Sex differences in the efficacy, safety and persistence of tofacitinib in patients with rheumatoid arthritis: a post hoc analysis of phase III and long-term extension trials Authors: Vibeke Strand, Hendrik Schulze-Koops, Samar Al-Emadi, Cassandra D Kinch, David Gruben, Rebecca Germino, Carol A Connell, Eduardo Mysler Published: 2025 ABSTRACT AbstractObjectivesEvaluate tofacitinib efficacy, safety and persistence by sex in rheumatoid arthritis (RA).DesignPost hoc analyses using data from phase III placebo-controlled randomised controlled trials (ORAL Scan, ORAL Sync and ORAL Standard).SettingORAL Scan, ORAL Sync and ORAL Standard were global, multicentre trials conducted across 111, 114 and 115 sites, respectively.ParticipantsThe trials enrolled adults with active RA and prior inadequate response to methotrexate (ORAL Scan/ORAL Standard) or ≥1 conventional synthetic or biologic disease-modifying antirheumatic drug (ORAL Sync). Post hoc analyses included 2265 patients (1870 female and 395 male).InterventionsPatients received tofacitinib 5 mg or 10 mg two times a day, adalimumab or placebo.Outcome measuresEfficacy outcomes to month 12 included American College of Rheumatology (ACR)20, 50 and 70 responses, Disease Activity Score in 28 joints (DAS28) (erythrocyte sedimentation rate (ESR))-defined low disease activity (LDA) and remission, DAS28 (C reactive protein (CRP)) ≤3.2 and <2.6 and normative thresholds for Health Assessment Questionnaire-Disability Index (HAQ-DI) and Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F). Regression analyses included exploration of possible non-linearity in the relationship between age and treatment response in males and females. Safety was evaluated at month 24. Additionally, Kaplan-Meier survival analysis of two open-label, long-term extension studies (NCT00661661 and ORAL Sequel) assessed time to discontinuation of tofacitinib treatment.ResultsAt baseline, female patients had similar DAS28(CRP and ESR), slightly higher HAQ-DI and lower FACIT-F scores versus male patients (n=395). ORs for active treatments (tofacitinib and adalimumab) versus placebo were generally >1 for ACR20, 50 and 70 responses, DAS28(CRP) ≤3.2 and <2.6, DAS28(ESR) LDA and remission and proportions of patients reporting FACIT-F scores ≥normative values and trends towards numerically higher ORs in males versus females were observed. Sex differences were not consistently observed for HAQ-DI normative threshold ORs and, for most outcomes, when comparing responses within active treatment groups by sex. Polynomial modelling suggested age did not impact treatment response differences. Safety and persistence were similar across sexes.ConclusionsIn post hoc analyses, tofacitinib was efficacious across both sexes, with higher responses in males observed particularly for more stringent composite endpoints and patient-reported outcomes. Findings are generally consistent with studies of other advanced RA therapies. Safety and persistence were similar across sexes. Interpretation is limited by the small proportion of male patients (<20%) and baseline differences between sexes.Trial registration numberNCT00847613, NCT00856544, NCT00853385, NCT00661661 and NCT00413699.Keywords: Patient Reported Outcome Measures, Safety, THERAPEUTICS, RHEUMATOLOGY STRENGTHS AND LIMITATIONS OF THIS STUDY This analysis was conducted using data from randomised controlled trials, eliminating potential treatment selection and channelling biases that may be associated with sex in observational datasets. This analysis was conducted using data from randomised controlled trials, eliminating potential treatment selection and channelling biases that may be associated with sex in observational datasets. These analyses were post hoc in nature, and the included studies were not designed to make comparisons between sexes. These analyses were post hoc in nature, and the included studies were not designed to make comparisons between sexes. As most enrolled patients were female, the sample sizes of male patients were small, particularly for some treatment groups, limiting interpretation. As most enrolled patients were female, the sample sizes of male patients were small, particularly for some treatment groups, limiting interpretation. The findings may be influenced by baseline differences between sexes, including greater physical impairment and fatigue observed in female versus male patients. The findings may be influenced by baseline differences between sexes, including greater physical impairment and fatigue observed in female versus male patients. ORAL Standard, which was not powered to make direct comparisons between tofacitinib and adalimumab, was the only study in this dataset that provided data for adalimumab. ORAL Standard, which was not powered to make direct comparisons between tofacitinib and adalimumab, was the only study in this dataset that provided data for adalimumab. Introduction In patients with rheumatoid arthritis (RA), differences between sexes (male vs female) in disease activity and treatment responses have previously been reported. Multiple registry studies report greater disease activity, physical function impairment,1 2 lower remission rates2 3 and lower likelihood of attaining low disease activity (LDA)3 in females versus males. Inferior responses to conventional synthetic disease-modifying antirheumatic drugs (csDMARDs)4 and tumour necrosis factor inhibitors (TNFi) in early versus established RA5 have been observed in females versus males, although differences were not always clinically significant. In a clinical trial of first-line TNFi therapy, male sex was a positive predictor of LDA/remission in patients with RA.6 Differences in disease presentation and treatment response may be functions of gender (ie, socially constructed), sex (ie, biological) or a mixture of both. Lack of sex/gender neutrality in the measures of disease activity and disability routinely employed in the assessment of RA may account for discrepancies between male and female patient responses.1 2 For example, Disease Activity Score in 28 joints (DAS28) tends to be higher in female versus male patients, as do core components of the American College of Rheumatology (ACR) response criteria (including swollen joint counts (SJC) and tender joint counts (TJC), erythrocyte sedimentation rate (ESR) and Health Assessment Questionnaire-Disability Index (HAQ-DI)),1 2 which may impact ACR response rates. Additionally, variability in patient experience of disease has been previously reported. Patient-reported improvements in fatigue, pain and other aspects of health-related quality of life (Short Form (36) Health Survey (SF-36) scores) are dampened in female versus male patients receiving advanced treatments for RA.1 7 Differences in the clinical management of patients with RA may further impact sex differences in responses. Notably, higher levels of discordance between patient and physician global assessments of disease activity have been reported in female versus male rheumatology patients, particularly for functional impairment, fatigue and pain, which are likely to be higher when reported by patients compared with physicians.8 Such discordance has been shown to negatively impact medical care, adherence to therapy and disease outcomes.9 Indeed, the mechanisms responsible for sex-based differences among patients with RA are incompletely understood, and X-linked genetic differences and sex hormones may be contributing factors.10 Hormonal changes associated with menopause may affect RA aetiopathogenesis.11 Postmenopausal female patients have been shown to have worse joint destruction and physical disability than premenopausal patients,12 which may, in turn, account for differing treatment responses. Dampened responses to biologic DMARDs (bDMARDs) have been demonstrated in older (aged >65 years) versus younger patients with RA and in females versus males.13 14 A trend towards a reduced response to tofacitinib in older (aged ≥65 years) females versus males was observed in a post hoc analysis of clinical trials;15 however, it is unclear whether this may be accounted for by differences in physiological changes, such as hormonal shifts associated with menopause or other genetic/environmental/social factors. Epigenetic changes, which represent the confluence of gene and environment interaction, accumulate with age and the resultant impact on the immune system, and in turn on the host response to the microbiome, may play an important role in RA pathogenesis and sex-related differences.10 16 Tofacitinib is an oral Janus kinase (JAK) inhibitor for the treatment of RA. An Italian real-world multicentre analysis evaluated sex differences in patients’ responses to tofacitinib and another JAK inhibitor, baricitinib.17 Although improvements in patient-reported pain (pain; Visual Analogue Scale (VAS)) were significantly greater in males versus females, overall, there were no differences in TJC and SJC, C reactive protein (CRP) levels or DAS28-defined remission or LDA between sexes. Here, we evaluated the efficacy and safety of tofacitinib in patients with RA stratified by sex in a post hoc analysis of placebo-controlled randomised controlled trials (RCTs). We also assessed potential associations between age and treatment response over time in males and females to understand if physiological shifts or other factors associated with age account for any separation of response between males and females. Methods Study design and patients In this post hoc analysis of pooled data from three phase III placebo-controlled RCTs (ORAL Scan (NCT00847613),18 ORAL Sync (NCT00856544)19 and ORAL Standard (NCT00853385)),20 we evaluated the impact of sex on tofacitinib clinical responses, including patient-reported outcomes (PROs), safety and persistence in patients with an inadequate response (IR) to DMARDs. Briefly, ORAL Scan, ORAL Sync and ORAL Standard were global, multicentre trials conducted across 111, 114 and 115 sites, respectively. Eligible patients were aged ≥18 years with a diagnosis of active RA, defined by ACR 1987 Revised Criteria; active disease classified as ≥6 tender/swollen joints (≥4 in ORAL Sync) and ESR >28 mm/hour (Westergren method) or CRP level >7 mg/L. Patients had prior IR to methotrexate (MTX; ORAL Scan/ORAL Standard) or at least one csDMARD or bDMARD (ORAL Sync). Patients received tofacitinib 5 mg or 10 mg two times a day, adalimumab 40 mg once every 2 weeks (ORAL Standard only as an active control) or placebo; all treatments were administered with csDMARDs (MTX in ORAL Scan/ORAL Standard). Placebo-treated patients were blindly advanced to receive tofacitinib 5 mg or 10 mg two times a day at month 3 (non-responders) or month 6 (all remaining patients). Persistence analyses included pooled data from two long-term extension (LTE) studies (NCT0066166121 and ORAL Sequel (NCT00413699))22 23 where patients received open-label tofacitinib 5 mg or 10 mg two times a day as monotherapy or with background csDMARDs. Patients were categorised as having received an average tofacitinib dose of 5 mg or 10 mg two times a day based on their average total daily dose over the course of observation (<15 or ≥15 mg, respectively). Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research. Clinical efficacy Outcomes at months 1, 3, 6 and 12 included ACR ≥20/50/70% response criteria (ACR20/50/70 response rates), DAS28(CRP) ≤3.2 and <2.6, DAS28(ESR)-defined LDA (≤3.2) and remission (<2.6), Clinical Disease Activity Index (CDAI)-defined LDA (≤10) and remission (≤2.8)24 and changes from baseline in DAS28(CRP), DAS28(ESR), CDAI, TJC (28-joint count), SJC (28-joint count), Physician Global Assessment of arthritis, CRP and ESR. Patient-reported outcomes PROs assessed at months 1, 3, 6 and 12 included changes from baseline in Patient Global Assessment of disease activity (PtGA), Pain (VAS 0–100 mm), HAQ-DI, Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F), and SF-36 physical (PCS) and mental (MCS) component summary scores and eight domain scores. Also assessed at the same timepoints were the proportions of patients reporting moderate (≥30%) or substantial (≥50%) improvements in Pain (VAS) scores and improvements in Pain ≥minimum clinically important differences (MCIDs; ≥10 (VAS 0–100 mm)).25 The proportions of patients with improvements ≥MCID in HAQ-DI scores (≥0.22),26 and scores ≥normative values (normative threshold: <0.25),27 improvements ≥MCID in FACIT-F scores (increase from baseline of ≥4.0)28 and scores ≥normative values (normative threshold: ≥43.5)29 were also reported at the same timepoints. Improvements ≥MCID in SF-36 PCS and MCS scores (≥2.5 for PCS and MCS),25 and scores ≥normative values (normative threshold: based on age- and sex-matched norms) were reported at month 3. Safety Safety was assessed through month 24 via the incidence of treatment-emergent adverse events (TEAEs), serious adverse events (SAEs), discontinuations due to adverse events (AEs), deaths and AEs of special interest (AESI), which included serious infection events, herpes zoster (HZ; including non-serious and serious events), opportunistic infections (OIs; excluding tuberculosis), malignancies excluding non-melanoma skin cancer (NMSC), NMSC, major adverse cardiovascular events (MACE) and venous thromboembolism (VTE) (refer to the online supplemental file 1 for the full list). Persistence Treatment persistence was assessed via three time-to-event outcomes: time to all-cause discontinuation, time to discontinuation due to AEs and time to discontinuation due to lack of efficacy. The proportions of patients remaining on tofacitinib (ie, drug continuation rates) were also calculated at 2 years and 5 years. Statistical analyses Binary endpoints were assessed using logistic regression at each visit; this approach was taken to keep the analyses as similar as possible to those of the individual RCTs, specifically by-visit analyses, using logistic regression to add independent variables needed to produce the estimates of rates for female versus male patients. Continuous endpoints were assessed using a repeated-measures linear model. Independent variables included in the model (fixed effects) were geographic region, RCT, sex, treatment group (and sex:treatment group interaction), and, for the longitudinal model only, baseline value of the outcome of interest and study visit (and all-way interactions between sex, treatment group and study visit). Also, for the repeated measures model, a covariance matrix was specified using a compound-symmetry form. If convergence failed, the form applied was autoregressive (first order). The TJC, Physician Global Assessment of arthritis, SF-36 PCS and SF-36 vitality domain score used an autoregressive covariance. There was no imputation for missing data. Comparisons between treatment groups or between sexes were assessed as ORs for binary endpoints and mean differences for continuous endpoints, with 95% CIs. Two sets of model-based comparisons were performed to assess differences between treatment groups and sexes. The figures in this report present ORs and mean differences for active treatments versus placebo, within each sex. Except for SF-36 (month 3 only), these analyses were performed at the timepoints with placebo data available, that is, months 1, 3 and 6. ORs and mean differences for female versus male within each treatment group were also produced; results for these comparisons are provided in online supplemental file 1. Except for SF-36 (month 3 only), these analyses were performed at months 1, 3, 6 and, for active treatments only, month 12. Two-sided p values were generated for each comparison; these were not corrected for multiple comparisons and should be considered as descriptive and not hypothesis-testing. For illustrative purposes and to facilitate the viewing of improvements at month 3 in SF-36 simultaneously across the eight domains, spydergrams were generated. These are plots of the mean values of the SF-36 domain scores (range: 0–100) at month 3, and at baseline, where at baseline the treatment group means were combined (a composite baseline), and the US population norms reweighted for the age-sex distribution of the data are also plotted for reference. On account of previously cited hypotheses regarding higher disease activity and treatment resistance in female patients with RA experiencing perihormonal shifts (such as menopause),2 30 regression models were expanded to include increasing order polynomial terms in age to explore possible non-linearity in the relationship of age with efficacy response to tofacitinib. Menopause status was not robustly collected in the phase III RCTs, so patient age was used as a proxy measure. Further details regarding this analysis are in the online supplemental file 1. Safety data are summarised descriptively. Kaplan‐Meier survival curves were used to compare treatment persistence (difference between the treatment discontinuation and/or end‐of‐study date and the first tofacitinib dose date +1 day) for each of the doses of tofacitinib (average 5 mg two times a day, average 10 mg two times a day) in the LTE studies. Ongoing patients were censored at the LTE study cut‐off dates (final data cut; March 2017), and completers were censored at the end‐of‐study date. In addition, for Kaplan-Meier analyses of discontinuation due to AEs or lack of efficacy, discontinuations due to other reasons were censored. For each analysis, 2‐ and 5‐year continuation (survival) rates were derived, with 100% of the study population considered, as all patients entered the LTE studies. Results Patients Overall, 2265 patients were included in the efficacy analyses; of these, 1870 (82.6%) were female and 395 (17.4%) were male. Demographics and baseline characteristics were generally comparable across sexes and treatment groups (online supplemental table S1). Some variability between males and females was observed, with females tending to have longer disease duration, lower CRP levels, slightly higher Pain (VAS) and HAQ-DI scores, lower FACIT-F scores and lower rates of prior TNFi use (online supplemental table S1). Clinical efficacy Through month 6, the proportions of patients with ACR20/50/70 responses were generally numerically higher (ORs >1) with tofacitinib or adalimumab versus placebo in both sexes, though these trends were generally more pronounced in males versus females (figure 1, online supplemental figure S1). Comparing ACR20/50/70 responses within each treatment group, a higher proportion of males than females had ACR20/50 responses at months 3 and 6 with tofacitinib 5 mg two times a day, with similar trends observed at some timepoints for tofacitinib 10 mg two times a day and adalimumab (online supplemental figure S1). ORs (active treatments vs placebo) for (A) ACR20, (B) ACR50 and (C) ACR70 response, stratified by sex (phase III RCTs).† *p<0.05, **p<0.001 and ***p<0.0001 for tofacitinib 5 mg or 10 mg two times a day or ADA versus placebo. Response rates by group are shown in online supplemental figure S1. ORs designated as ‘Inf’ were infinite owing to division by zero, as no male patients in the placebo group achieved an ACR70 response at month 1 (vs n=2, 9 and 1 male patients in the tofacitinib 5 mg two times a day, tofacitinib 10 mg two times a day and ADA groups, respectively). †Includes all randomised and treated patients from ORAL Scan, ORAL Sync and ORAL Standard. Patients (n=9) from an ORAL Standard site that was found to be non-compliant with study procedures were excluded. ‡At month 3, patients receiving placebo and not achieving ≥20% improvement in swollen and tender joint counts after 3 months were advanced in a blinded manner to their predetermined dose of tofacitinib; all remaining patients advanced to tofacitinib at month 6, and subsequent observations were excluded from the analyses. ACR20/50/70, American College of Rheumatology ≥20/50/70% response criteria; ADA, adalimumab; Inf, infinite; N, number of patients assessed at each visit; RCT, randomised controlled trial. Improvements from baseline in DAS28(CRP), DAS28(ESR) and CDAI were generally greater with tofacitinib or adalimumab versus placebo in both sexes, though these trends were generally more pronounced in males versus females (figure 2A and online supplemental figure S2). Similarly, the proportions of patients achieving DAS28(CRP) ≤3.2 and <2.6 (figure 2B,C), DAS28(ESR) LDA and remission (online supplemental figures S3A,B) and CDAI LDA and remission (online supplemental figures S3C,D) were broadly higher with tofacitinib or adalimumab versus placebo in both sexes (ORs >1), with these trends generally more pronounced in males versus females. LS mean differences or ORs (active treatments vs placebo) for (A) mean change from baseline in DAS28(CRP), (B) DAS28(CRP) ≤3.2 and (C) DAS28(CRP) <2.6, stratified by sex (phase III RCTs).† *p<0.05, **p<0.001, ***p<0.0001 for tofacitinib 5 mg or 10 mg two times a day or ADA versus placebo. Mean change from baseline in DAS28(CRP) and rates of achieving DAS28(CRP) ≤3.2 and <2.6 are shown in online supplemental figure S4. LDA and remission values for DAS28(CRP) (≤3.2 and <2.6, respectively) have not been validated54 but are commonly used in rheumatology.24 †Includes all randomised and treated patients from ORAL Scan, ORAL Sync and ORAL Standard. Patients (n=9) from an ORAL Standard site that was found to be non-compliant with study procedures were excluded. ‡At month 3, patients receiving placebo and not achieving ≥20% improvement in swollen and tender joint counts after 3 months were advanced in a blinded manner to their predetermined dose of tofacitinib; all remaining patients advanced to tofacitinib at month 6, and subsequent observations were excluded from the analyses. ADA, adalimumab; DAS28(CRP), Disease Activity Score in 28 joints, C reactive protein; LDA, low disease activity; LS, least squares; N, number of patients assessed at each visit; RCT, randomised controlled trial. For the changes from baseline within each treatment group, greater improvements from baseline in DAS28(CRP), DAS28(ESR) and CDAI were observed in males versus females at most timepoints with tofacitinib 5 mg two times a day; generally similar trends were evident with the other active treatments (online supplemental figure S4). Similar trends favouring males versus females were observed for the proportions of patients with DAS28(CRP) ≤3.2 and <2.6 and CDAI LDA and remission (online supplemental figures S5A–D and I–L). Higher proportions of males versus females across all active treatment groups achieved DAS28(ESR) LDA and remission at most timepoints (online supplemental figure S4E–H). Changes from baseline in clinical assessments (TJC, SJC, Physician Global Assessment of disease activity, CRP and ESR) were generally higher with active treatment versus placebo across timepoints and generally similar between sexes, with some trends noted in favour of males, especially with tofacitinib 5 mg two times a day (online supplemental table S2). Patient-reported outcomes Mean changes from baseline in selected PROs, including PtGA, Pain, HAQ-DI, FACIT-F and SF-36 PCS and MCS scores, were greater with active treatments compared with placebo and generally similar between sexes or slightly larger in males compared with females (online supplemental table S3). For HAQ-DI and FACIT-F, the proportions of patients reporting improvements from baseline ≥MCID or scores ≥normative values were generally greater (ORs >1) with tofacitinib or adalimumab versus placebo (online supplemental figure S6A,B and figure 3A,B). ORs for active treatments versus placebo were generally numerically larger in males than females for FACIT-F outcomes (online supplemental figure S6B and figure 3B). Within each active treatment group, there were generally no differences between the proportions of males versus females reporting HAQ-DI and FACIT-F improvements from baseline ≥MCID (online supplemental figures S7A and S8A,B), while a higher proportion of males versus females reported HAQ-DI and FACIT-F scores ≥normative values with both tofacitinib doses and adalimumab (online supplemental figures S7B and S8C,D). ORs (active treatments vs placebo) for (A) HAQ-DI scores ≥normative values (<0.25) and (B) FACIT-F scores ≥normative values (≥43.5), (C) moderate (≥30%) improvements in Pain and (D) substantial (≥50%) improvements in Pain; stratified by sex (phase III RCTs).† *p<0.05, **p<0.001, ***p<0.0001 for tofacitinib 5 mg or 10 mg two times a day or ADA versus placebo. Response rates by group are shown in online supplemental figures S7 and S8. †Includes all randomised and treated patients from ORAL Scan, ORAL Sync and ORAL Standard. Patients (n=9) from an ORAL Standard site that was found to be non-compliant with study procedures were excluded. ‡At month 3, patients receiving placebo and not achieving ≥20% improvement in swollen and tender joint counts after 3 months were advanced in a blinded manner to their predetermined dose of tofacitinib; all remaining patients advanced to tofacitinib at month 6, and subsequent observations were excluded from the analyses. ADA, adalimumab; FACIT-F, Functional Assessment of Chronic Illness Therapy-Fatigue; HAQ-DI, Health Assessment Questionnaire-Disability Index; MCID, minimum clinically important difference; N, number of patients assessed at each visit; RCT, randomised controlled trial. For improvements from baseline ≥MCID in Pain scores, ORs with active treatments versus placebo were >1 and generally similar in males versus females (online supplemental figure S6C), while trends favouring males were observed at most timepoints for moderate or substantial improvements in Pain scores (figure 3C,D). There were no consistent differences in the proportions of males and females reporting improvements ≥MCID or moderate/substantial improvements in Pain scores (online supplemental figure S8E–J). For SF-36 PCS and MCS, the proportions of patients reporting improvements from baseline ≥MCID or scores ≥normative values were generally greater (ORs >1) with active treatments versus placebo (online supplemental figures S9 and S10). ORs for active treatments versus placebo were numerically greater in males than females for MCS scores but broadly similar across sexes for PCS scores (online supplemental figure S9). Changes in individual SF-36 domain scores were similar between males and females, with a tendency towards higher domain scores in males across active treatments. Composite weighted baselines were higher for males versus females (online supplemental figure S11). Impact of age on response to tofacitinib To explore whether differences in physiological processes associated with age between male and female patients with RA have an impact on responses to tofacitinib, a series of statistical models was fit to the clinical response data across the age spectrum for each treatment group. Following assessment of goodness of fit for the set of polynomial models of mean change from baseline in DAS28(ESR) and DAS28(CRP), a consistent trend in treatment response based on age was not identified in males or females, suggesting that sex-dependent factors across the age spectrum are unlikely to explain differences in response to treatment for these outcomes in this clinical dataset (online supplemental table S4). The preferred models by timepoint and sex are in online supplemental figure S12. Safety Rates of TEAEs were numerically higher in females versus males receiving tofacitinib 5 mg two times a day, and generally similar between sexes in patients receiving tofacitinib 10 mg two times a day and adalimumab (table 1). By month 24, SAEs were reported in 107 (15.1%), 71 (10.2%) and 13 (8%) females receiving tofacitinib 5 mg two times a day, tofacitinib 10 mg two times a day and adalimumab, respectively, and in 17 (12.8%), 24 (17.5%) and six (14.3%) males (table 1). A total of 87 (12.3%) and 10 (7.5%), 88 (12.6%) and 10 (7.3%) and 17 (10.5%) and five (11.9%) females and males receiving tofacitinib 5 mg two times a day, tofacitinib 10 mg two times a day and adalimumab, respectively, discontinued due to AEs (table 1). Ten deaths occurred, six females (0.8%) and four males (3%) with tofacitinib 5 mg two times a day; three deaths, all male (2.2%), occurred with tofacitinib 10 mg two times a day; and three deaths, one female (0.6%) and two males (4.8%), occurred with adalimumab. Table 1Summary of safety up to month 24, stratified by sex and treatment group (phase III RCTs)*Patients with events, n (%)Tofacitinib 5 mg two times a dayTofacitinib 10 mg two times a dayADAFemales (N=707)Males (N=133)Females (N=698)Males (N=137)Females (N=162)Males (N=42)Treatment emergent AEs562 (79.5)85 (63.9)529 (75.8)107 (78.1)119 (73.5)30 (71.4)Serious AEs107 (15.1)17 (12.8)71 (10.2)24 (17.5)13 (8.0)6 (14.3)Discontinuations due to AEs87 (12.3)10 (7.5)88 (12.6)10 (7.3)17 (10.5)5 (11.9)Death6 (0.8)4 (3.0)0 (0)3 (2.2)1 (0.6)2 (4.8)AE of special interest Serious infections28 (4.0)6 (4.5)27 (3.9)6 (4.4)2 (1.2)1 (2.4) All HZ (non-serious and serious)†35 (5.0)7 (5.3)43 (6.2)5 (3.6)2 (1.2)3 (7.1) TB0 (0)0 (0)4 (0.6)2 (1.5)0 (0)0 (0) Opportunistic infection (excluding TB)3 (0.4)0 (0)5 (0.7)0 (0)0 (0)0 (0) Hepatic function test abnormal‡2 (0.3)0 (0)4 (0.6)2 (1.5)0 (0)0 (0) Liver function test abnormal‡3 (0.4)3 (2.3)6 (0.9)0 (0)0 (0)0 (0) Major adverse cardiovascular events§5 (0.7)0 (0)2 (0.3)3 (2.2)0 (0)3 (7.1) Malignancies excluding NMSC7 (1.0)1 (0.8)9 (1.3)1 (0.7)0 (0)1 (2.4) NMSC2 (0.3)5 (3.8)4 (0.6)2 (1.5)1 (0.6)1 (2.4) Lymphoma0 (0)0 (0)2 (0.3)0 (0)0 (0)0 (0) Gastrointestinal perforation0 (0)0 (0)2 (0.3)0 (0)0 (0)0 (0) Venous thromboembolism3 (0.4)0 (0)3 (0.4)1 (0.7)0 (0)0 (0)*Includes all randomised patients from ORAL Scan, ORAL Sync and ORAL Standard.†Among HZ cases in females receiving tofacitinib 5 and 10 mg two times a day, four cases in each treatment group were serious. No serious events of HZ were reported in males receiving tofacitinib 5 or 10 mg two times a day. No serious events of HZ were reported in females and males receiving ADA.‡Abnormal test findings were defined as an AE if the test result was associated with accompanying symptoms and/or required additional diagnostic testing or medical/surgical intervention and/or led to a change in study dosing or discontinuation from the study, significant additional concomitant drug treatment or other therapy and/or was considered to be an AE by the investigator or sponsor.§Defined as CV death and death due to acute MI, sudden cardiac death, heart failure, stroke, CV procedures, CV haemorrhage and other CV causes (eg, peripheral artery disease), but not death due to pulmonary embolism, non-fatal MI; and non-fatal stroke, including reversible focal neurological deficits with image evidence of a new cerebral lesion consistent with ischaemia or haemorrhage.ADA, adalimumab; AE, adverse event; CV, cardiovascular; HZ, herpes zoster; MI, myocardial infarction; N, number of patients randomised; n, number of patients with ≥1 event; NMSC, non-melanoma skin cancer; RCT, randomised controlled trial; TB, tuberculosis. Summary of safety up to month 24, stratified by sex and treatment group (phase III RCTs)* Includes all randomised patients from ORAL Scan, ORAL Sync and ORAL Standard. Among HZ cases in females receiving tofacitinib 5 and 10 mg two times a day, four cases in each treatment group were serious. No serious events of HZ were reported in males receiving tofacitinib 5 or 10 mg two times a day. No serious events of HZ were reported in females and males receiving ADA. Abnormal test findings were defined as an AE if the test result was associated with accompanying symptoms and/or required additional diagnostic testing or medical/surgical intervention and/or led to a change in study dosing or discontinuation from the study, significant additional concomitant drug treatment or other therapy and/or was considered to be an AE by the investigator or sponsor. Defined as CV death and death due to acute MI, sudden cardiac death, heart failure, stroke, CV procedures, CV haemorrhage and other CV causes (eg, peripheral artery disease), but not death due to pulmonary embolism, non-fatal MI; and non-fatal stroke, including reversible focal neurological deficits with image evidence of a new cerebral lesion consistent with ischaemia or haemorrhage. ADA, adalimumab; AE, adverse event; CV, cardiovascular; HZ, herpes zoster; MI, myocardial infarction; N, number of patients randomised; n, number of patients with ≥1 event; NMSC, non-melanoma skin cancer; RCT, randomised controlled trial; TB, tuberculosis. AESIs were similar between sexes with both tofacitinib doses and adalimumab (table 1), although patient numbers were comparatively lower in the adalimumab group. By month 24, serious infections were reported in 28 (4%), 27 (3.9%) and two (1.2%) females receiving tofacitinib 5 mg two times a day, tofacitinib 10 mg two times a day and adalimumab, respectively, and in six (4.5%), six (4.4%) and one (2.4%) males, correspondingly. OIs were reported in three (0.4%) females receiving tofacitinib 5 mg two times a day (pneumonia, HZ disseminated and oesophageal candidiasis) and five (0.7%) females receiving tofacitinib 10 mg two times a day (sialadenitis, cytomegalovirus viraemia, cryptococcal pneumonia, oesophageal candidiasis and HZ disseminated). Among the 35 (5%) HZ cases reported in females receiving tofacitinib 5 mg two times a day and 43 (6.2%) HZ cases reported in females receiving tofacitinib 10 mg two times a day, four cases in each group were serious (one case in the tofacitinib 5 mg two times a day group was disseminated). No serious cases of HZ were reported in males receiving tofacitinib (HZ reported in seven (5.3%) and five (3.6%) patients with 5 and 10 mg two times a day, respectively) or in any patients receiving adalimumab (HZ reported in two (1.2%) and three (7.1%) females and males, respectively). MACE was reported in five (0.7%) patients receiving tofacitinib 5 mg two times a day (all female; acute myocardial infarction (MI), cardiac arrest, cardiac failure, cerebrovascular accident and MI), five patients receiving tofacitinib 10 mg two times a day (two (0.3%) females (infarction and MI) and three (2.2%) males (cardiac failure acute, lacunar infarction and MI)) and in three (7.1%) males receiving adalimumab (acute MI, cardiac arrest and MI). Malignancies excluding NMSC were reported in seven females (1%) and one male (0.8%) receiving tofacitinib 5 mg two times a day, nine females (1.3%) and one male (0.7%) receiving tofacitinib 10 mg two times a day and one male (2.4%) receiving adalimumab (further details provided in online supplemental table S5). NMSC events were reported in two females (0.3%) and five males (3.8%) receiving tofacitinib 5 mg two times a day, four females (0.6%) and two males (1.5%) receiving tofacitinib 10 mg two times a day and one female (0.6%) and one male (2.4%) receiving adalimumab. Two (0.3%) females receiving tofacitinib 10 mg two times a day had lymphoma events. Three (0.4%) females receiving tofacitinib 5 mg two times a day had VTE events (pulmonary embolism (two cases) and venous thrombosis of the limb) and no males had VTE events. In the tofacitinib 10 mg two times a day group, VTE events were reported in three (0.4%) females (deep vein thrombosis, pulmonary embolism and venous thrombosis limb) and one (0.7%) male (pulmonary embolism) (table 1). Persistence Time to all-cause discontinuation and discontinuation due to AEs/lack of efficacy with tofacitinib 5 mg two times a day was similar between sexes (figure 4A, C and E). Time to all-cause discontinuation and discontinuation due to AEs was numerically longer in females versus males in the tofacitinib 10 mg two times a day group (figure 4B,D). Estimated 2- and 5-year tofacitinib continuation rates were similar between sexes (figure 4A–F) but were numerically higher in females versus males in the tofacitinib 10 mg two times a day group (figure 4B,D), which were likely driven by the longer time to discontinuation due to AEs in females noted above. Estimates of treatment persistence (months) in patients receiving average tofacitinib 5 or 10 mg two times a day in pooled LTE studies,* stratified by sex: time to all-cause discontinuation (A and B), time to discontinuation due to AEs (C and D) and time to discontinuation due to lack of efficacy (E and F). Time to discontinuation was defined as the difference between the treatment discontinuation and/or end-of-study date and the first tofacitinib dose date +1 day. Ongoing patients were censored at the cut-off date. Reasons for discontinuation captured in the all-cause discontinuation curves (panels A and B) other than AEs or lack of efficacy could include ‘lost to follow-up’, ‘no longer willing to participate’ or ‘other’. Patients were categorised based on the average tofacitinib dose (average tofacitinib 5 or 10 mg two times a day, based on average total daily dose <15 or ≥15 mg, respectively). *A3921041 (NCT00661661) and ORAL Sequel (NCT00413699); not all patients from the phase III RCTs were enrolled in the LTE studies; thus, the population for the persistence analysis was not the same as the population for the efficacy and safety analyses. AE, adverse event; LTE, long-term extension; N, number of patients included in the Kaplan-Meier persistence analysis; RCT, randomised controlled trial. Discussion In this post hoc analysis evaluating the impact of sex on tofacitinib efficacy and safety in patients with RA, tofacitinib and adalimumab were efficacious in both sexes and had similar safety outcomes, and persistence was similar between sexes for tofacitinib. Overall, greater responses or improvements were observed with active treatments versus placebo, and these treatment differences were generally more pronounced in males versus females. Responses or improvements from baseline within each active treatment group were broadly comparable or tended to favour males, particularly for more stringent composite endpoints (eg, DAS28(ESR)) and selected normative PRO thresholds (eg, physical function and fatigue). These findings may be influenced by the imbalance in sample size between females and males (>4:1) in our study and baseline differences observed between sexes, with females having, on average, longer disease duration, lower CRP levels, slightly higher Pain (VAS) and HAQ-DI scores, lower FACIT-F scores and lower rates of prior TNFi use. However, gender-based differences in sociocultural factors, adherence and reporting biases may also have contributed to divergences in the outcomes reported here.31,33 Across clinical efficacy outcomes, including ACR20/50/70 response rates, DAS28 and CDAI, trends varied between active treatment groups, but differences between sexes in response rates were generally more common with tofacitinib 5 mg two times a day versus tofacitinib 10 mg two times a day and adalimumab. The fact that fewer sex-dependent differences in efficacy were observed for adalimumab likely reflects the limited sample size of this group, although a mechanistic difference between the two treatments cannot be ruled out. For outcomes associated with DAS28(ESR) in particular, we consistently observed differences between sexes across active treatment groups, such that higher proportions of males versus females showed improvement from baseline in DAS28(ESR) and reported DAS28(ESR)-defined LDA and remission. Sex-based differences in disease activity and inflammatory markers (ESR and CRP) are well substantiated in the RA literature1 2 34 and have been previously observed with tofacitinib treatment in patients with psoriatic arthritis. A post hoc analysis of two tofacitinib RCTs in psoriatic arthritis showed generally similar efficacy between sexes with the exception of minimal disease activity, a treatment target that male patients were more likely to achieve.35 In the present study, numerically larger decreases from baseline in ESR were reported in males versus females across treatment groups, aligned with the greater improvements in DAS28(ESR) outcomes in males. To evaluate treatment efficacy in this placebo-controlled RCT cohort, we performed two sets of model-based comparisons to assess rates of response or improvement: males versus females, within each treatment group, and active treatments versus placebo, within each sex. There are conflicting data in the literature regarding potential sex differences in placebo responses. A literature review concluded that males responded more strongly to placebo treatment across 18 studies,36 while a review of meta-analyses reported that 3 out of 75 found a stronger placebo response in females, while others reported no significant sex difference,37 suggesting that effects may differ across outcomes or populations assessed. In our study, numerical trends in some endpoints showed larger responses in females than males receiving placebo, although the sample size of this treatment group was small. These differences in placebo responses appear to have contributed to the fact that sex differences were more evident when active treatments were compared with placebo, though generally similar findings were observed in both sets of model-based comparisons. Efficacy observed in tofacitinib-treated females (ACR20/50/70 response rates of 66%/40%/18% at month 6 for females receiving tofacitinib 5 mg two times a day) generally aligned with previous findings with advanced therapies (eg, adalimumab, etanercept, rituximab and abatacept) in mixed populations.38 39 In contrast, response rates for selected ACR20/50/70 outcomes were comparatively higher in tofacitinib-treated males than females (eg, 79%/53%/25% at month 6 in patients receiving tofacitinib 5 mg two times a day). While these findings may be influenced by the imbalance in sample size between sexes, with males representing only 17.4% of the population, the impact of sex on the markers of disease activity, rather than on disease activity itself, has been mentioned previously. These effects may be more likely to be observed within a tightly controlled RCT,1 2 where any potential gender bias in treatment or clinical management is mitigated. Intersex differences in PROs, including the impact of physical strength on patient-perceived functionality and reporting/self-assessment of fatigue and health-related quality of life, should also be considered when interpreting these findings. The impact of tofacitinib on PROs from these phase III RCTs has been previously reported.40,42 In contrast with real-world findings in treated patients with RA,1 2 17 only slight sex differences were observed in changes from baseline in Pain scores with active treatments. This difference may be reflective of the heterogeneity of real-world populations, compared with the stringently controlled RCT populations. The proportions of patients reporting moderate or substantial improvements in pain did not consistently differ between sexes in the active treatment groups, while the ORs for active treatments versus placebo generally favoured males, which may reflect the numerically greater proportions of females versus males reporting these improvements in the placebo group. Consistent with real-world data,43 females in our study reported slightly greater levels of physical impairment at baseline than males. It follows, and has been widely reported, that males experience greater changes from baseline than females in physical function1 and fatigue.44 When looking at minimal thresholds of improvement (≥MCID), males and females appear to improve to a similar degree. However, in this analysis, patients who reported substantial improvements in physical function and fatigue, for example, scores ≥normative values following tofacitinib treatment, were predominantly male. It is possible that differences in the perception of fatigue may partly account for between-sex variations, but the lack of objective measures precludes further investigation into this. Normative values can be considered a high threshold for measuring response to treatment, and, in contrast to HAQ-DI, it is interesting to note that only numerical differences were observed between sexes in the SF-36 physical functioning domain. This likely relates to variations in the information evaluated by these scales, the SF-36 focusing on physical function (eg, mobility activities such as walking specified distances and carrying groceries) and the HAQ-DI on the ability to perform activities of daily living (eg, dressing and eating). This further suggests that sex differences may be partly embedded in the measure of disease activity or PRO used, rather than actual differences in disease activity per se. Additionally, the presence of comorbid disease, for example, fibromyalgia, which is known to be more prevalent among female than male patients with RA, has been associated with worse scores on the SF-36 and HAQ-DI.45 Collectively, our findings suggest that it may be appropriate to redefine normative thresholds, MCIDs and target cut-points based on sex to allow more meaningful comparisons between sexes in terms of response to RA treatment. The average age of RA onset in women is lower than in men and often coincides with the timing of hormonal shifts associated with the onset of menopause.11 30 46 Given that RA joint destruction and associated physical disability may be negatively impacted by menopause,12 we examined the effects of age, which would capture typified periods of physiological/hormonal shifts during perimenopause, on efficacy outcomes following tofacitinib treatment. Polynomial modelling rejected the hypothesis that the response to tofacitinib (assessed via mean changes from baseline in DAS28(ESR) and DAS28(CRP)) declines more sharply in females versus males as a result of advancing age. These findings suggest that physiological processes or other factors (eg, social factors) associated with age may not play a major role in the trend towards a reduced response observed for older females (aged ≥65 years) versus males receiving tofacitinib as reported previously.15 Further research is needed using direct or quantitative measures to comprehensively explore the role of hormonal shifts in response to tofacitinib. Importantly, the impact of age on tolerability, treatment adherence and ultimately treatment response should be considered, particularly as a previous study showed higher rates of SAEs among females versus males in older adults (aged ≥65 years) receiving tofacitinib.15 Overall, safety and persistence data did not reveal a consistent pattern between males and females for tofacitinib and adalimumab through 24 months. Although females receiving tofacitinib 5 mg two times a day were more likely to report TEAEs and SAEs and discontinue treatment due to AEs, rates of treatment persistence did not show separation between sexes at this dose, and this is in contrast to the lower retention rates in females versus males taking bDMARDs reported in the literature.47 It is important to acknowledge that in the current setting, persistence data for both sexes were derived from LTE studies in patients selected from index studies based on their positive response to tofacitinib treatment and who, therefore, may have been inclined to continue treatment. Although we did not observe any difference between females and males in terms of discontinuation of the index trials, it is a limitation of the analyses that we did not account for a potential survivorship bias. Moreover, by taking part in the LTE studies, patients did not pay for their treatment, which may encourage greater persistence compared with real-world use in some countries. The higher rates of TEAEs in females treated with tofacitinib 5 mg two times a day observed here align with findings from the MetaGeM project, which examines sex-based differences in drug safety across different diseases.48 For AESI, event numbers were too low to draw conclusions; however, no obvious trend in the type of AESI was observed between sexes. Differences between sexes in the absolute risk of certain AESIs have previously been noted for RA therapies, including in the prospective ORAL Surveillance safety trial (NCT02092467), which randomised high CV-risk patients with RA to tofacitinib or TNFi treatment.49 In ORAL Surveillance, male sex was an independent risk factor for MACE and female sex was an independent risk factor for HZ across tofacitinib and TNFi treatment groups,50 51 consistent with trends in the general population.52 53 Limitations of this study include the post hoc nature and that studies were not designed to make comparisons between sexes. Additionally, the adalimumab dataset derives solely from ORAL Standard, which was not powered to make direct comparisons between tofacitinib and adalimumab. Furthermore, as most enrolled patients were female, there was a marked imbalance in sample sizes between the sexes. As a result, following stratification by sex, the sample sizes of male patients for some treatment groups were small, limiting interpretation. However, in line with our findings, RA patient data from the Veterans Affairs Rheumatoid Arthritis (VARA) registry and clinical trials show that sex differences in LDA and remission rates vary substantially across different measures of LDA/remission,3 and in contrast to our analysis, most patients in the VARA registry were male (89.8%). Their findings, therefore, substantiate the trends observed in our analysis and demonstrate that sex differences in response to RA treatment were likely not confounded by the low male sample size. In our analyses of associations between age and treatment response over time in males and females, age (in years) was a surrogate measure of menopausal status in female patients, as such data were not collected in these RCTs. As factors beyond menopause (eg, social or environmental factors) could have an influence on the treatment response with increasing age, the results of these analyses do not allow us to draw conclusions regarding the impact of hormonal influence on response to tofacitinib. A strength of this analysis is that the use of RCT data eliminates potential treatment selection and channelling biases associated with observational datasets. Females and males are likely to be clinically managed in a similar fashion in RCTs, which may explain differences compared with real-world studies. Conclusion Overall, the results of this post hoc analysis demonstrate that tofacitinib is efficacious in males and females, with similar safety outcomes and persistence. Greater responses and magnitudes of improvement for active treatments versus placebo were observed in males than in females for some endpoints, especially more stringent composite measures of disease activity and normative thresholds for selected PROs, including physical function and fatigue. Importantly, these findings may be influenced by the small proportion of male patients (<20% of the study population) and baseline differences in PROs across sexes, with higher HAQ-DI scores and lower FACIT-F scores observed in female versus male patients. These data may assist the clinician in determining sex-specific treatment targets for patients and may support future definition of sex-appropriate MCIDs. Ultimately, further studies are needed to understand the mechanisms underlying sex differences in response to RA treatment, with the purpose of achieving a more personalised and equitable management approach across men and women with RA. Supplementary material Acknowledgements The authors would like to thank the study patients and investigators. Medical writing support, under the direction of the authors, was provided by Kirsten Woollcott and Jennifer Arnold, CMC Connect, a division of IPG Health Medical Communications, and Eleanor Finn, on behalf of CMC Connect, and was funded by Pfizer, New York, New York, USA, in accordance with Good Publication Practice (GPP 2022) guidelines (Ann Intern Med 2022;175:1298–304). Footnotes Funding: This study was sponsored by Pfizer. Authors employed by Pfizer contributed to the development of the manuscript, with final decisions on content agreed by the full author group. Pre-publication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2023-082366). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants. All studies were conducted in accordance with the Declaration of Helsinki and International Council for Harmonisation Guidelines for Good Clinical Practice. The studies included in this pooled post hoc analysis were approved by the institutional review board and/or independent ethics committee for each study centre. Participants gave written informed consent to participate in the study before taking part. This informed consent was applicable to all medical research conducted using data from these studies, including this post hoc analysis. All of the data included in this analysis were covered by the original ethical review/approval process. Consequently, no additional ethical review/approval was required for this post hoc analysis. Data availability free text: Upon request, and subject to review, Pfizer will provide the data that support the findings of this study. Subject to certain criteria, conditions and exceptions, Pfizer may also provide access to the related individual de-identified participant data. See https://www.pfizer.com/science/clinical-trials/trial-data-and-results for more information. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Presented at: Data in this manuscript were previously presented at the ACR 2020 Convergence Meeting (Jones N, Strand V, Schulze-Koops H, et al. Arthritis Rheumatol. 2020; 72 [suppl 10]: Abstract 0825); the EULAR 2021 Annual Meeting (Jones N, Strand V, Schulze-Koops H, et al. Ann Rheum Dis 2021; 80 [Suppl 1]: Poster POS0652) and the PANLAR 2022 Annual Meeting (Jones N, Strand V, Schulze-Koops H, et al. J Clin Rheum 2022; 28 [Suppl 1]: Abstract 1217). Data availability statement Data are available upon reasonable request. References Supplementary Materials Data Availability Statement Data are available upon reasonable request.