Psoriatic arthritis (PsA) is a complex condition that still presents several unmet needs. Researchers have highlighted key areas of investigation in PsA, including tests to improve the assessment of disease activity and treatment response (1). The development of early predictors of treatment response would help tailor treatment to each patient and avoid inappropriate therapies or unnecessary therapeutic switches (2, 3).

The treatment of PsA includes conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), biological therapies (bDMARDs), and orally administered targeted small molecule drugs (tsDMARDs). Despite the availability of numerous drugs, ~40% of patients fail to respond to csDMARDs or bDMARDs (4).

Serum biomarkers have been studied as possible predictors of treatment response. Elevated levels of C-reactive Protein (CRP) are found early in 33–89% of PsA patients; thus, there is a significant percentage of patients without inflammation despite active disease (1). This is likely due to the fact that in SpA, joint and entheseal damage driven by local cytokines does not fully reflect the systemic context (5).

Among imaging techniques, ultrasound (US) is the most promising for the assessment of treatment response, as it is safer, easily accessible and time efficient. Despite numerous studies published on the use of US in PsA, its application in clinical practice remains a matter of debate (6, 7). A recent systematic literature review comprising 15 studies evaluated the role of US in the follow-up and prognosis of PsA. A major limitation that emerged was the variability in the definitions of elementary lesions and US scoring systems. Five different scoring systems were used across the selected articles (8), two of which considered different domains involved in PsA (joints, entheses, and tendons): (i) the Five Targets PwD score for psoriasis; (ii) the composite US score for evaluating inflammatory and structural pathology in PsA, including 13 and 22 joints (PsASon-Score 13 and 22, respectively) (9, 10). Both composite scores showed sufficient sensitivity to detect US variations following therapy. Moreover, it has recently been proposed to also use in patients with PsA the global synovitis scoring system EULAR-OMERACT “GLOESS,” which was developed in patients with rheumatoid arthritis (RA) (11). Although this scoring system allows to globally assess synovial severity across numerous joints, it does not evaluate the involvement of entheses and tendons. Thus, there are two main unmet needs when discussing the role of US in PsA follow-up. The first is to find a US score allowing us to monitor disease activity in all its domains, both in clinical practice and in clinical studies. The second is to find US predictors of treatment response. There are indeed few studies on the US response in PsA treatment.

In this context, the primary endpoint of our pilot study was to identify early US predictors of drug retention with bDMARDs/tsDMARDs at 6 months. Thus, we recruited patients with active PsA who initiated therapy with bDMARDs/tsDMARDs, or who switched/swapped to other bDMARDs/tsDMARDs (due to ongoing treatment failure). The secondary endpoints of this study were: (i) to identify differences between clinical and ultrasound responses; (ii) to identify differences in sonographic scores among patients treated with different therapeutic classes.

In this pilot study, we aimed to assess how US evaluations may help predict drug retention of bDMARDs and tsDMARDs therapies. For this purpose, the MIJET and 2MIJET ultrasound scoring systems have been proposed. By focusing on specific sites with a high prevalence of inflammatory activity, they enhance the integration of ultrasound with clinical examination in a time-efficient manner, potentially increasing specificity in detecting inflammatory activity within articular and peri-articular structures. Patients who continued therapy after 6 months, as deemed effective by clinicians, showed statistically significant improvement in the ultrasound assessment of the most clinically involved joints, MIJET and 2MIJET, as early as t3, with further confirmation at t6. In contrast, the variation in the unilateral GUIS ultrasound score was significant only at t6, identifying cResponders less quickly than the MIJET and 2MIJET score (Figure 4). Biologic therapy appeared to be associated with a faster ultrasound response compared to the clinical response. Indeed, the variation in clinical disease activity according to cDAPSA was not significant at t1 and t3 between cResponders and non-cResponders, unlike at t6. The HAQ and PsAID questionnaires did not show significant variations between cResponders and non-cResponders during the 6-month follow-up period, although a trend toward improvement in scores was observed among cResponders (Figure 4). This may be partly attributed to the relatively long average disease duration in our cohort, as well as the associated chronic damage and secondary pain that compromise the quality of life in these patients (15).

Some studies in the literature have previously confirmed the validity of ultrasound in assessing response to single biologic drug (16). A recent randomized controlled trial on 166 PsA patients demonstrated how changes in the US—according to the GLOESS US score—may appear as early as 1 week after initiating secukinumab therapy (17). It bears noting that the Authors also reported that the variation in GS of the GLOESS score was more sensitive vs. changes in the pD GLOESS, a finding that was corroborated in our study. Specifically, the variation in 2MIJET-GS, but not in 2MIJET-pD, was greater among cResponders vs. non-cResponders (Table 1). This could be partially explained by the lower baseline levels of the power Doppler signal observed in the entheses of our patients and the reduced pD signal often observed in larger joints, such as the knee.

In our study, achieving disease remission according to cDAPSA did not correlate significantly with the change in ultrasound scores, whereas achieving MDA correlated significantly only with the change in MIJET. Moreover, the change in ultrasound scores was not significantly associated with achieving values indicative of wellbeing in PROs measurements, except for the HAQ. This finding highlights the difference between clinical and ultrasound responses. Generally speaking, variations in MIJET and 2MIJET appear to correlate more with clinimetric indices than the GUIS score, although significance was reached in only a few items, likely due to the small sample size.

The statistical analysis of the distribution of mean ultrasound scores and clinimetric indices did not show significant differences between cResponders and non-cResponders in various assessments. This underscores that ultrasound and clinical variations are more reliable than individual baseline and subsequent assessments.

In our cohort, we noted a faster US response of MIJET and 2MIJET in cResponders treated with JAKi vs. TNFi and IL-17/12-23i, with a statistically significant difference already in Δt1-t0. Conversely, differences in early clinical response according to cDAPSA in Δt1-t0 were not significant among these treatment groups, although they tended toward significance. Consequently, the difference in response speed between pharmacological classes was more pronounced from an ultrasound perspective than from a clinical standpoint (Figure 3).

In comparing the baseline characteristics of patients, non-cResponders had greater enthesitis involvement, with a higher mean of the LEI. This finding aligns with the literature suggesting that enthesitis-predominant sub-phenotype is associated with greater disease activity and is more refractory to treatment (18, 19). Furthermore, this data may be influenced by overlapping conditions such as fibromyalgia or chronic nociplastic pain, comorbidities that could complicate the identification of inflammatory pain, thereby impacting the perceived treatment response. The added value of US assessment and the MIJET and 2MIJET scores lies in their ability to detect subclinical inflammation that may not be identified through clinical examination, as well as to highlight the absence of ultrasound findings suggestive of enthesitis (20). In cases of predominant entheseal involvement, ultrasound evaluation of the enthesis at the site of pain localization—rather than the systematic assessment of predetermined entheses as conducted with the Leeds Enthesitis Index (LEI)—provides a more accurate understanding of the condition of the enthesis associated with pain, both at baseline and during treatment response assessment. This approach may help minimize the influence of potential confounding factors, thereby facilitating a more tailored therapeutic strategy targeting the underlying cause of pain. In our study, only 6 of the 29 patients had an associated diagnosis of fibromyalgia (4 cResponders and 2 non-cResponders), and the small sample size did not permit a subgroup analysis, which could have provided valuable insights.

No patients enrolled in the study had to discontinue treatment due to AE. It is noteworthy that the two cases of hypertensive crisis occurred in patients undergoing JAKi therapy. These two patients already had hypertension, and adjustments to their antihypertensive therapy were sufficient to achieve blood pressure control.

There are several limitations associated with this study. Firstly, the sample size was relatively small. Future studies with larger sample sizes would allow more accurate analysis, especially when comparing specific patient subgroups (e.g., different treatment approaches). Biologic-naïve patients were 5/22 among the cResponders and 1/7 among the non-cResponders; this difference, although not significant, may have influenced the assessment of clinical and ultrasound response. The heterogeneity of therapies prescribed to patients in our cohort may have influenced the evaluated outcomes. The US evaluation was performed by a single sonographer. There are also limitations in patient selection. Specifically, the presence of enthesitis or tenosynovitis at the time of selection was assessed clinically. Moreover, ours was a single-center study with a homogeneous Caucasian patient population.

In conclusion, our pilot study highlighted how US may be an early predictor of drug retention. We were able to identify statistically significant US changes 3 months after treatment initiation among cResponders vs. non-cResponders. The changes in US scores observed after 1 month were also greater among cResponders, though they did not reach significance.

Considering the impracticality of evaluating numerous joints included in the main composite US assessment scores, our results indicate that assessing only the most clinically involved joints can be equally valid in predicting treatment retention rates at 6 months. Furthermore, the variations in the US scores of MIJET and 2MIJET identified Responders more promptly compared to the unilateral GUIS.

Our pilot study highlighted how including US assessment in PsA trials could allow for a more timely and objective evaluation. Indeed, there is a distinction between US response and clinical response, with the latter being more influenced by comorbidities and secondary pain. Additionally, ultrasound response to therapy appears to differ across drug classes. Repeating similar studies but with a larger sample size could help develop more accurate precision medicine.

The identification of early predictors of treatment response may help mitigate the side effects and the overall financial burden on the healthcare system of ineffective therapies.