Achievement and Maintenance of Disease Targets with Upadacitinib in Rheumatoid Arthritis: 2-Year Outcomes from the UPHOLD Real-World Study

Abstract

Introduction

Upadacitinib, an oral Janus kinase inhibitor, has shown a favorable benefit–risk profile in rheumatoid arthritis (RA) clinical trials. This analysis aimed to assess the long-term effectiveness and safety of upadacitinib through 24 months in the UPHOLD real-world observational study.

Methods

The coprimary endpoints were: (1) proportion of patients achieving disease activity score in 28 joints using C-reactive protein (DAS28[CRP]) remission (< 2.6) at 6 months; and (2) proportion of those patients who maintained remission at 12 months. Additional endpoints assessed through 24 months included: achievement and maintenance of DAS28(CRP) low disease activity (LDA; ≤ 3.2); maintenance of response by treatment strategy and prior therapy; disease category by visit; improvements in patient-reported outcomes (PROs); and achievement of glucocorticoid (GC)-free remission. Endpoints were analyzed by modified nonresponder imputation (mNRI) and as observed (AO). Exposure-adjusted event rates (events/100 patient-years [E/100 PY]) for treatment-emergent adverse events (TEAEs) were reported through May 15, 2024.

Results

In total, 1029/1715 (60.0%) patients completed the 24-month follow-up. DAS28(CRP) responses at 6 months were well maintained through 24 months (mNRI/AO remission, 62.9%/83.9%; and LDA, 65.7%/90.1%, respectively). Remission rates at 24 months were consistent between patients maintaining combination therapy and those maintaining monotherapy (81.1% and 87.4%, respectively; AO). Responses were also largely similar across prior therapy subgroups. Improvements in PROs achieved by 12 months were maintained through 24 months. Most patients receiving GCs at baseline who achieved GC-free remission maintained GC-free remission at 24 months. Among selected TEAEs of interest, herpes zoster, serious infection, and hepatic disorder occurred at 3.5, 3.1, and 2.6 E/100 PY, respectively. No new safety signals were identified.

Conclusions

Upadacitinib was effective for the long-term treatment of RA in real-world practice, with responses achieved at 6 months maintained through 2 years, and no new safety findings.

Trial Registration

NCT04497597.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40744-026-00826-5.

Key Summary Points

Why carry out this study?

Patients with rheumatoid arthritis (RA) require lifelong treatment to control disease activity and reduce the risk of progressive joint damage, and it is therefore important to evaluate long-term treatment efficacy and safety.

Real-world data are crucial for providing information on long-term safety, effectiveness, and treatment patterns of a drug in more diverse patient populations seen in routine rheumatology practice, outside the strictly regulated design of clinical trials.

The aim of this analysis was to explore whether the responses observed up to 12 months in interim analyses of the UPHOLD observational study in patients with RA treated with upadacitinib could be maintained through 2 years in a real-world setting.

What was learned from the study?

The final analysis of the UPHOLD study showed that responses to treatment, including remission (< 2.6) and low disease activity (≤ 3.2) per disease activity score in 28 joints using C-reactive protein, were maintained through 24 months, regardless of therapy strategy or prior treatment exposure.

In addition, there was a glucocorticoid (GC)-sparing effect of upadacitinib, as most patients who received GCs at baseline and achieved GC-free remission were able to maintain GC-free remission at 24 months.

These results provide valuable insight into the long-term effectiveness and maintenance of response with upadacitinib for the management of RA in real-world clinical practice across patient populations with different treatment strategies and prior treatment experience.

Introduction

International guidelines on the treatment of rheumatoid arthritis (RA), a chronic condition that can be associated with significant disability, pain, and reduced quality of life, recommend a treat-to-target (T2T) strategy aiming for remission as the treatment goal [1–3]. Currently available treatment options include conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), glucocorticoids (GCs), biologic (b)DMARDs (including tumor necrosis factor inhibitors [TNFi] and interleukin-6 inhibitors), and targeted synthetic (ts)DMARDs, namely Janus kinase (JAK) inhibitors [3]. Given the need for lifelong treatment to control disease activity and reduce the risk of progressive joint damage and morbidity in patients with RA, it is crucial to evaluate long-term treatment efficacy and safety [2, 4–6]. Although multiple treatment options are available, a considerable proportion of patients either respond inadequately or are unable to tolerate therapy, resulting in suboptimal disease control and failure to achieve sustained clinical remission [6, 7]. Additional insights are required to guide appropriate management strategies for such patients based on previous treatment history. In addition, recommendations on the use of GCs in RA are to use these at the lowest dose and for the shortest time possible due to safety concerns, although many patients remain on GCs for several years [1, 3, 8].

While clinical trials have strict inclusion and exclusion criteria that often exclude patients with comorbidities or receiving concomitant medications, real-world data may offer a more realistic picture of treatment effectiveness and safety in more diverse patient populations, providing important insights for clinical decision-making [9]. There is also an ongoing need for continued assessment of JAK inhibitors in RA, as data from long-term follow-up in clinical trials and beyond 1 year of follow-up in real-world populations are limited [10–13].

Upadacitinib, an oral, selective JAK inhibitor [14], has demonstrated efficacy and safety, with or without csDMARDs, across a range of patient types in the phase 3 SELECT RA clinical trial program [15–21]. However, there is limited understanding on how to optimally implement T2T principles with upadacitinib due to limited evidence both from controlled studies and in real-world practice [22]. To optimize the T2T approach with upadacitinib, further research is needed to understand the different treatment strategies (e.g., monotherapy vs. combination therapy, modification of concomitant medications) and their impact on patient outcomes. Real-world data on the effectiveness and safety of upadacitinib in RA with up to 1 year of follow-up reported meaningful improvements in clinical and patient-reported outcomes (PROs), regardless of prior treatment experience, line of therapy, or concomitant medications [23]. The UPHOLD observational study was designed to assess the real-world, long-term effectiveness and safety of upadacitinib in adults with moderate-to-severe RA. The 1-year interim analysis of UPHOLD showed that approximately half of patients had achieved a disease activity score in 28 joints using C-reactive protein (DAS28[CRP]) < 2.6 (defined as DAS28[CRP] remission in this study) at 6 months, and ~ 80% of those patients maintained remission at 12 months [24]. Extending real-world follow-up is necessary to generate valuable insight for clinical practice, including control of disease activity, safety, tolerability, and potential for reducing concomitant medications in the long term.

The aim of this final 2-year analysis of the UPHOLD study was to evaluate longer-term outcomes with upadacitinib in patients with RA, including in subgroups with different treatment history and treatment strategy, by assessing achievement and maintenance of response, safety, and concomitant GC-sparing effect.

Methods

Study Design

UPHOLD (NCT04497597; Upadacitinib treatment patterns, achievement of treatment targets and maintenance of response in patients with moderate-to-severe rheumatoid arthritis in real-world practice) is a noninterventional, prospective, open-label, multicountry, multicenter, postmarketing observational cohort study designed to assess achievement and maintenance of remission with upadacitinib over 24 months following therapy initiation in real-world practice. Primary analyses were performed at 6 and 12 months [24], with a follow-up period through 24 months (Fig. S1). The study ran from October 16, 2020 (first patient enrolled) to May 15, 2024 (last patient last visit) in 19 countries at multiple study sites (Supplementary Table S1).

Patient Population

The characteristics of the patient population and inclusion criteria have been previously described in detail [24]. Briefly, eligible patients were aged ≥ 18 years, had moderate-to-severe active RA, and were administered upadacitinib (either as monotherapy or in combination with csDMARDs) by their physician as per the locally approved label. The decision to initiate upadacitinib treatment was made prior to and independent of study enrollment. The use of concomitant antirheumatic drugs (e.g., GCs, nonsteroidal anti-inflammatory drugs [NSAIDs], other analgesics) was at the discretion of the treating physician. Prior use of tsDMARDs and/or bDMARDs was permitted. Patients were excluded if they had enrolled in a clinical trial of an investigational drug concurrently or within the last 30 days, received prior treatment with upadacitinib, or had contraindications to upadacitinib treatment per the locally approved label.

The UPHOLD study complied with the Strengthening the Reporting of Observational Studies in Epidemiology statement for observational studies [25] and was conducted according to the International Council for Harmonisation guidelines, local regulations, and guidelines governing clinical study conduct, and the Declaration of Helsinki. All patients provided written informed consent, and the study protocol and consent forms were approved by an institutional review board or independent ethics committee at each study site.

Effectiveness Assessments

The coprimary endpoints were (1) the proportion of patients receiving upadacitinib who achieved DAS28(CRP) remission (< 2.6) at 6 months, and (2) the proportion of patients achieving remission at 6 months who continued to receive upadacitinib and maintained remission (or had a ≤ 0.6-point increase in DAS28[CRP]) at 12 months. Additional effectiveness endpoints assessed through 24 months in this study included: proportion of patients with DAS28(CRP) remission at 6 and 12 months, who maintained remission (or had a ≤ 0.6-point increase in DAS28[CRP]) at 24 months; proportion of patients with DAS28(CRP) low disease activity (LDA; ≤ 3.2) at 6 and 12 months, who maintained LDA (or had a ≤ 0.6-point increase in DAS28[CRP]) at 24 months; proportion of patients with remission, LDA, moderate disease activity (MDA), and high disease activity (HDA) by DAS28(CRP), clinical disease activity index (CDAI), and simplified disease activity index (SDAI) cutoffs by visit; change from baseline in PROs, including Health Assessment Questionnaire-Disability Index (HAQ-DI; 0–3 scale; lower scores indicate less disability and better physical functioning), patient’s global assessment of pain over the past 7 days (PtPain; 0–10 numeric rating scale [NRS]; lower scores indicate less pain), morning stiffness severity over the past 7 days (0–10 NRS; lower scores indicate less severe morning stiffness), Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue (0–52 total score; higher scores indicate less fatigue and better functioning), and treatment burden questionnaire (TBQ; 0–130 total score; lower scores indicate lower treatment burden). The proportions of patients who maintained remission/LDA by treatment strategy (monotherapy vs. combination with csDMARDs), and the proportions of patients who achieved GC-free remission through 24 months among those receiving GCs at baseline with available data on total equivalent dose of prednisone, were also assessed.

Differences in selected endpoints were assessed by treatment strategy (monotherapy vs. combination therapy), and by prior treatment exposure in the following subgroups: ts/bDMARD-naïve (patients without prior use of TNFi or other bDMARD or tsDMARD); ts/bDMARD-experienced (prior use of any tsDMARD [JAK inhibitor] or bDMARD [TNFi or other bDMARD]); TNFi-only (prior use only of a TNFi); and tsDMARD-experienced (prior use of a tsDMARD).

Safety

Treatment-emergent adverse events (TEAEs) and selected TEAEs of interest were recorded through 24 months in all patients who received ≥ 1 dose of upadacitinib and are reported as exposure-adjusted event rates (EAERs; events per 100 patient-years [E/100PY]), where available, and as exposure-adjusted incidence rates (EAIRs; n/100 PY).

Investigator-reported clinically significant abnormalities in laboratory tests were recorded by visit through 24 months. Clinical significance was determined by the study investigator and defined as a newly observed unfavorable and unintended laboratory abnormality.

Statistical Analysis

The analysis sets have been previously described [24] and can be summarized as follows: all patients who signed the informed consent form and met eligibility criteria: enrolled analysis set (EAS); all patients who received ≥ 1 upadacitinib dose during the study: full analysis set (FAS; all safety evaluations); all patients within the FAS who completed 6 months of treatment and had DAS28(CRP) data available, and those who discontinued the study before 6 months: modified (m)FAS1 (first coprimary endpoint only); all patients within mFAS1 achieving DAS28(CRP) remission, who completed 12 months of treatment and had DAS28(CRP) data available, and those who discontinued the study between 6 and 12 months: mFAS2 (second coprimary endpoint only).

For the 24-month analyses, modified nonresponder imputation (mNRI) was employed, in which discontinuations for any reason before prespecified timepoints were treated as nonresponders. Effectiveness data were also analyzed as observed (AO) using prespecified analysis sets without imputation of missing data.

For all effectiveness analyses, the number of patients who achieved the assessed endpoint at 6 months, had data available at 6 and 12 or at 6 and 24 months, and continued to receive upadacitinib at 12 or 24 months (or who discontinued between 6 and 12 months or between 6 and 24 months for mNRI analyses), was used as the denominator.

Safety data were assessed up to the data cutoff date of May 15, 2024 for all patients who received at least one dose of upadacitinib (FAS). TEAEs were reported as EAERs and EAIRs, and investigator-reported abnormalities in laboratory parameters were summarized using descriptive statistics.

Ethics Approval

The study was conducted according to the International Council on Harmonisation guidelines and the Helsinki Declaration of 1964 and its later amendments. The trial protocol was approved by independent ethics committees and institutional review boards at each study site. Written informed consent was provided by patients ahead of study screening.

Results

Patient Disposition and Baseline Characteristics

Of the 1715 participants, 1713 patients were enrolled (EAS), and 1699 received at least one dose of upadacitinib and comprised the FAS. A total of 1029 (60.0%) patients completed the 24-month follow-up, and 686 (40.0%) prematurely discontinued the study before month 24. Among the primary reasons for discontinuation, the most common were lack of efficacy (n = 260; 15.2%), adverse events (n = 163; 9.5%), and loss to follow-up (n = 95; 5.5%) (Fig. S2).

Patient demographics and disease characteristics at baseline have been previously reported [24]. Briefly, the mean age (standard deviation [SD]) of the patients was 56.9 (12.4) years; the mean disease duration was 10.1 (9.1) years; and 62.3% had ≥ 1 cardiovascular risk factor at baseline (history of hypertension, diabetes mellitus, dyslipidemia, and smoking). A total of 735/1699 (43.3%) patients received GCs at baseline, with a mean (SD) total equivalent prednisone dose during the first week of 8.1 (7.6) mg/day (n = 455). Of 1524 patients who received prior RA therapy (not ongoing at study initiation), 64.3% had received ≥ 1 bDMARD and 18.3% ≥ 1 prior tsDMARD. A total of 101/1699 (5.9%) patients had been vaccinated against the varicella zoster virus (Supplementary Table S2).

Effectiveness

Disease Activity Outcomes

Of 1073 patients in the mFAS1, 497 (mNRI 46.3% [497/1073]; AO 55.3% [497/899]) achieved DAS28(CRP) remission at 6 months. Of 341 patients in the mFAS2, 270 (mNRI 79.2% [270/341]; AO 84.9% [270/318]) maintained DAS28(CRP) remission at 12 months. Responses achieved at 6 months were maintained through 2 years, with 261 patients with DAS28(CRP) remission at 6 months maintaining remission at 24 months (mNRI 62.9% [261/415]; AO 83.9% [261/311]; Fig. 1a).

Fig. 1.

Achievement and maintenance of a DAS28(CRP) remission (< 2.6), b LDA (≤ 3.2), and c mean DAS28(CRP) value per visit. ^amFAS1 denotes FAS patients who completed 6 months of treatment and had DAS28(CRP) data available or discontinued the study for any reason before 6 months. ^bmFAS1 patients with nonmissing DAS28(CRP) data. ^cmFAS2 denotes mFAS1 patients who achieved remission at 6 months, completed 12 months of treatment, and had DAS28(CRP) data available or discontinued the study for any reason between 6 and 12 months. ^dmFAS2 patients with nonmissing DAS28(CRP) data. ^eFAS patients who achieved remission at 6 months, completed 24 months of treatment, and had DAS28(CRP) data available at 6 and 24 months or discontinued the study between 6 and 24 months. ^fFAS patients who achieved remission at 6 months, completed 24 months of treatment, and had DAS28(CRP) data available at 6 and 24 months. ^gmFAS1 patients who achieved LDA at 6 months, completed 12 months of treatment, and had DAS28(CRP) data available or discontinued the study between 6 and 12 months. ^hmFAS1 patients who achieved LDA at 6 months, completed 12 months of treatment, and had DAS28(CRP) data available. ⁱFAS patients who achieved LDA at 6 months, had DAS28(CRP) data available at 6 and 24 months, and continued to receive upadacitinib at 24 months or discontinued the study between 6 and 24 months. ^jFAS patients who achieved LDA at 6 months, had DAS28(CRP) data available at 6 and 24 months, and continued to receive upadacitinib at 24 months. ^kPatients with nonmissing data. *Coprimary endpoints. Error bars represent 95% confidence intervals (a and b) or SD (c). AO as observed, DAS28(CRP) disease activity score in 28 joints using C-reactive protein, FAS full analysis set, LDA low disease activity, mFAS modified FAS, mNRI modified nonresponder imputation, SD standard deviation

DAS28(CRP) LDA followed a similar pattern, with 636 patients achieving LDA at 6 months (mNRI 59.3% [636/1073]; AO 70.7% [636/899]); 366 of those maintained LDA at 12 months (mNRI 83.6% [366/438]; AO 91.5% [366/400]), and 356 maintained LDA at 24 months (mNRI 65.7% [356/542]; AO 90.1% [356/395]; Fig. 1b). The mean (SD) DAS28(CRP) value decreased from 4.6 (1.22) at baseline to 2.9 (1.19) at 3 months and to 2.3 (1.02) at 24 months (Fig. 1c).

The overall proportion of patients achieving DAS28(CRP) remission increased from 5.1% at baseline to 54.3%, 60.1%, and 66.8% at 6, 12, and 24 months, respectively. For CDAI remission (≤ 2.8), this proportion increased from 0.6% at baseline to 20.2%, 28.1%, and 36.1%, respectively. For SDAI remission (≤ 3.3), this proportion increased from 0.6% at baseline to 22.5%, 28.5%, and 36.7%, respectively (AO; Fig. 2). The proportions of patients meeting the criteria for LDA per DAS28(CRP) (≤ 3.2), CDAI (≤ 10.0), or SDAI (≤ 11.0) cutoffs followed a similar pattern, increasing from 12.9%, 5.5%, and 5.8% at baseline to 81.8%, 76.8%, and 78.2%, respectively, at 24 months (AO; Fig. 2). The proportions of patients meeting the criteria for MDA and HDA per respective DAS28(CRP), CDAI, and SDAI cutoff values at baseline and at 6, 12, and 24 months (AO) are shown in Fig. 2 and Fig. S3.

Fig. 2.

Disease activity by visit (AO). Data are presented AO. The proportion of patients who achieved each level of disease activity was calculated using the number of patients with nonmissing data for the disease activity score as the denominator. The sum of the proportions may not total 100% due to rounding. DAS28(CRP) remission: < 2.6; LDA: 2.6 ≤ DAS28(CRP) ≤ 3.2; MDA: 3.2 < DAS28(CRP) ≤ 5.1; HDA: > 5.1. CDAI remission: ≤ 2.8; LDA: 2.8 < CDAI ≤ 10.0; MDA: 10.0 < CDAI ≤ 22.0; HDA: > 22.0. SDAI remission: ≤ 3.3; LDA: 3.3 < SDAI ≤ 11.0; MDA: 11.0 < SDAI ≤ 26.0; HDA: > 26.0. AO as observed, BL baseline, CDAI clinical disease activity index, DAS28(CRP) disease activity score in 28 joints using C-reactive protein, HDA high disease activity, LDA low disease activity, MDA moderate disease activity, mo months, REM remission, SDAI simplified disease activity index

Outcomes by Treatment Strategy

Of patients who initiated upadacitinib as combination therapy, 13.4% (91/677) had switched to upadacitinib monotherapy by 12 months, and 22.6% (119/527) by 24 months (AO).

Patients Who Maintained Their Initial Treatment Strategy. For patients with DAS28(CRP) remission (< 2.6) at 6 months, 106 of those who initiated upadacitinib as combination therapy (mNRI 79.1% [106/134]; AO 86.2% [106/123]) maintained remission at 12 months, and 90 (mNRI 62.5% [90/144]; AO 81.1% [90/111]) maintained remission at 24 months while staying on combination therapy. Of patients who initiated upadacitinib as monotherapy and stayed on monotherapy, 134 (mNRI 80.2% [134/167]; AO 85.4% [134/157]) maintained remission at 12 months, and 118 (mNRI 61.1% [118/193]; AO 87.4% [118/135]) maintained remission at 24 months (Fig. S4a).

For patients with DAS28(CRP) LDA (≤ 3.2) at 6 months, 151 of those who initiated upadacitinib as combination therapy (mNRI 83.9% [151/180]; AO 92.6% [151/163]) maintained LDA at 12 months, and 138 (mNRI 67.3% [138/205]; AO 90.8% [138/152]) maintained LDA at 24 months while staying on combination therapy. Of patients who initiated upadacitinib as monotherapy and stayed on monotherapy, 174 (mNRI 82.5% [174/211]; AO 91.6% [174/190]) maintained LDA at 12 months, and 153 (mNRI 63.0% [153/243]; AO 92.2% [153/166]) maintained LDA at 24 months (Fig. S4b).

Patients Who Switched Treatment Strategy. Of patients who initiated upadacitinib as combination therapy and switched to monotherapy, 27 (mNRI 73.0% [27/37]; AO 81.8% [27/33]) maintained remission at 12 months, and 42 (mNRI 65.6% [42/64]; AO 80.8% [42/52]) maintained remission at 24 months. Of patients who initiated upadacitinib as monotherapy and switched to combination therapy, 3 (both mNRI and AO 60.0% [3/5]) maintained remission at 12 months, and 11 (mNRI 78.6% [11/14]; AO 84.6% [11/13]) maintained remission at 24 months (Fig. S4a).

Of patients who initiated upadacitinib as combination therapy and switched to monotherapy, 37 (mNRI 82.2% [37/45]; AO 90.2% [37/41]) maintained LDA at 12 months, and 51 (mNRI 66.2% [51/77]; AO 82.3% [51/62]) maintained LDA at 24 months. Of patients who initiated upadacitinib as monotherapy and switched to combination therapy, 4 (mNRI 57.1% [4/7]; AO 66.7% [4/6]) maintained LDA at 12 months, and 14 (mNRI 82.4% [14/17]; AO 93.3% [14/15]) maintained LDA at 24 months (Fig. S4b).

Outcomes by Prior Therapy

Achievement and maintenance of response were similar across patient subgroups per prior therapy exposure. At 24 months, 77.5% (124/160) of patients in the ts/bDMARD-experienced subgroup, 80.6% (54/67) in the TNFi-only subgroup, 69.2% (27/39) in the tsDMARD-experienced subgroup, and 90.7% (137/151) in the ts/bDMARD-naïve subgroup maintained DAS28(CRP) remission (AO; Fig. S5). For DAS28(CRP) LDA, 86.1% (180/209) ts/bDMARD-experienced, 90.4% (75/83) TNFi-only, 85.1% (40/47) tsDMARD-experienced, and 94.6% (176/186) ts/bDMARD-naïve patients maintained LDA at 24 months (AO; Fig. S5).

Increases in the proportions of patients achieving remission and LDA per DAS28(CRP), CDAI, and SDAI cutoffs were also observed in all patient subgroups per prior treatment exposure (Fig. S6). The proportion of patients in the ts/bDMARD-experienced subgroup achieving DAS28(CRP) remission increased from 5.7% at baseline to 63.6% at 24 months; for CDAI remission (≤ 2.8), from 0.5% to 31.1%; and for SDAI remission (≤ 3.3), from 0.4% to 31.9%, respectively (AO; Fig. S6a). In the TNFi-only subgroup, the proportion of patients achieving DAS28(CRP), CDAI, and SDAI remission increased from 9.9%, 0.4%, and 0.4% at baseline to 73.3%, 41.1%, and 41.6%, respectively, at 24 months (AO; Fig. S6b). In the tsDMARD-experienced subgroup, 52.8%, 25.0%, and 26.0% of patients achieved DAS28(CRP), CDAI, and SDAI remission, respectively, at 24 months (AO; Fig. S6c). Of ts/bDMARD-naïve patients, 70.9%, 42.6%, and 42.6% achieved DAS28(CRP), CDAI, and SDAI remission, respectively, at 24 months (AO; Fig. S6d). The proportions of patients meeting the criteria for LDA, MDA, and HDA at baseline and at 6, 12, and 24 months per DAS28(CRP), CDAI, and SDAI cutoffs followed a similar pattern as in the overall population (Fig. S6).

GC-Free Remission Among Patients Receiving GCs at Baseline

Of the 735 patients who received GCs at baseline, 455 patients with available data on the mean total equivalent dose of prednisone were included in the subanalysis of remission and GC-free status at 6, 12, and 24 months. Among patients achieving remission per DAS28(CRP), CDAI, and SDAI at 6 months, 28.5% (45/158), 39.1% (25/64), and 41.0% (25/61), respectively, were also GC-free. Of those patients with GC-free remission at 6 months, 93.3% (28/30), 100% (13/13), and 84.6% (11/13) maintained GC-free DAS28(CRP), CDAI, and SDAI remission, respectively, at 12 months (Fig. S7a). Among patients achieving remission per DAS28(CRP), CDAI, and SDAI at 12 months, 41.0% (59/144), 49.2% (32/65), and 48.4% (30/62) were also GC-free. Of those patients with GC-free remission at 12 months, 100% (37/37, 20/20, and 17/17 for DAS28[CRP], CDAI, and SDAI, respectively) maintained GC-free remission at 24 months (Fig. S7b).

Among patients receiving GCs at baseline who stayed on low-dose GCs (< 5 mg/day), the proportions maintaining DAS28(CRP) remission and LDA did not change markedly through 24 months (remission, 30.3% [40/132] and 29.1% [32/110]; and LDA, 28.8% [53/184] and 26.4% [38/144] at 3 and at 24 months, respectively; AO).

Improvement in PROs

Improvements (mean change from baseline) in PROs observed at 3 months were maintained or further improved through 24 months. For PtPain, the change from baseline (SD) improved from − 2.5 (2.8) at 3 months to − 3.4 (2.8) at 24 months; for morning stiffness severity, from − 2.5 (3.0) to − 3.3 (3.1); for HAQ-DI, from − 0.4 (0.6) to − 0.5 (0.7); for FACIT-Fatigue, from 6.6 (10.8) to 9.5 (12.1); and for TBQ, from − 10.2 (23.1) to − 14.1 (26.2), respectively (AO; Fig. 3). Changes from baseline in each PRO were generally similar across subgroups by treatment strategy and prior therapy.

Fig. 3.

Change from baseline in PROs with upadacitinib over 24 months (AO). ^aOver the past 7 days using a 0–10 numeric rating scale. AO as observed, BL baseline, FACIT Functional Assessment of Chronic Illness Therapy, HAQ-DI Health Assessment Questionnaire-Disability Index, NA not applicable, PRO patient-reported outcome, SD standard deviation, TBQ treatment burden questionnaire

Safety

In total, 2922 TEAEs were reported in 1022 patients through 2 years, including 265 serious TEAEs in 199 patients. Among TEAEs classified by system organ class and preferred term, the most frequently observed events were infections (n = 617; 36.3%), including 288 events (17%) of coronavirus disease-2019 (COVID-19); 70 (4.1%) of herpes zoster; 62 (3.6%) of urinary tract infection; 57 (3.4%) of bronchitis; 45 (2.6%) of upper respiratory tract infection; 40 (2.4%) of influenza; and 40 (2.4%) of nasopharyngitis. Among selected TEAEs of interest, the most common were herpes zoster (3.5 E/100 PY), serious infection (3.1 E/100 PY), and hepatic disorder (2.6 E/100 PY). The EAER for malignancy excluding nonmelanoma skin cancer (NMSC) was 0.8 E/100 PY, for NMSC 0.5 E/100 PY, for venous thromboembolic events (VTE) 0.5 E/100 PY, and for major adverse cardiovascular events (MACE) 0.2 E/100 PY (Fig. 4). EAIRs are summarized in Fig. S8.

Fig. 4.

EAERs of TEAEs in patients with moderate-to-severe rheumatoid arthritis treated with upadacitinib. ^aReasonable possibility. ^bIncludes cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke. ^cIncludes VTE, other venous thrombosis, and arterial thromboembolic events. ^dIncludes deep vein thrombosis, venous thromboembolism, and pulmonary embolism. ^eExcluding herpes zoster and active tuberculosis. Safety was evaluated by assessing all TEAEs (adverse events occurring after the first dose of study drug and up to 30 days after the last dose of study drug) occurring in the FAS up to the data cutoff date of May 15, 2024. There were no reported events of gastrointestinal perforation. CI confidence interval, D/C discontinuation, E event, E/100PY events per 100 PY, EAER exposure-adjusted event rate, FAS full analysis set, MACE major adverse cardiovascular events, NMSC nonmelanoma skin cancer, PY patient-years, TEAE treatment-emergent adverse event, UPA upadacitinib, VTE venous thromboembolic events

A total of 17 TEAEs (0.65 E/100 PY) resulted in death in 14 patients. These included COVID-19 (n = 4), myocardial infarction (n = 2), multiple organ failure (n = 2), septic shock (n = 1), cardiorespiratory arrest (n = 1), chest infection (n = 1), respiratory distress (n = 1), pneumonia (n = 1), domestic accident (n = 1), metastatic renal cancer (n = 1), gastrointestinal bleeding (n = 1), and unknown (n = 1). Of the 17 adverse events reported as leading to death, two events were recorded in a single patient whose ultimate cause of death was reported as unknown. Additionally, one event was identified as a serious case of COVID-19; however, this event was reported as resolved prior to the patient’s death, and no further information is available to clarify why it was included in the deaths listing. Among all TEAEs leading to death, only one case of pneumonia and one case of multiple organ failure were considered by the investigator to have a reasonable possibility of being related to the study drug (Supplementary Table S3).

There were 22 events of malignancy excluding NMSC (0.8 E/100 PY) and 12 of NMSC (0.5 E/100 PY) (Fig. 4 and Supplementary Table S4).

The proportions of patients with investigator-determined clinically significantly abnormal laboratory test results are summarized in Table 1. The most commonly reported clinically significant abnormalities at 24 months were abnormalities in blood lipid levels (total cholesterol, 4.9%; low-density lipoprotein, 4.7%) and increased levels of markers of inflammation (erythrocyte sedimentation rate, 4.0% [28/695]; C-reactive protein, 3.4% [28/812]). The proportion of patients with abnormalities in absolute neutrophil count had increased from 0.3% at 12 months to 0.6% at 24 months, as had the proportion of patients with abnormalities in hepatic enzyme levels (alanine aminotransferase, 0.2–0.9%; and aspartate aminotransferase, 0.1–0.8%, respectively).

Table 1.

Proportion of patients with investigator-reported abnormal and clinically significant^a laboratory test results by visit (FAS)

Parameter, n/N (%)	Baseline	6 months	12 months	24 months
Hematology
Hemoglobin	32/1443 (2.2)	10/1059 (0.9)	4/907 (0.4)	4/794 (0.5)
Erythrocyte sedimentation rate	257/1293 (19.9)	52/946 (5.5)	39/776 (5.0)	28/695 (4.0)
Absolute lymphocyte count	5/1211 (0.4)	2/892 (0.2)	5/767 (0.7)	5/682 (0.7)
Absolute neutrophil count	3/1187 (0.3)	6/875 (0.7)	2/752 (0.3)	4/679 (0.6)
Chemistry
Serum creatinine	6/1359 (0.4)	5/979 (0.5)	9/827 (1.1)	4/743 (0.5)
Creatine phosphokinase	5/302 (1.7)	4/299 (1.3)	4/274 (1.5)	2/237 (0.8)
Alanine aminotransferase	7/1375 (0.5)	9/1022 (0.9)	2/874 (0.2)	7/774 (0.9)
Aspartate aminotransferase	3/1195 (0.3)	4/907 (0.4)	1/783 (0.1)	6/711 (0.8)
Blood lipid levels
Total cholesterol	28/646 (4.3)	30/570 (5.3)	20/469 (4.3)	22/451 (4.9)
Low-density lipoprotein	24/490 (4.9)	22/431 (5.1)	21/362 (5.8)	17/363 (4.7)
High-density lipoprotein	7/504 (1.4)	7/459 (1.5)	0/380 (0.0)	5/373 (1.3)
Triglycerides	14/569 (2.5)	7/512 (1.4)	12/425 (2.8)	8/396 (2.0)
C-reactive protein	354/1484 (23.9)	59/1094 (5.4)	37/911 (4.1)	28/812 (3.4)

FAS full analysis set

^aClinical significance was determined by the investigator and defined as a newly observed unfavorable and unintended laboratory abnormality

Discussion

Long-term data beyond 12 months on the maintenance of response with JAK inhibitors in real-world practice are limited [10–13, 24, 26]. To the best of our knowledge, UPHOLD is the first systematic evaluation of achievement and maintenance of response in patients with RA initiating a JAK inhibitor in a real-world setting. Previously published interim results from this study showed that ~ 80% of patients who achieved DAS28(CRP) remission at 6 months maintained remission at 12 months [24]. The final results of the UPHOLD study build on these previous findings and have shown that > 60% of patients achieving DAS28(CRP) remission at 6 months were able to maintain remission at 2 years. These data provide additional insight into the long-term benefits of upadacitinib for the treatment of moderate-to-severe RA in real-world clinical practice.

The current standard of care for first-line therapy in RA is csDMARDs, with bDMARDs and tsDMARDs usually reserved as second-line therapy for patients who have failed to respond or have contraindications to csDMARDs [1, 3]. The characteristics of patients who initiated upadacitinib in the UPHOLD study showed that most had long-standing disease (~ 10 years) with moderate-to-high disease activity at baseline; > 40% had evidence of structural damage (presence of erosions on X-ray); 64.3% had already received at least one bDMARD; and 18.3% had received at least one tsDMARD. Despite evidence of suboptimal disease control and inadequate response to previous treatments at baseline, achievement and maintenance of response were observed regardless of prior therapy exposure. Similar to results reported at 12 months [24], the highest response rates at 24 months were observed in patients not previously treated with tsDMARDs (ts/bDMARD-naïve and TNFi-only patient subgroups). This may be partly due to the fact that the TNFi-only subgroup had the lowest proportion of patients previously treated with ≥ 3 bDMARDs (8.1% vs. 30.0% and 37.3% in the ts/bDMARD- and tsDMARD-experienced subgroups), i.e., had the lowest proportion of patients likely to be treatment-refractory. Although response rates were numerically lower among patients with previous inadequate response to a tsDMARD (ts/bDMARD- and tsDMARD-experienced) compared with the other subgroups, > 55% were still able to maintain remission and > 60% were able to maintain LDA at 24 months (mNRI). These findings support treatment with upadacitinib irrespective of prior response to therapy, as patients who have been refractory to previous treatment may benefit from switching to a different mechanism of action [27], or even an in-class switch, as cycling JAK inhibitors may be more effective than switching to bDMARDs in RA [28]. The proportions of patients who maintained DAS28(CRP) remission at 12 and 24 months were consistent between patients who maintained combination therapy and those who maintained monotherapy; they were also comparable between patients who maintained combination therapy and those who had switched to monotherapy by 24 months.

Although GCs are valuable for the short-term (≤ 3 months) management of rheumatic diseases, including RA, chronic GC use (> 6 months) is not generally recommended due to safety concerns associated with cumulative exposure over time [1, 3, 8, 29]. RA studies like the SEMIRA trial [30] suggest that maintaining a low dose (≤ 5 mg/day) may offer better disease control than GC tapering and, in the absence of alternative treatments, may be considered; however, risks must be evaluated per individual patient-specific characteristics [31, 32]. The current consensus for GC treatment is to use as low a dose and for as short a time period as possible, due to safety concerns associated with long-term cumulative exposure [1, 3, 8, 29, 33]. Among patients receiving GCs at baseline who were in remission (per DAS28[CRP], CDAI, or SDAI) at 6 months while still receiving GCs, > 80% were able to maintain remission without GCs at 12 months; and of those in remission with GCs at 12 months, 100% remained in remission at 24 months without the need for GCs. These findings indicate that upadacitinib treatment was associated with a GC-sparing benefit in patients with RA.

The effectiveness outcomes from the UPHOLD study are in line with long-term efficacy data from clinical trials of upadacitinib, with 63–72% of patients achieving DAS28(CRP) < 2.6 and 35–36% achieving CDAI ≤ 2.8 (AO) in the long-term extensions of the SELECT-BEYOND (260 weeks) and BALANCE-EXTEND (312 weeks) studies [34, 35]. Effectiveness was also in line with other published real-world data for upadacitinib in RA [11, 23, 36]. Interim 6-month data from the Canadian CLOSE-Up study of patients with RA initiating upadacitinib reported that 63.5% of patients (191/301) achieved DAS28(CRP) remission, consistent with early reports from other real-world studies, including UPHOLD [11]. The German UPwArds study showed that 66.3% of patients (246/371) with RA who initiated upadacitinib achieved DAS28(CRP) remission (< 2.6) and 81.7% (303/371) met the criteria for DAS28(CRP) LDA (≤ 3.2) by 12 months, with a slight decrease in effectiveness noted among patients who had received prior treatment [37]. A separate study of data from the US CorEvitas RA registry utilized CDAI criteria to define remission and LDA, and found that 11.1% of patients met the criteria for CDAI remission (≤ 2.8) and 29.1% met the criteria for CDAI LDA (2.8 to ≤ 10) at 12 months [23]. This cohort included all patients treated with upadacitinib. When stratified by prior treatment, remission rates were lower for those having received prior TNFi treatment (5.2%), and higher for those initiating upadacitinib as first-line treatment (16%) [23]. Of note, comparisons with data from clinical trials and registries should be approached with caution given the differences in study design, geographic diversity, and baseline patient characteristics and comorbidities.

In UPHOLD, the observed improvements in PROs at 12 months were maintained through 24 months, while the change from baseline in FACIT-Fatigue and TBQ further improved from 8.9 to 9.5 and from − 13.8 to − 14.1, respectively, at 24 months. As for the effectiveness outcomes, improvements in assessed PROs also extended to all patient subgroups by treatment strategy and prior therapy, with a similar extent of improvement across subgroups. Patients remaining on upadacitinib at 12 months in the CorEvitas registry also reported sustained improvements in PROs, including change from baseline (SD) in mean HAQ-DI score of − 0.28 (0.55), PtPain of − 19.8 (28.4), and patient-reported fatigue using a 0–100 visual analog scale of − 17.0 (28.2) [23]. As for the effectiveness outcomes, numerically greater improvements were noted at 12 months for patients initiating upadacitinib as first-line treatment, with slightly less prominent improvement from baseline noted for those with prior TNFi treatment [23].

The safety profile of upadacitinib in this study was consistent with that reported in long-term clinical trial programs of upadacitinib in rheumatologic diseases (EAERs for serious TEAEs, hepatic disorder, malignancy excluding NMSC, herpes zoster, and MACE in RA: 12.4, 10.2, 0.7, 3.0, and 0.4 E/100 PY, respectively) [15], and with the known safety profile of upadacitinib across indications [38]. Overall, the EAERs for selected TEAEs of interest, including malignancy excluding NMSC, herpes zoster, MACE, and VTE at 24 months, were consistent with those reported at 12 months in the UPHOLD study (0.8 vs. 0.8; 3.5 vs. 3.1; 0.2 vs. 0.3; and 0.5 vs. 0.5 E/100 PY, respectively), with no new important identified safety signals. All patients with MACE (n = 5) and VTE (n = 11) had ≥ 1 cardiovascular risk factor (body mass index > 25.0, history of hypertension, diabetes mellitus, high-density lipoprotein cholesterol ≤ 40 mg/dl, low-density lipoprotein cholesterol ≥ 130 mg/dl, and current/former tobacco/nicotine use) [24]. An increased risk of gastrointestinal perforation has been suggested for JAK inhibitors, particularly in patients receiving concurrent treatment with GCs or NSAIDs, and those with associated risk factors, such as inflammatory bowel disease [39]. No gastrointestinal perforation events were reported in this study. The safety analyses from UPHOLD showed a favorable benefit–risk profile and did not suggest increased cumulative risk over time with long-term upadacitinib treatment. Other real-world safety data also indicated that JAK inhibitors overall carry an acceptable risk of TEAEs of special interest, including cardiovascular events, without notable differences between different JAK inhibitors [40], or between JAK inhibitors and bDMARDs [41]. However, cross-study comparisons must be approached with caution due to the differences in study design (e.g., randomized clinical trial vs. observational study), comparator arms (e.g., active comparator vs. placebo), and baseline patient characteristics such as age, which may be associated with higher rates of MACE, VTE, and malignancy events. The presence of risk factors for cardiovascular events, malignancy, and VTE should always be taken into consideration when considering the use of JAK inhibitors in RA [3, 42, 43].

Limitations of the study have been previously reported [24] and include the inherent biases associated with observational design, including: difficulties in objective interpretation of investigator-reported data; the relatively high discontinuation rate, which may introduce attrition bias; the overlap with the COVID-19 pandemic, which may have disproportionately influenced discontinuation, infection, and mortality rates; the lack of data on the proportion of patients who maintained disease control among the 40% who discontinued upadacitinib, as well as the missing data and potential recall bias associated with self-administered questionnaires. In addition, the use of DAS28(CRP) in patients receiving therapeutic agents that act through the interleukin-6 (IL-6) signaling pathway, such as IL-6 inhibitors and JAK inhibitors, has been questioned, given their impact on CRP levels. As a consequence of this impact, use of the DAS28(CRP) measure may allow for a higher swollen joint count within the definition of remission, and reductions in CRP may not always correspond to improvements in disease activity [44]. This limitation was partly addressed by including non-CRP-dependent disease activity indices, such as CDAI, in the analyses. As UPHOLD is a post-marketing observational study, patients were enrolled only after the treating physician’s decision to prescribe upadacitinib was made. Upadacitinib was used according to the locally approved label, which may have introduced bias due to differences in prescription coverage per region or patient-related cost constraints. Finally, a statistical evaluation of upadacitinib effectiveness or safety across countries (which could be influenced by differences in access to medical care) was not performed and an assessment of the maintenance of disease control in patients who discontinued upadacitinib treatment was beyond the scope of the current study. Despite these limitations, real-world evidence is crucial for exploring the pattern of use of RA treatments and providing information on the long-term safety and effectiveness of a drug in more diverse patient populations, outside the boundaries of strictly regulated clinical trials, thereby supporting clinical decision-making in routine rheumatology practice.

Conclusion

In the UPHOLD observational study, upadacitinib was effective for the long-term treatment of moderate-to-severe RA in real-world clinical practice, with its effectiveness and maintenance of effect remaining consistent through 24 months across patient populations with different treatment strategies and prior treatment experience. The effectiveness and safety profile of upadacitinib in a real-world setting was favorable and consistent with data from phase 3 clinical trials, with no newly identified safety signals. These findings may prove valuable for supporting clinical decision-making in routine rheumatology practice across a wide spectrum of patient, disease, and treatment characteristics.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Andrew Östör contributed to the study design and was an investigator in the study. Eugen Feist, Prodromos Sidiropoulos, Jérôme Avouac, Martin Rebella, Rajaie Namas, and Suzan Attar were investigators in the study. Tianming Gao conducted the statistical analyses. Andrew Östör, Eugen Feist, Prodromos Sidiropoulos, Jérôme Avouac, Martin Rebella, Rajaie Namas, Frank Buttgereit, Philip G. Conaghan, Ana B. Romero, Erin L. McDearmon-Blondell, Ivan Lagunes-Galindo, Tianming Gao, Aditi Kadakia, Tim Shaw, and Suzan Attar analyzed and interpreted the data and contributed to the critical revision of the manuscript. All named authors met the International Committee of Medical Journal Editors criteria for authorship for this article, take responsibility for the integrity of the work as a whole, and have given their approval for this version to be published.

Funding

AbbVie funded this trial and participated in the trial design, research, analysis, data collection, interpretation of data, and the review and approval of the publication. All authors had access to relevant data and participated in the drafting, review, and approval of this publication. No honoraria or payments were made for authorship. The journal’s Rapid Service Fee was funded by AbbVie.

Data Availability

AbbVie is committed to responsible data sharing regarding the clinical trials we sponsor. This includes access to anonymized, individual, and trial-level data (analysis data sets), as well as other information (e.g., protocols, clinical study reports, or analysis plans), as long as the trials are not part of an ongoing or planned regulatory submission. This includes requests for clinical trial data for unlicensed products and indications. These clinical trial data can be requested by any qualified researchers who engage in rigorous, independent scientific research, and will be provided following review and approval of a research proposal, statistical analysis plan, and execution of a data sharing agreement. Data requests can be submitted at any time after approval in the USA and Europe and after acceptance of this manuscript for publication. The data will be accessible for 12 months, with possible extensions considered. For more information on the process or to submit a request, please visit the following link: https://www.abbvieclinicaltrials.com/hcp/data-sharing.

Declarations

Conflict of Interest

Andrew Östör has served as a consultant and/or on advisory boards and/or undertaken clinical trials for AbbVie, GSK, Janssen, Lilly, Novartis, and Pfizer. Eugen Feist has received honoraria and research grants from AbbVie, BMS, Galapagos, Lilly, MSD, Novartis, Pfizer, Roche, and Sobi. Prodromos Sidiropoulos has received honoraria and research grants from AbbVie, Janssen, Lilly, MSD, Novartis, Pfizer, Roche, and UCB through the University of Crete Special Account for Research. Jérôme Avouac has received honoraria from AbbVie, AstraZeneca, Biogen, BMS, Fresenius Kabi, Galapagos, Janssen, Lilly, Medac, MSD, Nordic Pharma, Novartis, Pfizer, Roche-Chugai, Sandoz, and Sanofi; and research grants from BMS, Fresenius Kabi, Galapagos, Novartis (Dreamer), and Pfizer (Passerelle). Martin Rebella has received honoraria and/or support to participate in academic events from AbbVie, Bayer, Pfizer, and Roche. Rajaie Namas has received speaker fees, research grants, and advisory honoraria from AbbVie. Frank Buttgereit has received consultancy fees, honoraria, travel expenses, and grant support from AbbVie, Amgen (Horizon Therapeutics), Grünenthal, Novartis, and Sparrow. Philip G. Conaghan has received speaker fees or consultancies from AbbVie, Formation Bio, Genascence, Grünenthal, GSK, Kolon TissueGene, Levicept, Lilly, Moebius, Novartis, Pacira, Stryker, and Takeda. Ana B. Romero, Erin L. McDearmon-Blondell, Ivan Lagunes-Galindo, Tianming Gao, Aditi Kadakia, and Tim Shaw are AbbVie employees and may own AbbVie stock or options. Suzan Attar has received speaker fees, research grants, and advisory honoraria from AbbVie, Amgen, AstraZeneca, BMS, Gilead, GSK, Hikma, Janssen, Lilly, Novartis, Organon, Pfizer, Roche, Sandoz, Sanofi, and Takeda.

Ethics Approval

The study was conducted according to the International Council on Harmonisation guidelines and the Helsinki Declaration of 1964 and its later amendments. The trial protocol was approved by independent ethics committees and institutional review boards at each study site. Written informed consent was provided by patients ahead of study screening.