Association between musculoskeletal sonographic features and response to treatment in patients with psoriatic arthritis

Abstract

Objective

To investigate the association between musculoskeletal sonographic features and clinical features, as well as treatment outcomes, in patients with active psoriatic arthritis (PsA).

Methods

A prospective cohort study was conducted involving patients with active PsA. Disease activity was assessed clinically at baseline and 3–6 months after initiating therapy, with a Disease Activity Index for PsA (DAPSA) score calculated. A baseline ultrasound examination of 64 joints, 28 tendons and 16 entheses evaluated the following lesions: synovitis, peritenonitis, enthesitis, tenosynovitis, new bone formation and erosions. Total scores for each lesion and total inflammatory and structural scores were calculated. The association between baseline sonographic scores and treatment outcomes was assessed using Cox proportional hazards models (for drug persistence) and generalised estimating equation models for DAPSA change.

Results

A total of 135 treatment periods (107 patients) were analysed. Multivariable analysis showed that a greater reduction in DAPSA score at follow-up was associated with higher baseline synovitis (β −3.89), peritenonitis (β −3.93) and enthesitis structural scores (β −2.91). Additionally, the total inflammatory score independently predicted DAPSA change (β −5.23) regardless of the total structural damage score. Drug persistence was analysed in 105 treatment periods, revealing that a higher sonographic erosion score was associated with earlier drug discontinuation (adjusted HR 1.28, 95% CI 1.03 to 1.61).

Conclusion

The study results provide preliminary evidence supporting the utility of musculoskeletal ultrasound in predicting treatment response and drug persistence in PsA.

What is already known on this topic?

• While ultrasound is recognized for its accuracy in assessing inflammation and structural damage in the various domains of psoriatic arthritis (PsA), limited information exists on the predictive value of musculoskeletal sonographic features across different PsA domains.

What this study adds?

• Our study showed that the presence of sonographic articular and periarticular inflammation independently predicted improvement in Disease Activity Index for PsA score following treatment, regardless of the presence of joint damage, whereas the sonographic erosive damage was associated with earlier discontinuation of drug therapy.

How this study might affect research, practice or policy?

• Given the discordance observed between clinical and sonographic findings, ultrasound can serve as a valuable tool for risk stratification of patients with peripheral PsA, aiding in predicting their likelihood to respond and persist on prescribed medications.

Introduction

Psoriatic arthritis (PsA), a systemic inflammatory musculoskeletal and skin disease, manifests in diverse clinical forms. These include peripheral arthritis, axial disease, dactylitis, enthesitis, as well as skin and nail involvement. An accurate evaluation of inflammation across these domains is important, not only for enhancing our understanding of the immunopathogenesis of the disease but also for guiding treatment decisions and monitoring treatment response. Yet, the inherent limitation of physical examination hinders accurate detection, quantification and localisation of inflammation within the various musculoskeletal structures relevant to PsA.¹ In addition, the high prevalence of concomitant comorbidities such as obesity, osteoarthritis (OA) and central sensitisation (fibromyalgia) in this patient population further exacerbates the challenges associated with relying solely on physical examination for diagnosis and assessment.²

Ultrasound is a precise and sensitive imaging tool for evaluating musculoskeletal inflammation across various PsA domains including synovitis, tenosynovitis, peritenonitis (inflammation in tendons without tendon sheaths) and enthesitis.3,6 Additionally, ultrasound can detect structural lesions, such as bone erosions or new bone formation (NBF), frequently resulting from uncontrolled inflammation. The detection of these lesions not only aids in confirming the diagnosis of PsA but also provides valuable insights into disease severity.

The affordability and accessibility of ultrasound make it a cost-effective tool in clinical settings, with the added advantage of being radiation-free, allowing for repeated testing to monitor disease activity over time. Despite its clear advantages compared with physical examination, limited information exists on the relationship between sonographic musculoskeletal inflammation and clinical outcomes in PsA. Some studies have highlighted associations between sonographic enthesitis and poor clinical outcomes, including radiographic axial and peripheral damage,^{7 8} as well as damage progression.⁹ However, other studies have demonstrated weak or no association between sonographic and clinical enthesitis and patient-reported outcomes, such as pain score and physical function.¹⁰ Moreover, this divergence between clinical and sonographic PsA disease activity has also been observed in a subset of patients in minimal disease activity states, who were found to have subclinical sonographic musculoskeletal inflammation.^{1 11}

Of paramount importance is the limited information regarding the predictive utility of sonographic characterisation of various PsA features. Most ultrasound studies published so far have been cross-sectional, lacking longitudinal follow-up data. Consequently, the ability to interpret the importance of discrepancies between clinical and sonographic findings remains somewhat limited. A recent randomised trial showed that ultrasound synovitis scores could distinguish secukinumab from placebo as early as 1 week after treatment initiation, suggesting that early sonographic response might serve as a predictor of clinical treatment response.¹² Moreover, the presence of sonographic synovitis predicted the need for future advanced therapy in patients with early peripheral PsA.¹³ Nevertheless, substantial knowledge gaps persist regarding the overall predictive ability of ultrasound in PsA, particularly concerning the various sonographic features relevant to this condition.

The overall objective of this study is to assess the association between musculoskeletal sonographic features and clinical outcomes in PsA. Specifically, we aimed to assess whether the severity and location of sonographic inflammatory and structural articular and peri-articular lesions are linked to clinical manifestations of PsA. Additionally, we aimed to investigate whether these sonographic findings could predict future response to systemic disease-modifying anti-rheumatic drugs (DMARDs) in patients with active PsA.

Methods

Patients and settings

We conducted a prospective ultrasound study, nested within the University of Toronto PsA cohort. This cohort comprises individuals with a confirmed diagnosis of PsA by a rheumatologist. Participants are assessed by rheumatologists at 3- to 6‐month intervals, during which detailed information on demographics, musculoskeletal symptoms, medications, disease activity and laboratory tests is systematically collected according to a standard protocol.¹⁴ All gathered data are stored in a web‐based computerised database.

The present study included consecutive patients with active PsA enrolled in the University of Toronto PsA cohort between December 2017 and March 2023, who had been advised by their rheumatologist to change their systemic therapy due to active peripheral manifestations of the disease. These patients were invited to undergo a comprehensive musculoskeletal ultrasound assessment before and after initiating any new treatment. A patient could have contributed more than one treatment period to this study, each requiring a separate ultrasound assessment before the commencement of treatment.

The inclusion criteria comprised (1) a confirmed diagnosis of PsA that satisfied CASPAR criteria¹⁵ ; (2) active peripheral manifestation of PsA including peripheral arthritis, enthesitis or dactylitis and (3) about to initiate or switch to a new DMARD therapy. The decision to initiate therapy was made at the discretion of the rheumatologist, without any predefined clinical or sonographic criteria. The study was approved by the Research Ethics Board of Women’s College Hospital (REB # 2020–0071-E). All patients have provided written informed consent at the time of enrolment.

Clinical data

Clinical data were systematically collected within a 2-week timeframe from the ultrasound assessment, both before and after treatment initiation, as part of the University of Toronto PsA cohort. A rheumatologist, independent of the ultrasound evaluation, conducted the clinical assessment.

The collected clinical information included age, sex, duration of PsA, current and past use of DMARDs and physician global assessment of disease activity. Physical examination was performed by a rheumatologist and included the following items: the number of tender and swollen joints in 68 and 66 joints, respectively; the number of tender entheses (by Spondyloarthritis Research Consortium of Canada (SPARCC)); the number of tender dactylitic digits and psoriasis severity (by Psoriasis Area and Severity Index). Laboratory tests included high-sensitivity C reactive protein (CRP).

Patient-completed questionnaires to assess their level of pain (NRS 0–10); Patient global assessment of PsA disease activity (NRS 0–10) and physical function (by Health Assessment Questionnaire Disability Index (HAQ-DI)). Finally, Disease Activity Index for PsA (DAPSA) was calculated as a composite measure of disease activity.

Treatment outcomes

We analysed treatment outcomes among patients who initiated therapy with a conventional synthetic DMARD (csDMARD), biologic DMARD (bDMARD) or targeted synthetic DMARD (tsDMARD), provided they had attended a follow-up visit subsequent to treatment initiation. The choice of drug was determined by the treating rheumatologist in accordance with standard practice. Our primary outcome of interest included a change in DAPSA and attainment of DAPSA low disease activity (LDA) at 3–6 months. Additionally, we evaluated drug persistence, defined as the duration from initiation to discontinuation of the prescribed medication.

Ultrasound assessment

An experienced ultrasonographer (LE) conducted all scans using a MyLab Twice (Esaote) scanner, which was equipped with a 6–18 MHz linear array transducer (Esaote). To optimise vascularisation visualisation, a Power Doppler (PD) frequency ranging from 8.3 MHz to 10 MHz (adjusted according to body habitus), a pulse repetition frequency of 750 Hz and a wall filter of 2 were used. Each ultrasound examination was recorded and stored as a short DICOM video file. Subsequently, a single reader (LE), blinded to clinical information, reviewed and scored each scan.

We systematically scanned 64 joints, 28 tendons and 16 entheses following a standard protocol (see online supplemental table 1). Each anatomical site was scanned in a longitudinal plane, and if any abnormality was detected, we performed a transverse plane assessment. Bilateral scans were performed on the following joints: wrists, metacarpo-phalangeal (MCP) 1–5, proximal interphalangeal (PIP) 1–5, distal interphalangeal (DIP) 2–5, elbows, knees, tibiotalar, metatarsophalangeal (MTP) 1–5, feet PIPs 1–5 and DIPs 2–5. Additionally, bilateral scans of tendons included the second, fourth and sixth extensor wrist compartments, finger flexor tendons 1–5 and tibialis posterior tendons. Entheses were also bilaterally scanned at the following insertion points: supraspinatus to the humerus, common extensor tendon to the lateral epicondyle, triceps to the olecranon process, quadriceps tendon to the patella, patellar ligament to the patella and the tibial tuberosity, Achilles tendon and plantar fascia insertions to the calcaneus. Peritendon inflammation (PTI) assessment covered the extensor tendons over the MCP, PIP, DIP and MTP, PIP and DIP in the fingers and toes. Furthermore, the evaluation of NBF and bone erosion was conducted in 42 joints in the hand and foot joints.

Ultrasound scoring

The assessment of musculoskeletal inflammation encompassed a comprehensive evaluation of various sonographic lesions, including synovitis, tenosynovitis, PTI and enthesitis. In addition, the presence and extent of sonographic structural damage were evaluated across the following distinct domains including periarticular NBF, bone erosions and structural entheseal damage. Whenever possible, we used validated scoring systems16,20 (refer to online supplemental table 2) for detailed scoring.

Global scores were calculated for each sonographic lesion as the sum of Greyscale and PD scores from all pertinent joint, tendon or entheseal sites. The following global scores were derived: synovitis score (0–384), PTI score (0–224), tenosynovitis score (0–84), enthesitis inflammatory score (0–80), enthesitis structural score (0–128), articular bone erosion score (0–64) and articular NBF score (0–64). In addition, we calculated the total structural score by summing the scores for enthesitis structural, NPF and bone erosion, while the total inflammatory score was derived from the summation of scores for synovitis, PTI, tenosynovitis and enthesitis inflammatory score. Intrarater reliability of ultrasound scoring was assessed by having the sonographer re-score 10 sets of stored ultrasound scans (>1 year after the first scoring). Intrarater reliability was excellent with an intraclass correlation coefficient of 0.90 across all domains.³

Statistical analysis

Patient characteristics were analysed using descriptive statistics, presenting mean (SD) for continuous variables and frequency (%) for categorical variables. Pearson correlation was used to assess the association between continuous variables, and Spearman correlation was used for associations involving categorical variables with r >0.3 indicating at least a moderate correlation.

Since some patients contributed to more than one treatment period, generalised estimating equation (GEE) models were used to evaluate the association between baseline musculoskeletal sonographic scores and treatment outcomes. These models included linear regression GEE models for assessing DAPSA change (a continuous outcome) and logistic regression GEE models for evaluating DAPSA-LDA (a binary outcome). Cox proportional hazards models were used to analyse drug persistence, with the time at risk defined as the duration from treatment initiation to either treatment discontinuation or last cohort visit (censored).

The regression models included the following baseline ultrasound scores as covariates: synovitis, tenosynovitis, peritenonitis, enthesitis-inflammatory, enthesitis-structural, NBF and bone erosion. All ultrasound scores were standardised (mean=0, SD=1) to facilitate comparison across sonographic lesions, given different score ranges. Each ultrasound score was individually included in a separate regression model along with medication type (bDMARD/tsDMARD vs csDMARD) and prior exposure to bDMARD/tsDMARD (yes/no) as model covariates. Notably, due to the exploratory nature of the study and the small sample size, we directed our attention to effect size interpretation more than to formal p value testing. However, we provide both raw p value and false discovery rate (FDR)-adjusted p value to adjust for multiple testing.

Results

A total of 135 treatment periods involving 107 patients with PsA were analysed. The mean duration of follow-up was 320 (SD 319) days. Details regarding patient characteristics and ultrasound scores are presented in table 1 and online supplemental table 3, respectively.

Table 1. Baseline characteristics of study participants (n=135).

Variable	Mean (SD) or frequency (%)
Age (years)	47.7 (13.7)
Sex: female (%)	66 (49%)
PsA duration (years)	4.7 (6.8)
No. of prior DMARDs (%)
0	52 (38.5%)
1	33 (24.5%)
2+	50 (37%)
No. of prior b/tsDMARD (%)
0	90 (66.7%)
1	19 (14.1%)
2	13 (9.6%)
3+	13 (9.6%)
Tender joints (0–68)	6.5 (5.5)
Swollen joints (0–66)	4.7 (4.6)
Enthesitis (any)	81 (60%)
SPARCC	1.6 (1.7)
Dactylitis (any)	39 (28.9%)
Dactylitis count	0.5 (0.9)
PASI	4.5 (5.6)
hsCRP	7.6 (11.8)
DAPSA	29.1 (18.8)
Pain (0–10)	5.1 (2.4)
Patient global (0–10)	5.2 (2.4)
Physician global (0–10)	5.1 (1.7)
HAQ (0–3)	0.7 (0.5)

Medications analysed for treatment outcomes
	DAPSA change (n=87)	Drug persistence (n=105)
csDMARD	11	13
Methotrexate	9	11
Sulfasalazine	2	2
Start ts/bDMARD*
TNF inhibitor	41	44
IL-17 inhibitor	28	37
IL12/23 or IL-23 inhibitor	1	1
JAK inhibitor	6	10

b/tsDMARDbiologic and targeted Disease-modifying antirheumatic drugscsDMARDconventional synthetic DMARDDAPSADisease Activity Index for Psoriatic ArthritisDMARDsdisease-modifying antirheumatic drugsHAQHealth Assessment Questionnaire for rheumatoid arthritishsCRPhigh-sensitivity C reactive proteinILInterleukin (IL)JAKJanus KinasePASIPsoriasis Area and Severity IndexSPARCCSpondyloarthritis Research Consortium of CanadaTNFTumor Necrosis Factor

Correlations across ultrasound features

Moderate to high correlations were found between several sonographic features (figure 1). Notably, inflammatory articular lesions correlated with periarticular inflammatory lesions (eg, synovitis and peritenonitis). Similarly, correlation was found across structural articular lesions (eg, bone erosions and NBF). Entheseal inflammatory and structural scores exhibited a strong correlation. Furthermore, moderate correlations were found between synovitis and erosions, NBF and tenosynovitis, PTI and NBF, as well as between NBF and enthesitis structural score.

Figure 1. Pearson correlation coefficients (r) between ultrasound domains.

Correlations between clinical and ultrasound features

A substantial discrepancy was found between clinical and sonographic features (figure 2). Among the clinical parameters, swollen joint count and physician global assessment showed the strongest correlation with sonographic features, namely synovitis and PTI. Clinical enthesitis, as assessed by SPARCC, showed a moderate correlation with inflammatory enthesitis score. However, patient pain, patient global, tender joint count, CRP levels and dactylitis count showed weaker correlations with sonographic features.

Figure 2. Spearman correlation coefficients (r) between ultrasound domains and measures of disease activity. CRP, C reactive protein.

Ultrasound features treatment outcomes

Treatment outcomes in relation to ultrasound features were assessed across 87 treatment periods involving 73 patients, focusing on DAPSA response at 3–6 months (11 csDMARDs and 76 ts/bDMARDs). Among these, 54% of patients achieved DAPSA-LDA and the mean change in DAPSA score was −11.4 (SD 14.6).

An association was found between several sonographic features at baseline and the change in DAPSA score at 3–6 months (table 2). Specifically, a greater reduction in DAPSA score was associated with higher baseline scores of synovitis (adjusted β −3.89 (95% CI −7.09 to –0.68)), PTI (adjusted β −3.93 (95% CI −7.01 to -0.84)) and structural enthesitis (adjusted β −2.91 (95% CI −5.75 to –0.06)). When the analysis was restricted to users of TNF inhibitors (n=41), numerically higher effect sizes were observed for several sonographic features, including synovitis, PTI, tenosynovitis and structural enthesitis. No association was found between baseline ultrasound scores and DAPSA-LDA (online supplemental table 4).

Table 2. The association between baseline ultrasound scores and the change in DAPSA score at 3–6 months.^* GEE linear multivariable regression models.

Ultrasound feature	Adjusted regression models—all medications (n=87)			Adjusted regression models—TNF inhibitor users (n=41)
	Beta (95% CI)	P value	P value FDR	Beta (95% CI)	P value	P value FDR
Total synovitis score	−3.89 (−7.09 to -0.68)	0.02	0.07	−5.26 (−8.84 to -1.68)	0.004	0.009
Total PTI score	−3.93 (−7.01 to -0.84)	0.01	0.07	−6.11 (−9.85 to -2.38)	0.001	0.007
Total bone erosion score	−0.86 (−3.79 to 2.07)	0.56	0.56	−1.68 (−3.56 to 0.20)	0.08	0.11
Total NBF score	1.20 (−2.83 to 5.22)	0.56	0.56	−0.79 (−6.08 to 4.51)	0.77	0.77
Total tenosynovitis	−4.21 (−8.62 to 0.21)	0.06	0.11	−5.18 (−8.73 to -1.64)	0.004	0.009
Total enthesitis-inflammation	−2.28 (−5.15 to 0.57)	0.12	0.17	−1.32 (−4.68 to 2.04)	0.44	0.51
Total enthesitis-structure	−2.91 (−5.75 to -0.06)	0.045	0.11	−3.21 (−6.37 to -0.07)	0.045	0.07

Ultrasound scores are standardised.

^*Each model was individually adjusted for medication class and prior exposure to biologics. Change in DAPSA = DAPSA at 3–6 month minus DAPSA at baseline. Positive coefficients indicate an increase in DAPSA score from baseline, while negative coefficients indicate a reduction in DAPSA scores at 3–6 months.

DAPSADisease Activity Index for Psoriatic ArthritisFDRfalse discovery rateGEEgeneralised estimating equationNBFnew bone formationPTIperitenon inflammation

To ascertain whether sonographic inflammatory and structural ultrasound features independently predict DAPSA change, we integrated both summary scores into a single regression model (table 3). Consistent with the results for individual inflammatory features, our analysis revealed that a higher baseline total inflammatory score predicted a greater reduction in DAPSA scores, independently of the total damage score (adjusted β −5.23, 95% CI –11.34 to –1.99).

Table 3. The association between baseline total inflammatory and structural ultrasound scores and the change in DAPSA score at 3–6 months. GEE linear multivariable regression model.

Model covariate	Multivariable regression model—all medications (n=87)		Multivariable regression model—TNF inhibitor users (n=41)
	Beta (95% CI)	P value	Beta (95% CI)	P value
Total inflammatory ultrasound score	−5.23 (−9.01 to -1.46)	0.007	−6.67 (−11.34 to -1.99)	0.005
Total structural ultrasound score	1.32 (−2.70 to 5.33)	0.52	1.09 (−2.53 to 4.71)	0.55
Prior exposure to biologics (yes)	5.24 (−2.52 to 13.01)	0.19	7.44 (−0.92 to 15.81)	0.08
Medication class (ts/bDMARD vs csDMARD)	−4.74 (−13.41 to 3.93)	0.28	N/A

Change in DAPSA = DAPSA at 3-6 month minus DAPSA at baseline. Positive coefficients indicate an increase in DAPSA score from baseline, while negative coefficients indicate a reduction in DAPSA scores at 3-6 months. Ultrasound scores are standardised.

csDMARDconventional synthetic DMARDDAPSADisease Activity Index for Psoriatic Arthritists/bDMARDbiologic and targeted disease-modifying antirheumatic drugs

Given that DAPSA encompasses both physician-derived items, such as joint count, and patient-reported items, such as pain, we proceeded to investigate whether the association between various sonographic lesions differed across the five components of DAPSA: pain, patient global, tender joint, swollen joint and CRP levels. Using multivariable analysis, we aimed to assess which components of DAPSA were most influenced by sonographic features (figure 3). Our analysis revealed significant associations between baseline sonographic synovitis and reduction in pain (adjusted $\beta$ −1.28). Moreover, higher baseline sonographic PTI and tenosynovitis scores predicted a greater reduction in swollen joint count (adjusted $\beta$ −0.97 and −1.76, respectively). A higher baseline NBF score was associated with a lower reduction in CRP (adjusted $\beta$ 1.00). Notably, none of the ultrasound features was associated with changes in tender joint count. Similarly, sonographic enthesitis was not associated with a change in any DAPSA items.

Figure 3. The association between baseline ultrasound scores and the change in DAPSA components from baseline to 3–6 months; (3A) CRP; (3B) pain; (3C) swollen joint count and (3D) tender joint count. Each graph shows the β coefficient and the 95% CIs adjusted for medication type and prior exposure to biologics as generated by individual GEE linear regression models. CRP, C reactive protein; DAPSA, Disease Activity Index for Psoriatic Arthritis; GEE, generalised estimating equation; NBF, new bone formation; PTI, peritenon inflammation.

Ultrasound features and drug persistence

Drug persistence was analysed across 105 treatment periods (92 ts/bDMARD, 13 csDMARD), revealing that 40% of the treatments were discontinued during the follow-up period (end of study). Among the sonographic features analysed, baseline bone erosion score was significantly associated with earlier drug discontinuation (adjusted HR 1.28, 95% CI 1.03 to 1.61; table 4). After adjusting for total inflammation score, the effect size for erosion score remained similar but with a wider CI (HR 1.30, 95% CI 1.00 to 1.70).

Table 4. The association between baseline ultrasound features and drug persistence by Cox regression model (n=105, 42 discontinuation).

Ultrasound score	Individual models adjusted for drug class and prior exposure to biologics		Single multivariable model adjusted for class and prior exposure to biologics
	HR (95% CI)	P value	HR (95% CI)	P value
Total synovitis score	1.17 (0.89 to 1.55)	0.27
Total PTI score	1.01 (0.77 to 1.32)	0.92
Total erosion score	1.28 (1.03 to 1.61)	0.03	1.30 (1.00 to 1.70)	0.05
Total NBF score	1.10 (0.84 to 1.44)	0.48
Total tenosynovitis	0.85 (0.63 to 1.15)	0.30
Total enthesitis—inflammation	1.16 (0.85 to 1.60)	0.35
Total enthesitis—structure	1.09 (0.78 to 1.54)	0.61
Total inflammatory score	1.11 (0.84 to 1.48)	0.47	0.97 (0.70 to 1.36)	0.97

NBFnew bone formationPTIperitendon inflammation

Discussion

Our study sheds light on the correlation between sonographic and clinical features of PsA and their predictive value for treatment outcomes, offering novel insights into the clinical relevance of sonographic inflammatory and structural ultrasound findings in this condition. Several findings warrant further discussion. First, we found a discordance between clinical sonographic features of PsA, particularly concerning patient-reported outcomes. Notably, the highest correlations were observed between physician-derived indicators, such as swollen joint count and physician global assessment, and sonographic inflammatory features. In contrast, structural lesions and sonographic enthesitis showed an overall weak correlation with clinical features of PsA.

Second, we found moderate to strong correlations between intra-articular and periarticular sonographic inflammatory scores, including synovitis and peritenonitis, as well as synovitis and tenosynovitis. These associations may be explained by the synovio-entheseal complex hypothesis,²¹ which posits a close anatomical link between periarticular/entheseal inflammation and intra-articular (synovial) inflammation. For example, the close anatomical and physiologic proximity between the finger extensor tendon and its enthesis at the base of the phalanx, alongside the synovial tissue within the adjacent joint capsule, may explain the associations between articular and periarticular sonographic scores found in our study. Similarly, the correlation between sonographic enthesitis and synovitis scores aligns with this hypothesis and has been previously described by Macía-Villa²¹ and Balulu.¹⁰

Another important finding is the association between baseline ultrasound scores and future treatment response. Our study showed that higher levels of sonographic inflammation in peripheral joints, specifically synovitis and peritenonitis, along with a higher total inflammatory score at baseline, predict a more favourable short-term response to therapy, characterised by a greater reduction in DAPSA. Conversely, a higher burden of bone erosions was associated with decreased persistence on therapy. These findings highlight the potential utility of ultrasound at the point of care for identifying difficult-to-treat PsA and informing treatment decisions.

Our findings offer new insights into the musculoskeletal lesions underlying the clinical features of PsA. The construct of peripheral joint inflammation in PsA is multifaceted; clinically, it is manifested as swelling and tenderness on joint examination, whereas sonographically, various anatomical structures may be involved, including synovitis, peritenonitis and tenosynovitis. Husic et al reported that swollen joints correlated with synovitis and tenosynovitis, while peritenonitis alone did not show any association with other clinical features.¹ Similar to our findings, Dubash et al found a strong correlation between swollen joints and sonographic synovitis, which was higher than the correlation with tender joints; however, their study did not evaluate periarticular ultrasound inflammation, such as peritenonitis and tenosynovitis.²² Furthermore, our group has shown that distinct biological pathways underlie imaging phenotypes of PsA. Specifically, gene expression profiles differed among patients with predominantly periarticular inflammation (PTI predominant phenotype) and those with intra-articular inflammation (synovitis predominant phenotype) phenotypes.³ Thus, ultrasound provides information independent of physical examination, with potential pathophysiological implications.

We present novel insights into the predictive capacity of ultrasound regarding treatment response in PsA. Higher baseline levels of sonographic inflammation in peripheral joints including synovitis, PTI and tenosynovitis were associated with more favourable short-term response to therapy. These sonographic features of inflammation in peripheral joints were predominantly linked to improvement in objective components of DAPSA, such as swollen joint count and CRP levels. Furthermore, the total inflammatory score, encompassing all domains (synovitis, enthesitis, PTI and tenosynovitis) independently predicted short-term treatment response, irrespective of structural joint damage. This finding holds particular importance, especially considering the limited longitudinal studies evaluating the predictive ability of ultrasound in PsA. For instance, Sakellariou et al showed that the presence of baseline PD signal in the hand joints predicted the need for advanced therapies in patients with early PsA.¹³ Ruta et al showed that PD at baseline could predict flares in PsA patients who were in clinical remission.²³ Our results contribute important additional information regarding the clinical utility of ultrasound. Patients with objective evidence of sonographic inflammation, even in the presence of joint damage, are more likely to respond to therapy and should thus be prescribed effective treatments. Conversely, individuals with very low levels of inflammation may have lower chances of responding and could benefit from alternative treatment options, such as symptomatic pain management.

Another important finding is the association between sonographic structural damage and drug persistence. While none of the sonographic inflammatory lesions were associated with drug persistence, a higher erosion score predicted earlier drug discontinuation. Moreover, sonographic NBF was associated with worsening in several DAPSA score items including tender joints, pain and CRP levels. These findings underscore the role of joint damage and concomitant OA as contributing to treatment failure. Differentiating between OA and PsA can be challenging as they often coexist in the same patient due to shared disease features, such as periarticular NBF, or overlapping risk factors like obesity and physical trauma. Furthermore, active inflammation in PsA can result in early secondary OA. Differentiating between bone proliferation due to primary OA or due to PsA using ultrasound is challenging without considering other features, such as coexisting inflammatory features, erosions and clinical features.²⁴ Nevertheless, our findings suggest that the presence of sonographic structural erosive damage identifies patients with PsA that is more challenging to treat. This may stem from a more aggressive inflammatory process or other mechanisms driving patient symptoms. Such patients may benefit from closer monitoring and possibly from earlier intense therapy.

Our study’s primary limitations lie in its relatively small sample size, which precluded us from assessing whether sonographic features predict response to a specific class of medication. This type of analysis is important, as certain sonographic features may respond differently to specific classes of advanced therapy, thereby guiding treatment selection. Additionally, a limitation pertaining to the feasibility of implementing our study protocol in a clinical setting is the extended duration of the ultrasound protocol, which averaged approximately 1 hour. Identification of the different sonographic features in the clinical field may eventually not require this extended scanning and will only require identification of the symptomatic ultrasound features at the time of the clinical evaluation.

Nevertheless, this study represents one of the largest cohorts to date, incorporating detailed sonographic characterisations of inflammatory and structural lesions of PsA alongside comprehensive longitudinal clinical data. This wealth of information enables exploration of the clinical utility of ultrasound in patients with PsA.

In conclusion, our study offers novel insights into the added value of ultrasound characterisation in active PsA. The discordance between clinical and sonographic data suggests that the latter provides independent information about disease activity. Sonographic evidence of articular and periarticular inflammation, as well as overall inflammation levels, predict favourable short-term treatment outcomes. Conversely, the presence of erosive joint damage is associated with earlier drug discontinuation in PsA. The ability to detect and quantify inflammation and structural damage in the peripheral joints can inform risk stratification of patients regarding their likelihood to respond and persist on prescribed medications.

Data availability statement

Data are available upon reasonable request.