FormalPara Key Summary Points

Why carry out this study?

Treatment with tofacitinib, another Janus kinase inhibitor, relative to tumor necrosis factor inhibitors has been associated with a potentially increased risk of adverse events, such as serious infections, major adverse cardiovascular events, and malignancies in patients aged 50 years or older with at least one cardiovascular risk factor; however, there are limited long-term safety data available for these agents.

This integrated analysis of the phase 3 SELECT trials describes the safety profile of upadacitinib for up to 5 years of exposure in patients with psoriatic arthritis (PsA), ankylosing spondylitis (AS, including one phase 2/3 study), non-radiographic axial spondyloarthritis (nr-axSpA), and pooled axial spondyloarthritis (axSpA), including AS and nr-axSpA.

What was learned from this study?

Events of serious infection, herpes zoster, lymphopenia, and nonmelanoma skin cancer were observed at higher rates with upadacitinib 15 mg once daily than adalimumab in patients with PsA; slightly elevated rates for most of these adverse events were also seen with upadacitinib 15 mg in patients with PsA than those with AS and nr-axSpA.

Rates of key adverse events of interest, including major adverse cardiovascular events, malignancy excluding nonmelanoma skin cancer, and venous thromboembolic events were similar between upadacitinib 15 mg and adalimumab and across disease indications.

Upadacitinib 15 mg once daily demonstrated a generally consistent safety profile across disease states, with no new safety signals identified compared with previous reports.


Psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA), encompassing non-radiographic axial spondyloarthritis (nr-axSpA) and ankylosing spondylitis (AS), are inflammatory rheumatic diseases defined as spondyloarthritis (SpA) [1]. While these conditions have diverse manifestations, they also share certain clinical features, genetic factors, and disease mechanisms [1, 2]. Advances in understanding the pathogenesis of SpA have led to a new therapeutic option with targeted, small-molecule Janus kinase (JAK) inhibitors, which have also been assessed for treating different immune-mediated inflammatory diseases (IMIDs) [3, 4]. Individuals with IMIDs are more susceptible to developing infections and potentially other adverse events because of underlying health conditions and the use of immunomodulatory therapies, such as JAK inhibitors [5]. Recent results from the ORAL Surveillance study, evaluating the safety of tofacitinib, a different JAK inhibitor, relative to tumor necrosis factor (TNF) inhibitor therapy in a cardiovascular (CV)-risk enriched population with rheumatoid arthritis (RA), suggest an increased risk of major adverse cardiovascular events (MACE) and malignancy excluding nonmelanoma skin cancer (NMSC) with tofacitinib [6]. These findings have raised questions about the safety of JAK inhibitors with uncertainty remaining about whether these safety events are exclusive to RA or tofacitinib or may relate to a JAK inhibitor class effect [7]. Longer-term safety data across disease states are needed to better characterize the benefit-risk profiles of JAK inhibitors and guide clinical decision-making [8].

Upadacitinib is an oral JAK inhibitor that has demonstrated efficacy and safety in phase 3 clinical trials in patients with PsA [9, 10], AS [11, 12], and nr-axSpA [13], among other IMIDs. Across these approved indications, upadacitinib 15 mg once daily (QD) is the recommended treatment dose [14]. A long-term safety analysis of over 15,000 patient-years (PY) across RA, PsA, AS, and atopic dermatitis (AD) showed that upadacitinib was generally well tolerated with no new safety risks observed compared with previous reports [15]. Here, we present an integrated analysis of the safety profile of upadacitinib 15 mg QD across PsA, AS, and nr-axSpA from the SELECT trials, including pooled data from patients with axSpA and active comparator data from those treated with adalimumab.



This analysis included pooled safety data from five randomized, placebo-controlled trials of the SELECT upadacitinib clinical program: two in PsA [9, 10], two in AS [11, 12], and one in nr-axSpA [13] (Supplementary Materials Table S1).

Patients enrolled in each trial were ≥ 18 years of age with active disease. Briefly, patients with PsA had an inadequate response or intolerance to non-biologic (SELECT-PsA 1) or biologic (SELECT-PsA 2) disease-modifying antirheumatic drugs (DMARDs). In the AS trials, patients had an inadequate response to non-steroidal anti-inflammatory drugs (NSAIDs) or intolerance/contraindication for NSAIDs and were naïve to biologic DMARDs (SELECT-AXIS 1; phase 2/3) or had an inadequate response or intolerance to biologic DMARDs (SELECT-AXIS 2 bDMARD-IR study). Patients with nr-axSpA (SELECT-AXIS 2 nr-axSpA study) were either naïve to or had an inadequate response or intolerance to biologic DMARDs. All trials complied with the Declaration of Helsinki, the International Conference on Harmonization guidelines, and local regulations and were approved by independent ethics committees and institutional review boards. Each patient provided written informed consent.


Pooled safety data across trials were summarized for each disease and treatment group as follows: PsA (pooled upadacitinib 15 mg QD from SELECT-PsA 1 and 2 and adalimumab 40 mg every other week from the SELECT-PsA 1 trial), AS (pooled upadacitinib 15 mg QD from SELECT-AXIS 1 and SELECT-AXIS 2 Study 1), nr-axSpA (upadacitinib 15 mg QD from SELECT-AXIS 2 Study 2), and pooled axSpA (pooled upadacitinib 15 mg QD from the two AS trials and one nr-axSpA trial) (Supplementary Materials Figure S1). Across all studies, patients could have received upadacitinib in combination with conventional synthetic DMARDs, oral corticosteroids, or NSAIDs, if appropriate. Patients with axSpA could have received background therapy with oral corticosteroids and NSAIDs, and patients with PsA were permitted to use NSAIDs, oral corticosteroids, and up to two concomitant conventional synthetic DMARDs.

Safety Assessments

Patients who received ≥ 1 dose of upadacitinib 15 mg or adalimumab were included in this analysis. Safety data were evaluated up to the data cutoff date of 15 August 2022 for all trials, with a maximum exposure to the study drug of up to 5 years across treatment groups.

Treatment-emergent adverse events (TEAEs), adverse events of special interest (AESI), and laboratory values were assessed. TEAEs were defined as adverse events (AE) that occurred on or after the first dose of the study drug and within 30 days (upadacitinib 15 mg) or 70 days (adalimumab) after the last dose of study drug. Mortality assessment also included deaths that occurred beyond the 30 or 70 days after the last dose of study drug. The Medical Dictionary for Regulatory Activities version 25.0 was used to code TEAEs by system organ class and preferred term. Potentially clinically significant laboratory parameters (grade 2, 3, and 4 changes) were graded based on the NCI Common Terminology Criteria for Adverse Events (CTCAE), version 4.03, for PsA, AS, and nr-axSpA. Adverse events of laboratory abnormalities were investigator-reported and were not based on formal, objective criteria.

Cardiovascular (CV) events and venous thromboembolic events (VTEs) were adjudicated by an independent cardiovascular adjudication committee blinded to treatment allocation. Gastrointestinal perforations were adjudicated by an internal AbbVie adjudication committee comprised of gastroenterologists outside of the upadacitinib program. MACE was defined as CV death, non-fatal myocardial infarction (MI), and non-fatal stroke; VTEs were defined as fatal or non-fatal deep vein thrombosis (DVT) and pulmonary embolism (PE). Opportunistic infections were evaluated separately from active tuberculosis and HZ. A patient’s history of CV risk factors was assessed including history of a CV event, venous thromboembolism, hypertension, diabetes mellitus, tobacco/nicotine use (current or former), elevated low-density lipoprotein cholesterol (≥ 3.36 mmol/l), and lowered high-density lipoprotein cholesterol (< 1.034 mmol/l).

Statistical Analyses

TEAEs were summarized as exposure-adjusted event rates (EAERs; events per 100 patient-years [E/100 PY]); multiple TEAEs that occurred in a single patient were included in the numerator. In addition, exposure-adjusted incidence rates (EAIRs) and the number of patients with an event per 100 PY [n/100 PY]) were presented with exposure evaluated up to the onset of the first event. Associated 95% confidence intervals (CIs) for EAERs and EAIRs were computed using the exact method for the Poisson mean. The proportions of patients with potentially clinically significant laboratory parameters were reported.

The standardized mortality ratio (SMR) was estimated for the general population using World Health Organization country-age-gender-specific death data through 2016; 95% CIs were calculated using Byars’ approximation. Standardized incidence ratio (SIR) for malignancy excluding NMSC was estimated for the general population using age- and gender-adjusted cancer data in 2000–2018 from the Surveillance, Epidemiology, and End Results (SEER) database; 95% CIs were calculated following a Poisson distribution.


Patients and Exposure

A total of 1789 patients across PsA (n = 907), AS (n = 596), and nr-axSpA (n = 286) received ≥ 1 dose of upadacitinib 15 mg in the SELECT trials. In PsA only, 429 patients received ≥ 1 dose of adalimumab. Overall, 3689.4 PY of exposure with upadacitinib 15 mg were reported across PsA, AS, and nr-axSpA, with the longest exposure occurring within the PsA studies (2426.4 PY); 1146.6 PY of exposure were reported with adalimumab in PsA (Table 1). The maximum duration of upadacitinib treatment was longest in PsA (5.0 years, median 2.9 years) and shortest in nr-axSpA (2.2 years, median 1.0 years).

Table 1 Exposure and overview of treatment-emergent adverse events

Baseline characteristics generally reflected the respective disease state (Table 2). The AS and nr-axSpA populations were younger and a higher proportion of patients were female in the PsA and nr-axSpA populations. Most patients with PsA used concomitant non-biologic DMARDs, whereas most with AS and nr-axSpA used concomitant NSAIDs. Across treatment groups, 70%–84% of patients had ≥ 1 CV risk factor at baseline.

Table 2 Baseline demographics and disease characteristics

Overview of AEs

The rates of TEAEs with upadacitinib 15 mg ranged from 185.9 E/100 PY in AS to 232.9 E/100 PY in PsA (Table 1). Overall, the rates of TEAEs and serious TEAEs were highest in PsA and numerically higher with upadacitinib 15 mg compared with adalimumab; these rates were generally similar between AS and nr-axSpA. TEAEs leading to discontinuation were comparable across treatment groups and diseases with a numerically lower rate in AS. The most common TEAEs with upadacitinib 15 mg across PsA, AS, and nr-axSpA were upper respiratory tract infection, nasopharyngitis, and COVID-19 infection (Supplementary Materials Table S2). Most cases of COVID-19 infection were mild or moderate in severity (79% to 94%), and the study drug was interrupted in 65% to 77% of patients but rarely withdrawn (< 6%) (Supplementary Materials Table S3).

Rates of death were numerically higher with upadacitinib 15 mg versus adalimumab in PsA, which was attributed to an increased number of COVID-19-related deaths in patients receiving upadacitinib (10 of 20 deaths on upadacitinib were COVID-19-related; 8 of 10 were treatment-emergent). The majority of COVID-19-related deaths (8/10 [80%]) occurred prior to 2022 before global vaccination efforts were widely implemented. Of the 10 COVID-19-related deaths, three patients were fully vaccinated, one patient was partially vaccinated, and six patients did not report receiving any COVID-19 vaccine (Supplementary Materials Table S4). The mean age of COVID-19-related deaths on upadacitinib was 60 years old, and approximately half of patients received concomitant glucocorticoid treatment during their infection. Additionally, the majority of COVID-19-related deaths (7/10 [70%]) on upadacitinib occurred in patients with a body mass index ≥ 30. Beyond COVID-19-related deaths, the most common cause of death was related to CV disease. Two deaths were reported with adalimumab, one due to multiple injuries from a road traffic accident and the other due to COVID-19 pneumonia. In AS, one death occurred with upadacitinib 15 mg because of multiple injuries from a fall. No deaths were reported in nr-axSpA. An SMR analysis indicated that the mortality rate among patients with PsA treated with upadacitinib 15 mg was not higher than what would have been expected among the general population (SMR [95% CI]: 0.71 [0.40, 1.18] including COVID-19-related deaths; 0.33 [0.13, 0.69] excluding COVID-19-related deaths). SMR estimates were not calculated for AS, nr-axSpA, or pooled axSpA because of a low number of events.

Adverse Events of Special Interest

AESIs are presented as EAERs (Fig. 1) and EAIRs (Supplementary Materials Figure S2) by treatment group and disease state. Rates of serious infection with upadacitinib 15 mg were higher in PsA versus AS and nr-axSpA, largely driven by an increased rate of COVID-19 infections due to the pandemic; serious infection rates were lowest in nr-axSpA. In PsA, rates of serious infection were higher with upadacitinib 15 mg than adalimumab. Overall rates of serious infection including and excluding COVID-19 events were 3.9 versus 2.0 E/100 PY for PsA, 2.6 versus 1.1 E/100 PY for AS, and 0.9 versus 0.3 E/100 PY for upadacitinib 15 mg in nr-axSpA and 1.6 versus 0.9 E/100 PY for adalimumab in PsA. COVID-19 pneumonia was the most common serious infection across disease states followed by COVID-19 in PsA and AS and pyelonephritis in nr-axSpA. Opportunistic infections excluding tuberculosis and HZ were infrequently reported, with the highest rates observed with upadacitinib 15 mg in PsA. No opportunistic infections occurred in nr-axSpA or with adalimumab in PsA. Oesophageal candidiasis was the most common opportunistic infection reported across disease states (Supplementary Materials Table S5). No active tuberculosis cases were observed in this analysis of the SELECT trials.

Fig. 1
figure 1figure 1figure 1figure 1

Exposure-adjusted event rates for adverse events of special interesta. ADA adalimumab; AE adverse event; AESI adverse event of special interest; AS ankylosing spondylitis; axSpA axial spondyloarthritis; COVID-19 coronavirus disease 2019; CI confidence interval; CPK creatine phosphokinase; EMM extra-musculoskeletal manifestation; EOW every other week; IBD inflammatory bowel disease; MACE major adverse cardiovascular event; NMSC nonmelanoma skin cancer; nr-axSpA non-radiographic axial spondyloarthritis; PsA psoriatic arthritis; QD once daily; TB tuberculosis; UPA upadacitinib; VTE venous thromboembolism. aIncludes AS and nr-axSpA; UPA sample sizes include patients who switched from placebo. bOpportunistic infections excluding tuberculosis and herpes zoster. cPer the protocol, CPK elevation was not measured in the nr-axSpA study. dLymphoma AEs reported in patients with PsA treated with UPA 15 mg were transient abnormal lymphocyte morphology and were not confirmed to be true lymphomas. eMACE was defined as non-fatal myocardial infarction, non-fatal stroke, and cardiovascular death. fVTE includes deep vein thrombosis and pulmonary embolism (fatal and non-fatal). gIBD includes inflammatory bowel disease, colitis ulcerative, ulcerative colitis, Crohn’s disease, ulcerative proctitis, and proctitis. hUveitis includes uveitis, iritis, and iridocyclitis. iPsoriasis includes psoriasis, nail psoriasis, guttate psoriasis, rebound psoriasis, pustular psoriasis, paradoxical psoriasis, erythrodermic psoriasis, psoriasis area severity index, psoriasis area severity index increased, psoriasis severity index decreased, psoriatic arthropathy, and juvenile psoriatic arthritis. AESIs are presented as exposure-adjusted incidence rates in Supplementary Materials Figure S2

HZ with upadacitinib 15 mg occurred at similar rates among PsA, AS, and nr-axSpA, with higher rates in the upadacitinib 15 mg group compared with the adalimumab group in PsA. Most HZ events were mild or moderate in severity (90% in PsA, 96% in AS, and 100% in nr-axSpA) and rarely led to treatment discontinuation. Additionally, the majority of HZ events involved a single dermatome (61% in PsA, 78% in AS, and 100% in nr-axSpA) (Supplementary Materials Table S6). Several cases of post-herpetic neuralgia were reported in PsA (0.2 E/100 PY) and AS (0.2 E/100 PY), and one case of disseminated varicella-zoster virus infection was reported in AS (0.1 E/100 PY). Of the 67 cases of HZ occurring in patients with PsA, three patients (5%) showed ophthalmic involvement.

Rates of malignancy excluding NMSC were generally consistent across diseases and between upadacitinib 15 mg and adalimumab in PsA. No patterns pertaining to the types of malignancies were observed. In PsA, most malignancies other than NMSC occurred in one patient each; prostate cancer was reported in two patients. In AS, there was one case of metastases to the liver and one case of squamous cell carcinoma of the tongue, which occurred in a former smoker. Invasive ductal breast carcinoma was reported in a single patient with nr-axSpA. The SIR analysis (95% CI) indicated that malignancy (excluding NMSC) risk among patients in the PsA (0.94 [0.54, 1.53]) and pooled axSpA (0.58 [0.12, 1.69]) groups treated with upadacitinib 15 mg were not higher than what would be expected among the general population. NMSC was reported at similar rates across diseases (≤ 0.9 E/100 PY), with higher rates observed with upadacitinib 15 mg versus adalimumab in PsA (0.9 E/100 PY with upadacitinib 15 mg versus 0.2 E/100 PY with adalimumab). Events of NMSC were generally non-serious, and one event with upadacitinib 15 mg in PsA led to treatment discontinuation. Five cases of abnormal lymphocyte morphology were reported with upadacitinib 15 mg: three in the PsA group and one each in the AS and nr-axSpA groups; none of these events were confirmed to be lymphoma.

Adjudicated MACE and VTE occurred at similar rates (≤ 0.6 E/100 PY) across diseases and treatment groups. In PsA, six patients receiving upadacitinib 15 mg experienced MACE: one had fatal acute MI, one had sudden cardiac death, two had a non-fatal MI, and two had a non-fatal stroke. Three patients receiving adalimumab experienced MACE: two had a non-fatal stroke and one had non-fatal MI. One patient each with AS and nr-axSpA receiving upadacitinib 15 mg had a non-fatal stroke. All events of VTE were non-fatal, and PE was more common than DVT across diseases. One patient with PsA receiving upadacitinib 15 mg had concurrent PE and DVT. At least one cardiovascular and/or thromboembolic risk factor was identified in patients with MACE and VTE events.

Overall, events of inflammatory bowel disease (IBD) with upadacitinib 15 mg were rare across all diseases (≤ 0.3 E/100 PY); no IBD events were reported with adalimumab in PsA. Rates of uveitis were numerically higher in AS (2.8 E/100 PY; 26 events; 20 flares and 6 new onset) and lowest in PsA (0.2 E/100 PY; 4 events; 3 flares and 1 new onset). Psoriasis was the most common extra-musculoskeletal manifestation in PsA (≤ 7.9 E/100 PY); few psoriasis events were reported in AS (0.4 E/100 PY) and none in nr-axSpA.

Rates of hepatic disorders were higher in PsA versus AS and nr-axSpA and numerically higher with adalimumab versus upadacitinib 15 mg in PsA. Most hepatic disorders were non-serious, mild to moderate, transient transaminase elevations, which rarely led to study drug discontinuation. Across indications, no cases consistent with probable drug-induced liver injury were attributed to upadacitinib.

Anemia and neutropenia occurred at similar rates across diseases, and few were serious in nature. Mean decreases in hemoglobin and neutrophil counts were seen with upadacitinib treatment. Most changes in hemoglobin levels were transient and resolved with temporary treatment interruption. No clear association between low neutrophil counts and the occurrence of serious infections was found. Lymphopenia most frequently occurred in patients with PsA receiving upadacitinib 15 mg. Increases in CPK were observed across indications, although few events were serious and led to study drug discontinuation. No rhabdomyolysis was reported. Few grade 3 or 4 laboratory abnormalities were reported across diseases (Supplementary Materials Table S7). Of the grade 2, 3, and 4 aminotransferase disorders reported in PsA, a numerically higher percentage of patients receiving upadacitinib plus concomitant MTX experienced alanine and aspartate aminotransferase abnormalities compared to those receiving upadacitinib alone (Supplementary Materials Table S8). A similar trend was not observed in AS or nr-axSpA, although the sample size of patients receiving concomitant MTX was limited.


This integrated analysis of 1789 patients receiving upadacitinib for up to 5 years of exposure demonstrated a generally consistent safety profile across PsA, AS, and nr-axSpA; our findings were also largely consistent with the safety profile of upadacitinib in RA [9,10,11,12,13, 15, 16]. No new safety signals were identified based on previous reports. Similar rates of key AESIs, such as MACE, VTE, and malignancy excluding NMSC, were observed across diseases and treatment groups.

Variability in patient baseline characteristics and AE types reflected the different disease states and associated comorbidities [17, 18]. Younger age and a lower risk factor profile were observed in patients with AS and nr-axSpA versus PsA. Most patients with PsA used concomitant non-biologic DMARDs, whereas most with AS and nr-axSpA used concomitant NSAIDs, attributed mainly to disease-specific treatment recommendations and guidelines [19, 20].

In PsA, the safety profiles of upadacitinib 15 mg and adalimumab were generally consistent with previous reports. Rates of serious infection, HZ, lymphopenia, and NMSC were numerically higher with upadacitinib 15 mg compared with adalimumab; elevated rates for most of these AESIs were also found with upadacitinib 15 mg in PsA versus AS and nr-axSpA. An increased risk of infection and HZ has been associated with JAK inhibitor treatment in patients with rheumatic disease [5, 21, 22]. COVID-19 infections, especially in PsA, largely drove serious infection rates. The COVID-19 pandemic was ongoing when studies were conducted, likely contributing to the increased events of serious infection, but the impact of JAK inhibitors on the course of a COVID-19 infection remains uncertain [23, 24]. Most cases of COVID-19 were mild or moderate in severity, and the study drug was rarely withdrawn (< 6% of cases). Possible factors influencing the observed higher proportion of COVID-19-related deaths in upadacitinib-treated patients with PsA include higher mean age compared to those in the AS and nr-axSpA cohorts as well as higher rates of metabolic syndrome, non-biologic DMARD use, and CV risk factors. Other potential contributing factors include low rates of COVID-19 vaccination, geographical predominance in Eastern Europe where increased COVID-19 mortality was reported [25], and a potential protective effect of TNF inhibitor treatment for adverse COVID-19 outcomes [26]. Ultimately, the crucial role of COVID-19 vaccination in reducing risks should be underlined, and effective vaccination strategies are possible for patients receiving JAKi therapy [27]. Higher rates of HZ infections were found with upadacitinib versus adalimumab but were mostly non-serious and affected a single dermatome. Notably, although upadacitinib labeling recommends vaccinating patients against HZ before starting upadacitinib treatment, < 5% of patients across indications were vaccinated upon study entry. Overall, no consistent trends in changes in lymphocyte count with JAK inhibition have been reported [21]. Patients with IMIDs receiving biologics or JAK inhibitors may have a higher risk of NMSC, but no causal link has been established [28,29,30].

Results from the head-to-head ORAL Surveillance study demonstrated an increased risk of MACE and malignancy excluding NMSC with tofacitinib relative to TNF inhibitor therapy in a CV-risk enriched population with RA aged ≥ 50 years [6]. In the current analysis, rates of MACE, malignancy excluding NMSC, and VTE were similar across diseases and treatment groups. However, due to the limited numbers of these events reported in the patient populations evaluated here, an in-depth post hoc analysis of patients aged ≥ 50 years having similar cardiovascular risk factors as the ORAL Surveillance population was not possible.

Several limitations of this analysis should be considered. Placebo control arms were not included, and AS and nr-axSpA studies did not include an active comparator; moreover, adalimumab treatment was only included in one of the two pooled PsA studies. Findings from this analysis should be interpreted with caution as the duration of exposure was limited for AS and nr-axSpA. Data for upadacitinib were presented across PsA, AS, and nr-axSpA to provide a comprehensive safety overview of SpA indications; as a limitation, however, background medications and baseline characteristics of patients varied across disease indications (e.g., patients with AS and nr-axSpA tended to be younger and have a lower number of CV risk factors relative to patients with PsA). This study also did not include the characterization of safety events observed with upadacitinib relative to other approved indications (rheumatoid arthritis, atopic dermatitis, ulcerative colitis, and Crohn’s disease). Lastly, as an inherent limitation of long-term extensions, only patients who met the original entry criteria for the studies were included. Continued follow-up is warranted to characterize and understand the long-term safety profile of upadacitinib.


In this integrated safety analysis, higher rates of serious infection, HZ, lymphopenia, and NMSC were observed with upadacitinib 15 mg compared with adalimumab in PsA; patients with PsA had slightly elevated rates for most of these TEAEs than those with AS and nr-axSpA. Upadacitinib 15 mg demonstrated a generally consistent safety profile across PsA, AS, and nr-axSpA, with no new safety signals identified based on previous reports. The safety of upadacitinib will continue to be monitored in the ongoing clinical trial program and postmarketing reports.