Body mass index and achievement of minimal disease activity in psoriatic arthritis across different classes of advanced therapy

Abstract

Objectives

Obesity is a prevalent comorbidity in psoriatic arthritis (PsA). We aimed to evaluate the association between body mass index (BMI) and minimal disease activity (MDA) state in PsA and examine this across different drug classes.

Methods

In a longitudinal observational study using the Gladman Krembil PsA Program cohort, patients with available BMI measurement over follow-up were included. Univariable and multivariable (MV) generalized estimating equations and linear mixed models were used to assess associations between BMI and MDA (and its components) over time, adjusting for age, sex, anxiety/depression, fibromyalgia, smoking, treatment type and radiographic damage. Subgroup analyses evaluated this association across six drug classes: TNF-α inhibitors (TNFi), IL-17 inhibitors (IL-17i), IL-12/23 inhibitors (IL-12/23i), IL-23 inhibitors (IL-23i), Janus kinase inhibitors (JAKi) and phosphodiesterase-4 inhibitors (PDE4i).

Results

In 1291 patients (mean age 44.7 years, 56% male, mean BMI 28.8 kg/m²), higher BMI was independently associated with lower odds for MDA (MV, odds ratio [OR] 0.97; 95% CI: 0.94, 0.99) along with female sex, older age, smoking, fibromyalgia and radiographic damage. BMI was negatively associated with all MDA components except swollen joint count. Higher BMI at drug initiation was associated with reduced odds for MDA state in TNFi-treated patients excluding infliximab (MV, OR 0.94; 95% CI: 0.92, 0.97), while no significant effect was seen for IL-17i, IL-12/23i, IL-23i or JAKi. Longitudinal BMI assessment similarly showed reduced MDA odds with TNFi (excluding infliximab).

Conclusions

High BMI decreases the odds for MDA in PsA, mainly affecting subjective disease measures. This effect is most pronounced in patients treated with TNFi (except infliximab). This underscores the importance of weight management in optimizing treatment response.

Rheumatology key messages.

• Higher BMI was associated with lower odds of being in a state of MDA after accounting for demographic, psychosocial factors and comorbidities.

• BMI adversely affected nearly all MDA components except swollen joint count, suggesting a pronounced influence on patient-reported measures.

• The negative association was most apparent in TNFi-treated patients (excluding infliximab), but class-specific differences require confirmation in larger studies.

Introduction

Obesity is an important comorbidity associated with psoriatic disease (PsD) [1–3]. It is linked to an increased risk of developing psoriasis and transition from cutaneous psoriasis (PsC) to psoriatic arthritis (PsA) [4]. Additionally, obesity may also arise as a consequence of the disease due to reduced physical activity associated with musculoskeletal pain, secondary joint damage, and psychosocial stress associated with skin and joint disease [5–7].

Over the past decade, accumulating evidence has demonstrated that obesity is associated with poorer clinical outcomes in PsA [6–10]. It was associated with reduced odds of achieving a state of sustained minimal disease activity (MDA), driven largely by PsC and patient-reported outcome measures (PROMs) but not swollen joint counts (SJC) [8]. These findings suggest that obesity may disproportionately influence subjective elements of composite disease measures in PsD [8]. Conversely, an interventional study investigating the impact of weight loss in PsA demonstrated significant improvements, including a reduction in SJC [11]. Together, these observations raise important questions regarding whether obesity contributes directly to inflammatory disease activity or primarily affects PROMs through psychosocial pathways such as mood, pain perception and overall well-being [12]. This is particularly relevant given the emergence of effective weight-loss therapies, including glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) receptor agonists [13].

An additional area of uncertainty relates to whether obesity modifies treatment responses across different therapeutic classes. Findings have varied across studies regarding the effectiveness of TNF-α inhibitors (TNFi) in obese patients. While some studies indicated that obesity was associated with poorer response rates at follow-up [14–16], others found no significant effect of body mass index (BMI) on response to TNFi [17]. This could be due to different outcome measures used, along with assessment of response at a single time point at follow-up. Poor response to TNFi in obese patients could be due to lack of weight-based dosing, altered pharmacokinetics, TNF sink phenomenon [18], poor adherence [16] and prominent adipokine-mediated inflammation in obese individuals [19, 20]. Similar discrepancies have been observed with IL-17 inhibitors (IL-17i), with some studies showing no impact of BMI on response [21, 22] and others suggesting an improved drug retention in obese PsA patients treated with IL-17i compared with TNFi [23]. A post hoc analysis of phase 3 studies involving Janus kinase inhibitors (JAKi) such as tofacitinib, found no association between BMI and response [22]. However, this was underpowered to detect a meaningful effect of BMI on outcomes.

Despite these insights, there remains a gap in understanding the impact of obesity on disease activity in PsA. It is unclear whether the impact of obesity on MDA is driven by persistent inflammatory burden or comorbid factors such as anxiety, depression, fibromyalgia and osteoarthritis, which may influence patient-reported outcomes. In addition, most earlier studies were conducted in the TNFi era, and it remains uncertain whether these findings persist across newer biologic (b)DMARD and targeted synthetic (ts)DMARD classes with differing mechanisms of action [8]. Evidence regarding differential treatment responses across bDMARD and tsDMARD classes in patients with obesity is inconsistent [16, 17, 24]. We previously reported that TNFi treatment was associated with weight gain, whereas apremilast was linked to weight loss [7]. The implications of these weight changes for treatment response also warrant further investigation. Accordingly, the relationship between BMI and MDA would be best addressed longitudinally, accounting for relevant confounders, and across different therapeutic classes.

To address these gaps, we aimed to investigate the association between BMI and disease activity in PsA and to evaluate whether this relationship varies across different PsA treatment classes in a real-world cohort.

Methods

Setting

We analysed data from an ongoing observational cohort maintained by the Gladman Krembil Psoriatic Arthritis Program at the Schroeder Arthritis Institute, University Health Network, Toronto, Canada [25]. Clinical, radiographic and laboratory data have been prospectively collected since 1978 at 6- to 12-month intervals using a standardized protocol. Each visit includes a comprehensive history, physical examination, disease activity assessment, patient-reported outcomes and laboratory investigations. Radiographs of the hands, feet, pelvis, cervical spine and lumbar spine are obtained every 2 years. More than 99% of patients in the cohort meet the Classification Criteria for Psoriatic Arthritis (CASPAR) [26].

Patients

All patients enrolled in the cohort up to 15 October 2025 with at least one BMI assessment at any visit were eligible for inclusion. To examine the association between BMI and MDA across drug classes, we included patients who initiated a bDMARD or tsDMARD, remained on treatment for at least 6 months, and had a BMI recorded at treatment initiation. This was intended to ensure that patients had an adequate opportunity to experience therapeutic benefit. Patients who were managed exclusively with NSAIDs, those with pure axial disease, and those already receiving bDMARDs or tsDMARDs at the time of cohort entry were excluded. Pure axial disease was defined using radiographic evidence of axial involvement in the absence of peripheral arthritis. This approach was chosen because validated classification criteria for axial PsA were not available for much of the study period, MRI was not systematically performed across the cohort, and symptom-based instruments such as Bath Ankylosing Spondylitis Disease Activity Index have limited specificity for axial disease assessment in PsA [27].

Assessments

Demographic characteristics (age, sex, ethnicity and disease duration), BMI, comorbidities (hypertension, diabetes mellitus, hyperlipidaemia), smoking status (current smoker, ex-smoker, never smoker), clinical features (SJC [0–66], tender joint counts [TJC, 0–68], total entheseal count [TEC, 0–16] [28], dactylitis and nail involvement), disease activity (Psoriasis Area and Severity Index [PASI, 0–72], clinical Disease Activity index for Psoriatic Arthritis [cDAPSA, 0–154] [29], MDA [30]), patient-reported outcomes (patient pain [PP, 0–10], Patient Global Assessment [PGA, 0–10], Health Assessment Questionnaire [HAQ, 0–3], Short Form [SF]-36 [31, 32]), laboratory values (ESR, normal range 0–15; CRP, normal 0–5 mg/l]), radiographic features (number of joints with erosions, sacroiliitis, syndesmophytes, modified Steinbrocker score [mSS, 0–168] [33]) and treatment (NSAIDs, conventional synthetic DMARDs [csDMARDs], bDMARDs and tsDMARDs) were retrieved from the database. We also recorded information on anxiety and depression, defined as a score of ≤38 on the SF-36 mental component summary (MCS) and/or a score of ≤56 on the SF-36 mental health domain and/or physician diagnosis as defined previously [34, 35]; osteoarthritis, and fibromyalgia as per physician diagnosis.

Statistical analysis

Baseline characteristics

Patient characteristics at cohort entry were summarized using descriptive statistics. Continuous variables were expressed as mean (S.D.) or median (range), depending on distribution. Categorical variables were summarized as frequencies and percentages.

Association between BMI and MDA

We studied the association of BMI with MDA and its individual components (SJC [0–66], TJC [0–68], HAQ [0–3], PASI [0–72], TEC [0–16], PP [0–10] and PGA [0–10] as continuous variables) using univariable generalized estimating equations (GEE) for MDA and linear mixed models (LMM) for the components. A multivariable model was then fitted, adjusting for potential confounders identified through a directed acyclic graph (Figure S1) [36]. Confounders included time-independent variables (age at enrolment and sex) and time-dependent variables (anxiety/depression, smoking status, fibromyalgia, apremilast use, TNFi use and mSS). TNFi and apremilast use were included given their known effects on body weight [7].

Association between BMI and MDA across drug classes

We evaluated the association between BMI at drug initiation and MDA at follow-up visits across six drug classes: TNFi, IL-17i, IL-23 inhibitors (IL-23i), IL-12/23 inhibitors (IL-12/23i), JAKi and phosphodiesterase-4 inhibitors (PDE4i). We also studied the longitudinal association between BMI (time-varying) and MDA, both measured concurrently at follow-up visits (to account for drug-related and disease activity-related BMI changes) across the six drug classes. Analyses were performed using univariable and multivariable GEE.

The multivariable models were adjusted for time-independent variables (age at drug initiation, sex, line of treatment for bDMARDs and JAKi) and time-dependent variables (anxiety and depression, osteoarthritis, fibromyalgia, smoking and radiographic damage assessed by the mSS). A subanalysis for infliximab (weight-based dosing) was also done. Drug courses were used as the unit of analysis, since patients could receive multiple bDMARDs/JAKi. Line of treatment was for b/tsDMARDs, irrespective of mechanism of action (first-line, second-line, third-line or beyond).

Missing data were addressed by multiple imputation with chained equations (MICE) in R, and twenty ‘complete’ datasets were generated [37]. Variables requiring imputation included fibromyalgia (missing, 5%), osteoarthritis (6%), smoking status (17%), mSS (3%) and SF-36 (35%). Predictive mean matching was used for continuous variables, while baseline smoking status was imputed using logistic regression. The imputed datasets were then merged with biologic treatment and response data for analysis.

Statistical analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC, USA) and R version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria). A P-value of <0.05 was deemed statistically significant, and results were expressed as beta (β) estimate with 95% CI.

Consent and ethics

We obtained ethics approval from the University Health Network Research Ethics Board (08–0630). All patients provided informed consent at enrolment.

Results

Baseline characteristics

A total of 1291 patients were included in this study. The mean age at enrolment was 44.7 years (S.D. 13), with a male predominance (723 patients, 56%). The mean BMI was 28.8 kg/m² (S.D. 6.36). The median SJC was 2 (range, 0–42), and the median TJC was 4 (range, 0–64). The mean PASI score was 4.04 (S.D. 7), and the mean DAPSA score was 25.2 (S.D. 22.6). At baseline, 797 patients (61.7%) were on NSAIDs and 480 (37.2%) were receiving csDMARDs, while a minority were on targeted advanced therapies (Table 1).

Table 1.

Baseline data for patients with available BMI at enrolment, n = 1291.

Characteristic	Value
Demographics
Age at enrolment, mean (s.d.), years	44.7 (13.0)
Age at onset of PsA, mean (s.d.), years	39.0 (13.8)
Age at onset of PsC, mean (s.d.), years	28.9 (14.8)
Follow-up duration, median (range), years	11.01 (0–46.31)
Sex, male, n (%)	723 (56.0)
BMI, mean (s.d.), kg/m2	28.8 (6.36)
Clinical features
SJC (0–66), median (range)	2 (0–42)
TJC (0–68), median (range)	4 (0–64)
PASI (0–72), mean (s.d.)	4.04 (7.00)
Dactylitis, n (%)	318 (24.6)
Enthesitis, n (%)	293 (22.7)
Disease activity
DAPSA (0–154), mean (s.d.)	25.2 (22.6)
MDA, n (%)	264 (20.4)
Patient reported outcomes
Patient pain (0–10), median (range)	4 (0–10)
Patient global (0–10), median (range)	4 (0–10)
HAQ (0–3), mean (s.d.)	0.69 (0.65)
Laboratory details, n (%)
Elevated ESR	21.4 (20.1)
Elevated CRP	12.7 (18.8)
Treatment, n (%)
NSAIDs	797 (61.7%)
csDMARDs	480 (37.2%)
bDAMRDs	7 (0.5%)
tsDMARDs	26 (2.0%)

Abbreviations: bDMARD, biologic DMARD; cDAPSA, clinical Disease Activity in Psoriatic Arthritis; csDMARD, conventional synthetic DMARD; HAQ: Health Assessment Questionnaire; MDA, minimal disease activity; PASI, Psoriasis Area and Severity Index; PsA, psoriatic arthritis; PsC, psoriasis cutaneous; SJC, swollen joint count; TJC, tender joint count; tsDMARD, targeted synthetic DMARD.

Association of BMI with MDA and its components

In the unadjusted analysis, higher BMI was significantly associated with lower odds of MDA state (odds ratio [OR] 0.95; 95% CI: 0.94, 0.97). This association was also sustained in multivariable analysis (OR 0.97; 95% CI: 0.94, 0.99). Additional negative associations were observed with fibromyalgia, smoking and higher mSS, while male sex and TNFi use were positively associated with MDA achievement. Anxiety/depression, osteoarthritis and apremilast use were not significantly associated with MDA (Fig. 1). Higher BMI was linked to worse outcomes across several MDA subcomponents, including TJC, TEC, PASI, PP, PGA and HAQ, but not SJC, in both univariable and multivariable models (Fig. 2).

Figure 1.

Association of BMI with MDA state in unadjusted and in multivariable analyses using generalized estimating equations. Associations with other factors are shown for multivariable analysis. Abbreviations: MDA: minimal disease activity; mSS, modified Steinbrocker score; TNFi, TNF-α inhibitor

Figure 2.

Association of BMI with components of MDA as continuous variables using univariable and multivariable linear mixed models over follow-up with results reported as β coefficients and 95% CI. Abbreviations: HAQ, Health Assessment Questionnaire; MDA: minimal disease activity; PASI: Psoriasis Area and Severity Index; PGA, Patient Global Assessment; PP: patient pain; SJC, swollen joint count; TES, total entheseal score; TJC, tender joint count

Association between BMI and MDA across drug classes

An analysis of association between BMI and MDA across drug classes was based on 1102 treatment courses from 582 patients. Characteristics at drug initiation are presented in Table S1. Of these, 433 received TNFi (80 infliximab), 166 IL-17i, 71 IL-23i, 64 IL-12/23i, 32 JAKi and 57 PDE4i. The median duration of each course was 2.57 (range, 1–6.27) years.

In the univariable analysis, higher BMI at drug initiation was associated with lower odds of MDA state with TNFi (OR 0.93; 95% CI: 0.91, 0.96). This remained significant when analysed for TNFi when infliximab was excluded (OR 0.93; 95% CI: 0.90, 0.96). No significant associations between BMI at drug initiation and MDA were observed for infliximab, IL-17i, IL-12/23i, IL-23i, JAKi or PDE4i (Fig. 3).

Figure 3.

Association of BMI at drug initiation with MDA stratified by drug class using generalized estimating equations. Abbreviations: IFX: infliximab; IL-12/23i: IL-12/23 inhibitor; IL-17i: IL-17 inhibitor; IL-23i: IL-23 inhibitor; JAKi, Janus kinase inhibitor; MDA: minimal disease activity; OR, odds ratio; PDE4i, phosphodiesterase 4 inhibitor; TNFi, TNF-α inhibitor

In multivariable analysis, higher BMI at drug initiation was similarly associated with reduced odds of MDA state within the TNFi group overall (OR 0.95; 95% CI: 0.93, 0.98) and TNFi without infliximab (OR 0.94; 95% CI: 0.92, 0.97). In contrast, higher BMI was associated with better odds for MDA with apremilast (OR 1.08; 95% CI: 1.01, 1.16). The effect was not statistically significant for infliximab and other drug classes (Fig. 3).

Results were consistent when BMI was modelled as a time-varying variable to look for associations of BMI at follow-up visits with corresponding MDA statuses. Higher BMI was associated with reduced odds of MDA state within the TNFi class (OR 0.93; 95% CI: 0.90, 0.96) and TNFi without infliximab (OR 0.93; 95% CI: 0.90, 0.95) in the univariable analysis. In multivariable analysis, higher BMI was similarly associated with reduced odds of MDA state within the TNFi class (OR 0.95; 95% CI: 0.92, 0.97) and TNFi without infliximab (OR 0.94; 95% CI: 0.91, 0.96). The effect was not statistically significant for infliximab and other drug classes (Fig. 4).

Figure 4.

Longitudinal association of BMI with MDA stratified by drug class using generalized estimating equations. Abbreviations: IFX: infliximab; IL-12/23i: IL-12/23 inhibitor; IL-17i: IL-17 inhibitor; IL-23i: IL-23 inhibitor; JAKi, Janus kinase inhibitor; MDA: minimal disease activity; OR, odds ratio; PDE4i, phosphodiesterase 4 inhibitor; TNFi, TNF-α inhibitor

Discussion

In this study, we found that higher BMI was associated with significantly lower odds of MDA state. This association remained robust after adjusting for demographic factors, disease activity, comorbidities and radiographic damage. Higher BMI was also linked to worse outcomes across several MDA subcomponents, except SJC, highlighting a disproportionate effect on subjective outcome domains. The effect on reduced odds for MDA was most pronounced among patients treated with TNFi (excluding infliximab), with no significant associations for other drug classes.

When modelled cross-sectionally across all follow-up visits, BMI remained consistently associated with lower odds of achieving MDA; each 1-unit increase in BMI was associated with a 3–6% reduction in the likelihood of being in MDA. Multivariable adjustment adjusting for potential confounders yielded similar findings. In addition, we found that female sex, older age, peripheral damage, smoking and fibromyalgia were associated with reduced odds for MDA. Anxiety/depression was included in the multivariable models as a potential confounder. Therefore, the absence of a statistically significant association in adjusted analyses should not be interpreted as evidence of no clinical relevance, but rather within the context of the specified adjustment framework. When we studied associations between BMI and components of MDA, we found that BMI was associated with all components except for SJC. Thus, while MDA is an established treatment target in PsA, it is a multidomain composite outcome and should not be interpreted as a direct surrogate of inflammatory activity alone. In the present study, the association between higher BMI and lower likelihood of MDA was primarily driven by patient-reported subjective domains, whereas swollen joint count was not significantly associated with BMI. These findings suggest that obesity may influence overall disease burden through both inflammatory and non-inflammatory pathways.

These findings are consistent with earlier work from our centre, where individuals with obesity had approximately half the odds of achieving sustained MDA for 12 months compared with those of normal BMI [8]. This was driven by PsC and subjective outcome measures such as TJC, PP, PGA and HAQ, whereas SJC and TEC were unaffected [8]. Multivariable modelling similarly identified female sex, older age, longer disease duration, NSAID use and DMARD use as negative predictors of MDA. Comparable results have been reported in the ReFlaP multicentre cohort, where obesity reduced the likelihood of remission or low disease activity by 2.5- to 3-fold, again with minimal effect on SJC [6]. Observational data from European cohorts have also shown adverse effects of obesity on achieving MDA at 1 year [9, 38]. Post hoc analysis of the Study of Etanercept and Methotrexate in Combination or as Monotherapy in Subjects with Psoriatic Arthritis (SEAM-PsA) trial also alluded to poor responses in terms of MDA in patients with a BMI > 30 irrespective of the treatment arm [39].

To investigate potential differential effects across drug classes, we evaluated BMI at drug initiation and modelled BMI as a time-varying covariate. In both approaches, BMI was associated with reduced odds of achieving MDA among patients receiving TNFi, including those treated with non-weight-based TNFi agents. No significant effect was observed for infliximab, consistent with the hypothesis that weight-based dosing may mitigate BMI-related pharmacokinetic attenuation. Similar patterns were not observed for IL-17i, IL-23i, IL-12/23i, JAKi or PDE4i. These data suggest that BMI may influence the effectiveness of TNFi more strongly than other drug classes, though prospective confirmation is needed. The smaller sample sizes in newer therapeutic classes limited power, and the absence of statistically significant associations in these groups should not be interpreted as evidence of no effect.

The stronger negative association observed with fixed-dose TNFi compared with infliximab may reflect pharmacokinetic differences related to body weight. Higher BMI can alter drug distribution and clearance, potentially leading to lower effective exposure with fixed-dose regimens. In contrast, weight-based dosing strategies such as infliximab may partially mitigate this effect. Similar principles have been demonstrated in other immune-mediated diseases such as inflammatory bowel disease, where effect of body weight was observed for non-infliximab TNFis and therapeutic drug monitoring improves treatment outcomes [40, 41]. Furthermore, most TNFi treatment courses in our cohort occurred as first-line advanced therapy, reflecting historical treatment availability. Consequently, the number of patients receiving TNFi after prior exposure to newer biologic or targeted therapies was limited, restricting the power of treatment-line-specific subgroup analyses.

An observational study from the Danish and Icelandic biologics registry showed that obesity increased the risk for TNFi withdrawal and reduced the odds for a good or moderate EULAR response at follow-up. They found similar effects with infliximab as well. However, these findings were based on cross-sectional assessments at 6 months and lacked adjustment for key confounders [16]. Other studies have reported no association between obesity and TNFi response, although the use of DAS28, an inadequate PsA activity measure, may explain the discrepancy [17]. Data regarding IL-17i remain inconsistent. Some studies report poorer retention of secukinumab in patients with obesity [42]. On the other hand, findings from our cohort showed better drug survival of IL-17i in obese individuals [23]. Others have also reported an inverse correlation between DAPSA at 6 months and BMI. However, this was a cross-sectional correlation, without adjustment for potential confounders [21].

The present findings have potential implications in the evolving era of weight-modifying therapies. If higher BMI is associated with lower odds of achieving MDA, then interventions targeting BMI may represent an important adjunctive strategy in PsA management. Prior interventional work has shown that structured weight loss, through dietary interventions, can improve disease activity in patients with PsA and obesity [11, 43]. More recently, incretin-based therapies such as GLP-1 receptor agonists and dual GIP/GLP-1 agonists have demonstrated substantial and sustained weight reduction in obesity populations, with emerging evidence of benefits on pain and physical function in related musculoskeletal conditions such as knee osteoarthritis [13, 44]. In parallel, a recent phase 3 b clinical trial of combined ixekizumab and tirzepatide (TOGETHER-PsA study) suggested that GLP-1 receptor agonist use was associated with a higher combined end point of American College of Rheumatology response criteria (ACR50) and ≥10% weight reduction at 36 weeks. Improvement was prominent in the combination group across all components of ACR50 except PP. Additionally, retrospective analyses of two independent cohorts suggested improvement in disease activity measures (particularly PP) proportional to weight reduction [45]. This supports our current findings of significant impact of BMI on PROMs. Taken together, future studies are needed to employ comparator arms with weight loss interventions, pain phenotyping tools, early time point assessments and mediation analysis to disentangle the contributions of these medications towards weight loss, inflammation and central pain modulation.

The strengths of this study include the large, well-characterized real-world PsA cohort, the use of both baseline and time-varying BMI to account for drug-related changes, and the adjustment for multiple relevant confounders, including fibromyalgia, smoking and radiographic damage. Missing data were addressed using robust imputation methods. However, some limitations warrant consideration. As with all observational studies, causal inference is limited, and residual confounding (e.g. lifestyle factors, diet, socioeconomic status), reverse causality and selection bias (requirement of one BMI reading for inclusion) cannot be excluded. We did not have comprehensive medication-level data for non-rheumatic therapies (for anxiety, depression, diabetes mellitus, etc.) limiting detailed assessment of these confounders. Similarly, data regarding weight-modifying interventions such as metformin, GLP-1/GIP receptor agonists, structured dietary programs or bariatric procedures were not systematically captured throughout the study period. Although the uptake of GLP-1 receptor agonists was likely limited during much of the cohort follow-up, their potential impact on BMI and treatment response cannot be excluded. Missingness in SF-36 data may have resulted in some misclassification of anxiety/depression status but the use of a composite definition including physician diagnosis would have minimized it. Although fibromyalgia was included as a covariate, physician diagnosis likely underestimates the broader burden of nociplastic pain and central sensitization. These mechanisms may influence pain, fatigue, function and patient global scores, and could therefore contribute to the observed association between BMI and reduced MDA attainment. Future studies incorporating validated instruments such as the Widespread Pain Index/Symptom Severity Scale may better characterize these pathways. A further limitation relates to the definition of axial disease in this cohort. Axial involvement was ascertained using radiographic evidence, which may have limited sensitivity and specificity. In addition, MRI was not systematically available across the full study period, precluding a more sensitive assessment of active inflammatory lesions and early structural abnormalities. Therefore, some misclassification of axial involvement cannot be excluded, although pure axial disease represented only a small proportion of patients in our cohort. The relatively small numbers of patients treated with newer drug classes restricted the power to detect BMI effects in those groups. And finally, BMI is an imperfect surrogate of adiposity and does not distinguish between lean mass and visceral fat. In PsA, central adiposity and associated metabolic dysfunction may be more relevant drivers of inflammation and treatment response than BMI alone [46, 47]. Future studies incorporating waist circumference, imaging-based adiposity measures or body composition analyses may provide greater mechanistic insight. Phenotyping obesity may help distinguish inflammatory vs metabolic subtypes with different PsA trajectories and treatment responses.

In conclusion, in this large PsA cohort, higher BMI was independently associated with lower odds of MDA state and worse outcomes across multiple disease domains. This was particularly evident in patients treated with TNFi. These findings suggest that obesity may be an important, modifiable factor in PsA management and underscore the need to integrate weight optimization strategies alongside pharmacological therapy. Further research is needed to investigate the possible role of weight reduction on long-term outcomes with newer therapeutic drug classes.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Author contributions

D.P. and P.M. contributed to the conceptualization. P.M., D.P., M.Y. and R.J.C. contributed to the study design. All authors contributed to data acquisition, analysis and interpretation. All authors contributed to manuscript preparation or revision and approved the final version prior to publication. All authors agreed to be accountable for all aspects of the work.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: P.M., V.C.A., F.K., R.J.C., M.Y. and S.G. have no conflicts of interest. D.G. has received grants and/or consulting fees from Abbvie, Amgen, AstraZeneca, BMS, Eli Lilly, Fresensius Kabi, Johnson and Johnson, Novartis, Oruka, Pfizer, and UCB. V.C. has received research grants from AbbVie, Amgen and Eli Lilly and has received honoraria for advisory board member roles from AbbVie, BMS, Eli Lilly, Fresenius Kabi, Janssen, Novartis and UCB. His spouse is an employee of AstraZeneca. D.P. has received research support from AbbVie, Bristol Myers Squibb, Eli Lilly, Johnson and Johnson, Novartis, Pfizer, and UCB, consulting fees from AbbVie, Eli Lilly, GlaxoSmithKline, Greywolf Therapeutics, Johnson and Johnson, Merk, Moonlake, Novartis, Pfizer, and UCB, and speaker fees from AbbVie, Canon, Celltrion, Eli Lilly, Globemed, Johnson and Johnson, Medscape, Novartis, Peervoice, Pfizer, and UCB. He serves as a member of executive committee of ASAS, member of steering committee of GRAPPA.