Introduction
Pain, including inflammatory joint pain, may be debilitating in patients with rheumatic diseases, such as rheumatoid arthritis (RA) and psoriatic arthritis (PsA), and it is one of the main issues affecting quality of life.14 Pain can be categorised into three main types: nociceptive (localised pain derived from tissue injury), neuropathic (resulting from peripheral or central nerve damage) and nociplastic (indicative of dysregulation of nervous system pain-processing pathways; eg, hypersensitivity/central sensitisation).5 These three types are not always distinct, and the relative contributions to the overall sense of ‘pain’ may be heterogeneous between individuals, both within and between diseases. At the local joint level, inflammation of the synovium, caused by proinflammatory cells and cytokines, can lead to activation and sensitisation of afferent nociceptive fibres, which transmit ‘pain’ signals to the spinal cord. Multiple cytokines, such as interleukin (IL)–1β, IL-6, IL-10 and tumour necrosis factor (TNF)-α, which play important roles in mechanisms underlying pain and pain sensitisation, are directly or indirectly modulated by the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway.6
Rapid and sustained pain control is an important goal for patients with RA and PsA.13 However, pain in these conditions is multidimensional, reflecting both sensory and affective components, and control of inflammation does not always eliminate pain. Other directly acting analgesic drug effects would therefore be advantageous.1710 Some patients continue to experience pain even after attenuation of inflammation (referred to as residual pain, or pain that persists despite adequate control of inflammation and/or disease activity).1 Furthermore, the balance between direct effects of treatment on pain (by reducing sensitisation of nociceptive pathways) and indirect effects (eg, by reducing inflammation) remains unknown.
Upadacitinib, a JAK inhibitor approved for the treatment of RA, PsA and other inflammatory diseases, demonstrated significant efficacy in reducing disease activity and preventing joint damage compared with placebo or continued methotrexate (MTX) in phase 3 studies.1114 Improvement in pain outcomes compared with placebo has also been demonstrated with upadacitinib in RA, PsA and axial spondyloarthritis.15 16 Pain relief may be the result of direct effects on the nervous system or of suppression of inflammation (indirect effects on pain), and pharmacologic agents targeting discrete molecular pathways might have different direct effects on pain.
The first part of this post hoc analysis aims to assess pain after treatment with upadacitinib, adalimumab and placebo in patients with active RA or PsA who had attenuation of inflammation and in those with remaining inflammation through weeks 24 (PsA) and 26 (RA).1719 The second part uses mediation analysis to assess the direct and indirect effect of treatment with upadacitinib or adalimumab compared with placebo on pain in patients through weeks 24 (PsA) and 26 (RA). This analysis explores the relationship between the direct effect of treatment (the improvement in pain not attributed to changes in the surrogates of inflammation available for the analysis) and the indirect effect (the improvement in pain due to changes in inflammatory factors (mediators)).1719
Methods
Study design and participants
Adult patients (≥18 years old) with RA who had active disease despite stable MTX treatment were enrolled in the 48-week, randomised, double-blind, phase 3 SELECT-COMPARE study, with a 26-week placebo-controlled period.11 Patients on background MTX were randomised 2:2:1 to upadacitinib 15 mg one time per day, placebo or adalimumab 40 mg every other week. Patients with an insufficient response (<20% improvement in tender joint count (TJC) or swollen joint count (SJC) at weeks 14, 18 or 22 or a Clinical Disease Activity Index of >10 at week 26) received rescue treatment, switching from placebo to upadacitinib, from upadacitinib to adalimumab or from adalimumab to upadacitinib. All patients receiving placebo who did not receive rescue treatment were switched to upadacitinib at week 26.
Adult patients (≥18 years old) with PsA who had active disease at baseline and inadequate response or intolerance to ≥1 non-biologic disease-modifying antirheumatic drug (DMARD) were enrolled in the randomised, double-blind, phase 3 SELECT-PsA one study.14 Patients were randomised 1:1:1:1 to upadacitinib 15 mg one time per day, upadacitinib 30 mg one time per day, adalimumab 40 mg every other week or placebo. Upadacitinib 30 mg data are not presented here because the 30 mg dose does not have regulatory approvals in PsA.
The studies were conducted according to the International Council for Harmonisation guidelines and the principles of the Declaration of Helsinki. All patients provided written informed consent, and the study protocols were approved by an independent ethics committee or institutional review board at each site.11 14
Assessments
Patient-reported pain was measured by the Patient’s Global Assessment (PtGA) of pain (based on response to ‘How much pain have you had because of your condition during the last week?’) using a visual analogue scale (VAS; 0 mm (no pain) to 100 mm (most severe pain)) for RA and numerical rating scale (NRS; 0 (no pain) to 10 (most severe pain)) for PsA. Elicited pain was recorded as the TJC based on 28 joints (TJC28). At week 12 (for RA and PsA) and at week 24 (for PsA) and week 26 (for RA), least squares (LS) mean pain; LS mean change from baseline in pain; absolute mean pain; and ≥30%, ≥50% or ≥70% reduction in pain from baseline were assessed.20 21 Achievement of low disease activity thresholds for pain (VAS ≤20 mm for RA and NRS ≤2 for PsA) was also assessed, with thresholds chosen based on consensus cut-offs for RA.22 23
For the analysis of pain (defined as pain level at a specific visit), patients were grouped by attenuation of inflammation versus remaining inflammation at weeks 12 and 24/26 (PsA/RA). Attenuation of inflammation was defined as an SJC based on 66 joints (SJC66) of 0 and a C-reactive protein (CRP) <6 mg/L; remaining inflammation was defined as SJC66 >0 or CRP ≥6.
In RA, the indirect effect of treatment on pain was assessed as change in level of pain attributable to change in inflammation based on objective markers (erythrocyte sedimentation rate (ESR), CRP and SJC based on 28 joints (SJC28)). Direct effect of treatment on pain was defined as the remaining effect not attributed to the indirect causal pathways (figure 1A). In PsA, the indirect effect of treatment on pain was assessed based on change in inflammatory factors (itch, total enthesitis count, Leeds Enthesitis Index (LEI), and CRP), and the direct effect of treatment on pain was defined as the remaining effect not attributed to the indirect causal pathways (figure 1B).7 Two sensitivity analyses were also conducted in which only one measure of enthesitis was included: 1 excluding LEI and one excluding total enthesitis count.
Statistical analysis
Methods for analysing pain by visit differed based on study design. Last observation carried forward for data after rescue was used for RA, with confirmed observed case data used up to week 12. For PsA, observed case data were reported, with patients who received rescue therapy after week 16 excluded from the week 24 analysis.
For change from baseline in patient-reported pain analyses using PtGA of pain (1–100 mm for RA, 0–10 for PsA) or elicited pain using TJC28 between-group LS mean, 95% CI and nominal p values were based on an analysis of covariance (ANCOVA) model with treatment as a fixed factor and baseline value as a covariate; ANCOVA was performed to compare treatment effect at week 12 and weeks 24/26 (PsA/RA). For RA, SJC28 and CRP were also included as covariates for patients with remaining inflammation. For PsA, current DMARD use (yes/no) was also included as a fixed factor and for pain analyses of patients with remaining inflammation levels, SJC28 and CRP levels were additionally included as covariates.
For binary endpoint analyses, nominal p values were calculated by the χ2 test for RA and the Cochran-Mantel-Haenszel test adjusting for current DMARD use (yes/no) for PsA; 95% CIs for response rate and response rate difference were based on normal approximation for PsA. P values were generated for comparisons between upadacitinib versus placebo and versus adalimumab; nominal p values<0.05 are reported. Mediation analysis18 for the direct and indirect effects of upadacitinib versus placebo and adalimumab versus placebo on pain was assessed as improvement in PtGA of pain or TJC28 and was conducted using observed case analysis for week 2 and week 12 (RA) or 16 (PsA) and last observation carried forward for data after the switch to rescue treatment for the analysis on weeks 24/26 (PsA/RA).
Results
Pain analyses in patients with attenuated versus remaining inflammation
Among patients with RA, a higher proportion of those receiving upadacitinib achieved attenuation of inflammation at weeks 12 (18.1% (118/651)) and 26 (28.7% (187/651)) compared with those treated with adalimumab (11.3% (37/327) and 18.3% (60/327), respectively) or placebo (4.1% (27/651) and 6.3% (41/651)). Some differences in baseline characteristics were observed between treatment groups among patients with attenuation of inflammation compared with those with remaining inflammation (online supplemental table 1). Among patients with PsA, higher percentages of patients achieved attenuation of inflammation with upadacitinib (29.6% (120/406) at week 12 and 48.1% (169/351) at week 24) and adalimumab (27.1% (110/406) and 42.0% (144/343)) compared with placebo (11.7% (46/392) and 24.5% (66/269)).
Some differences in baseline characteristics were observed between treatment groups among patients with attenuation of inflammation (online supplemental table 1).
Among patients with RA and attenuation of inflammation, pain improvement was greater with upadacitinib (LS mean change (95% CI), –42.9 (–46.6 to –39.1)) compared with adalimumab (–34.7 (–41.1 to –28.0); p<0.05) or placebo (–33.1 (–41.0 to –25.2); p<0.05) at week 12. Improvements in pain were similar across groups by week 26 (figure 2A). Trends in pain improvement among treatment groups in patients with remaining inflammation were consistent with those with attenuation of inflammation at week 12, and greater pain reduction with upadacitinib versus placebo was maintained at week 26 (figure 2B).
In patients with RA and attenuated inflammation, only the upadacitinib treatment group had a mean PtGA of pain ≤20 at week 12; mean PtGA of pain ≤20 was achieved in all treatment groups at week 26 (figure 2C). Findings for LS mean pain score by visit were comparable with upadacitinib and adalimumab in patients with remaining inflammation (figure 2D). Similar trends have been previously reported in RA and ankylosing spondylitis.11 15
Among patients with PsA and attenuation of inflammation, improvement in PtGA of pain from baseline was greater with upadacitinib (LS mean change (95% CI), –2.7 (–3.2 to –2.3) at week 12 and –3.8 (–4.2 to –3.5) at week 24) versus placebo (–1.8 (–2.5 to –1.2) p<0.05 and –2.8 (–3.3 to –2.3) p<0.001). A similar response was observed with adalimumab (–2.8 (–3.2 to –2.4) and –3.6 (–3.9 to –3.2)) compared with upadacitinib (figure 2E). LS mean improvement in pain was lower in patients with remaining inflammation; however, LS mean improvement (95% CI) in PtGA of pain from baseline to weeks 12 and 24 was still greater for upadacitinib (–2.0 (–2.2 to –1.7) and –2.3 (–2.7 to –2.0)) versus placebo (–0.8 (–1.0 to –0.5) p<0.001 and –1.5 (–1.9 to –1.2) p<0.001) and similar to adalimumab (–2.0 (–2.3 to –1.8) and –2.5 (–2.8 to –2.2); figure 2F).
LS mean pain scores by visit are reported in figure 2G and H.
Among patients with RA and attenuation of inflammation, absolute mean PtGA of pain values at weeks 12 and 26 were 19.9 and 15.4, respectively, with upadacitinib; 28.0 and 18.9 with adalimumab; and 28.3 and 16.5 with placebo (figure 3A). Among patients with remaining inflammation, a similar trend was observed at weeks 12 and 26 (figure 3B). Among patients with PsA and attenuation of inflammation, absolute mean PtGA of pain values at weeks 12 and 24 were 3.0 and 2.2 with upadacitinib, 2.9 and 2.4 with adalimumab, and 3.7 and 3.3 with placebo (figure 3C).
A similar trend was noted among patients with remaining inflammation, although absolute mean pain values were higher (figure 3D).
For ≥30%, ≥50% or ≥70% reduction in PtGA of pain from baseline values,17 24 25 only the ≥50% reduction showed a significantly higher response with upadacitinib versus placebo at week 12 in patients with RA and attenuation of inflammation (p<0.05; figure 4A). No significant differences in response rates by treatment were observed at week 26 (online supplemental figure 1A). Among patients with RA with remaining inflammation, greater response rates were observed with upadacitinib versus placebo (p<0.001) or adalimumab (p<0.05) at week 12 (figure 4A) and with upadacitinib versus placebo at week 26 (online supplemental figure 1A) for ≥30%, ≥50% and ≥70% reductions in pain from baseline (p<0.01, p<0.001, p<0.001, respectively).
For patients with PsA, response rates for ≥30%, ≥50% or ≥70% reduction in PtGA from baseline with upadacitinib were higher than placebo at week 12 (p<0.01, p<0.05, p<0.05, respectively; figure 4B) and week 24 (p<0.01, p<0.01, p<0.001; online supplemental figure 1B) among patients with attenuation of inflammation. A similar trend was observed for adalimumab (figure 4B; online supplemental figure 1B). Although response rates for ≥30%, ≥50% or ≥70% reduction in PtGA at week 12 (figure 4B) and week 24 (online supplemental figure 1B) were lower in patients with remaining inflammation compared with patients with attenuation, response rates were higher for patients with remaining inflammation treated with upadacitinib compared with placebo.
A similar trend was observed for adalimumab (figure 4B; online supplemental figure 1B).
A greater proportion of patients with RA achieved PtGA of pain VAS ≤20 mm at week 12 when receiving upadacitinib versus adalimumab or placebo, both in patients who had attenuation of inflammation (p<0.05 for upadacitinib vs placebo) and those with remaining inflammation (p<0.001 for upadacitinib vs placebo; Online supplemental figure 2A). A numerically greater proportion of patients with PsA achieved PtGA of pain NRS <2 at week 12 when receiving upadacitinib versus placebo; the difference was statistically significant (p<0.0001) in patients with remaining inflammation. At week 24, a significantly greater proportion of patients with PsA achieved PtGA of pain NRS ≤2 when receiving upadacitinib versus placebo in patients who had attenuation of inflammation (p=0.0024) and those with remaining inflammation (p<0.0002; Online supplemental figue 2B).
Direct and indirect treatment effects on pain
For the analysis of direct and indirect treatment effects on pain, 1629 patients were included in the RA population (upadacitinib, n=651; adalimumab, n=327; placebo, n=651) and 1281 patients were included in the PsA population (upadacitinib, n=429; adalimumab, n=429; placebo, n=423). Disease characteristics were similar between treatment groups at baseline (online supplemental table 2). The direct effects on pain, assessed as improvement in PtGA of pain, were greater with upadacitinib versus placebo (both p<0.001) as early as week 2 and increased at weeks 12 and 26 in the RA population (all p<0.001; figure 5A).
This was also observed with adalimumab versus placebo at weeks 2, 12 and 26 (all p<0.001; figure 5A). The direct effect on pain was similar between upadacitinib and adalimumab at week 2 and increased to nearly double with upadacitinib between weeks 2 and 12, and was maintained at week 26, whereas with adalimumab the direct effect was approximately the same at each time point (figure 5A). Compared with placebo, upadacitinib and adalimumab had similar indirect effects on pain versus placebo at weeks 2, 12 and 26 in the RA population.
The greatest indirect effect of both upadacitinib and adalimumab on pain at week 26 was mediated through SJC28, followed by ESR and CRP.
In the PsA population, both upadacitinib and adalimumab showed greater direct and indirect effects on improvement in PtGA of pain compared with placebo at weeks 16 and 24 (all p<0.001; figure 5B). The direct effect on pain among patients with PsA for upadacitinib versus placebo (assessed as improvement in PtGA of pain) was 0.98 at weeks 16 and 24 and for adalimumab versus placebo was 0.83 and 0.82, respectively (figure 5B). Indirect effects on pain for upadacitinib were 0.57 and 0.62 at weeks 16 and 24, and those for adalimumab were 0.43 and 0.43, respectively.
The greatest indirect effect of both upadacitinib and adalimumab on pain at week 26 was mediated through itch and CRP.
In the PsA population, total effects and direct effects on improvement in PtGA of pain were greater with both upadacitinib and adalimumab versus placebo at weeks 16 and 24 (all p<0.001; figure 5B). Total effects on pain among patients with PsA assessed as improvement in PtGA of pain at weeks 16 and 24 for upadacitinib versus placebo were 1.51 and 1.56, respectively; and for adalimumab versus placebo were 1.24 and 1.28, respectively (figure 5B). The greatest indirect effect of both upadacitinib and adalimumab on pain at week 26 was mediated through itch and CRP.
In the RA population, the total effects of treatment on elicited pain as assessed by improvement in TJC28 were greater with upadacitinib and adalimumab than with placebo at weeks 2, 12 and 26 (all p<0.05). The direct effect on pain was similar between upadacitinib and adalimumab at week 2, increased to nearly double with upadacitinib between weeks 2 and 12, and was maintained at week 26. For adalimumab, the direct effect decreased between weeks 2 and 12 and weeks 2 and 26 (figure 6A).
Upadacitinib and adalimumab had similar indirect effects on TJC28 at weeks 2, 12 and 26 in the RA population, with the greatest indirect effect on pain (TJC28) being mediated by SJC28.
In the PsA population, improvement in elicited pain, as assessed by improvement in TJC28 as a measure of total effects, was also greater with upadacitinib and adalimumab versus placebo in the PsA population (all p<0.05; figure 6B). Total effects on pain assessed as improvement in TJC28 at weeks 16 and 24 with upadacitinib versus placebo were 2.16 and 2.30, respectively, compared with 1.35 and 1.71 for adalimumab versus placebo. Among patients with PsA, the indirect effects on pain at weeks 12 and 24 for upadacitinib and adalimumab were comparable, with upadacitinib having an effect between 1.5 and 2 times the effect of adalimumab.
The greatest indirect effect on pain was mediated by total enthesitis, followed by itch and CRP at weeks 16 and 24; unfortunately, earlier data for the PsA population were not available. Results of the sensitivity analyses were consistent (online supplemental figures 3 and 4).
Discussion
Pain is a cardinal feature of immune-mediated inflammatory diseases, such as RA and PsA, and negatively affects quality of life.13 A key goal of therapy is to reduce inflammation and thereby prevent tissue damage and alleviate key symptoms, such as pain. However, we confirm that residual pain may persist after inflammation is optimally attenuated.1 We analysed data from phase 3 placebo-controlled studies of upadacitinib and adalimumab in RA and PsA to better understand their relative effect on pain reduction, with and without attenuation of inflammation. Furthermore, we constructed a mediation analysis model to distinguish how much pain reduction was indirectly associated with reduction in inflammatory markers versus the direct effect of the therapy independent of its effect on the surrogates of inflammation employed in the models.
Patient-reported pain outcomes were evaluated using PtGA of pain and supported by patient-reported TJC28 assessments as a measure of elicited pain.
Our findings show that upadacitinib and adalimumab reduced pain relative to placebo with and without attenuation of inflammation. The magnitude of pain reduction was greater with upadacitinib versus adalimumab in patients with RA and similar between upadacitinib and adalimumab for patients with PsA. In the RA population, mean pain VAS scores of ≤20 mm represent a consensus cut-off when satisfaction with health is not negatively affected by pain.22 23 Patients receiving upadacitinib had mean VAS ≤20 in the attenuated inflammation group at week 12; mean VAS ≤20 was achieved in all treatment groups of patients with attenuated inflammation at week 26.
In the PsA population, a similar response was observed for mean PtGA of pain measured using NRS (0 (none) to 10 (most severe)), which dropped from 6.2 at baseline to 2.2 at week 24 in patients with attenuated inflammation receiving upadacitinib. In patients with remaining inflammation, upadacitinib demonstrated greater response rates for ≥30%, ≥50% or ≥70% reduction in PtGA of pain versus placebo in RA and PsA populations at weeks 12 and 24/26 (PsA/RA) and versus adalimumab in the RA population at week 12. Rates of achieving VAS ≤20 mm for RA and the analogous cut-off of NRS ≤2 for PsA also favoured upadacitinib for most comparisons, regardless of remaining inflammation.
Taken together, these results suggest that upadacitinib has the ability to reduce pain outcomes in patients with RA and PsA even with persisting inflammation and does so to a greater extent compared with adalimumab in patients with RA. Our results are consistent with previous studies that demonstrated better pain control with JAK inhibitors versus placebo or MTX.2629 In addition, some studies have also suggested that JAK inhibitors may provide a greater magnitude of pain relief than TNF inhibitors.26 29
A mediation analysis was used to assess the effects of upadacitinib or adalimumab versus placebo on pain improvement directly or indirectly via improvement in available surrogate measures of inflammation in the PsA and RA data sets. The mediators comprised CRP, ESR and SJC28 for RA and CRP, itch and enthesitis for PsA. Overall pain improvement in RA was greater with upadacitinib versus placebo compared with adalimumab versus placebo at weeks 12 and 26, driven mainly by direct effects of treatment on pain improvement with upadacitinib versus adalimumab (defined as the remaining effect not attributed to changes in inflammation based on objective markers).
Our findings suggest that in patients with RA, upadacitinib may be more effective in controlling pain than adalimumab and enable us to generate hypotheses to account for these observations. The broader cytokine inhibition with JAK inhibitor might account for a greater improvement in inflammation than anti-TNF, and therefore a correspondingly larger magnitude of improvement in pain as a generic feature of inflammation.6 However, indirect effects of upadacitinib and adalimumab on pain were generally similar at weeks 2, 12 and 26 in the RA population. An alternative hypothesis is that JAK inhibition may have a direct effect on the experience of pain that is in addition to that mediated by suppression of inflammation.
It is also possible that the observed differences in pain improvement between TNF and JAK inhibition could be related to undefined inflammatory factors that were not captured by the available inflammation surrogates employed in these mediation analyses (ESR, CRP and SJC28).
In patients with PsA, upadacitinib and adalimumab produced higher mean improvements in pain via inflammatory or non-inflammatory mechanisms compared with placebo. Total, direct and indirect effects on pain improvement were numerically greater with upadacitinib versus adalimumab when evaluating pain using PtGA of pain and TJC28. In our analysis, the greatest indirect effect of upadacitinib on pain was mediated by SJC28 in RA and itch and CRP in PsA.
Similar results have been observed with another JAK inhibitor, tofacitinib, in PsA.7 Different models have identified different mediators of pain, and further studies are needed to confirm these results. Available mediation analysis data from a phase 2 study in patients with PsA found that improvement in pain associated with treatment with the tyrosine kinase inhibitor deucravacitinib may be attributed to both direct and indirect mechanisms, with total observable indirect effects derived from the Psoriasis Area and Severity Index and SJC improvements.30 A post hoc analysis of the phase 3 COAST-V study in patients with active ankylosing spondylitis demonstrated that treatment with the IL-17A inhibitor ixekizumab reduced pain through mechanisms that cannot be attributed only to measurable inflammation reduction but through pain-relieving mechanism(s) that differ from those of adalimumab.19 Multiple cytokines have been implicated in the function of neuronal pathways from the periphery to the brain, including IL-6R, IL-1R, IL-10R, IL-17 and interferon-γR, some of which signal via the JAK/STAT mechanism and others of which can be indirectly modulated via that mechanism.6 Patients may experience nociceptive and nociplastic pain via a mixture of mechanisms and that pain experience may be multifactorial.3133 The availability of an expanded range of efficacious targeted therapies affords the opportunity to better understand the key pathologic drivers of differential responses in certain symptoms such as pain.
This study is limited by the fact that it is an analysis of data from two distinct diseases and from trials with different study designs conducted at different study sites; therefore, direct comparisons should only be made with caution between the data sets. In addition, this was a secondary analysis of randomised controlled trial data with no data beyond weeks 24/26, and no early data (eg, week 2) in the PsA population. Our indices of inflammation were approximate, subject to measurement error and incomplete in that they may not have captured all aspects of inflammation.
Other inflammation indices (eg, ultrasound, MRI, measurement of specific inflammatory mediators) might serve to further reveal indirect (anti-inflammatory) effects of upadacitinib and adalimumab on pain. Overall, the data available from these trials do not allow us to elucidate the mechanisms underlying observed ‘direct’ effects, which could be due to other direct actions on nociceptive pathways or contextual factors not adequately controlled by placebo. Alternatively, the mediation models may have missed other anti-inflammatory mechanisms for which surrogate measures were not available.
Our analyses assumed linear relationships between pain and inflammation, comprehensive measurement of inflammation and independence of direct and indirect effects on pain. In reality, these relationships might be more complex, including synergy and overlap of pain mechanisms and inflammatory mechanisms not captured by ESR, CRP and SJC. It is therefore possible to conclude that the effects that we defined statistically to be direct might yet be explained by indirect effects on inflammation.
Conclusion
Several relevant pathogenic cells and molecules have been identified as therapeutic targets in RA and PsA through observation of the effectiveness of biologic and small molecule targeted therapies. Whereas pain improvement in both RA and PsA is associated with inflammation resolution, it is apparent that some pain amelioration can occur independent of reduction in inflammation, apparently through direct pain modulation at a more central level. In the case of RA, our findings generate the hypothesis that the greater magnitude of pain relief observed with upadacitinib compared with adalimumab might be related to effects on inflammatory and non-inflammatory pain pathways.
Supplementary material