Introduction

Psoriatic arthritis (PsA) is a very heterogeneous disease with axial and/or peripheral joint involvement, and extra-articular manifestations, such as enthesitis or dactylitis. Tumor necrosis factor (TNF) blockers are the mainstay of PsA treatment [1].

Nevertheless, a proportion of patients do not benefit from the TNF blocking strategy. New insights in the pathogenesis of PsA have shown that new molecular pathways are pivotal in PsA, mainly the IL-23/IL-17 axis [2].

Ustekinumab (UST) is a fully human IgG1 monoclonal antibody that binds with high affinity to the shared p40 subunit of human IL-12 and IL-23, inhibiting their binding to the IL-12Rβ1 receptor on the surfaces of T cells, NK cells, and antigen-presenting cells. It is used for treating moderate-to-severe psoriasis and Crohn’s disease, and it has also been approved for PsA . Its safety and efficacy have been evaluated in two multicenter, double-blinded, placebo-controlled trials (PSUMMIT 1 and PSUMMIT 2). At week 24, there was a significant improvement in ACR 20 in PSUMMIT 1 (UST 90 mg 50%, UST 45 mg 42%, placebo 23%) [3] as well as in PSUMMIT 2 (UST 90 mg 44%, UST 45 mg 44%, placebo 20%) [4]. Ustekinumab was associated with remarkable improvements on other PsA domains, such as enthesitis, dactylitis, pain, quality of life, physical function, and inhibition of radiological damage [5].

Apart from clinical trials, data on the effectiveness of UST in patients from real-world clinical practice is scarce, coming from small cohorts of PsA or patients with prevalent skin disease [6,7,8,9,10].

The aim of this study was to assess the effectiveness and survival of UST among patients with PsA treated under routine clinical conditions.

Methods

Patients

All patients diagnosed with PsA according to CASPAR criteria [11] and starting UST due to rheumatological indication until February 2019 were included. Fifteen Spanish hospitals from Cataluña (12), Navarra (1), and Balearic Islands (2) participated in the study. Epidemiological and clinical data was collected through electronic medical records. All patients were treated with UST 45 or 90 mg subcutaneous injection, administered at weeks 0 and 4, then every 12 weeks, according to the manufactures instructions [12].

At baseline, patients’ demographic and clinical features were recorded, focusing on age, gender, comorbidities, PsA clinical subset, disease duration, and previous and/or actual therapies (including TNF blockers, conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), and corticosteroids). At baseline and month 1, 4, 7, and 12, clinical aspects of the disease, such as arthritis, psoriasis, enthesitis, dactylitis, and physical functions, were evaluated through outcome measures as follows: tender joint count (TJC), swollen joint count (SJC), psoriasis activity and severity index (PASI), clinical presence of enthesitis, and dactylitis. Erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels were also recorded. DAS28 ESR and DAS66/68 CRP were calculated. Remission was defined as DAS66/68 CRP < 2.6. An alternative concept of remission was taken, called strict remission, requiring the absence of SJC, TJC, dactylitis, and enthesitis. In addition, PASI should be < 2 and CRP < 1 mg/dl.

Statistical analysis

We analyzed patient demographics and disease characteristics at treatment initiation using standard descriptive statistics. Categorical data is expressed as percentage, whereas continuous variables are shown as median and range. Baseline differences between groups were compared by Mann–Whitney U test. Percentages were compared by χ2 test. Drug survival time was estimated using Kaplan–Meier (K-M) life-table method and differences across line of biologic treatment, UST dose, concomitant treatment (DMARD and corticosteroids), presence of obesity, enthesitis, dactylitis, and onychopathy were compared using the log-rank test. Time to discontinuation was defined as the time between ustekinumab initiation and last administration.

Estimated hazard ratios (HRs) of continuing ustekinumab were assessed by performing a multivariate Cox regression, adjusting for patient demographics (age, gender, BMI categories), disease characteristics (HAQ, clinical subset, PASI and DAS66/68 CRP, disease duration at baseline), co-therapy with corticosteroids or DMARD, and line of treatment with ustekinumab (biologic-naïve+ < 1TNF blocker/≥ 2 TNF blockers). Factors with p values < 0.10 were stepwise backward removed until the final multivariate model was selected. Paired t test or Friedman test was used to compare baseline to and 12 months SJC, TJC, DAS66/68 CRP, HAQ, and PASI within groups. Statistical analysis was carried out by SPSS Statistics 20.0 program.

Results

Baseline characteristics

Two hundred and one patients were included, 130 (64.6%) with UST 45 mg and 71 (35.3%) with UST 90 mg. Twenty-three patients increased UST dose from 45 mg to 90 mg every 12 weeks throughout the follow-up. Physician’s perception of joint inefficacy was the reason for increasing UST dose in all patients.

Table 1 shows demographics and clinical characteristics of the whole population. One hundred and seventy patients (84.5%) had oligo (55 patients, 27.3%) or polyarticular involvement (115 patients, 57.2%). Forty-six (22.8%) and 65 patients (32.3%) had dactylitis and enthesitis, respectively. At the moment of UST initiation, 70 patients (34.8%) were naïve to biological therapy and 131 patients (65.1%) were TNF blocker experienced. Fifty-five (27.4%), 48 (23.9%), 24 (11.9%), 7 (3.5%), and 5 (2.5%) patients had previously taken one, two, three, four, or five TNF blockers, respectively. Overall, 118 patients (58.7%) were naïve or had received one TNF blocker and 83 patients (41.2%) had received two or more TNF blockers before UST initiation. One hundred and three patients (51.2%) were taking concomitant DMARDs and 80 patients (39.8%) were taking corticosteroids. Sixty patients (29.9%) were obese. At the moment of UST initiation, median DAS66/68-CRP was 3.09 and median PASI was 3 (Table 1).

Table 1 Baseline characteristics. F Female, M Male, BMI body mass index, HBP high blood pressure, DM diabetes mellitus, DIP Distal interphalangeal, PASI Psoriasis Area and Severity Index, CRP C-reactive protein
Full size table

Significant decrease in joint and skin disease activity since the first month of treatment

There was a significant decrease in DAS28 ESR, DAS66/68 CRP, SJC, TJC, and PASI since the first month of treatment (Table 2). Improvements in skin outcome (PASI) started earlier than joint outcomes. The maximum response for PASI was reached between first and fourth month. On the other hand, improvements in joint outcomes and CRP were slower, and the maximum response was not reached earlier than fourth to seventh month.

Table 2 Disease activity and clinical domains involvement along the follow-up. CRP C-reactive protein, TJC Tender Joint Count, SJC Swollen Joint Count. Strict remission: absence of SJC, TJC, dactylitis, and enthesitis. In addition, PASI should be < 2 and CRP < 1 mg/dl
Full size table

UST 45 versus UST 90 mg

At baseline, there were not significant differences in clinical and epidemiological data between patients at 45 or 90 mg except for the percentage of obese patients (42% in 90 mg vs 25.6% in 45 mg, p = 0.023) and the disease activity as evaluated with DAS66/68 CRP (p = 0.044), both significantly higher in the group of UST 90 mg (Table 3).

Table 3 Baseline characteristics according to UST dose. F Female, M Male, BMI body mass index, HBP high blood pressure, DM diabetes mellitus, DIP distal interphalangeal, PASI Psoriasis Area and Severity Index, CRP C-reactive protein, TJC Tender Joint Count, SJC Swollen Joint Count
Full size table

There was a trend indicating faster joint responses in patients with UST 90 mg (Fig. 1). Accordingly, the SJC and the number of patients with onychopathy were significantly lower in patients with UST 90 mg at month 4 (p = 0.038 and p = 0.036, respectively). In general, these differences tended to disappear at month 7. On the other hand, skin outcomes such as PASI quickly improved regardless of UST dose, and highest responses were reached at first month.

Fig. 1
figure 1

Clinical responses over 12 months of follow-up. PASI Psoriasis Area Severity Index, CRP C-reactive protein, SJC Swollen Joint Count, DAS disease activity score

Full size image

Clinical response depends on the previous number of TNF blockers

Likewise, improvements in joint outcomes appeared earlier in naïve versus biological experienced patients. DAS66/68 CRP, TJC, and SJC were significantly lower at month 4 in biological naïve patients (p = 0.005, 0.004 and 0.014, respectively). These differences lowered from the seventh month, except for the number of TJC, still significant (p = 0.032). On the other hand, PASI was significantly lower in patients naive to biological therapy along the first 12 months of follow-up.

When comparing naïve/< 1 TNF blocker versus > 2 TNF blocker-experienced patients, significantly better responses were seen for the former group regarding SJC (p = 0.029) within the first month of therapy. At month 4, TJC (p = 0.0001), SJC (p = 0.002), and DAS66/68 CRP (p = 0.0001) were still significantly better for those patients naïve/< 1 TNF blocker experienced. These differences narrowed from the seventh month (Fig. 2).

Fig. 2
figure 2

Clinical responses according to previous lines of bDMARDs., SJC Swollen Joint Count, DAS disease activity score, CRP C-reactive protein, PASI Psoriasis Area Severity Index

Full size image

Rates of DAS66/68 CRP remission

Twenty-eight (14%), 63 (31.3%), 70 (34.8%), and 59 (29.3%) patients fulfilled remission criteria according to DAS66/68 CRP < 2.6 at months 1, 4, 7, and 12, respectively. No significant differences were found in the percentage of patients in remission between patients with UST 45 and 90 mg. On the other hand, those patients with fewer lines of previous biological therapy had better rates of remission: significantly higher number of patients naïve to biological therapy reached remission DAS66/68 CRP at 4th month (p = 0.02). These differences were maintained when comparing patients with < 1 versus > 2 TNF blockers (p = 0.001).

Rates of remission DAS66/68 CRP were not significantly different according to the presence of obesity, dactylitis, or onychopathy. However, patients with enthesitis had a significantly lower remission rate at the seventh month (50% versus 81.1%, p = 0.006). Likewise, patients with erosive disease had also significantly lower rate of DAS66/68 CRP remission at the fourth month (38.2% versus 67.6%, p = 0.006).

Ten (4.9%), 21 (10.4%), 31 (15.4%), and 26 (12.9%) patients fulfilled criteria for the previous defined strict remission criteria at months 1, 4, 7, and 12, respectively. No significant differences were found in the percentage of patients in strict remission according to UST dose, clinical, or epidemiological features.

In the multivariate analysis, the number of previous bDMARDs was the only factor independently associated with not reaching DAS66/68 CRP remission at the seventh month (OR 0.72, CI 95% 0.55–0.93).

Safety

Four patients (1.8%) interrupted the treatment due to adverse events. Three cases were reported as general malaise and headache associated with the injection and one case was reported as hypertensive crisis. No serious infections leading to the interruption of UST were communicated. Three cases were reported in the first 3 months of treatment and one case 7 months after the first UST injection.

Drug survival

After an average time of 17.7 months of follow-up (median of 12 months), 129 out of 201 patients (64.1%) continued with UST. Survival was numerically lower in patients with UST 45 mg (58.1%) than those with 90 mg (76.1%), although significant differences were not found (p = 0.147).

Seventy patients (34.8%) stopped UST after a maximum follow-up of 90 months. Most of them (51 patients, 25.3%) interrupted the therapy due to joint inefficacy. Eleven patients stopped UST due to skin failure (5.4%) and 4 patients interrupted the treatment after an adverse event (1.9%). Four patients were lost during the follow-up.

Although drug survival was globally better in patients with fewer lines of biological agents, no significant differences were achieved between naïve and TNF blocker user patients (75.7% and 58.5%, respectively, p = 0.053). When gathering naïve patients with those with failure to one TNF blocker, survival was significantly better than patients with > 2 TNF blocker failure (72.9% versus 52.4%, p = 0.003, Fig. 3).

Fig. 3
figure 3

Ustekinumab survival according to previous lines of TNF blockers

Full size image

Patients with enthesitis (72.3% versus 61%) and non-erosive disease (69.7% versus 64.1%) had numerically higher survival rate although significant differences were not found (p = 0.106 and p = 0.092, respectively).

Regarding concomitant treatment, neither the use of DMARD therapy (56.9% versus 72.2%, p = 0.151) nor corticosteroids (59.5% versus 67.8%, p = 0.118) was related to a better drug survival.

In the Cox regression analysis, 90 mg dose was independently associated with better survival (HR 3.205, CI 95% 1.371–7.496, p = 0.007). On the contrary, higher PASI (HR 0.907, CI 95% 0.859–958, p = 0.0001) and onychopathy (HR 0.458, CI 95% 0.225–0.932, p = 0.051) were independently associated with lower survival.

Drug dose increase

Twenty three patients starting with UST 45 mg changed to 90 mg along the follow-up due to clinician decision. Joint partial response was the reason for all changes.

Seventeen (74%) out of these 23 patients continued with the treatment after a median time of 18 months follow-up.

Discussion

Observational studies provide additional data regarding drug effectiveness and safety. Registries reflect real-world data in unselected patients from clinical practice and yield complementary information about long-term adverse events and profiles of patients who can benefit more from the drug. In this multicenter, retrospective study, we collected clinical and serological data from 201 PsA patients taking UST from 15 hospitals of the northeast Spain. We saw that UST was effective in improving skin and joint outcomes, with significant differences from the first month of therapy. Retention rate of UST 90 mg was 76.1% after a median time of 12 months follow-up. Additionally, the clinical response was faster in patients with 90 mg and fewer lines of previous biological therapies. One third of the patients interrupted the therapy, most of them due to joint inefficacy, and a very low rate of adverse events leading to discontinuation of UST was reported.

As far as we know, this is the largest study of UST in PsA patients in a clinical care setting. Up until now, few works have explored the effectiveness of UST in PsA patients from clinical practice. Apart from BIOPURE (n = 160 patients) [8], most of these studies had small sample size (34–65 patients) and the majority of patients used UST 45 mg. This is the first study where clinical outcomes were analyzed according to UST doses. The need for joint faster responses and its good safety profile have pushed clinicians to use UST 90 mg over 45 mg in most patients, despite the lack of published data showing superiority of UST 90 mg over 45 mg. Accordingly, the results shown in this study are very relevant. Although globally we did not find significant differences in clinical outcomes according to UST doses, responses were achieved earlier in patients with UST 90 mg, although differences were balanced after month 7. Therefore, the better retention rate for UST 90 mg (76.1% versus 58.1% for UST 45 mg) reflects the idea of a better perception of effectiveness for both patients and clinicians. However, these findings should be evaluated with caution due to the study design (open label) and the subsequent risk of bias.

We only included patients starting UST due to rheumatological indication. Despite all patients included in previous studies fulfilled the CASPAR criteria, in some of them, UST was prescribed by dermatologists. Therefore, the population included in our work can differ from previous studies since most of our patients had mild psoriasis.

From the analysis of the few published studies on UST from real clinical practice, it is deduced that the drug works on the different domains of PsA. The safety profile is good, and drug survival is much better in patients naive to TNF blockers. Regarding MDA, between 30 and 70% of patients achieved this stringent goal [6,7,8,9,10].

The most recent data from UST in clinical practice comes from PSabio, an on-going observational study taking place at 8 European countries, where PsA patients receive first-, second-, or third-line biologics (starting either ustekinumab [UST] or a TNF blocker), with 563 UST- or TNF blocker-treated patients enrolled until August 2017. In a previous analysis, the authors observed that starting treatment with either UST or TNF blockers did not significantly influence the rates of remission (Disease Activity in Psoriatic Arthritis (DAPSA), minimal disease activity (MDA), or very low disease activity (VLDA)) at 6 months [13].

Although we did not have available data for MDA and DAPSA, we used alternative indexes to assess disease activity. We calculated DAS using 3 variables (TJC, SJC over 66/68 joints, and CRP) for disease activity. Remission was evaluated according to DAS66/68 CRP < 2.6 (34.8% at the seventh month, which is in line with previous studies). Fifteen percent of patients also fulfilled even a more strict definition of remission, requiring the absence of TJC, SJC, enthesitis, and dactylitis, in addition to CRP < 1 mg/dl and PASI < 2.

As we expected, PASI maximum response was quickly achieved after UST initiation, even faster than clinical trials [14, 15], perhaps due to the lower skin activity of our patients. This quick response was achieved regardless of the dose and the number of previous biologics used. Joint outcomes yielded contrasting results, and maximum responses appeared between the fourth and seventh month, reflecting a different kinetic of response for skin and joint. The latter sharply contrasts with data on TNF blockers, where outcomes are achieved 2 months after the initiation of the therapy, without differences between joint and skin. This gap in the onset of action between TNF blockers and UST stresses the relevance of adequately selecting the patients for UST. It is essential to consider this issue for the management of patient and clinician’s expectations.

Patients with fewer previous lines of biological therapies had higher survival, along with better and faster responses than very refractory patients, something already known from TNF blockers [16].

Animal models such as the mouse model of Sherlock et al. [17] and clinical studies such as ECLIPSA [18] suggest that enthesitis could be a good target for UST. Accordingly, our patients with enthesitis had a better retention rate, reinforcing the previous data of effectiveness of UST in enthesitis.

Based on the data sheet, UST 90 mg should be recommended in patients over 100 kg. Accordingly, rate of obesity in our study was significantly higher in patients with 90 mg. A recent subanalysis of PSUMMIT trials yielded similar responses rates for UST 90 mg regardless of patient weight [19]. This data could be of potential interest since TNF blockers have demonstrated less efficacy in obese patients [20,21,22]. However, we did not find a better response according to BMI in our study.

Only 4 patients stopped the treatment due to adverse events (1.8%), underlining the good safety profile of UST. Data extracted from PSOLAR observational study capturing real-world data from patients with psoriasis receiving systemic therapy [23] showed that cumulative incidence rates of adverse events for UST were numerically lower or comparable with other biologics and non-biologics cohorts in the registry. Rates of serious infections were significantly lower for UST when comparing with other biological therapies, whereas no significant differences were found with non-biological therapies [24]. Our data confirms the results from the PSOLAR observational study and place UST as a good option for patients with special safety concerns.

One of the most challenging issues in the treatment of PsA is to find biomarkers of good clinical response. According to the results of our study, we would preferably recommend UST in naïve patients or with few previous lines of TNF blockers, patients where skin or enthesis are the main affected domains of the disease and finally patients where safety is a special concern. As safety seems the same regardless of the dose [25], UST 90 mg should be the first election for all patients.

This study has some limitations. The retrospective data collection limits the strength of the conclusions. Also, it would have been of interest to dispose data regarding DAPSA, MDA/VLDA, or PsAID. However, these outcomes have been developed recently and the collection of data comes from 2014, which made impossible to obtain this information. However, we used PASI, BSA, DAS28 ESR, and DAS 66/68 CRP (although not validated in PsA) as the main skin and joint outcomes and the presence of dactylitis and enthesitis was also recorded. Hence, all the PsA domains were adequately covered. Likewise, we used 2.6 as remission cut off for DAS 66/68 CRP index, which has not been validated either. The bias of dose selection is another limitation for the interpretation of the results, since 90 mg was used not only in patients over 100 kg but also in those with high disease activity according to the clinician opinion.

Overall, data extracted from this observational study suggests that UST is effective for patients with PsA in a clinical care setting. Patients with UST 90 mg and fewer previous lines of biologics achieved better and faster responses. Global drug survival after a median time of 12-month follow-up was higher among UST 90 mg users. Finally, a low rate of UST discontinuation was observed, mostly due to joint inefficacy and less than 2% due to adverse events.