For JAKi, there were also reductions, showing changes from baseline in 3 studies,63,72,80 but none reported the significance of the changes. A significant reduction was reported in the study of tadekinig alfa, an IL-18i.58
Discontinuation due to AEs. The proportion of patients stopping treatment due to AEs was available for 7 interventions (TCZ,4,43,44,49,55,57,60,69 ANK,42,53,54 CNK,73,76 IFX,44 ETN,44 JAKi,72 IL-18i58; Table 2). Larger datasets were available for TCZ, ANK, CNK, and IL-18i, where rates of between 6% and 56% were reported; there was no apparent trend between treatment type and rate of discontinuation, but a statistical comparison is difficult with the limited available data. The interventions with smaller datasets (≤ 10 patients) showed discontinuation rates of 0% (ETN) and 20% (IFX and JAKi).
Previous bDMARD use subgroup analysis. The subgroup analysis comparing studies with < 50% of patients with previous bDMARD use and those with ≥ 50% previous bDMARD use is shown in Supplementary Figure S5 (available with the online version of this article). All the CNK studies had ≥ 50% of patients with previous bDMARD use, whereas, where reported, only 3 of the 11 TCZ studies and none of the ANK studies met this criterion. In the combined analysis of studies of TCZ, ANK, and CNK, for studies where < 50% of patients had previous bDMARD treatment, complete remission was 81% (95% CI 69-89%, I2 42%), and for studies where ≥ 50% of patients had previous bDMARD treatment, it was 74% (95% CI 23-97%, I2 73%). No significant difference was reported between groups (P = 0.61). However, CS discontinuation was 69% (95% CI 33-91%, I2 71%) and 33% (95% CI 9-72%, I2 24%), respectively, in studies where < 50% and ≥ 50% of patients had previous bDMARD treatment, showing a trend toward a better outcome in studies with patients with less previous bDMARD treatment (P = 0.05).
DISCUSSION
This systematic review identified 44 studies published in the last 10 years of conventional, biologic, and novel treatments for AOSD, including 3 RCTs of bDMARDs (TCZ, ANK, CNK) and non-RCTs evaluating a number of early and more recent treatments. Findings from the metaanalyses are consistent with those conducted previously of bDMARDs for AOSD,22,83 but analyses include 8 additional studies published more recently. Despite the absence of conclusive findings in the RCTs, there remains a large body of evidence suggesting that pharmacological treatments, particularly bDMARDs, can benefit patients with AOSD in both the short and long term.
Direct comparison of different drugs’ effectiveness is limited to only 1 RCT comparing ANK and csDMARDs.33 Indirect comparisons are complicated by heterogeneity of the available studies, particularly for line of treatment, where different drugs have been assessed in different lines of therapy. In this case, estimates of their effectiveness are likely to have been biased by the point in the pathway when the drug was prescribed. For example, CS and csDMARDs were used in the early stages of the pathway for possibly less severe or monocyclic patients compared to bDMARDs. Similarly, for ANK vs CNK, ANK tended to be used in earlier lines of treatment; therefore, there may be bias in estimates of relative effectiveness.
In this review, the effect of line of therapy was investigated by comparing studies in which most patients had received previous bDMARDs with studies where most patients had not (Supplementary Figure S5, available with the online version of this article). There was no significant difference between these subgroups for complete remission, but there was a trend toward a reduced rate of CS discontinuation for studies where a higher proportion of patients had received previous bDMARD treatment (P = 0.05).
The included studies had limitations, including the quality of their methods and reporting, which affect the interpretation of their results. The 3 RCTs were all assessed as having a high ROB, largely because they were underpowered. Some trials failed to recruit sufficient participants to meet sample size requirements due to the rarity of the condition. An absence of significant effect cannot be wholly attributed to a lack of power. However, trials showed a trend toward positive effects, and it is possible that, with adequate sample sizes, these trials may have shown significant effects.
Overall, 17 out of 40 case series studies were assessed as having a high ROB. In 9 cases, it was not possible to ascertain all study methods because publications were abstracts or letters. Further, despite providing some indication of the effectiveness of interventions, uncontrolled case series do not provide definitive evidence of treatment effectiveness.
Conclusions were limited by the small sample size and high degree of statistical heterogeneity. In addition, heterogeneity of the study methods (eg, length of follow-up), patients (eg, pattern of disease), and intervention characteristics (eg, line of therapy) makes direct comparison of outcomes between treatments prone to bias.
Another limitation of the studies in this review was that the authors recorded different types of outcome measures. Some reported the cumulative rate of remission or response over a particular time period, and some reported the rate at a specific timepoint in follow-up. This makes overall interpretation difficult and adds to the large heterogeneity across studies. In the sensitivity analysis, this confounding effect was reduced because cumulative rates were usually reported in studies with longer follow-up. Indeed, the heterogeneity across studies decreased in this analysis with the removal of studies with short follow-up.
A final limitation of the included studies was that different definitions of complete remission and partial response were used. However, to increase comparability across studies and interpretability of the results, the definitions were standardized, where possible, using clinically agreed definitions.
Due to high levels of heterogeneity as well as small sample sizes, caution is needed when drawing conclusions from the metaanalyses and other available data. Heterogeneity in the metaanalyses may have been further increased due to the inclusion of both case series and RCT data. However, because of the paucity of available data, RCT data was not excluded.
This review was limited to studies published since 2012, as the first RCT of bDMARDs was published in 2012.33 This limit was applied to increase homogeneity across study characteristics. However, this year limit may have caused underrepresentation of early older treatments (NSAIDs, CS, and csDMARDs). Nevertheless, the review included studies reflecting current clinical practice.
Finally, the conclusions of this review are limited by the positioning of studies in the patient pathway. As discussed, in this review, different interventions tended to be used in different lines of treatment (eg, CS as first-line treatment and biologics as third- or fourth-line treatment), and patient characteristics are therefore also different. Where reported, it appears that patients treated with first-line treatments tended to have mainly systemic disease, whereas patients treated with third- or fourth-line treatments tended to have mainly articular disease. This limits conclusions about the use of biologics in patients with systemic disease early in the treatment pathway.
Future research should be reported with standardized outcome definitions. Due to the high variability in outcome definitions, it is currently difficult to compare the results of studies, and, in this review, we chose to classify studies based on the actual outcome measure, rather than the reported outcome. A set of clinically agreed and consistently used definitions would provide more robust results from literature reviews and metaanalyses, and would facilitate the interpretation and application of the results in clinical practice. Cumulative incidence rates should be preferred to prevalence rates because they increase precision when comparing studies with different lengths of follow-up.
Future studies should take into account and stratify for the main disease manifestations of AOSD (eg, articular vs systemic or polycyclic vs monocyclic course of disease). With these outcomes, the underlying disease characteristics could aid the selection of the appropriate treatment approach. Another recommendation is the use of sufficient follow-up in future studies to ensure that most remission cases are included. Further research could also include investigation of the efficacy of biologics when given as monotherapy compared to when given with adjuvant csDMARDs.
Finally, for the comparison of treatments, prospective studies in similar populations are important to avoid confounding from line of treatment and other patient characteristics. In the absence of head-to-head trials, placebo-controlled trials in similar populations would allow for the indirect comparison of treatments.
In conclusion, evidence from case series appears to support the case for bDMARDs for AOSD, particularly TCZ, ANK, and CNK. However, the magnitude of effect and comparative effectiveness of different treatments is still uncertain. Standardization of outcomes and outcome definitions should drive future research protocols, with studies of adequate duration. To reduce bias in studies comparing interventions, treatments should be given at similar stages in the treatment pathway.
ACKNOWLEDGMENT
We would like to thank Deborah Watkins, Jacoby Patterson, and Maria Cikalo for their help with screening records and data extraction, and Deborah Watkins and Joshua Bracewell for their assistance in preparing the manuscript. We also thank Rachael McCool, Emily Gregg, and Charlotte Graham for their review of the manuscript for academic content and style.
Footnotes
This work was supported by Novartis Pharma GmbH. DM’s work is funded by the Leeds NIHR Biomedical Research Centre.
Payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events have been received by DM (Novartis and Sobi), EF (AbbVie, BMS, Galapagos, Lilly, Medac, Novartis, Pfizer, R-Pharm, Sobi, and Sanofi), and PR (AbbVie, BMS, and Janssen). EF has received grants from Lilly and Pfizer, and has participated on a Data Safety Monitoring Board or Advisory Board for UCB, Galapagos, Lilly, Novartis, and Sanofi. VCG, KT, and HR are employees of Novartis Pharma GmbH. VCG has stock or stock options in Novartis Pharma GmbH. MC, ME, PM, NH, JM, and SG are employees of YHEC who were commissioned by Novartis to undertake this work.
- Accepted for publication January 15, 2024.
- Copyright © 2024 by the Journal of Rheumatology
This is an Open Access article, which permits use, distribution, and reproduction, without modification, provided the original article is correctly cited and is not used for commercial purposes.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.
- 19.
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵